jam007
strength
FACULTY OF MANAGEMENT AND LAW SCHOOL OF MANAGEMENT
E X E C U T I V E M B A
2015/16
MAN4215M
Business Research
MODULE STUDY BOOK
Study Book: Business Research
2 Bradford MBA
Copyright © University of Bradford 2006, 2007, 2008, 2009, 2010, 2011, 2012, 2013, 2014, 2015, 2016
First published 2006 Seventh edition 2012 Second edition 2007 Eighth edition 2013 Third edition 2008 Ninth edition 2014 Fourth edition 2009 Tenth edition 2015 Fifth edition 2010 Eleventh edition 2016 Sixth edition 2011
MBABRINT-SB-11-2016
MAN4215M
University of Bradford Richmond Road Bradford BD7 1DP
Bradford University School of Management
Director of Studies Jonathan Muir
Programme Leader Ian Fouweather
Programme Administrators Gavin Turner [email protected] (Distance Learning MBA) Emma Pheasant [email protected] (Dubai and Bradford Executive MBA)
Module Leader Ian Fouweather [email protected]
Bradford University School of Management
Emm Lane Bradford BD9 4JL Tel: 01274 234936 Website: www.bradford.ac.uk/management
This Study Book may not be sold, hired out or reproduced in part or in whole in any form or by any means whatsoever without the publisher’s prior consent in writing.
Bradford MBA 3
Contents
Introduction to the Module 7
Your Module Leader 7
Overview of Module and Module Descriptor 7
Assessment criteria 8
Support for your learning 11
Approach to studying business research 11
Developing good academic practice 14
Module feedback from previous students 15
Unit 1: Business Research and Research Methodologies 17
Introduction 17
Objectives 18
What is (business) research? 18
Nature and purpose of business research 19
Two research paradigms – positivism and interpretivism 21
Positivist methodologies 24
Interpretivist methodologies 26
Mixed Methods Research 28
Summary 29
Additional learning resources 30
References 30
Unit 2: Developing and Planning Systematic Approaches to Business Research 31
Introduction 31
Objectives 31
Sequence of stages 35
Summary 41
References 42
Unit 3: Searching and Reviewing the Literature 43
Introduction 43
Objectives 43
Why search the literature? 44
What is the literature? 44
Systematic literature searching 45
Study Book: Business Research
4 Bradford MBA
Ethical implications – three points 50
Citations and references 50
Reviewing the literature 51
Summary 53
Additional learning resources 54
References 54
Unit 4: Ethics in Research and Data Collection 55
Introduction 55
Objectives 55
What ethical issues are there in research? 56
Data collection 59
Surveys 60
Issues with questions for questionnaires 61
Data collection for qualitative research 63
Interview preparation 66
Summary 66
References 67
Unit 5: Data Collection and Analysis 69
Introduction 69
Objectives 70
Data collection 70
Focus groups 70
Delphi method 72
Systems and agent based modelling 73
Qualitative Data Analysis 78
Summary 81
References 82
Unit 6: Sampling and Summary Statistics 85
Introduction 85
Objectives 86
Obtaining Samples from populations 86
Summarising and describing data 93
Statistical distributions 97
Summary 99
References 100
Unit 7: Hypothesis Testing, Correlation and Regression 103
Introduction 103
Contents
Bradford MBA 5
Objectives 103
Hypothesis testing 104
Correlation 108
Linear regression 113
Summary 117
References 119
Unit 8: Presenting Findings 121
Introduction 121
Objectives 121
Presenting data graphically 122
Presenting data in a tabular form 130
Writing a research report or MBA management project 132
Summary 138
Additional learning resources 139
References 139
Unit 9: Revision 141
Introduction 141
Objectives 141
Summary 143
Appendix: Model Answers to Activities 145
Unit 1 145
Unit 2 148
Unit 3 150
Unit 4 151
Unit 5 152
Unit 6 153
Unit 7 154
Unit 8 156
Bradford MBA 7
Introduction to the Module
Your Module Leader
Ian is a lecturer at the School of Management working within the Operations and Information Management (OIM) Group. His main areas of teaching are the Philosophy of Research, Operations Management, Research Methods and Data Analysis Software.
Ian’s passion is philosophy, with a special interest in how language shapes how we understand organizations and our wider environment. He is also an associate of the Groupe Ecole Superieure de Commerce et de Management (Poitiers) and Quaid-i-Azam University in Islamabad and has lectured at the University of New York (Prague).
In addition to his academic career Ian works as an independent management consultant with a focus on processes, strategy, IT implementations and change management. He has published research on Big Data, teaching practice, supply chains, internet technologies, virtual networks, qualitative data analysis, organisational learning and creativity, lean thinking, consulting, philosophy and language and statistical process control.
Overview of Module and Module Descriptor
The module is designed to develop individual competences in research that will enable you to solve business problems. You will develop the skills and knowledge to critically appraise a business problem, explore and identify appropriate approaches to solve such a problem: developing a research strategy, devising a plan, and implementing it successfully.
Upon completion of this module, you will be able to:
demonstrate knowledge of systematic approaches to problem solving
demonstrate a critical understanding of the role of information/data in informing business decision-making
demonstrate knowledge of a wide range of data collection and analytical techniques, including: data collection, sampling, statistical analysis, the interpretation of qualitative data, interviewing skills/techniques, and how to disseminate research findings
apply ethical considerations to research activities
Study Book: Business Research
8 Bradford MBA
critically appraise research/scholarship to support the formulation of solutions to business problems
scope a project and develop a research plan that is achievable within the defined constraints of time, cost and quality
identify and select appropriate methods of data collection and analysis to solve a range of business problems
demonstrate originality when addressing and solving business problems
report findings and the consequences of business research that are appropriate for the intended management audience.
Assessment criteria
The module assessment is split into two components.
Part 1: Essay
A 1,000 word essay critically analysing one of five academic papers.
Details of the five academic papers and further instructions on the assessment are available on the Business Research Blackboard site under the menu item entitled ‘Assessment’ and then ‘Paper Critique’.
This assignment makes up half the module assessment.
Part 2: Learning Log
Units 3 and 8 of the Business Research Study Book are each associated with individual activities that make up part of the assessment of your learning.
You MUST complete both the activities and submit them as a Learning Log which constitutes 50% of the module summative assessment.
As the name suggests, the idea of the Learning Log is for you create a single document containing the two prescribed tasks – each task demonstrating your knowledge and understanding of certain Business Research related issues.
It is worth noting that the Learning Log tasks tend to correspond closely with the various online tutorials and Study Book tasks. As such you are strongly encouraged to complete these activities as well.
Introduction to the Module
Bradford MBA 9
For further details of the Learning Log tasks, go to: Blackboard > Module Site > Assessment > Distance Learning Assessment > Learning Log Tasks.
The complete Learning Log should not exceed 1,000 words. Both tasks will be given equal weighting when assessed.
The two parts of the assessment must be handed in as a single document via Turnitin (see ‘Assessment’ on the Business Research Blackboard site). The standard coursework assessment criteria pro forma will be used to evaluate your work and provide the basis for your written feedback and marks. You should familiarise yourself with the criteria on this document, as your work will be assessed against this. You should also note that each category is not equally weighted. In particular, you must demonstrate an understanding of the theoretical concepts and frameworks in your chosen business research topic. Thus more credit will be given to analytical material than to simply descriptive material.
Appendices, diagrams, tables and figures are not included in the word count. Exceeding the limit will mean that your work is subjected to a penalty. There is an ‘urban myth’ that there is a ±10% tolerance – and whilst some colleagues may suggest otherwise, it is a myth. The limit is a maximum.
Each piece of work should be typed, double line spaced and single side printed. It should have a title page which includes a word count and a signed statement of authenticity from you stating the work being submitted is your own (for example: I certify that this assignment is the result of my own work and does not exceed the word count noted below, signed.......)
Your student identification number along with page numbers must be included within the footer on every page of the document you submit for formal assessment.
Appendices should contain specific types of analyses (such as financial, breakeven, capacity, cost, etc.) and information that supports and is relevant, but would be too detailed for the body of the report.
Please note: all assignments must be submitted as a single file and the appendices must not exceed 10 pages.
Written analysis should be of the same quality that you would provide to the management of a business, albeit in an academic style.
Describing the theory does not demonstrate that you understand it – we are looking for something that shows that you can apply the theory in a particular context.
Study Book: Business Research
10 Bradford MBA
All work must be adequately supported by academic/published material and correctly attributed to the source using the Harvard system. A link to the School of Management’s guidelines for this can be found on Blackboard under ‘Assessment’.
The standard coursework pro forma will be used to provide project feedback. You should familiarise yourself with the criteria on this document, as your work will be assessed against this. You should also note that each category is not equally weighted. In particular, you must demonstrate an understanding of the theoretical concepts and frameworks in your chosen topic. Thus more credit will be given to analytical material rather than simply descriptive material. The pro forma is available on Blackboard under ‘Assessment’.
Submitting the assignment
All assignments will be submitted electronically via the module Blackboard site: go to Assessment → Dubai/Manila/MDIS Assessment → Assignment Submission.
Click on ‘View/Complete’.
You will then be taken to a submission page.
The First and Last name boxes are automatically filled in. Check that your details are correct.
In the submission title box, you should label your work using your UB number and the module code e.g. 0123456789 MAN1234M (you can find the MAN code on the title page of this Study Book). It is imperative that you upload your assignment in the above format – failure to do so may result in examiners being unable to mark your work.
Click the Browse button to upload your file. Navigate to your file and click Open.
Click Upload. Wait while your file is uploaded to the server.
The next page gives you the opportunity to review your submission. At this point you have not yet submitted, and can return to the submission page to start again if you so wish. If you are happy that this is the correct paper and you want to continue to submit, scroll to the bottom of the page and click Submit.
You will then be emailed a receipt to your university email address, which will include your assignment identification reference.
Introduction to the Module
Bradford MBA 11
For further information on how to submit your assignments using Turnitin, go to the ‘How To’ section of Blackboard (Under ‘My Organisations’) and see ‘How to Submit an Assignment Electronically’.
Support for your learning
General guidance on the support available for students from information services can be found at: http://www.bradford.ac.uk/information-services/.
Approach to studying business research
As an executive student you will be studying this module at a time and place that fits around your work, social and family commitments; however, it is strongly advised that you progress steadily through the module. In this way, you will leave yourself plenty of time to prepare for the assessment.
Reading around the subject
There is no substitute for reading around the subject; this does not necessarily mean just the material offered to you. Typically the business section of a broadsheet newspaper will have articles and news that you should be able to relate to the subject, no matter what country you are living in. Following up a subject area by researching it on the web will also make a valuable addition to your knowledge. It is your responsibility to make a note of any difficulties you may be having. Your tutor is your first point of contact. Your tutor is there to help you but will not know to do so unless you make your problems known. You can also ask questions in your group via the module’s General Discussion Board on Blackboard.
Business Research is a very practical subject; commercial organisations and academics undertake research all the time. It is disseminated on TV and the internet. Reflecting on what you see within the workplace and what sort of research you see and read about through media channels will help reinforce the learning that you are doing. This will broaden your horizon of the subject as well as showing where and how the subject connects to aspects of business that you will cover in your MBA programme. Without someone doing research on organisations, academics could not develop theory and business schools as we know them would not exist.
Textbook
Throughout the module, you will need to refer to the module textbook:
Study Book: Business Research
12 Bradford MBA
Collis, J. and Hussey, R. (2014) Business Research: Practical Guide for Undergraduate and Postgraduate Students, 4th edition. Basingstoke: Palgrave Macmillan.
The textbook forms an essential part of your study. You will be asked to read the chapters specified in each unit to familiarise yourself with the concepts and issues covered.
The text book is available online in a digital format through the University’s library webpages at http://catalogue.brad.ac.uk/record=b2219511~S1
Module Study Book
This Study Book will provide you with an insight into the subject of Business Research. In order to develop your understanding of Business Research you will be prompted to complete a number of exercises and consult a variety of information sources (e.g. academic and non-academic papers). It is strongly advised that you follow the prompts provided and engage with all the materials. The materials include:
Study Book activities
As you progress through the Study Book, you will be asked to complete a series of short activities. Completion of these activities is absolutely essential if you are to develop a good understanding of the subject. Simply reading the textbook and the Study Book will not be sufficient.
Model answers to the activities, where applicable, are provided in the Appendix.
Live tutorials on Blackboard Collaborate virtual classroom platform
During the module, you will be required to attend four live, online tutorials conducted by a module tutor. These online tutorials will provide you with an opportunity to engage in detailed, real-time discussions on key issues and concepts with other students and academics. The subject and materials for each live online tutorial are outlined in the Study Book. You will be given details of the times and dates of these tutorial sessions once the module has commenced.
If you have never used the Blackboard Collaborate virtual classroom platform before, it is essential that you familiarise yourself with the platform and test your computer setup before you attend the first tutorial. To test your sound settings, please go to the Blackboard Collaborate Support site at http://bit.ly/1etRu2Y. For further support materials and information about
Introduction to the Module
Bradford MBA 13
Blackboard Collaborate, please see the Blackboard Collaborate Support site at http://bit.ly/1jBpgUc.
Multiple-choice questions
You will have access to a comprehensive bank of multiple-choice questions in Blackboard, allowing you to monitor your understanding of basic concepts and get instant feedback at the end of each unit.
Additional textbooks
Additional sources of relevant information are contained in the following books; you are not expected to purchase these books:
Anderson, D.R., Sweeney, D.J. and Williams, T.A. (2005) Contemporary Business Statistics with Microsoft Excel: with CD-Rom and Infotrac, Mason, OH: South Western College.
Bell, E. and Thorpe, R. (2013) A Very Short, Fairly Interesting and Reasonably Cheap Book about Management Research. London: SAGE.
Blumberg, B., Cooper, D.R. and Schindler, P.S. (2014). Business Research Methods 4th edition. Maidenhead: McGraw-Hill Education.
Bryman, A. and Bell, E. (2015) Business Research Methods 4th edition. Oxford: Oxford University Press.
Davies, M.B. and Hughes, N. (2014) Doing a Successful Research Project: Using Qualitative or Quantitative Methods 2nd edition. Basingstoke: Palgrave-Macmillan.
Easterby-Smith, M., Thorpe, R., and Jackson, P. R. (2012) Management Research 4th edition. London: Sage.
Johnson, P. and Duberley, J. (2000) Understanding Management Research. London: Sage.
Kumar, R. (2014) Research Methodology: A Step-by-Step Guide for Beginners 4th edition. London: Sage
Levene, D.M. and Stephan, D.F. (2014) Even You Can Learn Statistics 3rd edition. Englewood Cliffs, NJ: Pearson Prentice Hall.
Lomas, R. (2011) Mastering your Business Dissertation: How to Conceive, Research and Write a Good Business Dissertation. London: Routledge.
Study Book: Business Research
14 Bradford MBA
Ng, W., and Coakes, E. (2014) Business Research: Enjoy Creating, Developing and Writing Your Business Project. London: Kogan Page Publishers.
Sekaran, U. and Bougie, R. (2016) Research Methods for Business: a Skill Building Approach 7th edition. Chichester: John Wiley and Sons.
Silverman, D. (2015) Interpreting Qualitative Data: Methods for Analysing Talk, Text and Interaction 7th edition. London: Sage.
Saunders, M.N.K., Lewis, P. and Thornhill, A. (2016) Research Methods for Business Students 7th edition. Harlow: FT Prentice Hall.
The Bradford University School of Management library and your local library hold publications and databases relating to industries, organisations and economic trends. The internet also gives you the possibility to access libraries elsewhere in the world.
Developing good academic practice
Harvard referencing style
Students will be required to provide references to the sources used to produce work. This shows what students have read, supports the arguments and acknowledges the work of others.
The referencing system used in this programme is called Harvard. The reference consists of two parts:
1. A citation in the text. This appears next to the information you have used. It consists of the family name of the author followed by the year of publication. Each citation is matched to a reference.
2. The reference goes in a reference list at the end of your work. The list is in alphabetical order. It contains the full details of all of the sources
referred to in the text.
For details on how to create your reference list, go to: http://www.bradford.ac.uk/library/help/referencing/.
A note about referencing in the Study Book
This Study Book provides you with a model for citing literature and presenting reference lists. Citations in the body of the units and in the References section at the end of each unit, follow the University of Bradford version of the Harvard referencing system. However, please note that the references provided in the grey boxes at the start of each unit and
Introduction to the Module
Bradford MBA 15
at intervals throughout use a different convention, with hyperlinks embedded behind the title of the item (rather than given separately at the end of the reference) and no access date included. This is simply to increase the flow of text in the grey boxes.
Module feedback from previous students
At the end of the module you will receive an email with a link to a questionnaire, requesting you to comment on the module. The feedback you submit regarding your learning experience will help the module development team in their ongoing efforts to ensure the best possible learning experience for all students.
Previous student feedback on the module includes the following comments:
“It is very structured and easy to understand.”
“Relevant assessments with helpful, considered feedback. Easy to see how the topic can be applied both at work and when I come to do my MBA research project.”
“Indispensable for the dissertation at the end of the [MBA].”
“The module provided me with an insight into the importance of research in business contexts.”
“The module has given me an overview of how to conduct business research: the type of data to look for, how to get such data, how to analyse the data and report my findings.”
Bradford MBA 17
Unit 1:
Business Research and
Research Methodologies
Key reading:
1. Collis and Hussey (2014), Chapters 1, 3 and 4
2. Trochim, W.M.K. (2006) Deduction and Induction, Web Centre for Social Research Methods, 20 October (Blackboard > Module Site > Module Materials > Unit 1) – Activity 1.3
Key audio/video: 1. Introduction to Business Research (Blackboard > Module Site > Module
Materials > Unit 1) – Activity 1.1
2. Unit 1 audio and slides (Blackboard > Module Site > Module Materials > Unit 1 > Unit 1: Audio and Slides) – Activity 1.6
Other: 1. Unit 1 multiple-choice questions (Blackboard > Module Site > Formative
Exercises > Multiple-Choice Questions > Unit 1) – Activity 1.8
Introduction
Read: Collis and Hussey (2014), Chapter 1
In this first unit, we set the scene by considering what business research is, its nature and purpose, and provide examples of the sort of things we might attempt to achieve from undertaking research in a business context. Initially we will follow the research classification introduced and discussed
in the Collis and Hussey (2014) textbook. We will then look more closely at the two alternative research approaches of induction and deduction and learn how these are linked to philosophical perspectives. Finally, we will consider several important research methodologies that may be relevant for the project you will complete at the end of the MBA programme.
Study Book: Business Research
18 Bradford MBA
Objectives
By the end of this unit, you should be able to:
understand the purpose and nature of business research
classify research approaches
understand the objectives of qualitative and quantitative research
understand the spectrum of research philosophies encompassed by positivism and interpretivism
appreciate the need for credible findings based on collected data
select an appropriate research methodology to address your research problem.
Activity 1.1 – watch and reflect
Watch: Introduction to Business Research (Blackboard > Module Site > Module Materials > Unit 1).
The film provides an overview of the issues and ideas raised and discussed in this module.
What is (business) research?
There are numerous definitions of the term ‘research’ in the literature; many of these are subject-specific. A poet was once asked what poetry was. He replied that he could not define it but knew it when he read it. Research in business is often like this; it comes in many forms and it is likely that you will have done it in your workplace, perhaps without being conscious of it.
Davies and Hughes (2014: 15) acknowledge that whilst ‘[d]efinitions of research are legion’’ they provide three which embrace the key aspects of most research projects and all are relevant to research undertaken on an MBA programme. These are:
1. ‘… gathering data in a strictly organised manner’
2. ‘… testing a stated idea or assertion (the hypothesis) to see if the evidence supports it or not. It may involve putting in place experimental practices and comparing them with other controlled or current practices,
but can also employ simple data-gathering procedures’
Unit 1: Business Research and Research Methodologies
Bradford MBA 19
3. ‘… engaging in planned or unplanned interactions with or interventions in parts of the real world, and reporting on what happens and what they
seem to mean’.
Nature and purpose of business research
Although Davies and Hughes (2014: 15) provide some working descriptions of business research, there is, as Collis and Hussey (2014) assert, no perfect definition that everybody can agree upon. However, there is a consensus on three fundamental aspects of the research process and its outcome. These are:
1. It is a process of enquiry and investigation.
2. It is progressed systematically and is methodological (i.e. is conducted such as to follow a set of prescribed steps).
3. It increases knowledge (at least for the researcher).
Looking at these and appealing to the most naive understanding of what you are trying to achieve by conducting research, the nature of research necessitates the systematic and methodological collection of data that will at some point be analysed. To realise the third point from Collis and Hussey (2014), research needs to address a specific research question such that the investigation leads to increased knowledge. All research therefore shares a common objective of creating knowledge. For this to be meaningful and of benefit, the findings from the research must be both plausible and acceptable to an independent audience. From these observations it is possible to classify research according to:
1. its purpose
2. the process by which it was conducted
3. the outcome(s)
4. the approach adopted to address the research question.
These categories suggest that there are four distinct classes, or types, of research. For this taxonomy, Collis and Hussey (2014) suggest representative examples of research that can be classified by its purpose (1).
They also discuss the two approaches, qualitative research and quantitative research (or a mixed approach, combining both these), that are used to address the research question, i.e. the research process (2).
Study Book: Business Research
20 Bradford MBA
In addition, they distinguish between the two main types of outcome of research, applied research and basic research (3). Here, basic research is academically motivated (driven by the objective of creating knowledge), whereas applied research addresses practical concerns. Because of the practical objectives of MBA projects, most are likely to be applied research.
Finally, Collis and Hussey (2014) describe the two approaches that are based on the logic of the research, namely deductive and inductive research (4). We will see later that these are predicated on the research philosophy underpinning the investigation. Inductive research is concerned with building theories and models from empirical data. For MBA projects, this data is often derived from interviews with informed or relevant individuals. This is to be contrasted with deductive research, whose objective is to work with extant theories and to validate these.
Read: Collis and Hussey (2014), sections 1.2, 1.3, 1.3.1 and 1.3.2: 1–6
Activity 1.2 – stop and think
Task 1: Reflecting on research you have observed or been involved in (in your workplace maybe) note down alternative examples for the four categories of ‘research by purpose’. (See Collis and Hussey 2014, Table 1.2: 4).
Task 2: ‘Inductive research is almost entirely associated with qualitative research, with deductive research being almost entirely associated with quantitative research’. Do you agree with this proposition? Provide a brief justification for your answer.
Activity 1.2 – stop and think answers:
Unit 1: Business Research and Research Methodologies
Bradford MBA 21
From the sequence of starting with data, leading on to the building of a theory or theories, inductive research is a research-before-theory approach and is viewed as a bottom-up approach to research. Deductive research, on the other hand, starts with a theory, or premise, leading to the acquisition of data and finally, validation of the theory. This theory- before-research approach is viewed as a top-down approach to research.
Activity 1.3 – read and reflect
Read: Trochim, W.M.K. (2006) Deduction and Induction, Web Centre for Social Research Methods, 20 October (Blackboard > Module Site >
Module Materials > Unit 1)
Note that research in business often uses mixed methods combining both inductive and deductive research. For example, you may initially pursue deductive research but find that the theory you are attempting to validate does not hold. However, from your findings, you may wish to develop a new theory of the object (often described as the phenomenon) under investigation. This would be inductive research. Alternatively, you may start with inductive research and, having developed a novel theory, collect appropriate data and test the theory in an attempt to validate it. This would be deductive research.
Although such an approach is appealing because if its completeness, it is best not to set out to undertake a mixed-methods research-based project for your MBA dissertation or for any work-related research where results and feedback are required in restrictive timescales. You may wish to speculate, using logical reasoning, if you are not able to validate an extant theory from the literature.
As an example, you set out to show that offshoring some of your core activities (e.g. customer enquiries) would be of financial benefit to your organisation. If your findings contradict this, it might be more expedient to speculate where losses would occur instead of now initiating an alternative model and doing deductive research to validate this.
However, as a note of caution, it is always prudent to look at your design and particularly, your data collection method, before looking to challenge established theories.
Two research paradigms – positivism and interpretivism
Which of the two approaches to conducting research, namely inductive or deductive, that you pursue will, in part, be determined by the way you
Study Book: Business Research
22 Bradford MBA
think about the world and your own views upon what research should look like. These are often described as your research philosophy, or research paradigm. A research paradigm is a philosophical framework that guides and influences the way that you will conduct your actual research. In simple terms, there are two contrasting approaches and these are generally considered to be mutually exclusive, being based on the researcher’s belief of what reality is and how this can be measured. These two paradigms are defined as positivism and interpretivism.
A positivist is a researcher who believes that reality is objective (existing independently of the researcher) and can therefore be observed and measured, and that social (and therefore, business) phenomena follow rules or laws that are predictable. Positivists’ thinking is underpinned by logic, objectivity and rigour and, as a consequence, they are often likened to natural scientists; dealing in laws, theories and predictive models to try to explain and understand social phenomena.
An interpretivist, on the other hand, believes that reality is subjective and socially constructed by the researcher/observer. Interpretivists do not believe that social phenomena can be predicted accurately or that predetermined laws hold for the social sciences and therefore, the business researcher.
In reality, there are numerous alternative research philosophies (and associated paradigms) bounded by these two main philosophies, forming a quasi-continuum of possibilities. Many researchers who are interested in business and organizations may often adopt a mixed approach that seems to fit between these two opposing approaches.
The main consequence of this for a research project (including your final MBA management project) is that you will need to make a choice.
If you believe that you will be able to measure the phenomenon of interest objectively, and that others would get similar measures regardless of their philosophical persuasion (i.e. the measures are robust and reproducible), you are likely to take a positivist stance. This means that you will look to pursue your research using a methodology that perhaps ultimately
supports a quantitative investigation. Furthermore, you are also likely to be doing deductive research in which you could be developing and testing hypotheses (e.g. a supposition/proposition that you may have).
If this is not the case, the chances are you will be looking to do qualitative research and, most likely, inductive research. Here, you would collect empirical data, possibly from secondary sources, e.g. data that has already been collected from a different study from yours but that may be relevant. Alternatively you may use primary sources (e.g. collect data specifically for your research, such as interviews), and develop theories of the phenomenon of interest.
Unit 1: Business Research and Research Methodologies
Bradford MBA 23
Those who do combine deductive and inductive research, leading to a composite positivist/interpretivist research paradigm, are viewed as pragmatists and are said to be following a pragmatist paradigm. Some researchers may prefer the term realist to describe an approach that sits between the two, accepting that there is an objective reality, but our knowledge of it may be subjective.
Activity 1.4 – read and reflect
Read: Collis and Hussey (2014), Chapter 3, sections 3.1 to 3.4.7: 42–54
Question 1: The fundamental difference between a positivist and an
interpretivist is in the way they view reality. Explain what the different views are.
Question 2: Which research paradigm are you most likely to be pursuing if your study is exploratory in nature? Justify your reasoning.
Question 3: Your colleague is planning to pursue a mixed methods paradigm and is claiming his approach is underpinned by the pragmatism research philosophy. You observe that he is actually only using one research paradigm, namely the interpretivist paradigm, whilst attempting to validate his findings by replicating his investigation using different data and methodologies (triangulation). Do you agree with his claim? Justify your choice.
Activity 1.4 – read and reflect answers:
Received wisdom suggests that once you have opted for a research paradigm, this will guide your choice of research methodology, the way
Study Book: Business Research
24 Bradford MBA
that you will conduct your research and what methodology or research strategy you will use. In reality, in projects, the type of data you will have access to or are able to collect will often influence your choice of methodology, not the philosophical position! However, you will at least be implicitly adopting a philosophical position if this is the case.
There are numerous standard methodologies that may be used when carrying out projects. You are not required to have a deep understanding of all of these methodologies on this module. However, you need to be aware of the main methodologies so that you can make informed choices when deciding on, and planning for, a project.
Positivist methodologies
The most usual methodologies associated with the positivist paradigm are:
1. experimental studies
2. surveys
3. cross-sectional studies
4. longitudinal studies.
Experimental studies are focused, primarily, on practical applied research, or problem-solving, and tend to be associated with basic research targeted at academically related enquiry. In contrast, surveys are a very popular choice. They address the need to know or understand the current state of a phenomenon, e.g. the relative buoyancy of a particular business sector, or the attitude to outsourcing a business function in a company, or the relationship between important variables, e.g. the extent to which company size and longevity affect average share price. The former is termed a descriptive survey, with the latter an analytical survey.
Cross-sectional studies are studies that appraise a phenomenon with data collected from different contexts/settings but at (approximately, at least) the same time. Comparisons can then be made related to the different contexts, e.g. a comparison of per employee productivity between a small to medium (size) enterprise (SME), and a large enterprise, in a particular business sector. For this type of study, the data must be collected over a short period of time so that the phenomenon will not change appreciably over this period. It is often likened to a ‘snapshot’, as you gain a view at a particular time.
This is to be compared to longitudinal studies, although longitudinal study methodology can also be used for an interpretivist study. If a cross-
Unit 1: Business Research and Research Methodologies
Bradford MBA 25
sectional study is analogous to a ‘snapshot’ from a camera, longitudinal studies can be equated to a moving picture. A variable, or group of variables representing a phenomenon of interest, are measured repeatedly over an extended period. The transient variation in these is then the subject of the study. Due to the extended length of time required for this methodology, however, it may only be suitable if historic data (or at least a large portion of the data required is historic), is available, or can be made available for analysis. This point is particularly pertinent for time- limited research such as your MBA project.
Activity 1.5 – stop and think
Read: Collis and Hussey, Chapter 4, sections 4.1 to 4.3: 59–64
Question 1: Why do you think experimental studies are rarely used on MBA management projects?
Question 2: What is the fundamental difference between a descriptive survey and an analytical survey?
Question 3: Both cross-sectional studies and longitudinal studies have a time dimension associated with them. State what this is and explain how this supports the fundamental objectives of the two separate methodologies.
Activity 1.5 – read and reflect answers:
Study Book: Business Research
26 Bradford MBA
Interpretivist methodologies
Read: Collis and Hussey (2014), Chapter 4, sections 4.4 to 4.4.7: 64–71.
Compared to positivist methodologies, there is much more variation in the range of approaches adopted by interpretivists. This isn’t surprising, because if you believe that reality is subjective, then the methods you would want to use would depend upon how you think about reality. Common methodologies associated with the interpretivist paradigm are:
1. Hermeneutics
2. Ethnography
3. Participative enquiry
4. Action research
5. Case study
6. Grounded theory
7. Gender and ethnicity studies.
The objective of hermeneutics is to understand events that have taken place, as represented and recorded in text, in their historical and social context. It involves an evolving and unfolding process of discovery leading to re-appraisal and possible re-analysis of the data in what is termed a hermeneutic circle. For example, we might want to look at how the behaviour of the leaders of businesses is shaped by the portrayal of leaders in the media and academic literature.
The objective of ethnography is to understand and interpret the ‘social world’ of a specific group of individuals (or an individual) as seen through their ‘eyes’. The idea is to get ‘up close and personal’ with people, sharing aspects of their lives and their experiences so that we come to have insight into and understanding of their lives. Often individuals and groups are observed in a natural setting using participant observation, usually over a prolonged period; this latter point often makes this an unrealistic methodology for certain research projects, e.g. MBA projects. However academic researchers who have more time often engage in ethnographic studies within the workplace.
Participative enquiry is a novel approach to research in which all participants in a natural setting are actually involved in both data collection and in setting the evolving direction of the research. There are three main approaches to this methodology: co-operative enquiry, participatory enquiry and action research. The latter is a particularly powerful
Unit 1: Business Research and Research Methodologies
Bradford MBA 27
inductive approach to business research and is often highly suited to MBA projects (especially for those studying part-time or undertaking distance learning, as they often have access to their organisation).
The objective of action research is to identify and implement appropriate change management processes in organisations. The research usually takes place in an isolated setting whereby necessary change is identified, implemented, monitored and, typically, refined in an iterative fashion. This is a particularly effective method for change management where processes are entrenched or poorly defined.
A case study is the exploration of a simple phenomenon, or case, such as a group or section within a company, a business, a process or event. The exploration of these takes place in situ, with a view to gaining a deep understanding of the phenomenon, particularly within a certain, important, context. Comparative case studies are used to compare and contrast aspects of business activities between organizations. Arguably what defines a case study is that it draws on multiple sources of data. Both primary and secondary data will be used and it is often taken from within the organization and externally. Often, interviews, surveys, analysis of internal secondary data (potentially including emails, marketing materials, websites, company accounts and procedural documents) and observations will be employed to create a sense of what is going on in the organization. Case studies have been increasingly used to understand businesses and organizations, starting in the 1970s when researchers sought to develop techniques that went beyond ethnographic research. It is an ideal way of bringing together lots of diverse data and can in fact often utilize both quantitative and qualitative data (See Mixed Methods on the next page). Although a very intensive form of research, a Case Study can, if you are not overly ambitious, be a useful way of researching the sorts of problems often explored in MBA projects.
Grounded theory was developed as a counter to what was perceived as the limitations of much positivist research. As such, grounded theory is interpretivist research, where the objective is to develop theories about a phenomenon based on interpretivist reasoning and procedures from data derived directly from the phenomenon. There are three suggested stages (Silverman, 2015) that lead, ultimately, to the identification of categories
that can be used as broader, more general analytical frameworks.
With society rightly demanding greater corporate social responsibility, fairness, inclusion and diversity in the workplace and beyond, studies on social stratification phenomena are becoming increasingly attractive and relevant. Feminism, gender studies in general, and ethnicity studies are therefore likely to become increasingly important.
Study Book: Business Research
28 Bradford MBA
Mixed Methods Research
Read: Collis and Hussey (2014), Chapter 4, section 4.5: 71–72
Finally, it is important that you are aware of the possibility of using a mixed approach. It is possible to use a number of methods within a single research project, in order to create knowledge. Arguably, mixing methods would be problematic from a positivist standpoint, since if the correct data was collected, it would not require a mixture of methods to answer the research question. Nevertheless, many researchers do employ mixed methods, irrespective of the paradigm they come from and their research philosophy. It is possible to mix different quantitative or qualitative
methods, or to employ both types of data in a single piece of research. We have already encountered triangulation. This is a basic approach in which a research question is addressed in multiple ways that are considered independent. This could be achieved, say, by different researchers researching the same phenomenon independently, or the same researcher using different data sources and different methods. If the resultant findings from these independent sources tally, we have evidence to validate the consistent findings. Like a cartographer (map maker), triangulation sets out to observe a phenomenon from different positions (using different techniques) and, by looking from each, the research is able to ‘map’ (understand) the phenomena more accurately than is possible by taking readings from just one point. However, as Collis and Hussey (2014) point out, not all mixed methods research employs triangulation, and it is possible to mix methods in a variety of ways.
As with all research, it is always important to remember the purpose of the research. More methods may simply complicate your research, and will almost certainly require more time and resources.
With so much choice, it is often difficult to determine the most appropriate methodology to use for a project. The following checklist of issues may help you in this regard. It depends on:
1. the research question you are trying to solve
2. your research philosophy (e.g. positivist)
3. the existing body of knowledge (e.g. does theory exist?)
4. the type/quality of data you will be able to collect
5. the volume of data
6. the resources available: – personal
Unit 1: Business Research and Research Methodologies
Bradford MBA 29
– computing systems – finance
7. scope
8. return on investment
9. timescales.
Summary
In this unit we have defined and looked at the nature and purpose of business research. Working from the taxonomy provided by Collis and Hussey (2014), we then discussed the two main research approaches, induction and deduction, and saw how these are consequences of the research philosophies of positivism and interpretivism, respectively. Finally, we looked at several of the main research methodologies that are regularly used in business research and that may be useful when you consider and design a project, including your MBA management project.
Activity 1.6 – audio and slides
Listen to: the audio for this unit and view the accompanying slides at the same time. (Blackboard > Module Site > Module Materials > Unit 1 > Unit 1: Audio and Slides)
Activity 1.7 – review activity
Read: Collis and Hussey (2014), section 3.5: 54–55
Question 1: What is a pragmatic paradigm?
Question 2: The main difference between a cross-sectional study and a longitudinal study is that the former is an investigation at a particular time, whereas the latter is at a sequence of time points. With this in mind, how does a longitudinal study therefore differ from a mere sequence of cross-sectional studies in the time domain, and why?
Activity 1.7 – review activity answers 1 and 2:
Study Book: Business Research
30 Bradford MBA
Activity 1.8 – multiple-choice questions
The multiple-choice questions for this unit will enable you to test your knowledge and understanding of the key concepts covered. (Blackboard > Module Site > Formative Exercises > Multiple-Choice Questions > Unit 1)
Additional learning resources
Additional resources are available if you wish to learn more about the issues covered in this unit (Blackboard > Module Site > Module Materials > Unit 1 > Additional Learning Resources).
You can also go to www.socialresearchmethods.net/kb/dedind.php to look at a schema and description of inductive and deductive research.
References
Davies, M.B. and Hughes, N. (2014) Doing a Successful Research Project: Using Qualitative or Quantitative Methods 2nd
edition.
Basingstoke: Palgrave-Macmillan.
Silverman, D. (2015) Interpreting Qualitative Data: Methods for Analysing Talk, Text and Interaction 5th edition. London: Sage.
Trochim, W.M.K. (2006) Deduction and Induction, Web Centre for Social Research Methods, 20 October. www.socialresearchmethods.net/kb/dedind.php Accessed 10 June 2016.
Bradford MBA 31
Unit 2:
Developing and Planning
Systematic Approaches to
Business Research
Key reading: 1. Collis and Hussey (2014), Chapters 1, 2 and 6
2. Davies, M.B. and Hughes (2014) Ten steps to get you off to a good start, in Doing a Successful Research Project. Basingstoke: Palgrave MacMillan (Blackboard > Module Site > Module Materials > Unit 2) – Activity 2.1
Key audio: 1. Unit 2 audio and slides (Blackboard > Module Site > Module Materials >
Unit 2 > Unit 2: Audio and Slides) – Activity 2.6
Other: 1. Unit 2 multiple-choice questions (Blackboard > Module Site > Formative
Exercises > Multiple-Choice Questions > Unit 2) – Activity 2.7
2. Unit 2 live online tutorial (Blackboard > Module Site > Collaborate) – Activity 2.8
Introduction
Read: Collis and Hussey, sections 1.5–1.8, 2.1, 2.3–2.5 and 2.7
In this unit, we argue the need for tackling business research, and any form of business problem-solving, in a systematic and planned way. We initially look at what business research is and introduce some basic rules when undertaking a research project. The sequence of stages that is typically followed as a part of the research process is then discussed.
Objectives
By the end of this unit, you should be able to:
Study Book: Business Research
32 Bradford MBA
identify the research problem
formulate the research question
appreciate the need to schedule research activities
understand the research process
appreciate practical considerations involved when engaging in a research project.
Regardless of the precise nature of your chosen investigation, there are basic rules that you should consider adopting when undertaking a research project, including your MBA management project. These are:
1. Keep it simple (the principle of parsimony).
2. Do not try to do too much (reserve your ambition to getting it right – quality vs. quantity).
3. Be patient and realistic (do things properly at each stage – yet proportionately to the importance of the task).
4. Break down your problem into a solvable, sequence of manageable tasks (i.e. schedule).
5. Plan, and execute each stage (task) of the investigation with sufficient resources to do a good job (time management, persons, etc.).
Activity 2.1 – read and reflect
Read: Davies, M.B. and Hughes N. (2014) Ten steps to get you off to a good start, in Doing a Successful Research Project. Basingstoke: Palgrave MacMillan (Blackboard > Module Site > Module Materials > Unit 2)
Activity 2.2 – stop and think
Consider a project that a colleague has decided to undertake. They may have conducted a survey of a business sector that they are presently employed in, e.g. IT sales to small-to-medium size enterprises (SMEs). They are especially interested in the type of after-sales service the IT providers offer and whether this is considered as a value-add by the SMEs.
Identify what issues could be present here that could hamper successful completion of the project. What important time overheads would need to be planned for? Write your comments in the following box:
Unit 2: Developing and Planning Systematic Approaches to Business Research
Bradford MBA 33
Activity 2.2 – stop and think answer:
After identifying the research topic of interest, the starting point for any research project is the research problem, which, after careful consideration, leads on to a research question. It is this research question that you will then set out to address and possibly solve. (Note: All research is designed to answer a research question.)
The following is a company-based research problem that is typical of the type of problems that MBA students might consider:
A marketing consultant for a consortium of MSEs (micro and small enterprises) wishes to establish the level of sustainability of MSEs in the UK. The company is looking to secure European Union (EU) grants to increase sustainable activity in EU-based organisations.
The research question here is:
‘What is the level of sustainable activity for UK-based MSEs?’
There are several issues that you should consider when addressing the research question and the following are some for this sustainability problem:
a. What is sustainability (for/of a company)?
b. How is sustainability measured?
c. How are you going to measure it?
Study Book: Business Research
34 Bradford MBA
d. What will the level of sustainability be compared to?
e. What specific questions will you address to answer the research question?
f. Which MSEs are you going to approach?
g. How will you approach them?
h. How many will you approach?
i. How will you collect the responses/data?
j. How will these be stored?
k. What type of analysis will you conduct?
l. How will the findings be reported?
Before you can start to plan your activities – if appropriate – break down the activities into manageable tasks. You will need to produce a list like this of the issues, and consequent activities that you will need to conduct to answer your research question.
Activity 2.3 – read and reflect
Read: Collis and Hussey (2014), sections 6.1 to 6.5: 97–107
Appealing to your current and previous work experiences and training, you are to consider what steps you would take to answer the list of questions related to the MSE sustainability project outlined above. The sequences of steps, or activities, you may consider include:
A = doing a thorough literature review.
B = searching directories, databases; a literature review; research knowledge (if appropriate, informed colleagues); a small-scale (pilot) study.
C = searching databases; research knowledge.
Note down your responses/comments for the following questions.
Question 1: Identify which option (A, B or C) you would undertake to address each of the issues (a–l) previously introduced.
Unit 2: Developing and Planning Systematic Approaches to Business Research
Bradford MBA 35
You should note that there is a sequence and structure to the activities associated with establishing the level of sustainability of MSEs. This is usual and it is necessary to schedule this in a way that allows you to develop the research and move systematically to a conclusion.
Question 2: Look at the list of options that you have considered in order to address the main research issues (a–l, previously introduced). Identify any other main issues that you believe will aid in answering the research question and provide a brief explanation for their (or its) inclusion.
Activity 2.3 – read and reflect answers:
Sequence of stages
As implied by the sequencing of scheduled activities, it is recognised that efficient research is conducted in a prescribed sequence of steps or stages. Davies and Hughes (2014) suggests that for research projects, including student dissertations, this sequence is:
1. Planning stage
2. Data collection stage
3. Data analysis stage
4. Write-up stage.
These stages can be refined and broken down into a sequence of smaller, more distinct steps (see Collis and Hussey 2014) but these stages are generally considered as the appropriate sequence. We will see later that it is often prudent or even necessary to revisit earlier stages in the light of your findings. This does not compromise the integrity of the basic
Study Book: Business Research
36 Bradford MBA
sequence though; it is just an expedient way to proceed on the basis of evolving knowledge.
The planning stage
From the above, the first stage in the systematic formulation of a project is the planning stage. Initially you will have identified a topic which typically leads to a broad research problem.
If the research project in question is your MBA management project, then after you have identified the topic / research problem and searched through the appropriate literature, your research question will be
developed. Continuing to focus our attention on the MBA research project – under a positivist paradigm (see Unit 1), Black (2002) suggests that testable hypotheses should accompany this research question. As an example for your MBA project, you may be interested in the broad research problem of which factors influence the successful adoption of new technology. Here, the research question could be: ‘Do positive perceptual factors lead to the take-up of new technology?’ It is most likely that you would want to scope your investigation for a particular business sector, say, or to reduce the generality in some other way to enable you to tackle your project with a realistic prospect of completing it on time! The initial review of existing research might have led to the following two testable research hypotheses (H1 and H2):
H1: Perceived usefulness (PU) is positively related to a positive attitude towards take-up of new technology.
H2: Perceived ease of use (PEU) is positively related to a positive attitude towards take-up of new technology.
Before proceeding at this planning stage you will need to ask yourself the obvious, namely, does the research question (and the testable hypotheses, if relevant) lead to a solution to the research problem? If it does not, you will need to revisit your initial deliberations and formulate an alternative that does. Once you are satisfied that the proposed project will be addressing the research problem appropriately, you then need to conduct a full literature review to gain a deep understanding of current thinking on your topic (we will be looking at ways to do this in a later unit). From this, you will be able to identify data requirements, to plan your subsequent data collection and, if relevant, to draft any questions you may need answers to. Collis and Hussey (2014) provide a schema for the sequence of steps associated with identifying the research question or questions and an abridged version of this is given in Figure 2.1 below. An informal algorithm for the sequence of steps taken at the planning stage is also exhibited in Figure 2.2.
Unit 2: Developing and Planning Systematic Approaches to Business Research
Bradford MBA 37
Figure 2.1: Schema for sequence of steps at the planning stage
Items
Figure 2.2: Informal algorithm for planning stage of business research
The data collection stage
If relevant to your research, you will need to conduct sampling to identify potential participant(s). You can then collect data directly from these
1. State purpose
2. Formulate specific
questions/hypotheses
3. Are questions relevant
to solving problem?
No
Yes
4. Conduct literature
review
5. Does solution
already exist?
6. Continue in
Planning stage
No
Yes
Identifying research question(s)
(Source: Collis and Hussey, 2009)
Conduct literature review
Identify data requirements
Plan data collection approach
Draft questions (if relevant)
Pilot data collection (if feasible)
Piloting your data collection helps with the following: – identifying inconsistencies – identifying illogicalities – identifying general feasibility of approach – may determine required sample size / timescales for data
collection – could solve your problem!
Study Book: Business Research
38 Bradford MBA
identified sources. You should set time limits prior to commencing the data collection and record responses periodically. In addition, if engaged in a medium to large survey, consider undertaking an interim analysis to avoid surprises or potential findings that are counter to your expectations (or to those based on the literature). It can also alert you to an obvious lack of sufficient data or systematic cases of missing data from the respondents.
However, if your research is inductive (see Unit 1) and the objective is theory-building or answering a specific research problem, data collection could take the form of interviews and you should record and reflect on all data at the time of collection so that a picture emerges. An informal algorithm for the suggested steps to take at the data collection stage is exhibited in Figure 2.3.
Figure 2.3: Informal algorithm for the data collection stage of business research
The data analysis stage
The data analysis stage should also be conducted in a preferred order. Again, you can see this in the accompanying algorithm in Figure 2.4 below. This is for positivist, deductive research. For inductive research the Exploratory Data Analysis (EDA) and statistical analyses steps would not be relevant, as you would be analysing the results without recourse to quantitative analyses.
[If relevant, conduct sampling to identify potential participants]
Collect data directly from source(s) – Set time limits – Record responses periodically
[If a survey – follow up with reminders to initial non-responders]
Consider undertaking interim analyses to avoid: – surprises/counter-indications – obvious lack of sufficient data – ambiguities in data collection instrument (e.g. interview questions /
questionnaire) – systematic missing data
Write up a draft literature review
Write up your data collection instrument and collection method
Unit 2: Developing and Planning Systematic Approaches to Business Research
Bradford MBA 39
Figure 2.4: Informal algorithm for the data analysis stage of business research
The writing-up stage
The final stage of writing up your project will be considered, in detail, in Unit 8.
Activity 2.4 – read and reflect
Read: Collis and Hussey (2014), sections 1.5, 1.6, 1.7 and 1.8: 9–19
This will consolidate the points previously raised and enable you to see the characteristics of a ‘good’ research project:
Collis and Hussey (2014: 11) state that “[y]ou should not make decisions about your data collection method without also deciding which analytical method you will use”. To demonstrate this, you might be collecting data from interviews. This data will probably be ‘rich’ in information but could be highly particular to the interviewee. This is to be contrasted with very specific questions that are often present in a questionnaire; here, responses from each participant can be directly compared. The former (interviewing) is more subjective in that respondents may not even address the same narrow issue. This inductive approach will not necessarily lend itself to statistical analyses whereas the latter (questionnaire) is more likely to.
The number of responses is also likely to vary between approaches, with relatively few interviews being conducted in comparison to a survey questionnaire that could be answered by substantially more participants.
Record / finish recording responses
Conduct Exploratory Data Analysis (EDA)
Undertake statistical analysis (if relevant) – Fit appropriate statistical model to data
Consider your results carefully
Consolidate your findings
Relate findings to the Business Problem
Write up a draft analysis
Write up a draft of findings and initial conclusions
Discuss findings and conclusion(s) with supervisor/sponsor/colleagues
Study Book: Business Research
40 Bradford MBA
In the former, regardless of the nature of the data obtained, there still may not be sufficient responses to conduct a suitable statistical analysis. However, even for questionnaires, some statistical methods require a large sample before they can be used. As you can see, issues regarding the research paradigm, the sampling method used and the size of the response are all factors that need to be borne in mind in conjunction with the analytical method to be used.
Another important point that Collis and Hussey (2014) raise is that of timing. For all research projects, writing up takes time. If the required writing up time is not set aside the final report may not fully reflect the quality of the research actually conducted.
Activity 2.5 – stop and think
In the past, you have probably been interviewed and/or participated in a survey, where you filled in a questionnaire.
Question 1: Describe the main differences between the data you would typically obtain from conducting an interview and those you would get from a questionnaire.
Question 2: Which of these two approaches is typically the more labour- intensive data collection activity? Justify your answer.
Activity 2.5 – stop and think answers:
Read: Collis and Hussey (2014), sections 2.1–2.4 and 2.6: 22–30 and 35– 38
This will enable you to understand some additional practical issues that will confront you when undertaking a research project.
Unit 2: Developing and Planning Systematic Approaches to Business Research
Bradford MBA 41
Summary
In this unit, we have argued that business research should be conducted in a systematic and planned way. A sequence of four basic steps was proposed and you have investigated these in detail. Practical issues related to these have also been identified.
Activity 2.6 – audio and slides
Listen to: the audio for this unit and view the accompanying slides at the same time (Blackboard > Module Site > Module Materials > Unit 2 > Unit 2: Audio and Slides).
Activity 2.7 – multiple-choice questions
The multiple-choice questions for this unit will enable you to test your knowledge and understanding of the key concepts covered (Blackboard > Module Site > Formative Exercises > Multiple-Choice Questions > Unit 2).
Activity 2.8 – live online tutorial
Consider the following scenario: A colleague wishes to conduct a survey within the business sector they are presently employed in – IT sales to small-to-medium size enterprises (SMEs).
Your colleague is especially interested in investigating whether the type of after-sales service that the IT providers typically offer SMEs is seen to add value.
In order to tackle your colleague’s IT sales problem you have decided to conduct mixed methods research. Initially you will use an inductive approach to establish what types of after-sales provision are of interest to the SMEs and whether these services are likely to lead to sales of IT to SMEs. From this, you aim to develop a theory related to types of IT after- sales provision and their effect on future sales. You then propose to undertake deductive research to validate this research theory.
Task: Prior to the online tutorial session, consider the above scenario and draw up a plan outlining the steps you would take when conducting this research project.
Bring your ideas and answers to the online tutorial session and we will consider the various approaches put forward. As a note of encouragement – there being no one definitive process – your tutor will be interested in the varying ideas that you put forward.
Study Book: Business Research
42 Bradford MBA
Live online tutorial: To access the tutorial, please go to Blackboard > Module Site > Collaborate. To test your sound settings for the Blackboard Collaborate virtual classroom, please go to the Blackboard Collaborate Support site at http://bit.ly/1etRu2Y. For further support materials and information about Blackboard Collaborate, please see the Blackboard Collaborate Support site at http://bit.ly/1jBpgUc.
References
Black, T.R. (2002) Understanding Social Science Research 2nd edition. London: Sage.
Davies, M.B. and Hughes, N. (2014) Doing a Successful Research Project 2nd edition. Basingstoke: Palgrave-Macmillan.
Bradford MBA 43
Unit 3:
Searching and Reviewing
the Literature
Key reading:
1. Collis and Hussey (2014), Chapter 5
Key audio: 1. Unit 3 audio and slides (Blackboard > Module Site > Module Materials >
Unit 3 > Unit 3: Audio and Slides) – Activity 3.5
Other: 1. Unit 3 multiple-choice questions (Blackboard > Module Site > Formative
Exercises > Multiple-Choice Questions > Unit 3) – Activity 3.6
Learning log: 1. Unit 3 learning log (Blackboard > Module Site > Assessment > Distance
Learning Assessment > Learning Log Tasks > Unit 3) – Activity 3.7
Introduction
In this unit, you will learn how to search literature and how to conduct a literature review. This will help you to identify possible sources of secondary data, i.e. data that is relevant to your research but which has been collected for another project. It will also make you aware of the current thinking and knowledge about your research project and help you to establish your research question. It may also suggest the appropriate research methodology you should use to address this research question. In addition, the importance of cataloguing relevant results from your search is emphasised. Finally, we will discuss a skill that is useful when
compiling any project but particularly your MBA project, namely, how to cite and reference the literature appropriately.
Objectives
By the end of this unit, you should be able to:
search the relevant literature
review the literature
Study Book: Business Research
44 Bradford MBA
identify the main research question for your project
identify potential secondary data sources
record references and provide appropriate citations
write the literature review.
Why search the literature?
When embarking on a research project it is frequently the case that one of the first steps will be a search and review of the existing information/ literature relating to the subject area. In this unit, we will demonstrate that
the search needs to be systematic, and the subsequent review critical. By doing this you will learn what the current state of knowledge and thinking is about your topic, possibly revealing gaps in knowledge that you may wish to address with your research question. You will also be able to see what methodologies have been used to tackle the research problem to date and this is likely to inform you about the approach you should also be pursuing.
What is the literature?
Strictly speaking, literature is any source of secondary data that is relevant to your research project topic. These sources are numerous and comprise, but are not limited to:
Libraries, journals, e-journals, conference papers and proceedings, trade/professional journals/magazines, newspapers, web search engines, the internet, JSTOR, Wikipedia, items found via searching on Google Scholar, company reports, government documents (including websites), social networks, FAME database and the ABS (Association of Business Schools) list.
As you can see, the list is not exhaustive. However, not all sources are equally reliable and you need to factor this in when searching and reviewing the literature. To give an example to justify this assertion: consider a newspaper article that is relevant to your research project. How have the claims made within this article been validated? Are the statements made by an unduly interested party or are they subjective reporting by an affected journalist? For example, would you expect a CEO for an electrical retailer to declare that the market in electrical retailing is about to collapse or, despite predictions, would he/she be more bullish? Even published academic papers can be ranked for currency and impact, if not credibility. When searching journal articles, the recommendation is to use the CABS (Chartered Association of Business Schools) list to determine the importance of the article based on the rating of the journal the research is reported in. Journals included in the list are rated from ‘1’
Unit 3: Searching and Reviewing the Literature
Bradford MBA 45
through to ‘4’, denoting national quality journals through to those considered to be world class, respectively. Ideally, you would wish to concentrate on journals that are rated at the ‘3’ level, which is for top international quality journals, and the world class ‘4’ category journals. Indeed, Ryan et al. (2002) suggest that you concentrate on articles that are published in the top two or three journals in the field of research under investigation, an even greater restriction.
As a researcher you have to think about the audience for your research and use literature that is credible to that audience. For your MBA project, that audience may mean the organization you are researching and also the academics who will read and assess your project report. It is important that you support your own research with literature that makes your findings convincing.
Systematic literature searching
As previously stated, the main purpose of searching the literature is to develop a deeper understanding of your research topic and to see how best to pursue your investigation. In addition, you are likely to see what approaches and methodologies have been used by others researching in the same, or a similar, field. You could find a data collection instrument such as an interview protocol or questionnaire that has already been developed, tested and proven to be reliable. You may even identify a source of secondary data that would be sufficient for your investigation.
As with other aspects of your research, you should search the literature in a systematic way, initially with a broad scope and with a priori set ideas of what is relevant (potentially captured by key words if it is an academic piece of research). This initial search will usually be conducted either on the internet, including using the electronic library facility at Bradford – the web catalogue. The important thing is to have a checklist of what you are actually looking for. The following are three basic things to look for (there may be others, given your specific research question):
1. What is the purpose of the study reported / information provided; what does this add to that of the other information/literature available and
how does it help with my research?
2. How was the research conducted? Is this different from the other studies and if so, what are the advantages that I can take from this?
3. How do the findings inform me and help with my own investigations?
Once you have identified relevant content you should always record the information present from each source. Any web addresses need to be noted and stored, soft copies of the articles can be obtained and stored,
Study Book: Business Research
46 Bradford MBA
and hard copies indexed and placed in files. Collis and Hussey (2014) provide a suggested procedure for the overall process.
When carrying out the initial literature search you will probably see text that you might want to quote in your project report. It is always a good idea to copy and paste (or just reproduce) this and store it in a separate ‘Quotes’ document or spreadsheet, accompanied by the reference and page number (see later for format). Experience suggests that otherwise you will spend considerable time trying to locate it when you actually need it for your report.
Activity 3.1 – read and reflect
Read: Collis and Hussey (2014), subsection 5.2.2: 77
Study the suggested procedure for conducting a systematic literature search and then answer the following questions.
Question 1: The authors claim in the penultimate item in their list that when you start to recognise the references cited in other works, you are near the end of your first search. Why is this likely to be the case?
Question 2: Why would this be your ‘first’ search?
Question 3: Having identified in the literature researchers (possibly several) who have contributed substantially to your chosen research topic, it may be expedient to search their institutional, professional or private/social websites. Why?
Activity 3.1 – read and reflect answers:
Unit 3: Searching and Reviewing the Literature
Bradford MBA 47
From the suggested procedure for systematically searching the literature in Collis and Hussey (2014) there are several issues present, namely:
scoping (and context); key word(s) searching; and, by extension, finessing or refining the search.
Scoping
It is important to scope your search, otherwise you may end up with an unmanageable number of selected articles to investigate, especially if you have the right key word(s) – see next subsection.
Scoping may relate to time relevance. For example: if interested in the implementation and take-up of a new technology or a recently advocated business practice, say, you would probably only be interested in articles/papers that are a limited number of years old. You would therefore look for a cut-off time for data gathered in your search.
You may also be restricting your investigations to a particular location. An example here might be a research question about agricultural, migrant workers in a region in Australia, or mine workers in Chile.
Your investigation might also need to be scoped to take account of your interest in a single discipline with a multi-conceptual approach, such as looking at the role of training (single discipline) in change management (multi-conceptual).
Such an expedient approach is often necessary to make a thorough
search of the literature relevant and tractable.
Key word searching
Having decided where to search and how to focus the search, you need to specify what to search for. To achieve this you should identify specific key words (or phrases) that describe or represent the main aspects of your topic. It is these that you search for in the literature or surf the internet for. You must be flexible here and look for alternative spellings (e.g. UK versus US English), synonyms, phrases or non-standard English. You will
Study Book: Business Research
48 Bradford MBA
probably have to update your initial key words to either broaden the search, if too narrow, or tighten it up, if too broad. You may also choose to restrict your search to key words in certain sections of the articles (e.g. the abstracts/introductions only, or the body of the text only). Many databases and search engines now have “cited by” links. These can be used to find newer articles that have cited articles that you have found to be interesting and/or relevant.
Searching literature is an ongoing iterative process of refinement which should continue until you are happy with the literature that you have selected. As a rough guide, when you find that you are coming up with articles that you have already located, or are sifting through very old articles, these are indications that the process is nearing completion, or that your search terms are not appropriate. If, upon changing the search terms, nothing else of relevance appears, you may be reaching the point at which you stop searching. However, remember that however thorough you are, you may find that on another occasion you find relevant literature that you had previously missed.
Activity 3.2 – read and reflect
Read: Collis and Hussey (2014), subsections 5.2.3 to 5.2.4 and section 5.3 (intro only): 77–83
This will help reinforce your understanding of this subject area.
In Collis and Hussey (2014), subsection 5.2.4, ‘Defining key words’, the authors introduce Boolean operators that are used in ProQuest, a particularly useful online academic database that is currently available on the Bradford University library website. This use of proximity and adjacency search options is typical and usually available in many search engines that you are likely to encounter (especially academically- orientated search engines). The syntax may vary slightly between packages but use of the ‘Help’ facility should enable you to address any discrepancies.
Although the use of reference manager software packages can be appealing (p.83), it may not be needed on a project. Many students do, however, find these to be of considerable benefit but unless you are familiar with a particular package you will need to factor in the potentially time-consuming activity of learning how to use it! A typical word processing document is often all that is required. However, a word of warning: don’t delete it! You would not be the first person to delete this very important document. Sadly, its true importance will become apparent only after the event.
Unit 3: Searching and Reviewing the Literature
Bradford MBA 49
Activity 3.3 – stop and think
You are conducting research to establish the take-up of a new technology. A colleague mentions that the Technology Acceptance Model (TAM) may be appropriate and you decide that TAM is the first model to look at to see how to progress your research and the likely methodology that would be appropriate.
Task 1: Go to the academic search engine, Google Scholar [Google Scholar], and select TAM articles using ‘TAM’ as the key word for your search. Note how many papers and articles are selected.
Task 2: Repeat this but with the full name (Technology Acceptance Model) as the keywords. With the inclusion of the full name for the TAM, what difference did you notice and why do you think this difference is evident?
Task 3: Look at a few of the articles in detail and decide on a specific aspect or context that you wish to investigate further (e.g. something specific to mobile computing, say). Choose a suitable key word or phrase to narrow your search and search again, using this. Note the reduction in the number of selected articles. If your key word was too specific, you would have narrowed your search too much. If it is still too general, this would not have reduced the redundant articles selected that much.
Task 4: Continue with this process for a few iterations to experience the overall effect.
Activity 3.3 – stop and think findings:
Study Book: Business Research
50 Bradford MBA
Ethical implications – three points
1. Before you undertake searches of the literature you MUST ensure that you have appropriate permission to access the articles. As a general rule, if the articles are posted on the internet or any other obvious public domain by the rightful owner (e.g. the author, publisher or institute) there should not be a problem. Furthermore, any articles accessed using the University e-journal facility will also be acceptable. However, it is important to note point 3 below regarding acknowledgement of sources. If you find material on the web that you do not have permission to use, you need to be very careful before you use it. Just because material has been put in the public domain, it does not mean the owner
wishes it to be available. If in doubt, seek advice.
2. In relation to the literature search associated with your MBA project: if you have been given academic help in your search (referral to, or access to, specialist databases; provided with, or alerted to the presence of, specific articles, etc) you should consider formal acknowledgment of this. Check with the person providing the help to
see what their expectations are.
3. If your research is informed by an article in the literature (or by any other individual or source, for that matter) you should acknowledge this with a formal acknowledgement, citation and/or reference in your report. Please note – if you do not include this, in your MBA project it could lead to a charge of plagiarism and would compromise the integrity of your MBA. For more general projects it is simply good practice to do
this.
Read: Collis and Hussey (2014), section 5.5: 92–93
Citations and references
It is important that you cite all articles that you have based your work or your arguments on, in the text of your report. This is done tersely so that the flow of the discussion is not interrupted. Full details are then provided in a separate References section after the body of text of your report. For some projects such as research dissertations and particularly theses, some authors provide an additional Bibliography section in which an extensive list of relevant publications is listed, for completeness. This will contain references to important and relevant papers from the literature but that are not necessarily cited in the actual research reported. You may or may not wish to do this for more general reports.
There are numerous formats used in the literature for both the citations and the bibliographical information about an article, present in the Reference or Bibliography sections. Most follow an author-date or a
Unit 3: Searching and Reviewing the Literature
Bradford MBA 51
number-based system. Arguably, the most popular system used in management is the Harvard referencing system. This is an author-date based system and there are, regrettably, several marginally different variations of this basic system that are in use by journals.
Please note – all assessed work at Bradford including dissertations and theses requires you to use the standard Harvard system and so it is important that you are familiar with this for both citing and referencing publications.
An explanation of the Harvard system, examples of citations in the body of your text and reference formats and conventions using this system are given in the textbook. However it is essential that you refer to University’s Referencing Guide for Harvard which can be found at: http://www.bradford.ac.uk/library/help/referencing/
Read: Collis and Hussey (2014), subsections 5.3.1 and 5.3.2: 84–87
Reviewing the literature
Once you have conducted your initial literature search and catalogued all of the relevant documents you have identified, you will need to undertake a review of the literature. This is a critical appraisal and evaluation of this literature. Assuming you have managed to capture all of the important articles, this will enable you to comprehend, more fully, issues regarding the research topic, the current state of knowledge and thinking, and what methodologies have been used in previous studies to address these issues. Furthermore, as it is usual for research articles to also report the limitations of the study and to suggest future work, these will be of particular importance to you. When conducting academic research, a review of the literature will enable you to see where the gaps are, or any limitations in the scope of previous studies. This will help you to decide on your research question, or highlight that you need to amend it. It will also inform a decision about the research methodology you adopt and shape the overall project design. Finally, it will help you place your findings in context.
The process of further searching the literature and updating your review, periodically, will be necessary throughout your project. When doing this, you will need to adopt a systematic, analytical approach and to record important features of the research reported. Indeed, for large projects or shorter duration projects with an extensive literature, many researchers develop a network of the primary citations they have identified so that they are able to grasp the overall relationships between these articles (Ryan, et al. 2002). If your project fits into the latter category, you may find such an approach useful, although this would be rather untypical for Bradford MBA projects.
Study Book: Business Research
52 Bradford MBA
Read: Collis and Hussey (2014), sections 5.4 to 5.4.2: 87–89
Focusing specifically on the MBA project, you will normally be expected to write up a literature review and include it as a chapter in your report. This is a standard requirement and enables you to demonstrate your subject knowledge and to place your study in context. It is crucial that you are able to demonstrate a complete and comprehensive understanding of the subject area. It will also help to place your findings in context, making it apparent to the reader what contribution your study has made.
Collis and Hussey (2014) remind us that you need to be flexible when searching for relevant articles and this is particularly so for company- based projects that look, on first inspection, to have no parallels in the literature. However, it is extremely unlikely that this is the case for an MBA project, particularly as we encourage students to pursue relevant, practice- based research. If you are struggling here, look at the suggestions in Collis and Hussey (2014) and then, if you are still at a loss, discuss this with your supervisor.
Read: Collis and Hussey (2014), subsection 5.4.3: 89–92
Activity 3.4 – stop and think
Your MBA colleague has a postgraduate qualification in Chemical Engineering and has decided to conduct a project on the feasibility of introducing a new and innovative paper-reprocessing plant in one of the recent European Union member countries. Your colleague argues that she has the specific expertise in this new technique (knowledge gained when she studied her Masters of Technology at Georgia Tech). Further to this, she has practical work-related experience as a result of working in such a plant in the US. She will also be able to attract EU start-up money to pursue this entrepreneurial activity. When asked about the current information available in this area she claims there is no relevant literature for such a project. What suggestions would you give her to enable her to access suitable literature?
Activity 3.4 – stop and think answer:
Unit 3: Searching and Reviewing the Literature
Bradford MBA 53
Summary
In this unit, we have looked at how and where to search the literature to inform your research. We have focussed on ways in which you can tackle this important task efficiently and emphasised the need to continue this process, periodically, throughout the duration of your study. We have also seen how, by analysing the identified literature to conduct a literature review, you are then in a position to raise your research question. From this it is also possible to scope your research and identify the appropriate research methodology to pursue.
Activity 3.5 – audio and slides
Listen to the audio for this unit and view the accompanying slides at the same time. (Blackboard > Module Site > Module Materials > Unit 3 > Unit 3: Audio and Slides)
Activity 3.6 – multiple-choice questions
The multiple-choice questions for this unit will enable you to test your knowledge and understanding of the key concepts covered. (Blackboard > Module Site > Formative Exercises > Multiple-Choice Questions > Unit 3)
Study Book: Business Research
54 Bradford MBA
Activity 3.7 – learning log
Learning log: work through the learning log activities for this unit (Blackboard > Module Site > Assessment > Distance Learning Assessment > Learning Log Tasks > Unit 3)
You MUST save all of your workings in your Business Research Learning Log document. Your answers to the various activities will be submitted later as a part of your assessed Learning Log. You are advised to back this document up and store the back-up in a secure place.
Additional learning resources
Additional materials are available if you wish to learn more about the issues covered in this unit. (Blackboard > Module Site > Module Materials > Unit 3 > Additional Learning Resources)
References
Ryan, B., Scapens, R.W. and Theobald, M. (2002) Research Methods and Methodology in Finance and Accounting 2nd edition. London: Thompson Learning.
Bradford MBA 55
Unit 4:
Ethics in Research and
Data Collection
Key reading:
1. Collis and Hussey (2014), Chapters 2, 7, 10
Key audio/video: 1. Unit 4 audio and slides (Blackboard > Module Site > Module Materials >
Unit 4 > Unit 4: Audio and Slides) – Activity 4.5
2. How to relate to people in business (2008) Interview Skills, Part 1, YouTube, 10 February (Blackboard > Module Site > Module Materials > Unit 4) – Activity 4.4
3. How to relate to people in business (2008) Interview Skills, Part 2, YouTube, 10 February (Blackboard > Module Site > Module Materials > Unit 4) – Activity 4.4
Other: 1. Unit 4 multiple-choice questions (Blackboard > Module Site > Formative
Exercises > Multiple-Choice Questions > Unit 4) – Activity 4.6
2. Unit 4 live online tutorial (Blackboard > Module Site > Collaborate) – Activity 4.7
Introduction
In this unit, you will look at ethical issues in research. This will include ethical issues arising from interacting with, and behaviour towards,
participants who may be taking part in a research study. Standards that should be expected in the way that you conduct a research project will also be explored. You will review survey questionnaires and consider how best to formulate these. Finally, you will study participant observation and interviewing techniques as ways of collecting qualitative data.
Objectives
By the end of this unit, you should be able to:
Study Book: Business Research
56 Bradford MBA
appreciate acceptable ethical requirements when undertaking research
conduct ethical research when undertaking a research project
design and construct appropriate questions for survey questionnaires
understand the two types of observation available for data collection
conduct purposeful interviewing.
What ethical issues are there in research?
Whilst we are all aware of the requirement for research in the natural sciences to be conducted with the highest ethical standards, Coolican (2014, p.294) observes that it is ‘...difficult to conduct much research at all without running into ethical arguments’. He argues that it is important that we remember the social responsibility associated with all research and remain vigilant to ‘..malpractice, abuse, thoughtlessness, charlatanism and lack of professionalism’. We have already seen that ethics is important when searching the literature and reporting (see Ethical Implications: Three Points, in Unit 3). To see why this obvious imperative extends to the social sciences and therefore business research, we need to reflect on the claim by Ghauri and Gronhaug (2010) that: ‘Ethics are moral principles and values that influence the way a researcher or a group of researchers conduct their research activities.’ A consequence of this is alluded to by the credo of the Coalition of Americans for Research: ‘Do no harm’. The question is, harm to whom, i.e. who are the stakeholders in research? In the case of an MBA project, for example, the stakeholders are likely to include: yourself as the researcher, the company you are conducting your research for, its companies, the University of Bradford, Bradford alumni, and those who participate in your research (e.g. interviewees and those completing questionnaires).... The list can be exhaustive!
Given the number of people and organizations who could be affected, we need to explore the nature of the ethical issues that should be considered when doing a research project. The following is a table (Table 4.1) of some, not all, of the main issues that may be relevant when conducting a research project:
Unit 4: Ethics in Research and Data Collection
Bradford MBA 57
Table 4.1: Ethical issues associated with business research participants
1 Preserving participants’ anonymity 2 Exposing participants to mental stress 3 Questioning detrimental to the participants’ self-interests 4 Subjecting respondents to a health hazard 5 Involving participants without their knowledge or without their
permission 6 Deceiving participants 7 Use of coercion 8 Depriving participants of their rights
In addition to considerations related to interactions with the participants, there are also ethical considerations associated with the logistics of research. Some of the major ones are listed in Table 4.2 below:
Table 4.2: General ethical issues in business research
1 Access to organisations’ employees, management 2 Access to organisations’ data 3 Access to subject specialists/consultants 4 Access to any relevant research literature 5 Storage of potentially sensitive data 6 Consideration of legal/regulatory requirements 7 Publication of findings
To address the more general concerns regarding integrity and propriety when conducting business activities, several professional bodies, governments and institutes have introduced codes of conduct. These are used as guidelines and a framework for conducting research in an acceptable way. Although broader and more encompassing, all of the issues identified and listed in Tables 4.1 and 4.2 are subsumed within these broad guidelines. A representative sample of the bodies that have introduced such codes of conduct and abiding principles is given in Table 4.3 for completeness. Dates are not always supplied as these bodies tend to review and update their guidelines periodically. This table also includes
important corporate governance ethical guidelines.
Study Book: Business Research
58 Bradford MBA
Table 4.3: Ethical guidelines and codes of practice
1 Social Research Association’s Ethical Guidelines (2002) 2 British Psychological Society’s Code of Conduct 3 American Psychological Society’s Ethical Principles for
Psychologists and Code of Conduct 4 British Sociological Association’s Statement of Ethical Practice
Corporate Governance Ethical guidelines: 5 Sarbanes-Oxley Act (2002) 6 Cadbury: Code of Best Conduct (1996) 7 Financial Reporting Council: The Combined Code on Corporate
Governance
For those of you who may be considering your MBA management project, it is worth noting that business research at the School of Management is designed to comply with the principles contained in Table 4.1, with the overarching aim of using proportionality in contentious situations. Following the UK’s Engineering and Science Research Council (ESRC) framework, ethical judgements are to be made that are proportionate to the risk of offending any stakeholder or compromising these principles.
Read: Collis and Hussey (2014), sections 2.5, 2.5.1, 2.5.2 and 2.5.3: 30- 35
Activity 4.1 – stop and think
Question 1: You are currently interviewing a participant who gave informed consent to take part in a research study. However, it is evident that the person you are interviewing wants to withdraw from the interview. You need them to finish as the information you have from them already, and are about to obtain, is required for the research project. What action should you take?
Question 2: You propose to observe the work practices of several employees of a particular company. Your concern is that if the employees know they are being observed, their behaviours will change. The company suggests that you observe them without their knowledge. It argues that as the research is not detrimental to their careers, this would not be unethical behaviour. It also agrees to provide a complete debriefing for the selected employees after the observations have been completed. Based on ethical reasoning, what is your response likely to be?
Unit 4: Ethics in Research and Data Collection
Bradford MBA 59
Activity 4.1 – stop and think answers:
We have seen that there are ethical considerations associated with the logistics of research, and several of these were listed in Table 4.2. You will note that there are also issues about access to data. We will now leave ethics and look at ways in which data is often collected in business research.
Data collection
There are two fundamental types of data, namely: primary data and secondary data. The former is data collected for the (usually sole) purpose of your research, whereas the latter is existing data collected for some other purpose – but that may be relevant for your study.
When conducting positivist, quantitative research, in most cases you will need to collect data from the population that you are interested in. You will usually take a representative sample of this population, as the population may be too large to deal with; it would be too costly to collect data from the entire population, or it would take too long to obtain this. However, whatever the reason for taking a sample from the population, the main consideration is to ensure that the sample is taken appropriately and is sufficiently large, so that it accurately represents the population. If it is, you can often make reasonably accurate predictions about the population from
Study Book: Business Research
60 Bradford MBA
the sample. We will see how in Units 7 and 8. In this unit, we are interested in how to collect the data.
When conducting interpretive, qualitative research, it is usual to collect data by interviews with relevant individuals. There are other important data collection approaches though, such as observation, focus groups and the Delphi method, and we will be discussing some of these later in Unit 5.
Surveys
A survey is a particular methodology for collecting data (and is one of the most used for MBA projects). There are two types of survey: descriptive surveys and analytical surveys.
Descriptive surveys are designed to provide an accurate account or representation of a phenomenon of interest and are typically qualitative. An example would be a bank surveying customers to see how well received their new online customer relationship management system is.
Analytical surveys, on the other hand, support quantitative, positivist research and are used extensively to investigate or validate relationships and dependencies between variables of interest. An example would be a survey to establish whether a relationship exists between what a potential user believes is the usefulness of a proposed new technology (perceived usefulness) and their subsequent usage (actual use) once the technology has been implemented.
Questionnaires
A questionnaire is the usual data collection instrument for a survey. Questionnaires are used to collect demographic data such as personal information like age, qualifications, level of seniority, etc. from individuals; or about the organisation, e.g. size, number of employees, business sector, etc. They are also used to measure perceptions of respondents, often using the Likert scale. The semantic differential scale is also popular here. In both cases, the data obtained is usually treated as numerical data as the categories in both scales are considered to have equal increments across the complete range. Typically, Likert measurements are obtained from responses to a stated proposition or statement and include a set of response categories similar to the following, to measure the belief of the responder:
Likert type statement: ‘The cost of the technology limits its take-up.’
1 = Strongly Disagree, 2 = Disagree, 3 = Neutral, 4 = Agree, 5 =Strongly Agree
Unit 4: Ethics in Research and Data Collection
Bradford MBA 61
Semantic differential questions: ‘I feel:’
Sad (1) 5 Happy (10)
Both scales have extreme anchors (Strongly Agree and Strongly Disagree for Likert-type scales, and Sad through to Happy (say) for semantic differential scales). Although we exhibit the original and most used 5-point Likert scale, recent research demonstrates that it is better to use a larger scale and you are advised to look at a 7-point scale as an alternative. It is also advisable to keep the spread of your semantic differential scale to a range of 10, at most.
Read: Collis and Hussey (2014), subsection 10.5.4: 215–217
Activity 4.2 – stop and think
In the set textbook, you are encouraged to design rating scales (e.g. Likert, semantic differential, ...) with the lowest accord with the statement (Likert) or most negative view (semantic differential) being linked to the lowest numerical category. However, this is not advocated when using ranking scales. Why is there this discrepancy?
Activity 4.2 – stop and think answer:
Issues with questions for questionnaires
If using a questionnaire instrument to collect your data, it is important that you avoid some of the pitfalls that result in ambiguous, poorly comprehended and poorly answered questions.
Study Book: Business Research
62 Bradford MBA
Double negatives
Double negatives are often too convoluted for respondents to understand. Indeed, for some people, a ‘Yes’ is equivalent to others responding ‘No’. A typical example of an unnecessary double negative question is:
‘I do not believe that my company has not introduced ethical guidelines for conduct in the workplace.’
Such questions also tend to be too wordy and, as such, encourage questionnaires to be handed in incomplete, which could make then unusable.
Reverse coding/logic
Reverse coding is not the issue; it is a possible solution. The problem is that when confronted with a sequence of ‘positive’ questions, the respondent can get so used to just ticking in the ‘agree’ or ‘strongly agree’ boxes that they tend to switch off and do not critically appraise the questions as closely as one would like. A possible solution is to split the sequence up with questions that have their logic reversed. The respondent will then need to consider the question more deeply and respond with ‘disagree’ or ‘strongly disagree’. If you wished to check for consistency in the answers, you could always ‘reverse’ a question later in the questionnaire and check that the answers to the two questions were comparable.
Example: A sequence of ‘positive’ questions followed by: ‘My company is not amongst the most effective in our business sector’.
Double barrelled
Double-barrelled questions are always problematic and should always be avoided. They have two separate clauses that are co-joined by the coordinating conjunction, ‘or’. This means that either or both clauses must be correct for a confirmatory response. However, you will never know which! An example follows:
‘My company is efficient or has a high return on investment (ROI)’.
In such cases, you should consider splitting each phrase into two separate questions or, at the very least, use the ‘AND’ conjunctive. However, using ‘AND’ does not enable you to establish if just one of the options, only, is true.
These are some of the more obdurate design issues. Collis and Hussey (2014) also deal with more standard design issues such as whether questions should be open, allowing the respondent to answer with a
Unit 4: Ethics in Research and Data Collection
Bradford MBA 63
detailed written response, or closed, permitting the selection of one, or several, predetermined responses, and what the implication of these are, when multiple-choice questions will be suitable, etc. Once these design issues have been dealt with, you are then faced with the problem of where to get your responses from and the troubling issue of whether the responses received are representative of the population. It is usual to deal with this by looking at late responses. Again, Collis and Hussey explain what some of the options are here.
Read: Collis and Hussey (2014), sections 10.5, 10.5.1, 10.5.2 and 10.5.3: 210–214
Activity 4.3 – stop and think
Question 1: Explain why open questions are usually associated with qualitative research.
Question 2: It is expedient to limit the use of a ‘do not know’ category in closed questions to those where this may be applicable. Why?
Activity 4.3 – stop and think answers:
Data collection for qualitative research
We now turn our attention to methods used for the collection of qualitative data. We will initially look at participant observation and then look at interviewing. We will continue this topic into Unit 5.
Study Book: Business Research
64 Bradford MBA
Participant and non-participant observation
There are two types of observation that are used for data collection: non- participant and participant observation. In non-participant observation, the observer is disassociated from the group of individuals or single individual being observed. These could be e.g. a department in an organisation, a subgroup, a retail store, a service centre, a regional branch of an organisation, a chief executive officer (CEO), etc. The observation could be open, with the observer noting (possibly using a diary method) and recording activities and dialogue between group members, external colleagues or customers; or covert, using concealed technology such as video recorders or tape recorders. In both cases, the objective of observation is to gain understanding of the underlying motives, values and the particular practices of the observed subjects.
In participant observation, the researcher is fully involved in the process under consideration or the activities of interest. The researcher should have ethical approval to participate and the subjects should be made aware of the objectives of the study and be willing participants. Sometimes, the full nature of the research is not revealed until after the period of participation and observation has concluded and, in certain cases, participants will not even be conscious that such a study is taking place. Here, for the subjects to know what is being researched could alter their behaviour such as to compromise any findings. This phenomenon is termed demand characteristics. The concept of proportionality in ethics comes in here but all participants should be made aware of the research that has been conducted and given suitable reassurances after the event. It is seldom considered ethical to engage in covert observation within an organizational setting, irrespective of whether you have gained permission from the organization(s) concerned, even if you as the researcher feel it is essential for your research.
Read: Collis and Hussey (2014), section 7.8: 148–150
Interviewing
Interviewing is arguably the most common method of obtaining data from a respondent or group of respondents. The interviewees are asked questions so that the researcher can find out what they think about certain topics of interest, their attitudes, behaviour, previous responses, and their judgments about selected issues. These can be obtained in face-to-face contact, telephone and Skype conversations, or through video conferencing. Whilst these data collection methods can be used in positivist research, they are more likely to be for an interpretive paradigm, or within pragmatist or realist research. Here, the purpose is exploration and discovery. Questions are likely to be open-ended, allowing the interviewees to express themselves in their own words. Their responses are likely to lead to follow-up questions that enable the interviewer to
Unit 4: Ethics in Research and Data Collection
Bradford MBA 65
explore initial responses more deeply. Here you will be using probing questions to stimulate discussion on relevant aspects for elaboration of previous observations that have been made. You will also seek clarification and confirmation of your understanding of previous statements made by those you are interviewing.
Interviews can be relatively unstructured and this is most appropriate when a topic is poorly understood and the objective is to gain a fundamental understanding of basic issues. The other extreme is a structured interview. Here precise questions have been prepared in advance and each interviewee will be asked the same questions, usually in the same order. Comparisons are more readily made using structured interviews than unstructured ones in which it is sometimes very difficult to see any common ground between interviewees. This structured approach is the most typically used in positivist research. Semi-structured interviews use a mix of structured, predetermined questions, with an additional use of unprepared questions, raised in response to answers given in the interview.
In order to increase the chances of having a successful interview, and thus obtaining rich data that will substantially inform your research, there are certain steps that can be taken. These can be behavioural by the interviewer or relate to the nature of the types of question asked. Collis and Hussey (2014) discuss these and provide a table of the different types of interview questions that can be asked and where they are most likely to be useful. Interviewing is a personal activity and it is possible that the style and way that the interviewer performs may have an undue influence on the interviewee. This can lead to a lack of reliability or bias in the responses received. Collis and Hussey (2014: 208) provide a checklist of ways to conduct the interviews that could reduce this possibility. However some researchers would argue that adopting such a systematic and uniform approach fails to understand the dynamics of an interview and imposes other biases on the interviewees, by limiting the scope for the interviewee to respond as they wish.
Read: Collis and Hussey (2014), sections 7.3–7.3.4 and 10.4.2: 133–139, and 207–209
Activity 4.4 – watch and reflect
Watch: How to relate to people in business (2008) Interview Skills, Part 1 and Interview Skills, Part 2, YouTube, 10 February (Blackboard > Module Site > Module Materials > Unit 4).
These films offer insight into some of the issues discussed in this unit with practical suggestions to increase the quality of your interviews.
Study Book: Business Research
66 Bradford MBA
Interview preparation
From the video clips you can see that the essence of a successful interview is good preparation and a clear understanding of what you want to achieve. Prior to constructing your interview schedule that you will use to guide the interview or questionnaire, you must:
identify a population of potential participants to sample from
know how you are going to sample from this population (the sampling method and the mode of access)
have identified the variables that you need data for so that you can address your research question or research questions (questionnaires)
Again, the whole process of establishing the instrument (interview schedule or questionnaire) and getting data using this instrument needs to be implemented in a systematic and structured way. Collis and Hussey (2014) detail a typical schedule for this activity.
Read: Collis and Hussey (2014), sections 10.4, 10.4.1: 204–207
Summary
This unit has dealt with ethical issues that you will need to consider when you undertake a research project. You have also looked at data collection, and particularly, some design issues associated with questions to be present in a questionnaire. You also looked at measures obtained from Likert-type scales and the semantic differential scale. Finally, you studied several approaches to obtaining qualitative data, such as observation, interviewing and questionnaire, and worked through a checklist for an appropriate schedule for implementing the latter two.
Activity 4.5 – audio and slides
Listen to the audio for this unit and view the accompanying slides at the same time. (Blackboard > Module Site > Module Materials > Unit 4 > Unit 4: Audio and Slides)
Activity 4.6 – multiple-choice questions
The multiple-choice questions for this unit will enable you to test your knowledge and understanding of the key concepts covered. (Blackboard > Module Site > Formative Exercises > Multiple-Choice Questions > Unit 4)
Unit 4: Ethics in Research and Data Collection
Bradford MBA 67
Activity 4.7 – live online tutorial
Read: Case study I and Case study II (Blackboard > Module Site > Module Materials > Unit 4)
Consider that you are the head of the research ethics panel at the University; read the two scenarios present in Case study I and Case study II and note down your answers to the following questions:
Question 1: Would you approve the accountancy project in Case study I? You should justify your answer.
Question 2: Would you approve the mystery shopper project in Case study II? You should justify your answer.
These questions will form the basis of the online live tutorial, so be ready to discuss your answers to these questions. You will be informed as to when this tutorial will take place.
Live online tutorial: To access the tutorial, please go to Blackboard > Module Site > Collaborate. To test your sound settings for the Blackboard Collaborate virtual classroom, please go to the Blackboard Collaborate Support site at http://bit.ly/1etRu2Y. For further support materials and information about Blackboard Collaborate, please see the Blackboard Collaborate Support site at http://bit.ly/1jBpgUc.
References
Berman, J. (2008) How to relate to people in business. Interview Skills, (Parts 1 and 2), YouTube, 10 February. https://www.youtube.com/watch?v=EB2sA0vXk7E and https://www.youtube.com/watch?v=gxBi5FJl55o. Accessed 27 May 2016.
Coolican, H. (2014) Research Methods and Statistics in Psychology 6th edition. London: Psychology Press.
Ghauri, P. and Gronhaug, K. (2010) Research Methods in Business
Studies 4th edition. Harlow: Pearson Education Limited.
Bradford MBA 69
Unit 5:
Data Collection and
Analysis
Key reading:
1. Collis and Hussey (2014), Chapters 7, 8
2. Skulmoski, G.J., Hartman, F.T. and Krahn, J. (2007) The Delphi method for graduate research, Journal of Information Technology Education 6: 1–21 (Blackboard > Module Site > Module Materials > Unit 5) – Activity 5.3
3. Williams, R. (2005) Soft Systems Methodology, The Kellogg Foundation, December (Blackboard > Module Site > Module Materials > Unit 5)
Key audio/video: 1. University of Limerick (2010) Focus Group Tutorial, YouTube,
26 September (Blackboard > Module Site > Module Materials > Unit 5) – Activity 5.1
2. Unit 5 audio and slides (Blackboard > Module Site > Module Materials > Unit 5 > Unit 5: Audio and Slides) – Activity 5.6
Other: 1. Unit 5 multiple-choice questions (Blackboard > Module Site > Formative
Exercises > Multiple-Choice Questions > Unit 5) – Activity 5.8
Introduction
In this unit, we will continue to look at data collection. You will look in detail at focus groups and the Delphi method for the collection of, mostly, qualitative data. We will also explore an alternative approach using soft systems methodology that both collects data and proceeds to problem solving. This methodology has taken advantage of computer technology and has become absorbed within Agent Based Modelling. We will then explore qualitative analysis, with emphasis on content analysis. In order to obtain data from a population, it is necessary to sample from this population. You will investigate some of the major sampling approaches that are regularly used in research projects. Finally, you will explore some
Study Book: Business Research
70 Bradford MBA
of the options available for presenting summary findings for qualitative data.
Objectives
By the end of this unit, you should be able to:
appreciate approaches to collecting data in support of qualitative research
plan a focus group
facilitate a Delphi method for data collection
use soft systems methodologies to analyse business problems
analyse qualitative data using the content analysis approach.
Data collection
You have looked at how to obtain qualitative data from interviews or questionnaires. Whilst these are two of the main ways in which to get qualitative data, there are several alternative methods for data collection that are frequently used in business research. In this unit you will concentrate on focus groups and the Delphi method, as they are important methods that you may wish to consider when conducting a research project (including your MBA management project). Finally you will explore content analysis, as this is an important method for analysing qualitative data.
Focus groups
The objective of a focus group is to obtain detailed information about a topic or phenomenon of interest from knowledgeable, affected participants. There are several ways in which the focus group members can ‘meet’ but the basic ideas and mechanisms are best demonstrated by discussing the standard approach. Here, a small, informed or affected number of participants meet to have a group discussion under the control of a facilitator or group leader. The facilitator is often, but not necessarily, the researcher. The ideal group size should be between 5 and 12 individuals (Collis and Hussey (2014) recommend 5 to 10), and the meeting should not last longer than about one and a half hours.
The role of the facilitator is critical in creating a relaxed and ‘democratic’ environment in which feelings and attitudes about the research topic of interest to the researcher are explored by the participants. It is important that the views of the most junior members are treated with the same respect and given the same level of importance as those of the most senior. The facilitator will continue to ensure that the research topic is
Unit 5: Data Collection and Analysis
Bradford MBA 71
addressed but will also follow up with prompts to stimulate discussions on emerging themes and beliefs. The consequence is that a rich and complete picture of the phenomenon or problem emerges. To fully achieve this, the interactions and cues between individuals should be noted, as well as the verbal exchanges. Interactions are particularly important as they often lead to understanding amongst the participants that was not previously recognised or articulated. This is one of the main strengths of focus groups.
Whilst focus groups are predominantly used for qualitative, inductive research, they can and are used in positivist research: for example, to explore the limitations of a planned survey or to aid interpretation of the results obtained from a survey. Collis and Hussey (2014) also itemise some of the main purposes that focus groups are used for and discuss several practical ways in which to conduct successful sessions.
Activity 5.1 – watch and reflect
Watch: University of Limerick (2010) Focus Group Tutorial, YouTube, 26 September (Blackboard > Module Site > Module Materials > Unit 5)
This film provides some tips and ideas for running a focus group.
Read: Collis and Hussey (2014), sections 7.5, 7.5.1 and 7.5.2: 141–144
Activity 5.2 – stop and think
Question 1: State three positive reasons for using a focus group.
Question 2: State three reasons for not using focus groups.
Question 3: What are the main advantages of using a focus group in preference to a sequence of interviews?
Activity 5.2 – stop and think answers:
Study Book: Business Research
72 Bradford MBA
Delphi method
The Delphi method is a collection of allied methods that are regularly used for problem-solving and decision-making to resolve, or explore, fundamental issues in research. Similar to focus groups, there are several different ways in which the method can be implemented and, as there is a potential decision component, some approaches to the Delphi method use statistically informed decision-making. However, no matter what the precise sequence of steps actually taken, as Okoli and Pawlowski (2004: 26) suggest, ‘The Delphi method is a versatile research tool that researchers can employ at various points in their research. Use of the Delphi method for forecasting and issue identification/prioritization can be valuable in the early stages, particularly in selecting the topic and defining the research question(s)’.
When used as in the latter capacity, it is a particularly useful research tool for exploring and developing the research question(s), and in the formulation of theories in inductive research. The method is designed to collect highly relevant opinions from a number of ‘experts’. We will now discuss this research-oriented approach.
The basic research question or problem is given to the selected experts. For inductive research it is preferred to have a small number of experts (in deductive projects, approaching 200 experts have been used!). These are
typically independent individuals who are remote and unaware of the identities of the other participants. The researcher then receives and collates the initial opinions from the experts. Summaries of these are then sent out to the experts, who are invited to reflect on them and respond on the basis of these summaries. Revised responses are then received from the experts and the process continues until an overall consensus is reached or after a predetermined number of iterations have taken place.
At the end of the process, a rich picture emerges and you should have identified most of the major issues or you may have a suitable research
Unit 5: Data Collection and Analysis
Bradford MBA 73
question(s), depending on your objective. However, similar to focus groups, sensitive, quality facilitation is paramount if the experts are to be non-defensive, reflective and engage in the exercise. Also it is important to remember that the Delphi method reveals what experts think about a particular phenomenon, and whilst the collective view of experts may be important, in some cases collecting data from other sources may be necessary.
Activity 5.3 – read and reflect
Read: Skulmoski, G.J., Hartman, F.T. and Krahn, J. (2007) The Delphi method for graduate research, Journal of Information Technology
Education 6: 1–21 (Blackboard > Module Site > Module Materials > Unit 5)
Read the Executive Summary to revisit the concept of the Delphi method.
Systems and agent based modelling
Soft systems modelling
We have just looked at ways in which rich, qualitative data can be collected. This, in conjunction with the literature, will help in the formulation of theories associated with the research phenomenon of interest. Often these phenomena are from so-called hard systems: systems that are well defined and easily conceptualised. These systems typically comprise distinct and recognisable entities such as: health systems, computing systems (software and hardware), etc. These are to be contrasted with soft systems, on the other hand, which are highly complex, interrelated systems that are mostly difficult to conceptualise (see Figure 5.1). As such, these soft systems typically have no obvious identifiable components or subsystems and, consequently, no obvious single unique view of the system can be formed by all investigators. As an example, a research question for a soft system could be:
“Can an organisation’s current core competencies and expertise be exploited to compete successfully in alternative markets?”
The rationale for soft systems is based on the premise that human beings form an integral part of systems in the real world (for example organisational, operational, computer systems), and that a systems development methodology MUST embrace all the stakeholders who are involved in or affected by the development process. These will include: users, IS/IT professionals, managers, shareholders, and customers. Herein lies the complexity. These various stakeholders will have conflicting goals, objectives and possibly ulterior motives, and may even view the
Study Book: Business Research
74 Bradford MBA
system differently (perceptions and attitudes). To demonstrate this point, the objectives of the Common Agricultural Policy of the EEC were to:
1. Increase productivity (in the agricultural industry)
2. Safeguard jobs for workers in the agricultural industry
3. Provide the best possible service to customers.
This can be formulated as a classic multi-criteria optimisation problem with conflicting objectives (Lawrence et al. 2000), using a hard systems approach. However, to achieve 1 (increase productivity), say, would compromise 2 (safeguarding jobs) and would not necessarily lead to achieving 3 (providing the best possible service to customers). Only a compromise solution is possible and for this to be agreeable, all stakeholders should be involved in the formulation of the ‘solution’, with any possible, unforeseen complications being identified and accommodated.
Given this, and the need to deal with problems that will be caused by the unpredictability of human activities, soft system methods are seen by some researchers as more likely to deal with this complexity in a way that ‘resolves’ the conflicting objectives for more stakeholders, than solutions obtained from formal hard systems. In essence, it could be argued that a soft systems methodology (SSM) emphasises human involvement in such systems with their behaviour being modelled as a part of a systems analysis that is understandable to non-technical experts. This makes the process and, particularly, any interrelationships more understandable to the researcher and stakeholders alike (Figure 5.1).
The basic objective is to learn and refine an evolving understanding of a complex business problem using a process of a constrained, yet broad, exploration of several potentially conflicting views of the problem. As you will see later, the desired outcome is to facilitate the users in developing a deeper understanding of the problem faced by an iterative process of learning and analysis (see Figure 5.2). SSM is strictly a framework for investigating learning, unlike hard systems modelling which is seen, in comparison, as a set of incontrovertible rules. A seven-stage procedure for
conducting the SSM was introduced in 1999 (Checkland, 1999) and this has formed the basis for most contemporary SSMs.
Unit 5: Data Collection and Analysis
Bradford MBA 75
Figure 5.1: Soft systems view versus hard systems view of the world
(Source: Checkland 1999)
Study Book: Business Research
76 Bradford MBA
Figure 5.2: Schema for soft systems approach
(Source: Checkland 1999)
Read: Williams, R. (2005) Soft Systems Methodology, The Kellogg Foundation, December (Blackboard > Module Site > Module Materials > Unit 5)
It is important that you gain a good understanding of SSM if you wish to analyse such complex systems as a part of your MBA project.
Unit 5: Data Collection and Analysis
Bradford MBA 77
Activity 5.4 – stop and think
Question 1: Why is it essential to develop root definitions reflecting the viewpoints of the problem for all the stakeholders?
Question 2: One way that you can make comparisons between the conceptual models and that in the real world is to model the real world with the conceptual model and note any discrepancies. Suggest a way in which this can be achieved.
Activity 5.4 – stop and think answers:
Agent based modelling
Agent based modelling (ABM) is becoming widely used by organizations and institutions to solve complex problems. Strictly speaking it doesn’t solve problems in a traditional sense, rather what it does is model what could happen (or is likely to happen) given a particular set of conditions and one or more agents. Agents are simply decision-making units (which can be a single person or an organization) that are deemed to follow a set of rules. Although the computer models that are used to carry out ABM are constantly being developed and becoming increasingly complex or sophisticated, it is a form of problem solving that is derived from soft systems thinking. It acknowledges that agents (people) cannot be excluded from social systems (such as an organization), so if we want to understand things within an organization we need to include people in our thinking and consider how their decisions and actions will shape things. From this perspective ABM allows us to understand complex dynamic and evolving systems in ways traditional approaches don’t.
As an example, imagine we want to know what impact lowering the selling price of our products will have on our business. Classical economics would suggest that this isn’t that complicated, but as you will appreciate,
Study Book: Business Research
78 Bradford MBA
the reality is quite complex and the impact on our business is far from clear. To try and understand what will happen, we need to think about what our customers will do, what other potential customers will do, what our competitors will do, as well as thinking about other changes which are going on in the environment. ABM tries to capture this. It doesn’t give us a definite answer (say, volumes will rise by 10% and revenues by 6%); rather researchers running the model are able to consider various scenarios and quantify the likelihood and impact of them.
In the current business environment we tend to think things are changing ever more rapidly and in ways that we cannot imagine in advance. In such a world, businesses are increasingly using soft systems thinking and ABM to make decisions that allow them to solve a basic problem all businesses face: “What should we do next?”. Such approaches do not give definitive answers; the computers do not deliver a definitive answer. Some researchers argue that this isn’t research, but ultimately all research aims to create greater understanding that can shape our actions. Thinking this way allows us to see soft systems thinking as an approach to doing research.
Qualitative Data Analysis
Read: Collis and Hussey (2014), Chapter 8, 8.2 to 8.6: 154–172
Once you have your qualitative data from sources such as transcriptions of interviews or focus groups, broadcasts, audio recordings, newspaper articles, internal company reports, etc, you will want to extract the information content from these sources to inform your research.
In many ways, analysing qualitative data is much more challenging than analysing numerical data. Whilst we can employ statistical software and computer programs, making sense of qualitative data is much more labour intensive. With qualitative data, the researcher has to interpret the data, whether an interview, a written statement, or something we have observed, and then establish what the data means. Compared to the interpretation of numbers, this can appear rather subjective and problematic. Thus in order to try and make the interpretation more
objective, or at least credible, a range of structured techniques have been developed that bring transparency and rigour to the process.
Content analysis is a common way to achieve this, whereby text is transformed into quantitative data that can be readily analysed and presented using simple summary statistics and basic statistical tests (see Units 7 and 8). It is important to ensure that only relevant sections in large documents are analysed so that your task is manageable and that your findings are not specious, with text being included out of context.
Unit 5: Data Collection and Analysis
Bradford MBA 79
Initially, you will need to establish the coding units that you will use to ‘categorise’ the text and then an accompanying coding frame. The results of your searches for the coding units are then entered into the coding frame, along with selected demographic information. This document then lends itself to a simple frequency analysis and this is a usual analysis resulting from content analysis, although relationships and meaning/emphasis are also often developed and explored using this approach. Examples of a coding units document and coding frame are given in Figure 5.3 and Figure 5.4.
Figure 5.3: Coding units
Coding unit Example
Words/ phrases
1. dispute in minutes of management/union meeting; 2. increased dividend in circulars released to
shareholders
Theme 1. Search for examples of: agreement in Senior Management Committee Meeting minutes
2. Search for examples of: association of sustainability with profit
Item 1. Newspaper reports on oil prices 2. Company reports on CSR
Time 1. Proportion of time allocated to staff development by CEO in annual general meetings
2. Time allocated to business matters in terrestrial and cable TV newscasts
Figure 5.4: Coding frame
Date of
article
Page
number
Length
(words)
Author CICS and direct
funding from
Government
Name of firm
2/11/11 206 2103 FBloggs Receive funding
from Yorks.
Forward + EU
Regen. initiative
+ Sports Council
Training Skills
(Rochdale)
… … … … … …
The method is particularly attractive for MBA projects if secondary data is readily available for the chosen topic, as it removes the time-consuming
Study Book: Business Research
80 Bradford MBA
overheads associated with collecting primary data and possible validation issues with the instrument that you have developed for this. However, as Silverman (2015, p.118) points out, there is an established underpinning theory that can be used to validate findings using this method and these findings could also be ‘trite’ (i.e. trivial). There are issues about the degree of coverage of the topic, and particularly, what should be included and what should be excluded. These can lead to findings being challenged as partial, confounded or incomplete. For example, should various ethnic groups, social classes, or those with particular religious or political persuasions be excluded from the analysis? If so, how representative are the findings? If present, are data from these groups likely to differ from the remaining groups, thus providing distorted frequency counts?
Provided you are aware of these potential difficulties and consider presenting and contrasting appropriate findings for inclusive and exclusive groupings, this can be a very powerful analysis tool, illuminating some unexpected and interesting relationships.
An alternative approach is Discourse Analysis. This approach is more in keeping with hermeneutic approaches. Rather than focussing on the words employed, discourse analysis is much more concerned about what a piece of text means and how that meaning is created. It often explores the relationship between the author, the text and the reader and does not assume that we can interpret a text without understanding its producer and its consumer. Thus it challenges the idea that data can ever be interpreted objectively, and so it can appear to question our established ideas about what research is. However it is important to remember that questioning objectivity does not mean that we cannot produce credible and insightful research that helps us to understand organizations and business.
Activity 5.5 – stop and think
Question 1: Provide an example of a research topic based in your own company, or in that of a previous employer, that could be investigated using content analysis.
Question 2: What documents would you use for this investigation?
Question 3: Would this be a primary source or a secondary source of data?
Question 4: Provide an example of a research topic based in your own company, or in that of a previous employer, that could be investigated using discourse analysis and explain why you think it would be a suitable approach.
Unit 5: Data Collection and Analysis
Bradford MBA 81
Activity 5.5 – stop and think answers:
Summary
In this unit, you have looked at two of the major qualitative data collection approaches. You have also explored soft systems methodology that
both collects data and provides a systematic and inclusive approach to problem-solving. Finally, you have investigated the content analysis approach for analysing, again, predominantly qualitative data.
Activity 5.6 – audio and slides
Listen to: the audio for this unit and view the accompanying slides at the same time. (Blackboard > Module Site > Module Materials > Unit 5 > Unit 5: Audio and Slides)
Study Book: Business Research
82 Bradford MBA
Activity 5.7 – review activity
Question 1: What ways have you or your current or previous organisations used to collect data?
Question 2: What would be the greatest difficulty if you were to conduct a Delphi method for your research?
Question 3: List a few of the disadvantages of using content analysis, exclusively, to analyse the transcript from your focus group discussion.
Activity 5.7 – review activity answers:
Activity 5.8 – multiple-choice questions
The multiple-choice questions for this unit will enable you to test your knowledge and understanding of the key concepts covered. (Blackboard > Module Site > Formative Exercises > Multiple-Choice Questions > Unit 5)
References
Checkland, P.B. (1999) Systems Thinking. New York, NY: Oxford University Press.
Unit 5: Data Collection and Analysis
Bradford MBA 83
Lawrence, K.D., Reeves, G.R. and Klimberg, R. (2000) Applications of Management Science, Volume 10: Multi-criteria Applications. Greenwich, CT: JAI Press.
Okoli, C. and Pawlowski, S.D. (2004) The Delphi method as a research tool: an example, design considerations and applications, Information and Management, 42: 15–29.
Silverman, D. (2015) Interpreting Qualitative Data: Methods for Analysing Talk, Text and Interaction 5th edition. London: Sage.
Skulmoski, G.J., Hartman, F.T. and Krahn, J. (2007) The Delphi method for graduate research, Journal of Information Technology Education, 6: 1– 21. https://blackboard.brad.ac.uk/bbcswebdav/xid-2696445_2 Accessed 13 July 2016.
University of Limerick (2010) Focus Group Tutorial, YouTube, 26 September. www.youtube.com/watch?v=TGsDGFeff5I Accessed 26 May 2016.
Williams, R. (2005) Soft Systems Methodology. Battle Creek, MI: The Kellogg Foundation.
Bradford MBA 85
Unit 6:
Sampling and Summary
Statistics
Key reading:
1. Collis and Hussey (2014), Chapters 10 and 11
2. Easton, V.J. and McColl, J.H. (editors) (1997) Statistics Glossary, STEPS (Blackboard > Module Site > Module Materials > Unit 6) – Activity 6.1
3. Explorable (2015) Snowball Sampling, explorable.com (Blackboard > Module Site > Module Materials > Unit 6) – Activity 6.1
4. Explorable (2015) Cluster Sampling, explorable.com (Blackboard > Module Site > Module Materials > Unit 6) – Activity 6.1
5. Trochim, W.M.K. (2008) Non-Probability Sampling, Web Center for Social Research Methods (Blackboard > Module Site > Module Materials > Unit 6) – Activity 6.1
Other: 1. Unit 6 audio and slides (Blackboard > Module Site > Module Materials >
Unit 6 > Unit 6: Audio and Slides) – Activity 6.5
2. Unit 6 multiple-choice questions (Blackboard > Module Site > Formative Exercises > Multiple-Choice Questions > Unit 6) – Activity 6.6
3. Unit 6 live online tutorial (Blackboard > Module Site > Collaborate) – Activity 6.7
Introduction
Read: Collis and Hussey (2014), Chapters 10 and 11
In this unit, we introduce the main methods for sampling for both qualitative and quantitative data. The rationale for the varying sampling methods is explained and typical examples given. Having then obtained your data it is usual to provide descriptive summary statistics. We introduce some of the main summary statistics used and discuss these,
Study Book: Business Research
86 Bradford MBA
and also more general features of your data such as the distribution and, particularly, the important Normal (or Gaussian) distribution.
Objectives
By the end of this unit, you should be able to:
understand the requirement to obtain samples from a population
choose a suitable sampling method to obtain relevant data
undertake non-probability sampling
undertake probability sampling
understand the main summary statistics used to describe your data.
Obtaining Samples from populations
Most problems that we encounter in business will have a unique population of affected entities. These could be individuals such as employees, managers, general stakeholders, etc, or even inanimate in nature, such as organisations, business sectors, countries, and departments in companies. We usually cannot collect data from all of these members due to financial cost, time taken, timeliness of any findings being available, or the data to be taken from the selected members leads to its destruction (destructive testing in operations management and quality assurance) and a multitude of other reasons. I am sure that you can identify additional specific reasons in your own organisation.
However, provided we take a sufficiently large sample, i.e. a subset from the population, we can usually make quantifiable predictions of the properties of interest in the population by using statistical techniques. If the sampling approach used is appropriate, and the sample obtained is sufficiently large, this is usually accurate enough for our purposes. The objectives of sampling are exemplified in the Principles of Sampling, namely:
avoid bias in the selection of data
achieve maximum precision in inference from the sample, i.e. get as much information as possible from a sample of a given size.
Probability and non-probability sampling
If we sample from the population such that each sampled item (e.g. a person or organisation) has a known probability of being selected, we call this sampling probability sampling. A trivial example of this is when we sample at random, giving all members of the population an equal chance
Unit 6: Sampling and Summary Statistics
Bradford MBA 87
of being selected. Here, for a population of n members or items, the probability of selecting the first item is clearly 1/n, for each of the items.
If n = 250, we could sample, at random, 5 items (say) from the population and calculate the average value of the characteristic of interest from these 5 (e.g. the salary of the employees). This is highly desirable as we are able to make these estimates for population values (such as the average value, say) in an unbiased way. By unbiased we mean that if we were to repeat our sampling of a set number of objects many times, and to calculate the sample values for the property of interest (e.g. the mean value), the average for these sample values would be the same as for the population value.
For biased sampling, the value of the average of the values we calculate from our samples would differ from the population value. So, if we measured the height of 20 male employees selected at random from a particular retail store, using a faulty tape measure that started at 2 cm, the average heights we would obtain would be 2 cm greater, on average, than that of the population. This sample therefore produces biased results due to a faulty instrument. More usual faulty instruments in business research are poorly-designed or -worded questionnaires or questions in interviews.
Biased samples can also be obtained when the method of collection is deficient. This is the case with non-probability sampling methods. However, such methods are often very useful in business research and can produce results that are sufficiently accurate for our purposes, especially if you are not looking to make general claims about your findings. The main types of non-probability sampling of interest to business researchers are: convenience sampling, purposive sampling, quota sampling, and snowball sampling. Some authors use alternative names for some of these basic approaches.
Non-probability sampling
Convenience sampling is sampling where the subjects selected are convenient for the researcher. They could be: family relatives or personal friends, neighbours, work colleagues, even MBA colleagues. You would expect these potential participants to be similar to yourself in attitude, circumstance, social class, and many other attributes that could influence the characteristics (variables) of interest in your research topic and so, not necessarily representative of the population. Such sampling is often used as a quick and cheap way to obtain a ‘ballpark’ view of the issue(s) of interest at the initial stages of your investigations. Thus, results cannot be generalised to the population.
Purposive sampling is where you have a belief that certain types of participant will be more representative of your population of interest.
Study Book: Business Research
88 Bradford MBA
Consequently, you look to include such members, either exclusively or predominantly, in your sample. This could be something simple like females or people living in a particular neighbourhood. However, there can be an element of subjectivity in selecting the desired proportions, and this could lead to bias. The two major methods that employ this approach are judgement sampling and quota sampling.
Judgement sampling is used when only a limited number of groups have the information that is required for the investigation at hand. For example, if you were researching the experience of females and the so-called ‘glass ceiling’ you would clearly want to restrict your sample to that of senior female executives, ideally, in notorious male-dominated industries or business sectors, and those in contention for senior positions. Whilst the generalisability of any findings is contentious, this remains one of the best ways of dealing with such highly specialised issues.
In quota sampling, you decide beforehand that you want your sample to be comprised of certain numbers of members of specified groups. For example, you might decide to require half of the sample that answer your questionnaire to be males over 45 years of age who have been married for more than 5 years, with one third to be females who are under 35 years old and who have been married for at least 3 years, and the remaining one sixth to be females, 35 years or older and who are presently, unmarried. These proportions are often set in an effort to match population demographics, to capture desired groups or even to ensure representation of small, traditionally under-represented groups, i.e. purposively determined. Once you have determined what sample size you want, you can then set target numbers for these groups.
Quota sampling would then entail collecting data from a particular population until you have the number of participants for these categories. This data collection often takes place in ‘convenient’ locations, such as supermarket stores, town centre meeting places, office locations, etc. Collis and Hussey (2014: 200) also allow for the possibility that the sample for these quotas can be selected by probability sampling, although this is rarely the case for management projects.
With snowball sampling, you conduct an interview or submit a questionnaire (say) to somebody or several individuals who are knowledgeable about, or relevant to, the phenomenon you are interested in. At the same time, you get them to supply you with details of others who are similarly affected and you then approach them, in turn. This is particularly attractive if you are investigating something that few have knowledge or experience of.
Unit 6: Sampling and Summary Statistics
Bradford MBA 89
Probability sampling
If you are looking to generalise your findings to the whole population, you should ideally be looking to obtain your sample from the population using probability sampling approaches. The main approaches are: (simple) random sampling, systematic sampling, stratified sampling, and cluster sampling. All methods have a random selection aspect that makes them probability sampling methods, and enables findings from analysis of data collected using these approaches to be generalised to the complete population. This property is termed external validity.
(Simple) random sampling is achieved by the ‘simple’ expedient of sampling from the population at random such that each item (e.g. each person) present in the population has an equal chance of being selected. Here, a sampling frame is established where each member of the population is labelled conceptually from 1 to n, where n is the size of the population. If your population were a large department in a retail store with 67 employees (n=67) and you wished to have a simple random sample of 10 employees from this department, we can select a number between 1 and 67 and take the corresponding employee. This process would continue, with any repeat numbers being ignored, until the full 10 employees were selected. Such sampling is an example of sampling without replacement; once an ‘item’ is selected it is removed from the population and so cannot be selected again. This is the usual approach and all the probability methods we discuss will implicitly use this. If sampling is with replacement, allowing the possibility of several occurrences of the same items (employees here) in the sample, this should be stated in the reported methodology and findings.
It is not always possible to have a sampling frame for each population. When we have a (usually long) list, we can consider taking a systematic sample. Here the list is divided up into a sequence of smaller lists. First, you divide the population size by your required sample size. This gives the size of each of the sub-lists, with the probable exception of the last sub-list that is typically smaller. You then notionally divide the original population into this set of sub-lists and select a random number from 1 to the size of the sub-lists. The corresponding element in each of the sub-lists is then selected, with the usual exception of the last sub-list that is usually
smaller. Selection is made from this sub-list at random with each item being given an equal chance of inclusion.
The list could be, for example, a list of residents on a housing estate or members of a residents’ social club. Strictly, randomness is only present in the initial selection and the last, if the last sub-list is smaller than the other sub-lists. Once these are known, everything else is completely determined. A schema for systematic sampling compared to (simple) random sampling is given in Figure 6.1. Note the ‘systematic’ way in which the sample is built up with every ‘m
th ‘ item from the list. Also note you can
Study Book: Business Research
90 Bradford MBA
take details straight from the list and so you do not need to construct a sampling frame for this sampling method.
Figure 6.1: Systematic sampling and simple random sampling
Stratified sampling is a particularly important probability sampling method that addresses the requirement to maximise precision in inference by getting high levels of information from a sample of a particular size. However, it is only of use if the population exhibits certain characteristics. When the population is comprised of strata where all the members of each stratum are highly similar or identical in some property that is not actually of interest to you as the researcher, but may have an impact on the variable of interest to you, this may well be the sampling method to use.
As a trivial example, if you were interested in executive remuneration in a particular business sector, gender is (sadly) likely to be an issue. In most economies, females tend to get less pay for doing a similar job even if they are demonstrably of equal competence to their male counterparts. You would therefore want to have a sample with a desired proportion of females present. In fact, if they were substantially underrepresented in the workforce, you may even decide to increase their presence in your sample. The former would be classified as a proportionate sample and the latter as a disproportionate sample. Having split the population into the relevant strata, random sampling then takes place relative to the sizes of these strata to obtain a proportionate or disproportionate sample, as
Random Sampling Systematic Sampling
Unit 6: Sampling and Summary Statistics
Bradford MBA 91
required. If you wanted to ensure that small groups that are usually left out of surveys are included, you can not only guarantee their presence but represent them disproportionately. For this to increase precision you require that the within-group/stratum variation is small and the between- strata variation is large, otherwise there is little point in using this approach. A schema for stratified sampling is exhibited in Figure 6.2.
Figure 6.2: Stratified sampling
Similar to stratified sampling, cluster sampling notionally divides the population up so that each member of the population is in one, and only one, cluster (stratum for stratified sampling). For stratified sampling, this clustering is based on certain characteristics that the members exhibit. In the case of cluster sampling, this is usually a consequence of area such as households in a particular district.
However, unlike stratified sampling, you will not be attempting to minimise the within-group variation and maximise the between-group variation. In fact, you are hoping to achieve the exact reverse. You want to increase the within-group (cluster) variation and reduce the between-group variation, such that each cluster looks like a microcosm of the population. This being the case, you are at liberty to sample one or a small number of the clusters only instead of the entire population. As these are localised, it reduces sampling costs and the time to obtain your sample. If you take each member of the cluster(s) then you do not even require a sampling frame. However, this method is not as precise as stratified sampling and, usually, simple random sampling.
For both stratified sampling and cluster sampling, you are not restricted to taking simple random samples from the strata or clusters (if not taking all of the members of the cluster). You can use other, more convenient sampling methods here and you could even look to multistage sampling in support of cluster sampling.
Read: Collis and Hussey (2014), section 10.2: 196–201
Study Book: Business Research
92 Bradford MBA
Activity 6.1 – read and reflect
Read: Easton, V.J. and McColl, J.H. (editors) (1997) Statistics Glossary, STEPS
Read: Explorable (2015) Snowball Sampling, explorable.com
Read: Explorable (2015) Cluster Sampling, explorable.com
Read: Trochim, W.M.K. (2008) Non-Probability Sampling, Web Center for Social Research Methods
(Blackboard > Module Site > Module Materials > Unit 6)
These resources provide comprehensive definitions and examples of the various sampling techniques.
Activity 6.2 – stop and think
Think of situations in your workplace or a previous workplace where you would use the following sampling methods outlined. Describe the research they would provide data for and explain why they would be especially suitable for this research:
a) Quota sampling
b) Stratified sampling
c) Cluster sampling
Activity 6.2 – stop and think answers:
Unit 6: Sampling and Summary Statistics
Bradford MBA 93
Summarising and describing data
The data emanating from the research carried out on your chosen sample is typically summarised by using descriptive statistics. These descriptive statistics are numbers that describe properties of the sample in some way. Typically, these are sample estimates of corresponding population values. The latter are termed (population) parameters. If you have sampled your data from the population appropriately, you could then use the sample statistics to estimate the true value of the population, which is what you are really interested in.
Note that even certain qualitative data can be analysed quantitatively using statistical reasoning. This is usually achieved by analyses of frequency data noting the number of times a particular case occurs, such as when, for example, a respondent states a preference for a particular strategy over others. We have already looked at content analysis and this could be suitable for this situation, or it might be more appropriate to conduct a hypothesis test such as a chi square test if we have two such variables involved and we will look at these later, after we have seen how to summarise data using descriptive statistics.
Read: Collis and Hussey (2014), sections 11.1 to 11.2: 226–227
Note: SPSS (an acronym for Statistical Package for the Social Sciences) is a highly popular statistical software package used to support data analysis in business. The set textbook by Collis and Hussey (2014) uses this package to demonstrate much of the statistical analysis and data presentation. Whilst many of you might find this illuminating, knowledge of any software package, including SPSS, is not a requirement for the module. For those of you who have access to this package and are interested in learning how to use it, as completely optional activities, you may wish to read the extra sections in Collis and Hussey (2014) where SPSS is discussed in detail.
Summarising data
The way we present data depends on the level or scale of the data measured (Salkind 2014: 159) as well as the nature of the variables we are/were interested in. There are four different types of measurement level for variables and, in order of ascending precision, these are: nominal (or categorical), ordinal, interval, and ratio.
Variables measured on the nominal or categorical scale are used to signify membership of various categories. So, you may have three categories (e.g. 1 = micro organisations, 2 = small-to-medium sized organisations, 3 = large organisations) and the data for these would be measured as categorical data. The next highest level of precision is measurement on the ordinal scale. Here we are only interested in the
Study Book: Business Research
94 Bradford MBA
ranking, or order, of the data. Collis and Hussey (2014) give a suitable example for nominal categories that also have a natural order associated. Ordinal variables can also be used for data which is from a highly skewed or asymmetrical distribution. For example, if you looked at annual declared profit for organisations that were trading in different business sectors or markets but that were of similar size, you could have some companies with profits of £1,000s, consistent with their business sector/trading market. However, others could be presenting £100,000s of profit, and yet others with £Ms, again as a consequence of the types of profit that can be made in the associated business sector or trading market. Given these disparities, it might be more appropriate to represent these using a variable measured on the ordinal level scale.
The next level of precision is the interval scale and this is a more usual continuous quantitative variable but without a natural zero point. This means that comparisons of ratios are meaningless. An example of such a variable is temperature. If we measured the temperature of the freezing point of water, we would get 32 ºF on the Fahrenheit scale, 0ºC on the Celsius scale and 273 K on the Kelvin scale; three different measurements. There are several other candidates such as the Rankine scale (491.4R) for precisely the same phenomenon! Finally, there is the highest measurement scale, namely, the ratio scale. This is the ideal scale to work with as there are few limitations here. It is the measurement scale you would use for profit, ROI, return on assets (ROA), etc.
Nominal variables are not considered as quantitative variables despite often having numerical measures. Ordinal variables can be either with interval or ratio variables, both being numerical (quantitative) variables.
Read: Collis and Hussey (2014), section 10.3, 10.3.1, 10.3.2 and 10.3.3: 201–203
For inferential statistics, where you conduct statistical analyses of your sample data, Collis and Hussey (2014) quote Kervin (1992: 727) as stating that you need both a random sample, and the concept of sampling distribution. As the distribution is so important for conducting statistical tests, you will often see distributional characteristics presented as summary descriptive statistics for the sample data. We therefore need
to know what a statistical distribution is and which descriptive statistics are often used to represent this. We will discuss this, along with the other main descriptive statistics that are used to summarise data, in the following section.
Summary statistics – central tendency
For quantitative data, it is usual to present measures for the central tendency or location for the data, the range over which the data are spread, especially variation about the mean measure of central
Unit 6: Sampling and Summary Statistics
Bradford MBA 95
tendency, and measures for the way that the data are distributed. Central tendency is a single measure and is typically given by the:
mean
mode
median.
The mean is the average value, the mode is the most frequently occurring, and the median is the central value when all values are placed in numerical order.
When we are grouping our data (see Collis and Hussey 2014: 236), we
are concerned with the modal group and, as with ungrouped data, there may be several groups with the same number of occurrences. Such data are said to be multimodal. When dealing with finite data, the median is the middle value of the values when ranked in ascending order or the mean of the middle two observations if there is an even number present. Generally, when looking at a reasonable sized data set, say 20 observations or more, if the data is reasonably symmetrical, we would look to the sample mean as the usual central tendency value to provide. If the data is asymmetrical, we call this skewed, and the median is often the most appropriate statistic to represent these data. The mode is particularly relevant for frequencies of occurrence of nominal (categorical) data.
Summary statistics – dispersion
Another statistical property of interest is the degree of variation that is present in a sample. This is measured in several ways but the most used, because of its statistical properties, is the dispersion about the mean, central tendency measure. This is usually measured by the standard deviation or its squared value, the variance.
The most trivial measure of variation is the range. This is simply the difference between the largest and smallest values for the variable, in your sample. This can be refined by reporting the Inter Quartile Range (IQR). Here we are only interested in the range for the middle 50% of the data when arranged in numerical order. The first quartile (Q1) is the value for
the observation that has 25% of the observations smaller in the sample when ranked in ascending order of magnitude. The median is therefore, by definition, at the second quartile (Q2) as 2 x 25% = 50% of the observations are smaller than this value. The third quartile (Q3) is given by that observation in the sample that has 75% of the observations with smaller values. This means that Q3 also has 25% of the observations that are larger. As a consequence, the interval delimited by (Q1, Q3) contains 50% of the ordered observations and is the IQR. The semi inter quartile range (SIQR) is half of this range.
Study Book: Business Research
96 Bradford MBA
When looking for measures of dispersion about the mean value, as mentioned above, the most popular statistic is the standard deviation. This is only an appropriate measure for interval and ratio data. It is attractive due to its usefulness when analysing data statistically. It also has the advantage of including all observations in the population or sample, if dealing with a sample.
An interesting analogous measure of dispersion for samples is the standard error. Formally, the standard error is the standard deviation for the sampling distribution for the sample means. To see this, consider taking a sample from a population and calculating its standard deviation as a measure of its spread about the sample mean. We obtain a particular value. This process can be repeated several times for samples of the same size and the sample standard deviations noted. As each time we take a sample, at random, we are generating a random value and the variable associated with this, the sample mean, has its own distribution. This distribution has a population standard deviation and this is the standard error (SE). Its value is given by:
SE =
Where SD is the standard deviation for the underlying population and n is the fixed sample size.
Read: Collis and Hussey (2014), sections 11.5 to 11.6.2: 244–250
Activity 6.3 – stop and think
Question 1: What is the point of providing a measure of central tendency from your sample data?
Question 2: Why might it be best to use the median as a measure of central tendency when you have highly skewed data?
Question 3: What are the respective limitations of using the range and IQR to describe dispersion in data?
Question 4: The sample standard error for the turnover sample data is: 39.83 (to 2 decimal places). (See Collis and Hussey 2014, Table 11.6: 250) Why is the corresponding sample standard deviation = 1119.45 (to 2 decimal places)?
Unit 6: Sampling and Summary Statistics
Bradford MBA 97
Activity 6.3 – stop and think answers:
Statistical distributions
A statistical distribution is a characteristic of a particular population and is a measure of the way data is distributed across the population. The most used (probability) distribution is the normal (or Gaussian) distribution. This is a bell shape, symmetrical distribution that has the highest probability of occurrence near to the central value for the distribution. It also has tails. Many distributions are considered to be approximated by a corresponding normal distribution, for the purposes of forecasting and prediction in business modelling.
When a distribution is asymmetrical it is said to exhibit skewness. A skewed distribution is therefore one that is either positively skewed with a positive (right hand) tail that is substantially greater and more dispersed that its left hand tail, or is negatively skewed, with a negative (left hand) tail that is substantially more dispersed that its right hand tail. Such dispersion necessarily requires certain relative relationships between the three main measures of central tendency (mean, mode and
median).
In addition to the data being skewed, it might also be highly peaked about its central tendency value or it could be flatter here with far longer tails. This degree of flatness versus ‘peakedness’ is typically measured by the kurtosis exhibited by the sample data and this, together with a measure for skewness, is often reported, particularly when you are interested in whether your data is likely to have come from a Normal (Gaussian) distribution.
Study Book: Business Research
98 Bradford MBA
An example of data which has been sampled from a population that follows a Normal distribution is exhibited in Figure 6.3. The data displayed has a mean value of 100 units and a standard deviation of 20 units. The corresponding measure for skewness, S = 0.10 and that for kurtosis, K = –0.36. The population parameters for a Normal distribution are both zero for these statistics. The sample values are close to these population values and this demonstrates an important point: sample statistics are most unlikely to be exactly those of the population but you would expect them, on average, to be close, particularly if you had a large sample size.
Read: Collis and Hussey (2014), sections 11.7 to 11.7.1: 251–252 and section 11.5.2: 244–245
Activity 6.4 – stop and think
Question 1: Why is the skewness coefficient zero for a normal distribution?
Question 2: Is the median for a positively skewed sample greater or less than the mean value and why?
Question 3: What proportion of observations would you expect to see, on average, in the positive (right-hand) tail of a normal distribution, with values greater than 2 standard deviations from the mean value?
Following on from these questions, provide a rough sketch of the sampling distribution for the mean values for samples of four observations, taken randomly from the normal distribution depicted in Figure 6.3 [with = 100 and standard deviation () = 20]. Draw this over the curve in Figure 6.3. Only try to reflect the basic features.
Figure 6.3: A normal distribution with a (population) mean 100 and standard deviation of 20
15013512010590756045
Unit 6: Sampling and Summary Statistics
Bradford MBA 99
Activity 6.4 – stop and think answers:
Summary
In this unit you have looked at a number of ways to sample data collected from a population. These sampling methods have been divided into two categories: probability and non-probability sampling. You then looked at the advantages and disadvantages of these approaches. Finally, you looked at some of the major summary statistics that are typically used in business research projects and outlined in the business research literature as well as the features of the important Normal distribution.
Activity 6.5 – audio and slides
Listen to the audio for this unit and view the accompanying slides at the same time. (Blackboard > Module Site > Module Materials > Unit 6 > Unit 6: Audio and Slides)
Activity 6.6 – multiple-choice questions
The multiple-choice questions for this unit will enable you to test your knowledge and understanding of the key concepts covered. (Blackboard > Module Site > Formative Exercises > Multiple-Choice Questions > Unit 6)
Study Book: Business Research
100 Bradford MBA
Activity 6.7 – live online tutorial
Read: Attendance (Blackboard > Module Site > Module Materials > Unit 6)
This file contains a graph (a bar chart) of the number of times that various randomly selected employees have been late attending formal meetings in the previous 18 months. This is plotted against the number of years of employment in the company by the employee. There is also an accompanying table of selected summary statistics.
Have a look at this information and try and tackle the activities outlined below. Don’t worry if you can’t complete all the activities or you think that you may not have got the correct answers to the tasks – the key thing is to have a go at the tasks.
Task 1: Look at the table and the graph in detail. Confirm the calculation of the: a) sample variance, b) sample standard error, c) IQR and d) range, from relevant data in this table. Also note the skewness coefficient and kurtosis values and relate these to the graph provided.
Task 2: Provide an example from your present or previous organisations of research problems where data collection using the following would be appropriate:
1. stratified sampling
2. cluster sampling
Why are these appropriate?
Your answers to Task 1 and Task 2 will form the basis of an online live tutorial, so be ready to discuss these during the online live tutorial. You will be informed as to when this tutorial will take place.
Live online tutorial: To access the tutorial, please go to Blackboard > Module Site > Collaborate. To test your sound settings for the Blackboard Collaborate virtual classroom, please go to the Blackboard Collaborate
Support site at http://bit.ly/1etRu2Y. For further support materials and information about Blackboard Collaborate, please see the Blackboard Collaborate Support site at http://bit.ly/1jBpgUc.
References
Easton, V.J. and McColl, J.H. (editors) (1997) Statistics Glossary, STEPS. http://www.stats.gla.ac.uk/steps/glossary/alphabet.html Accessed 13 July 2016.
Unit 6: Sampling and Summary Statistics
Bradford MBA 101
Explorable (2015) Cluster Sampling, explorable.com. https://explorable.com/cluster-sampling?gid=1578 Accessed 13 July 2016.
Explorable (2015) Snowball Sampling, explorable.com. https://explorable.com/snowball-sampling Accessed 13 July 2016.
Kervin, J.B. (1992) Methods for Business Research. New York: Harper Collins.
Salkind, N.J. (2014) Exploring Research New International Edition. Harlow: Pearson Education.
Trochim, W.M.K. (2008) Non-Probability Sampling, Web Center for Social Research Methods. http://www.socialresearchmethods.net/ kb/sampnon.php Accessed 13 July 2016.
Bradford MBA 103
Unit 7:
Hypothesis Testing,
Correlation and
Regression
Key reading: 1. Collis and Hussey (2014), Chapter 12
Key audio/video: 1. UCMSCI (2008) Hypothesis Testing, YouTube, 17 November
(Blackboard > Module Site > Module Materials > Unit 7) – Activity 7.4
2. Unit 7 audio and slides (Blackboard > Module Site > Module Materials > Unit 7 > Unit 7: Audio and Slides) – Activity 7.5
Other: 1. Unit 7 multiple-choice questions (Blackboard > Module Site > Formative
Exercises > Multiple-Choice Questions > Unit 7) – Activity 7.7
Introduction
Read: Collis and Hussey (2014), Chapter 12
In this unit, you will see how to formulate a (statistical) hypothesis test. You will also learn how to conduct a test for association between two nominal variables (what we call a
2 test). Finally, you will see how we
measure any correlation (relationship) between two variables (e.g. age and the purchase of a Ferrari) and establish, and validate, a simple regression model.
Objectives
By the end of this unit, you should be able to:
understand the general procedure for undertaking a hypothesis test
conduct a 2 test for association between two nominal variables
Study Book: Business Research
104 Bradford MBA
understand what a correlation is between two variables and how to interpret the correlation coefficient
understand what a regression model is, and how to validate and interpret this.
Previously, we have distinguished between deductive and inductive research and noted that induction is about theory making, whereas deduction involves validation of theories. As such, deduction is mostly quantitative in nature and follows the positivist paradigm. Research is about answering a research question or questions and for positivist research, this is tackled by producing subordinate, testable hypotheses that, when answered, will provide answers to these research questions
(Black 2002).
In this unit we will initially consider statistical hypothesis testing. This is more usually termed as a form of inferential statistics: we are able to draw inferences from the tests that are undertaken. Here, the research hypotheses that are raised to ‘operationalise’ the research question are used so that we can conduct statistical tests on the data that we have collected to check the plausibility, or to validate these research hypotheses.
To achieve this, two separate statistical hypotheses are created, based on these research hypotheses, and it is these that are actually tested. The results from these will then provide answers to the research hypotheses. The two statistical hypotheses are conventionally denoted as: H0 and H1. H0 is termed the null hypothesis and is pronounced, ‘H sub-nought’. This is ideally used to represent the current situation, or status quo. The counter to this, H1, is the alternative hypothesis, pronounced, ‘H sub- one’. Often we drop the ‘sub’ and say, ‘H nought’ and ‘H one’, respectively. This all sounds most daunting on initial reading but you will soon be used to using these concepts and terms readily, just as a rather handy vocabulary for describing the tests that you need to conduct.
Hypothesis testing
Hypotheses testing is used to test claims about population values (i.e. parameters). As an example, we might have a research question such as: ‘The annual (population) mean profit for the organisation is greater than £10M’. You would make this the alternative hypothesis, with the counter to this being the null hypothesis. We would then assume that the null hypothesis was true and, based on data we had collected, test if H0 is consistent with this assumption.
As we are dealing with statistics, we set a level of probability for H0 being true, and discount this assumption if the probability of obtaining the estimate of the hypothesised value of the (population) parameter
Unit 7: Hypothesis Testing, Correlation and Regression
Bradford MBA 105
calculated from the data we have sampled is too small. As we have chosen H1 such that it is counter to H0, rejecting H0 means that we are accepting the claim contained in H1, and vice versa.
Following the usual convention, and denoting the mean profit of an organisation for the population by the Greek letter, pronounced ‘mu’, for our hypothetical example:
H0: ≤ 10
H1: > 10
Here, the alternative hypothesis, H1, expresses the research question as a testable statistical hypothesis: the population mean, , has a value greater than 10 (£10M). The null hypothesis is the counter to this such that all possibilities are covered, so that whichever hypothesis is accepted leads to the other being rejected.
In general there are three possible alternative paired hypotheses that deal with all possible research questions. These are detailed below for the problem we are looking at by:
1. H0: ≤ 10; H1: > 10
[≤ is notation for less than or equal to, > means greater than]
2. H0: ≥ 10; H1: < 10
[≥ is notation for greater than or equal to, < means less than]
3. H0: = 10; H1: ≠ 10
[≠ means not equal to]
There are a couple of important points that you should note here:
i) In all cases, ‘equality’ is reflected in the null hypothesis (≤ , ≥ , or =).
ii) The alternative hypothesis, H1, for case 1 and case 2 deals with ‘directional’ inequality ( < , > ) with case 3 dealing with strict inequality (≠).
When conducting hypothesis tests we look for the most exacting case for H0 and this is achieved when we have strict equality, hence the inclusion of the possibility of equality in all three cases. Case 1 and case 2 are one- tailed or one-sided tests. Case 3 represents a two-tailed or two-sided test. To see this, we need to look at case 3 in detail. Here, H1 gives us the two options of the population mean annual profit, being either less than 10 (£10M) or greater than 10 (£10M). In the other two cases, one option
Study Book: Business Research
106 Bradford MBA
only is accommodated; either greater than £10M for case 1, or less than £10M for case 2.
As we mentioned beforehand, when we conduct a test we proceed under the assumption that H0 is correct. There are also two distinct classes of tests: non-parametric tests and parametric tests.
If you are dealing with interval or ratio quantitative data and you know what distribution you have sampled from, you will be testing to see if the parameter from the population has a particular value or exhibits certain properties. Provided the variance (or equivalently, the standard deviation) is also constant (homogeneous) over the range of values for the variable, you are likely to be conducting a parametric test.
If, however, your data is measured on the nominal or ordinal scales, if you do not know the distribution the population follows, or if the variance for the distribution is not constant (heterogeneous) the appropriate test will be a non-parametric test.
Regardless of whether the test is a parametric test or a non-parametric test, all tests lead to the calculation of a test statistic from the data collected. The probability of obtaining this by chance alone, or a more extreme test statistic value, given that H0 is true (and that the additional assumptions of distribution and constant variance also hold), is calculated. We call this the P-value for the test. This is compared to a significance level that we decide on prior to calculating the test statistic. The convention is to denote this significance level by the Greek letter, (pronounced ‘alpha’).
If the P-value is smaller than, or equal to, we have evidence from our sample that the probability of obtaining the test statistic is rather low and therefore not consistent with H0 being true. The significance level allows us to set a level of confidence on whether H0 is likely to be correct, or otherwise. It therefore forms the basis of our inferences and decision- making. is usually set at 0.05 or, equivalently, 5%.
It is important to understand that this ‘1-in-20 rule’ (0.05 = 1/20) is completely arbitrary. If the findings from tests that you are conducting would have substantial implications for your organisation, you may wish to
work to a tighter significance level, 0.01, say. On the other hand, if actions following from a significant result, i.e. rejecting H0 and accepting H1, would result in relatively unimportant non-radical changes, you may look to run with something far less tight, such as 0.1 (or 10%). A scheme for the steps that need to be taken when conducting a hypothesis test is exhibited in the list below:
1. Establish H0 and H1, based on the research question to be tested
2. Treat H0 as true
Unit 7: Hypothesis Testing, Correlation and Regression
Bradford MBA 107
3. Decide on the significance level () (typically 5%)
4. Make distributional assumption (e.g. population is normally distributed)
5. Calculate test statistic (from sampled data)
6. Under H0, estimate probability of getting test statistic value or a more extreme value (P-value)
7. Compare P against significance level for test (
8. If P≤ reject H0 and accept H1, otherwise accept H0
A note of caution needs to be borne in mind when undertaking hypothesis tests, though. If H0 is true, the P-value from the test will be the true probability of obtaining a value of the test statistic or a greater, more extreme value, by chance, from your sample. If you are unfortunate you will obtain such an extreme value with a probability of less than . (Actually, the probability of this happening is given by the P-value.) You would therefore erroneously reject H0 and accept H1. Such an error is called a type I error. Similarly, when we accept H0 in error, by chance alone, consequently rejecting H1 when this is actually true we are making a type II error.
Read: Collis and Hussey (2014) sections 12.1, 12.2, 12.2.1, 12.2.2, 12.3.2 and 12.4.1: 259–270 (ignore the specific details for tests using SPSS)
Activity 7.1 – stop and think
Now that you have gained an initial understanding of some of the fundamental issues associated with hypothesis testing it is important to put your learning into practice. With this in mind it is ESSENTIAL that you tackle the following questions.
Question 1: You conduct a t-test with the significance level at 0.05. The resulting P-value is 0.034. Would you accept or reject H0?
Question 2: If your significance level was set at 0.01 and your P-value is 0.034, would you accept or reject H0?
You conduct a 2 test for association between two nominal variables. One
of the variables has three distinct categories; the other has two categories. The two variables are: Educational attainment (high school diploma, university degree, postgraduate degree) and Gender (male, female). The data are for senior managers in the HRM divisions of large, high- technology companies in the West Yorkshire region of the UK:
Study Book: Business Research
108 Bradford MBA
Question 3a: What are the dimensions and number of cells for this test?
Question 3b: One of the cells has an associated expected count of 2 with all the remaining cells having counts in excess of 10. Is it acceptable to continue to test these data for association using the
2 test?
Question 3c: What are H0 and H1 for this test?
Question 3d: The resulting P-value was 0.052 and you have set = 0.05. What do you conclude from this test?
Activity 7.1 – stop and think answers:
Correlation
To date, we have looked at univariate data, or data associated with a single variable. We are now going to discuss bivariate data, where we have two observations on each subject (or case) we are interested in. An example of this is the annual pre-tax profit that a company declares and the company turnover. Data for the profit and the turnover are both univariate data but as they are associated with a single company, we have multivariate data and in this case, bivariate data.
Collis and Hussey (2014) state that most statistics used in the social sciences (and therefore business and management) are based on linear models. Put simply, we attempt to identify relationships between the two variables that we have observed and to fit linear (straight line) models to the data. If a good fit is obtained for the data, we are then in a position to predict values for one of the variables, given the value for the remaining
Unit 7: Hypothesis Testing, Correlation and Regression
Bradford MBA 109
variable. This is the main reason for establishing the linear relationship in the first place.
Before we look at a linear model for bivariate data, a simple regression model, we will look at how we can measure how closely one variable is associated with another. By ‘associated’, we mean that we are interested in whether changes in values for one of the pair of variables is mirrored by changes in the other. Using the language of Collis and Hussey (2014), this does not necessarily suggest that there is a dependency between these variables such that one can be seen as the independent variable and the responding variable as the dependent variable. It could be that both are dependent on a third variable, or several variables. As this (or these) variables changes, the two variables of interest would also change sympathetically.
To help us focus, let us look at a simple problem: the management of a particular organisation believes that there is a relationship between the age of their employees and the number of times these employees are late attending formal meetings. Fifty randomly selected employees are monitored, covertly, for a period of 15 months and the results are plotted on a scatter plot exhibited in Figure 7.1.
Before we look at the graph, we need to consider whether age and lateness are likely to be causally related, and if so, which is the likely independent variable. On reflection, lateness cannot be the independent variable but it is not obvious that age, per se, is causing lateness if an association is seen to be present. Another consideration is to speculate whether any possible association would be negative or positive. By this we mean: as age increases, would we expect lateness for meetings to go down or up? Actually, there is a body of literature on this topic and the suggestion is that there is a causal association, termed a relationship, and that typically, as age increases, employees tend to be more punctual, on average. The ‘on average’ clause is actually rather important. This is not always the case for each observation, but this is the overall, average effect.
We can now look at the graph and can see an overall negative trend, consistent with the literature. An additional observation is that this negative
trend does appear to follow a linear (straight line) relationship. We will come back to this point later. However for now we are interested in measuring association between these variables (Age and Number of Late Attendances for Meetings).
The usual way to measure association is to calculate the Pearson correlation coefficient between these variables. This is a measure of how well two variables are linearly related. If the linear relationship is negative, as in our case, the correlation coefficient is negative. However, the correlation coefficient is positive if the linear relationship is positive. In addition, the measure is scaled such that all values are between -1 and 1.
Study Book: Business Research
110 Bradford MBA
A value of zero is for two variables that are completely independent of each other. A value of -1 is for a perfect negative linear relationship; the correlation coefficient is 1 for a perfect positive linear relationship. The closer that the correlation coefficient is to -1, for a negative linear relationship, the closer the scatter about a perfect linear relationship will be to the line describing this relationship. Similarly, the closer the correlation coefficient is to 1 for a positive relationship, the closer the scatter will be to a notional positive straight line. The sample correlation coefficient for the association between age and number of late attendances for the 50 employees in our problem is measured as -0.981 which is a ‘high’ negative value, reflecting the close negative linear association between the two variables.
As the scatter about the notional straight line increases, the correlation coefficient value will decrease in magnitude. In our problem, the increase in scatter, as exhibited in Figure 7.2, will lead to such a reduction (correlation coefficient = -0.899).
You should note that you will be dealing with sample values and so two completely independent variables will not have a sample correlation of exactly zero. The value should be small, however. If the data are either ordinal, or interval (or ratio) but quite skewed, the Spearman correlation coefficient is the preferred coefficient to use. Similar restrictions and rules of interpretation apply to those for the Pearson correlation coefficient.
Figure 7.1: Number of late attendances for meetings and age (years)
706050403020
16
12
8
4
0
Age
N o . o f la
te A
tt e n d a n c e s
Unit 7: Hypothesis Testing, Correlation and Regression
Bradford MBA 111
Figure 7.2: Number of late attendances for meetings and age (years)
Read: Collis and Hussey (2014), sections 12.5, 12.5.1, 12.5.2 and 12.5.3: 270–275
Activity 7.2 – stop and think
Now that you have gained an initial understanding of some of the fundamental issues associated with correlations it is important to put your learning into practice. With this in mind it is ESSENTIAL that you tackle the following questions.
Question 1: There are four scatter plots exhibited in the panel overleaf. (Figure. 7.3). Each has a particular sample correlation coefficient value. Identify the relevant values for the four samples from the following list and justify the choices that you make
[1.000, 0.848, 0.960, and 0.994]
All values are presented to 3 decimal places (3 d.p.).
706050403020
16
12
8
4
0
Age
N o . o f la
te A
tt e n d a n c e s
Study Book: Business Research
112 Bradford MBA
Figure 7.3: Bivariate Data and Correlation
Question 2: Why is it best to use the Spearman correlation coefficient as a measure for linear association, when the data are highly skewed?
Activity 7.2 – stop and think answers:
1.000.750.500.250.00
5
4
3
2
1
1.000.750.500.250.00
5
4
3
2
1
Y1
x
Y2
Y3 Y4
Unit 7: Hypothesis Testing, Correlation and Regression
Bradford MBA 113
Linear regression
Linear regression is a method for producing a model that enables us to predicate the expected value for a single variable, based on the value(s) of another single variable, or set of variables. The variable that is being predicted is called the dependent or response variable, with the variables that we are able to make the predictions from called, variously, the independent, explanatory or predictor variable(s). Traditionally, the dependent variable is denoted by the letter, y, and the independent variable by the letter, x. If several predictors are present, these are often denoted by the letter x but with the addition of a subscript, e.g., x1, x2 and x3 for three separate independent variables. When there are several
predictors present, the resulting regression is termed a multiple regression. This is to be contrasted with a simple (linear) regression where only a single predictor variable is present.
For multiple regression models, the independent variables should not be correlated, hence the term ‘independent’. If this is not the case, the variables are multicollinear and, with multicollinearity present, you should not attempt to fit a regression model to such data. As Collis and Hussey (2014: 274) observe, when a strong correlation exists between a pair of predicator variables, only one of these should be present in the final regression model. However, the suggestion that the correlation coefficient should be either greater than 0.9 for a positive relationship, or less than –0.9 for a negative relationship, is perhaps rather stronger than most would run with. The cautious correlation coefficient value of more extreme than ±0.60 is perhaps a more realistic benchmark. Finally, it is assumed that changes in the dependent variable are a consequence of changes in the predicator variables, i.e. that the regression model is causal.
It we look again at Figure 7.1 for the scatter plot of Number of Late Attendances plotted against Age, we can see the ‘best fit’ line that has been superimposed. This ‘best fit’ is a regression line that has a negative slope, or gradient, with random scatter about this fitted line. The ‘best fit’ is usually obtained by minimising the sum of the squared distances from the estimated line and the data, the so-called Ordinary Least Squares (OLS) fit. It is usual to get a computer package to calculate this for you. Such a facility is even available in MS Excel. The best fit for the data in Figure 7.1 is given in the regression table in Table 7.2:
Study Book: Business Research
114 Bradford MBA
Table 7.2: Regression table for regression of Late Attendances versus Age Predictor Regression
Coefficient t-value P-value Conclusion
Intercept 24.87 50.35 <0.0005 Reject H0, Accept H1
Age –0.37 –34.79 <0.0005 Reject H0, Accept H1
The resulting regression equation obtained from the output exhibited in Table 7.2 is given by:
Late = 24.87 –0.37 Age (1)
where Late is the dependent variable, representing the average number of times that employees of a particular age are late when attending formal meetings, a constant term, 24.87 which is called the intercept, a negative gradient value of –0.37 and Age which represents the age of the employees, in years.
Generally, the intercept is the average value for the dependent value when the independent, or predicator, variable is zero. Often this is meaningless for the problem and this is such a case. You can easily confirm the value by putting Age = 0 in equation (1):
Late = 24.87 –0.37 0 = 24.87
However, employees cannot be in their first year of life! You should leave the value in though, as this is constructional and will enable you to estimate the average value for a given age that is in the correct range. As an example, if you wanted to see what the average value of late attendances for an employee who is 30 years of age is, you merely substitute the age (30) in the regression equation (1):
Late = 24.87 –0.37 30 = 24.87 – 11.10 = 13.77
You can see that the gradient value is the amount that the dependent value changes in response to a unit increase in the predictor variable (here, this is a reduction of 0.37 late attendance times, on average, for an increase in age of one year for the employees). You should remember that this is an average value only and that it is calculated from your data. This may not be a general equation that would apply to other organisations. In research, however, we would like to generalise the nature of the relationship (a simple regression equation with a negative slope/gradient) to a broader population.
Unit 7: Hypothesis Testing, Correlation and Regression
Bradford MBA 115
Before we leave the topic, those who are keen-eyed will have noticed P- values in the regression table and statistical hypotheses (H0 and H1) quoted. When you fit a regression model to sample data, as we have just done, if no relationship is present, we still do not expect to get calculated intercept and gradient values that are exactly zero. These are the population values here, and so the values we would get, on average, if we obtained a large number of samples and calculated the respective intercept values and gradient values.
To emphasise the point, each sample is most unlikely to have corresponding calculated regression coefficient values of zero if this is the case for the population, but the average of these would be zero. If the population values for the gradient coefficients are zero, this simply means that a regression model does not exist and no straight line relationship is present.
We therefore have a dilemma: you obtain non-zero values for the gradient coefficients but do not know immediately whether these are just random values with no underlying linear relationship present, or whether these are from a linear relationship. This is where hypothesis testing can be used. We conduct separate tests for each of the regression coefficients (the intercept and each of the gradient values for the predicator variables in your model).
For a simple regression model we hypothesise:
H0: The intercept value = 0
H1: The intercept value ≠ 0
H0: The gradient value = 0
H1: The gradient value = 0
with the significance level, = 0.05
From Table 7.2 we have P < 0.0005 for both the test for the intercept and that for the gradient. As this is substantially lower than the significance level (= 0.05), we reject H0 and accept H1, indicating that the values for both are not zero and so each term should be included in the final
regression equation.
If either had not been significant, they would have been omitted from the final model. In fact, if that for the gradient had not been significant we would have had evidence, from our data, that no linear relationship was present. When we do have significant regression coefficients it is customary to use the calculated values for the coefficients as the estimated values in the final model. This is usually termed the empirical model to reflect the fact that the parameter estimates have been taken from the sample values. It should be noted that it is assumed that the variation about the OLS line is assumed to be normally distributed. This
Study Book: Business Research
116 Bradford MBA
enables us to conduct statistical tests knowing the distribution of the test statistic. As the standard deviation for each of the regression coefficients is usually unknown, computer packages use an alternative, marginally more conservative, distribution than the Normal distribution, the student t-distribution, to calculate the corresponding P-values.
Read: Collis and Hussey (2014), section 12.7: 281–282
Activity 7.3 – stop and think
Question 1: A multiple regression model was fitted to sample data and the resulting regression table is exhibited in Table 7.3. This is an extension of
the previously discussed problem of late attendance at company meetings but with the predictor variable of Age supplemented by the Number of years of employment in the company and Gender. Based on the output here, what is the final empirical model you would select if you were testing at the = 0.05 significance level?
Question 2: Given the predictor variables present in the multiple regression problem of late attendance that you have just analysed, do you think there may be problems with the fitted model? Justify your answer.
Table 7.3: Regression table
Predictor Regression Coefficient
t-value P-value
Intercept 20.33 25.24 <0.01
Age -0.37 -34.79 <0.01
Years of company employment -3.68 -153.76 <0.01
Gender 0.02 0.93 0.36
Activity 7.3 – stop and think answers:
Unit 7: Hypothesis Testing, Correlation and Regression
Bradford MBA 117
Activity 7.4 – watch and reflect
Watch: UCMSCI (2008) Hypothesis Testing, YouTube, 17 November (Blackboard > Module Site > Module Materials > Unit 7)
This film offers insight into some of the issues discussed in this unit.
Summary
In this unit, you have gained knowledge and practical experience of hypothesis testing. You also looked at how the correlation between two variables is measured and have explored the topic of regression modelling. Here, you have gained practical experience of conducting hypothesis tests for association between nominal variables and for the regression coefficients to establish the resulting empirical model.
Hypothesis testing is an essential part of business as it enables you to make comparisons, so that you can see how you are doing and compare this to others. Regression is also hugely important as it enables you to make predictions and forecasts with far greater accuracy than mere speculation and guesswork, increasing your competitive advantage.
Activity 7.5 – audio and slides
Listen to: the audio for this unit and view the accompanying slides at the same time. (Blackboard > Module Site > Module Materials > Unit 7 > Unit 7: Audio and Slides)
Activity 7.6 – review activity
Question 1: In hypothesis testing, it is important to establish the significance level, prior to calculating the test statistic. Why do you think this is best practice?
Activity 7.6 – review activity answer (Question 1):
Study Book: Business Research
118 Bradford MBA
Question 2: There are two separate regression equations present in Figure 7.4. The positive regression is that for the first 10 points. The second regression equation is for the combined data for these ten observations and an additional observation that is an outlier. Explain why the equation now has a negative gradient.
Figure 7.4: Lines of Regression
Activity 7.6 – review activity answer (Question 2):
Activity 7.7 – multiple-choice questions
The multiple-choice questions for this unit will enable you to test your knowledge and understanding of the key concepts covered. (Blackboard > Module Site > Formative Exercises > Multiple-Choice Questions > Unit 7)
20015010050
70
60
50
40
30
20
x
Y
Line of Regression
based on first 10 points
Outlier
Line of Regression based on
first 10 points PLUS the
outlier
Unit 7: Hypothesis Testing, Correlation and Regression
Bradford MBA 119
References
Black, T.R. (2002) Understanding Social Science Research 2nd edition. London: Sage.
UCMSCI (2008) Hypothesis Testing, YouTube, 17 November. www.youtube.com/watch?v=BX9iMIC6mcg&list=PLD471E07F82185195 Accessed 26 May 2016.
Bradford MBA 121
Unit 8:
Presenting Findings
Key reading: 1. Collis and Hussey (2014), Chapters 11 and 13
Key audio/video: 1. JackBarnett21 (2011) I-Teach-u How to Write a Management Report,
YouTube, 5 February (Blackboard > Module Site > Module Materials > Unit 8) – Activity 8.3
2. Unit 8 audio and slides (Blackboard > Module Site > Module Materials > Unit 8 > Unit 8: Audio and Slides) – Activity 8.5
Other: 1. Unit 8 multiple-choice questions (Blackboard > Module Site > Formative
Exercises > Multiple-Choice Questions > Unit 8) – Activity 8.6
2. Unit 8 live online tutorial (Blackboard > Module Site > Collaborate) – Activity 8.7
Learning log: 1. Unit 8 learning log (Blackboard > Module Site > Assessment > Distance
Learning Assessment > Learning Log Tasks > Unit 8) – Activity 8.8
Introduction
In this unit you will see how data can be presented, both graphically and in tabular forms, in reports. You will also look at a typical report writing strategy (Torrance et al. 1992) and standard layouts for reports (including that for your MBA management project).
Objectives
By the end of this unit, you should be able to:
present data in graphical form
present data in tabular form
recognise the value of report writing strategies
produce a report that has an acceptable structure
Study Book: Business Research
122 Bradford MBA
understand the required layout for reports.
Presenting data graphically
Qualitative data
In many research projects, qualitative data is collected. This can be either demographic data or data collected to address the research question. Examples of qualitative demographic data are: gender (of participant), size of company (micro, SME, or large), geographical region, business sector, etc.
Other non-demographical qualitative data could be: political party (allegiance), company financial turnover (in ranges), preferred strategic approach, etc. The boundaries between such demographical and research relevant qualitative data are often quite diffuse, depending on the area of investigation. It is usual to present demographic data to demonstrate the characteristics of the sample of population that has been investigated. It affords the reader a greater understanding of the background, context and limitations of the research being reported.
In this section, we are going to look at ways in which such data can be presented in a graphical form. Graphical presentation is often preferred over the alternative tabular form as this is seen as ‘telling a thousand stories’, facilitating identification of patterns in the data and as a crude check of how representative the data is of the population.
Charts are a popular way to present frequency data. Such data is transformed into quantitative data, representing the number of occurrences (frequencies) for the categories of nominal (or ordinal) variables. The bar chart in Figure 8.1 displays the number of customers buying items from the clothing departments from five retail stores as a set of bars with identical width. These are labelled as A, B, C, D, and E for convenience, so that these store names do not clutter up the chart. Full names can always be provided in an accompanying legend. This is a simple, visual image that provides immediate relative information about the profile of an aspect of sales.
Additional information regarding the ownership (and presumably the way of purchasing) or otherwise of a store card for the retailer can also be introduced by using a composite bar chart. An example of this, within each store, is presented in Figure 8.2, and for card ownership, in Figure 8.3. The most appropriate chart will depend on whether you are focusing on the store, per se, or on the ownership of store cards. An alternative presentation of these data is exhibited in the stacked composite bar chart in Figure 8.4. However, this really does only work well in colour as shades of grey are often difficult to distinguish.
Unit 8: Presenting Findings
Bradford MBA 123
These charts are to be contrasted with that of the humble, much used pie chart in Figure 8.5. Finally, an additional way to present these data is by use of a comparative dot plot (Figure 8.6). Here each dot represents a pre- set number of observations (here, it is 3). The representation is self- evident.
Figure 8.1: Bar chart: Number of customers buying in the clothing department of five retail stores (A, B, C, D, E) using a store card
Figure 8.2: Composite bar chart: Number of customers buying in the clothing department of five retail stores (A, B, C, D, E) using a store card
EDCBA
100
75
50
25
0
Store
C o u n t
Store
Own Store Card
EDCBA
Y esN
o Y esN
o Y esN
o Y esN
o Y esN
o
80
60
40
20
0
C o u n t
Store
Study Book: Business Research
124 Bradford MBA
Figure 8.3: Composite bar chart: Number of customers buying in the clothing department of five retail stores (A, B, C, D, E) using a store card
Figure 8.4: Stacked (composite) bar chart: Number of customers buying in the clothing department of five retail stores (A, B, C, D, E) using a store card
Own Store Card
Store
YesNo
EDCBAEDCBA
80
60
40
20
0
C o u n t
Store
Unit 8: Presenting Findings
Bradford MBA 125
Figure 8.5: Pie chart
Figure 8.6: Dot plot
EDCBA
No
Yes
Store
C o u n ts
Each symbol represents up to 3 observations.
Store
Study Book: Business Research
126 Bradford MBA
Quantitative data
You always have the option of grouping quantitative data into selected groups and displaying the frequencies using the above charts. There is a loss of information here, though, and alternative graphs are more able to represent the full range contained in the sample. Again, for univariate data where the data are for a single variable only, a box plot is an ideal way to represent your data. A comparative box plot of the sales data for the purchases made in the clothing department of store A is exhibited in Figure 8.7a, for those not owning a store card and those who do.
Figure 8.7a: Box plot: Sales in clothing department of store A using a store card
The central ‘box’ represents the extent of the sales, in pounds sterling (£), of the middle 50% of sales by value. The band in the central box is the position of the median of the sample; the lower edge to the box is set at the first quartile (Q1) and the upper side to the box at the third quartile (Q3). The tails stemming from the box above and below Q3 and Q1, respectively, are the upper and lower tails of the sample distribution (see Figure 8.7b).
YesNo
35
30
25
20
15
Own Store Card
S a le
s [ £ ]
Store A
Unit 8: Presenting Findings
Bradford MBA 127
Figure 8.7b: Box plot: Sales in clothing department of store A using a store card
Clearly, these represent the extent of the upper and lower 25% of the data, respectively, with the upper most extreme point being the maximum sales value and the lower most extreme point for those who own the store A card being the minimum sales value. However, for those who do not own the store A card, there are two extreme low values for sales denoted by the two asterisks. These are called outliers as they are some way away from the bulk of the other observations. The lower value here is the minimum sale in store A for those without a store card. Comparative box plots can also be used to view the entire range of sales for all of the stores as exhibited in Figure 8.8.
Another illuminating way to present time series data graphically is a time series plot (TS plot). This is somewhat self-explanatory. The assumption is that data is periodically available and an example of such data is given in Figure 8.9 for a TS plot of worldwide cinema attendance (in billions) each year for the years 1999 to 2010. For bi-variable data where we have two observations for each unit of interest (case), we have already seen a scatter plot.
YesNo
35
30
25
20
15
Own Store Card
S a le
s [ £ ]
Store A
Outliers Lower Tails
Upper Tails
Q1 Q1
Q3 Q3
Median Median
Study Book: Business Research
128 Bradford MBA
Figure 8.8: Box Plot: Sales in clothing department by stores (A, B, C, D, E) using a store card
Figure 8.9: Time series plot of cinema attendance (billions) against year
Store
Own Store Card
EDCBA
YesNoYesNoYesNoYesNoYesNo
80
60
40
20
0
S a le
s [ £ ]
Store
20 10
20 09
20 08
20 07
20 06
20 05
20 04
20 03
20 02
20 01
20 00
19 99
1.62
1.56
1.50
1.44
1.38
Year
A tt e n d a n c e
Unit 8: Presenting Findings
Bradford MBA 129
Read: Collis and Hussey (2014), subsections 11.4.4, 11.4.5: 237–243
Activity 8.1 – stop and think
Question 1: Based on the box plot in Figure 8.8, which store has the highest sales in their clothing department to non-owners of a store card and which the least to card owners?
Question 2: Why do the tails in a box plot cover 50% of the ordered observations?
Question 3: Will the median always be contained within the box of a box
plot? Justify your answer.
Question 4: Based on the box plot in Figure 8.8, do you think that it is likely that the average sales in the clothing departments of each of the stores are the same? Justify your answer.
Activity 8.1 – stop and think answers:
Study Book: Business Research
130 Bradford MBA
Presenting data in a tabular form
Frequency data can be successfully presented in a tabular (table) form. As above, frequencies are the number of occurrences, or counts, for specific categories of a particular variable. The data will be taken from a nominal variable, ordinal variable or from categories used to group interval and ratio data. As an example for integer data (Buglear 2010), the number of emails sent by 22 office workers is grouped into frequency data and exhibited in tabular form (Table 8.1). This can also be done for interval and ratio data. As an example of the latter, we look at the volume of nail varnish measured in 10 ml bottles (Buglear 2010). Here, the data is ratio data (see Table 8.2). Categories are exhaustive in that they cover the
complete range of possible values for the variable being measured. The range of each of the groups is typically reported to a fixed number of decimal places.
In Table 8.2, this is to two decimal places (2 d.p.), e.g. 9.80, 9.89. This means that any observed value for the variable being measured is truncated to 2 decimal places and then allocated to the relevant group. As an example for the nail varnish data, an observation of 9.9987 ml would be assigned to the second group for measurements between 9.90 ml to 9.99 ml as the remaining digits after the first two decimal places are truncated, leaving 9.99 ml which is the upper limit (to 2 d.p.) for this group. A measurement of 10.00006 ml (assuming we can measure this accurately) would be assigned to the next group as, to 2 d.p., it is contained in the range (10.00 ml to 10.09 ml).
Table 8.1: Number of emails sent by 22 office workers
Number of messages sent Frequency
0–19 11
20–39 4
40–59 4
60–79 2
80–99 1
(Source: Buglear 2010)
Unit 8: Presenting Findings
Bradford MBA 131
Table 8.2: Nail varnish content
Nail vanish (ml) Frequency
9.80–9.89 3
9.90–9.99 7
10.00–10.09 11
10.10–10.19 5
10.20–10.29 3
10.30–10.39 1
Activity 8.2 – stop and think
Question 1: What are the advantages of displaying data graphically over displaying it in a tabular form?
Question 2: What type of data can be displayed meaningfully by a:
a) box plot?
b) dot plot?
c) pie chart?
Question 3: Can you provide an example of data from a management setting for each of a), b) and c)?
Activity 8.2 – stop and think answers:
Study Book: Business Research
132 Bradford MBA
Writing a research report or MBA management project
As an apprehensive postgraduate student who was about to write his thesis, I was told by a senior academic that, ‘... unless you write up your research and findings, it is as though it was never conducted!’. As nobody would be any the wiser unless a document or report was made available, this statement could not be more true, although this is not what I wanted to hear at that time. In honesty, few actually enjoy the process of writing up research and findings. It is therefore especially important to have a developed strategy for this process. This will help to ensure suitable attention to detail and commitment. Such a strategy, which Collis and Hussey (2014: 298) refer to was developed by Torrance et al. (1992) based on their empirical research into the writing experiences of 228 post graduate students at ten UK Universities. Table 8.3 provides a template built developed from their analysis (Torrance et al. 1992: 159). Initially, a brainstorming session is advocated. This should lead to ideas and topics that will be included in the final report. The authors then suggest that you look to the literature currently available to identify and copy relevant notes and discussions. These are then ordered to provide a coherent sequence for your work. Next, you need to use some creative thinking, possibly aided by constructing a diagram such as a ‘mind map’ to represent the links and connections between these ideas. You should then be in a position to develop a plan for the contents of the research project report and the ordering that you will be using. Now you are able to actually start to write the first draft, concentrating more on structure and logical flow than on grammatical precision. After this, if writing a project for a supervisor or your MBA project report, respond appropriately to their comments and constructive feedback. If not, re-read the project, ideally after giving yourself a short break so that you can view the work afresh, and then update accordingly.
Unit 8: Presenting Findings
Bradford MBA 133
Table 8.3: Report writing template relevant for MBA project and other directly supervised research study
Report Writing Strategy
1. Brainstorm for ideas that might be included in report. (Do not worry about order or relative importance at this stage.)
2. Extract, unedited, notes from relevant literature. (Remember to reword OR fully cite any direct quotations in final report.)
3. Put notes in order for ideas to follow a logical sequence.
4. Construct a ‘mind map’ to represent links and antecedents between ideas.
5. Produce a plan for the content of the first draft and order of appearance of materials.
6. Write the first draft, paying particular attention to logical flow and sequencing, more than grammar and formatting.
7. Update the draft in response to your supervisor’s feedback and your re-reading. (Give it a few days before this, if possible.)
(Source: adapted from Torrance et al. 1992, p.159)
When writing up research you will be targeting a specific body, be it a particular journal, a university examining body, a professional body, newspaper etc. As such, you should always be aware of any formats that you will be required to comply to. The general structure for a report to be submitted for academic purposes, such as a postgraduate qualification or publication in an academic journal, will be similar to that exhibited in Table 8.4.
Table 8.4: Traditional structure for a research report/dissertation
Abstract (include Acknowledgements if appropriate)
Introduction
Background (optional if not covered in introduction)
Literature Review
Methods/Methodology
Results
Discussion
Conclusion
References
Focusing on the end of course MBA management project, the structure for this piece of work can typically be broken down to a more detailed sequence of requirements. This finer level of detail is exhibited in Table 8.5 below.
Study Book: Business Research
134 Bradford MBA
Table 8.5: Detailed format for a research project or supervised research report
Introduction
Background
Literature Review
Research Question
Theoretical Background/Framework
Hypotheses [to be tested]
Methods/Methodology
Study Design
Population and Sample
Variables
Data collection
Results / Data Analysis
Discussion
Conclusions/ Recommendations [+Future work?]
References/Bibliography
It is essential that you always meet any stated format requirements such as: a preface, acknowledgements, recommendations, future work, table of contents, etc.
The style of your writing is also important. You should always write concisely and as clearly as possible and reflect the expectations of your intended audience.
In traditional academic writing you should normally avoid:
1. Personal pronouns (we, I, you, ...) and their possessive forms (our, my, your, ...)
2. Hedging (appears, may be, might (possibly be), ... )
3. Clichés/colloquialisms (‘ ... at this moment in time’; ‘ ... at the end of the day’; ‘ ...when all is said and done’, ...)
4. Rhetorical questions (‘Why do we need a manager?’)
5. Split infinitives (‘to boldly go’ (boldly = adverb; to go = infinitive verb))
In terms of academic writing, it is usually considered best to adopt a style that reinforces objective reasoning. Consistent with this you are generally advised to use the passive voice whereby you describe the action and not the actor. As an example: ‘The organisation needs to control its spending on auxiliary devices if ...’ can be reworded to, ‘Spending on auxiliary devices needs to be controlled (by the organisation) if ...’. Here,
Unit 8: Presenting Findings
Bradford MBA 135
not only is the actor (the organisation) diminished in importance, implicit reference can even be used.
Despite these general guidelines, it has to be acknowledged that in recent years interpretive researchers have adopted a less depersonalised and objective style of writing. Personal pronouns may appear, especially where the researcher is writing about her/his own experience or feelings. Similarly the findings of the research may be offered cautiously, acknowledging that alternative interpretations may be possible, and for some researchers asking questions that encourage reflection from the audience may be appropriate. The importance of Critical Reflection is now recognised by many academics and academic writing can reflect this. Ann Cunliffe’s (2004) influential paper illustrates that in some circumstances the researcher’s subjective voice is brought to the fore. Cunliffe uses personal pronouns and tells her audience about her practice: ‘I often use the previous activity to highlight the difference between writing in reflective and critically reflexive ways. I use one of two approaches…’ (Cunliffe 2004: 418). She also asks her readers questions (And answers them): ‘Are Critically Reflexive Journals Just Naval Gazing? The answer is no. I expect students to take their reflexive questioning and assess possibilities for change (Cunliffe 2004: 422).
However such approaches are only to be used with more interpretive research methodologies and need to be used with caution. Despite the appearance of more subjective and personal writing, most research is still presented objectively, and in all cases it is expected that you write with care and precision. The perceived credibility of your research will depend upon the quality and credibility of your writing. If you are considering deviating from a traditional report writing style for your MBA project, seek guidance from your supervisor and always err on the side of caution.
You should also avoid ambiguities and get the tenses that you use right (past tense for things that have happened; present tense for results and labelling and legends for Figures and Tables).
Also be careful when using acronyms. As a general rule, the first time an acronym is used it should be written in full with the acronym in brackets. After this the acronym can be used in isolation. As an example on page
103 of the study book the acronym IQR is first introduced in full‘…by reporting the Inter Quartile Range (IQR)’. Subsequently the acronym appears five times without explanation. Even if you think yoyr audience will be familiar with the acronym, it is expected that you will “explode” it when it is first used.
Finally, again on writing style, it would be most unusual for non-standard English such as contractions or slang (e.g. gotcha, no way, yes way, lol, der…?) to be acceptable. Some contractions (e.g. don’t, won’t, can’t) might be considered acceptable in some contexts, but it is recommended that these are also avoided.
Study Book: Business Research
136 Bradford MBA
Final thoughts on research writing
The most important thing to remember when you write a research report is something that is arguably a key part of all our writing. When we write, we are trying to communicate a particular message to an audience, or more precisely a reader. You need to try and write in a way that your reader will both understand and find acceptable. In the previous section I have outlined some of the expectations of your reader. She will expect you to write in a professional scholarly way. She will expect a well-structured piece of work that shows you are well read, have undertaken a piece of informed research within an organization or company, and that through the process you have generated knowledge that is both insightful and
credible.
As with any narrative, the introduction needs to provide an outline of what will follow and generate sufficient interest that the reader wants to read on. I’d always suggest that you write the introduction at the end of the
process, or certainly after the bulk of the writing is complete.
After the introduction, you need to provide the details of your research, with the chapters demarcating the research process into smaller sections or episodes. You also need a structure, or a plot, so that the reader can link the episodes together.
Finally the work needs to conclude, establishing what was important in the research, what knowledge you have created and also what wasn’t so good. Remember no research is perfect, our understanding of the world is always incomplete and always open to revision. Whilst ‘hedging’ is not to be encouraged, there will be limits to what you can claim for your research. Being a little self-critical and acknowledging the limitations of
your work will not lose you any marks and will probably gain you some.
So to conclude, try and make sure that you communicate your research effectively. If nobody reads and understands what you have achieved, the research has not been effective.
Finally:
1. Always stick to the in-house style and ensure that you meet any word count restriction that may be in operation (this latter point definitely applies to the management project).
2. Working from a previous and well regarded ‘published’ report of similar stature is often a guide for style, structure, level of detail and content.
3. Ideally, start the introductory section, or introduction chapter, last as this should give a broad overview of the research conducted and, possibly, the findings. These will not be fully known until they have been
reported.
Unit 8: Presenting Findings
Bradford MBA 137
4. If you have a very comprehensive list of the relevant literature with many articles not being cited in the text, you may wish to consider
providing these in a bibliography section.
The standards usually set for a research report are further itemised in Collis and Hussey (2014) (Table 13.5: 316).
Read: Collis and Hussey (2014), sections, 13.1 to 13.5.1: 298–317
Activity 8.3 – watch and reflect
Watch: JackBarnett21 (2011) I-Teach-u How to Write a Management
Report, YouTube, 5 February (Blackboard > Module Site > Module Materials > Unit 8)
Activity 8.4 – stop and think
From your own experience of writing general reports, why do you think that it is a good idea to follow a predetermined writing strategy?
Activity 8.4 – stop and think answers:
Study Book: Business Research
138 Bradford MBA
Summary
In this unit you have considered how best to represent data both graphically and in a tabular form and several main approaches have been presented. In addition, you have looked at the report writing strategy, introduced by Torrance et al. (1992), and standard report structures.
Activity 8.5 – audio and slides
Listen to the audio for this unit and view the accompanying slides at the same time (Blackboard > Module Site > Module Materials > Unit 8 > Unit 8: Audio and Slides).
Activity 8.6 – multiple-choice questions
The multiple-choice questions for this unit will enable you to test your knowledge and understanding of the key concepts covered (Blackboard > Module Site > Formative Exercises > Multiple-Choice Questions > Unit 8).
Activity 8.7 – live online tutorial
In this tutorial session we will go over some key issues that you will need to address when you critique an academic paper as part of the module assessment.
Prior to the tutorial please note down any questions or queries that you may have relating to the critiquing of an academic paper. Your questions and queries can then be addressed in the tutorial session.
Live online tutorial: To access the tutorial, please go to Blackboard > Module Site > Collaborate. To test your sound settings for the Blackboard Collaborate virtual classroom, please go to the Blackboard Collaborate Support site at http://bit.ly/1etRu2Y. For further support materials and information about Blackboard Collaborate, please see the Blackboard Collaborate Support site at http://bit.ly/1jBpgUc.
Activity 8.8 – learning log
Learning log: work through the learning log activities for this unit (Blackboard > Module Site > Assessment > Distance Learning Assessment > Learning Log Tasks > Unit 8).
You MUST save all of your workings in your Business Research Learning Log document. Your answers to the various activities will be submitted
Unit 8: Presenting Findings
Bradford MBA 139
later as a part of your assessed Learning Log. You are advised to back this document up and store this in a secure place.
Additional learning resources
Additional materials are available if you wish to learn more about the issues covered in this unit (Blackboard > Module Site > Module Materials > Unit 8 > Additional Learning Resources).
You might also want to access the following items yourself:
The mechanics of writing (Chapter 8, Lomas 2011) gives some particularly useful and highly practical insights into writing reports.
Chapter 14 in Sekaran and Bougie (2016) discusses the research report and gives interesting examples of graphical output.
A specimen report containing tabular output is exhibited in the Appendix of Sekaran and Bougie.
References
Cunliffe, A. L. (2004) On becoming a critically reflexive practitioner. Journal of Management Education, 28(4): 407–426. http://jme.sagepub.com.ezproxy.brad.ac.uk/content/28/4/407.full.pdf+html Accessed 26 May 2016.
JackBarnett21 (2011) I-Teach-u How to Write a Management Report, YouTube, 5 February. www.youtube.com/watch?v=UwjpN4KSn_Y Accessed 26 May 2016.
Sekaran, U. and Bougie, R. (2016) Research Methods for Business: a Skill Building Approach 7th edition. Chichester: John Wiley and Sons.
Torrance, M., Thomas, G.V. and Robinson, E.J. (1992) The writing experiences of social science students. Studies in Higher Education 17(2): 155–167.
http://web.a.ebscohost.com.ezproxy.brad.ac.uk/ehost/detail/detail?sid=a69 be189-06cb-48d8-989c- c0688086d865%40sessionmgr4002&vid=3&hid=4212&bdata=JnNpdGU9 ZWhvc3QtbGl2ZQ%3d%3d#AN=9706306512&db=pbh Accessed 26 May 2016.
Bradford MBA 141
Unit 9:
Revision
Key video: 1. Conclusions to Business Research (Blackboard > Module Site >
Module Materials > Unit 9) – Activity 9.1
2. Unit 9 Audio (Blackboard > Module Site > Module Materials > Unit 9 >Unit 9: Audio and Slides) – Activity 9.2
Introduction
In this unit, we will review the module, emphasising several important points and key issues. You will be reminded of the need for a structured approach for tackling research. This will be influenced by the research paradigm you will be pursuing, which can be established during the crucially important literature review. Issues and approaches to data collection and the overarching requirement to interact with others and to conduct your research following acceptable ethical guidelines will also be reinforced. The various methodologies available to the researcher and basic analyses of data are also included in our discussions. Finally, the structure of a typical report and ways in which you can present your findings will be discussed.
Activity 9.1 – watch and reflect
Watch: Conclusions to Business Research (Blackboard > Module Site > Module Materials > Unit 9)
This film recaps the key issues discussed throughout the module.
Objectives
By the end of this unit, you should be able to:
review the steps need to conduct and report a successful research project.
We start off research with a topic and, typically, a business problem. This could be a theoretical, conceptual problem or a practical business problem. Regardless of its nature, we need to formulate a research
Study Book: Business Research
142 Bradford MBA
question and it is this that we then address to provide a solution to this business problem. To achieve this you may develop a theory or you may validate an existing one that relates to your research question. The former is termed inductive research and the latter, deductive research. Business research is therefore a process of solving this research question, while adopting a particular approach. There are several issues that need to be addressed before a decision is reached as to the most appropriate research paradigm to pursue. The way that the research is then pursued relates to a variety of factors, not least associated with your philosophical stance regarding your view of information, and includes the nature of the data that can be obtained and the current level of reported knowledge of the topic.
No matter which paradigm you will be adopting, we saw that all research requires a structured and systematic approach to be employed. This will lead to an efficient and demonstrable solution. To further ensure that an appropriate approach is pursued, the role of searching and critically appraising the literature was seen to be a crucial and ongoing process.
Once the methodology was decided, you then studied several of the major data collection methods. These are classified into probability and non- probability sampling methods and the reasons for, limitations and advantages of both were explored for selected approaches. The requirements necessary to ensure the full benefits from the main probability sampling methods of Stratified and Cluster sampling were also itemised.
We also looked at the esoteric but important problem-solving approach of soft systems methodology for those interested in solving highly complex problems with several conflicting objectives to achieve, for various stakeholders.
For those looking to analyse qualitative data, you further explored content analysis. This is a systematic way of identifying key issues, concerns and themes present in textual articles. For those with quantitative data, we then looked at how you can summarise your data and how to conduct hypothesis tests.
Finally, you looked at ways in which you can present your findings. We introduced and discussed a typical writing schedule and revealed a template for the basic structure of a report. This was finessed for the type of content that may be present in a MBA project report – something that you will be required to produce at the end of the programme.
As a final and personal note, research in business is considered to be one of the most engaging and enthralling experiences for a practising manager. It gives you the opportunity to express yourself creatively and to elicit findings that can be of considerable benefit to your organisation and
Unit 9: Revision
Bradford MBA 143
society as a whole. It can also substantially alter your standing in your organisation and increase your career opportunities.
Summary
In this unit, we have reviewed the content of the module. We have emphasised certain points, most notably that business research is about solving a research question and that the way forward is often pointed to by a critical review of the literature. There is a need, also, to be systematic and to plan your research activities.
Activity 9.2 – audio and slides
Listen to the audio for this unit and view the accompanying slides at the same time (Blackboard > Module Site > Module Materials > Unit 9 > Unit 9: Audio and Slides).
In this your Module Leader recaps on key aspects of the Module and considers the assessment activities, the sort of things that are likely to earn you a higher mark and the things to avoid when tackling the assignment.
Bradford MBA 145
Appendix:
Model Answers to Activities
Unit 1
Activity 1.2
Task 2: Yes. Inductive research results in the building of theories. This usually requires investigation of the literature, detailed discussion with informed and involved individuals, and might even require observation. These activities typically produce rich qualitative data that needs to be analysed before theories can be developed.
Deductive research is involved in validating existing (extant) theories and making predictions based on associated models. (These are usually achieved by collecting quantitative data and fitting models based on the
theories, to these data. If the model provides a good fit, we have evidence
in support of the theory on which the model is based.)
Activity 1.4
Question 1: A positivist believes that reality is independent of the beliefs and views of the humans. A consequence is that business phenomena will follow laws that we are able to capture by theories. Theories therefore provide the basis for explanation and prediction.
An interpretivist believes that social reality is actually highly subjective and is a consequence of the perceptions and world-view held by the researcher. As a result, the very act of conducting business research to investigate a social reality requires interaction with the researcher, affecting the phenomenon under investigation.
Question 2: You are most likely to be following an interpretivist paradigm. Exploratory research is about discovery and not about validation of existing theories based on previous research.
Question 3: No. For your colleague to be pursuing a mixed methods paradigm he would need to be using a combination of positivist and
Study Book: Business Research
146 Bradford MBA
interpretivist research. However, he is actually only using a single paradigm, but triangulating.
Activity 1.5
Question 1: Experimental studies require control of several variables and this is an unrealistic situation for typical business research problems.
Question 2: Descriptive surveys seek to describe a phenomenon at a particular time (usually, current). This description is typically qualitative and could, for example, describe the impact of an organisation’s brand on a particular group of consumers. Analytical surveys, on the other hand, are
undertaken to establish, or confirm, various hypothesised relationships between variables. Descriptive surveys are associated with qualitative research whereas analytical surveys support positivist, quantitative research.
Question 3: Cross-sectional studies are likened to a ‘snapshot’. They analyse a phenomenon of interest (e.g. the sales for the regional offices of a large national retailer) at a particular moment. This is to be contrasted with longitudinal studies where interest is in the dynamic behaviour of a variable, set of variables, or a case over time. The variable could be company profit, the set of variables, attitudes of staff over a turbulent time of change in an organisation (e.g. loyalty, trust, commitment, etc), the case could be a particular company or a business sector.
For cross-sectional studies, taking data at a time point or over a short period, relative to the natural dynamics of the phenomenon of interest, the current status can be measured for that time. By contrast, to observe dynamic behaviour, it is necessary to observe the phenomenon, or case, over time and to take data at several specified time points. The duration for the set of observations must be sufficient for the phenomenon to change appreciably.
Activity 1.7
Question 1: The two diametrically opposed research paradigms of positivism and interpretivism (constructivism) are generally associated with deductive or inductive research, respectively. These naturally lead to quantitative research, when conducting positivist research, or alternatively, qualitative research when doing interpretivist (constructivist) research. These are termed mono-method approaches.
However, in the real world and particularly in the highly complex world of management and business studies, often such narrow, single-focused approaches are not sufficient to address these complexities. It might be
Appendix: Model Answers to Activities
Bradford MBA 147
necessary to address some research questions, given the current state of knowledge, using a mixed methods approach. Such approaches are associated with the pragmatic paradigm. Here, collecting data, either simultaneous or sequentially, is achieved by use of methods that are drawn from both the quantitative and qualitative traditions to address the specific research question(s) of interest. In essence, a pragmatist is therefore somebody who uses whatever data collection approaches are deemed appropriate to answer the research question.
Question 2: For a longitudinal study, the emphasis and analyses are on the changes, over time, between the various sequences of times. For a cross-sectional study, the emphasis is on analysis within each time epoch to understand status at any particular time, not the change or changes across the time domains.
Study Book: Business Research
148 Bradford MBA
Unit 2
Activity 2.2
The major, initial obstacle would be to ensure that your colleague can actually get appropriate data. They would be seen as an employee representing a competitor and, as such, the other companies in this business sector may consider any information to be sensitive and ‘company confidential’. Your colleague will therefore need to establish that they can get sufficient data before it is even worth spending time planning
for the project. If it is reasonable to believe that suitable data could be collected and appropriate assurances provided, it would then be necessary to plan and allocate time for the various activities of the project:
If your colleague is to conduct interviews, there will be access issues depending on the mode of contact (telephone, face-to-face, email) and possible travel and time availability issues. The questions to be asked and the time taken to conduct and record the interviews will also need to be budgeted for.
If a questionnaire survey, time associated with developing the questionnaire, and possibly validating this, needs to be planned for. Sufficient time will need to be allocated to identifying potential participants and for responses to be received and non-responders followed-up. Parallel activities can also be scheduled here, while waiting to receive responses. Time will also need to be allocated for analysis of the data, particularly if interview data. If new techniques are to be used here, this will also have a potentially considerable overhead. If any of these were overlooked or given insufficient time, this could jeopardise your colleague’s ability to complete the project on time and/or to an acceptable standard.
Activity 2.3
The suggestion is: A = {a, b, c, d, e}; B = {f, g, h, i}; C = {j, k, l}.
Any additional activities that you may have included will need to be introduced into these lists.
Activity 2.5
Question 1:
Interviews: Rich, discursive, highly subjective, non-comparative.
Appendix: Model Answers to Activities
Bradford MBA 149
Questionnaires: Specific, comparative, perceptual and factual, structured.
Question 2: Interviewing is typically the most labour-intensive. Given the considerable time constraints imposed on many research projects, you would only be able to complete a few quality interviews and so be limited to a fundamentally non-quantitative project with little or no statistical analysis present.
Study Book: Business Research
150 Bradford MBA
Unit 3
Activity 3.1
Question 1: The items cited are clearly the critical / most influential articles, hence others referring to them consistently. Other articles are perhaps more peripheral and less informative for your research.
Question 2: Searching the literature should be an ongoing activity throughout the duration of your research so this should be the initial
search.
Question 3: They may cite other important articles that you have not identified, either written by themselves (with co-authors) or by others. There may also be relevant publications that are not universally available such as university working papers, papers in progress, etc.
Activity 3.3
Task 1: Searching with the keyword TAM will probably capture all the Technology Acceptance Model articles as this is the accepted standard abbreviation. However, it will also include many additional articles with ‘tam’ as a substring in their titles and anything with TAM as an author or somebody with this as a substring of their name or that of their institute. Use of the full name will remove these additional and superfluous articles in the resulting search list.
Task 2: It should have reduced the number of articles selected!
Activity 3.4
There will certainly be a body of literature on the paper-reprocessing technology as this is taught at her previous university. There will also be an extensive literature on start-up venturing in novel technologies. There will also be a body of literature regarding the economic climate and opportunities in the country of her choice although this may be in EU and relevant Government reports. These are usually freely available though.
Appendix: Model Answers to Activities
Bradford MBA 151
Unit 4
Activity 4.1
Question 1: You should have offered them the option of withdrawing, without providing an explanation, prior to the interview. If you sense that they are becoming distressed or uncomfortable, you should then remind them that they can withdraw from the interview if they would like to. To continue without this intervention would be seen as possible coercion and could cause them distress. If they were not notified prior to the start that
they had the option of withdrawal at any stage, continuance could also be argued to be depriving them of their rights.
Question 2: Provided you are persuaded that they would not cause any subsequent distress, from an appeal to proportionality, this should be acceptable.
Activity 4.2
Question 1: Ranking data are not numerical data (they are strictly ordinal data) and so do not need to follow this convention.
Activity 4.3
Question 1: Open questions allow the respondent to express their attitudes, feelings and beliefs. Their answers are therefore discursive, usually cannot be compared directly to replies from other respondents and, as such, allow the researcher to develop a view and deeper understanding of the subject of interest. This is therefore more likely to support inductive, qualitative research.
Question 2: Use of the ‘do not know’ category in closed questions will usually discourage respondents from fully engaging in the questions as it provides a ‘get-out clause’. This tends to result in an increase in ‘missing data’ as ‘do not know’ data cannot be used in most analyses.
Study Book: Business Research
152 Bradford MBA
Unit 5
Activity 5.2
Question 1: Efficient and timely data collection method; highly interactive between knowledgeable participants; can change direction of discussion leading to reflection and considered debate by experts.
Question 2: Could be difficult to time/schedule all experts; very reliant on quality of facilitation; can be difficult to present responses and interpret
any findings.
Question 3: Direct interactions with and between experts and the flexible and dynamic nature of the facilitated discussions.
Activity 5.4
Question 1: Necessary to capture the views of all stakeholders as you are looking at a multi-criteria problem that you are attempting to resolve for all parties.
Activity 5.8
Question 2: The following list is in no particular order of difficulty. Some issues may not be relevant for the research and there may be others that are topic- or environment-specific:
Identification of participants; facilitating the process (especially for over- defensive participants!); ensuring timely responses from all participants; editing and interpreting the responses from all participants in appropriate timescales to broadcast findings/summaries; judgement of convergence (if present).
Question 3: Complete loss of non-verbal cues; several themes emerging from various participants – could be contradictory; evolving dynamic consensus would be lost (no time considerations); dominant participant(s) biasing possible findings.
Appendix: Model Answers to Activities
Bradford MBA 153
Unit 6
Activity 6.3
Question 1: The sample data is all that you have! If you have sampled appropriately (i.e. without introducing bias) then these data should, on average, be representative and so the sample value should reflect that of the population.
Question 2: Generally, the mean value is sensitive to skewed values. The median, however, is more robust to these extreme observations and so likely to provide a better, less extreme, estimate.
Question 3: i) The range only uses two observations. In addition, these are the two
most extreme observations. If these are outliers that are inconsistent with the remaining data, this will lead to an unrealistic measure.
ii) The IQR also only uses two observations. If the sample distribution has long or short tails, this range will not capture this consistently.
Question 4: (The sample standard error for the turnover sample data is: 39.83 (to 2 decimal places) (see Collis and Hussey 2014, Table 11.6: 250). Why is the corresponding sample standard deviation = 1119.45 (2 decimal places.)?)
The standard error (SE) is given by:
This gives 1119.45/√790 = 39.83 (2 d.p.)
Activity 6.4
Question 1: The normal distribution is a symmetrical distribution and so has no skewness.
Question 2: It will be less than the mean value as you are adding on larger values when estimating the mean. The median is unaffected by these values.
Question 3: Approximately 1-in-40 or 2.5%. For your sketch, it should be more peaked than the supplied curve. The mean value will be in the same position but the tails will be shorter. The standard deviation for the sampling distribution will be half that of the supplied curve (√4 = 2).
Study Book: Business Research
154 Bradford MBA
Unit 7
Activity 7.1
Question 1: You would reject H0 and accept H1 at the = 0.05 significance level; 0.034 is less than 0.05 and so the test statistic obtained has too low a probability to accept H0.
Question 2: You would accept H0 and reject H1 at the = 0.01 significance level (alternatively, fail to reject H0); 0.034 is greater than 0.01
and so the test statistic obtained has a high enough probability to accept H0.
Question 3: a) 3 2; 6 cells.
b) This should be acceptable as the expected counts are only less than 5 (but greater than zero) for 16.67% of the cells which is less than 20% which is the usual lower boundary.
c) H0: Educational attainment is not associated with Gender.
H1: Educational attainment is associated with Gender.
d) Accept (fail to reject) H0. Educational attainment is not associated with Gender.
Activity 7.2
Question 1: Y1= 1.000; Y2=0.994; Y3=0.960; Y4=0.848 [Y1 – observations are on the regression line; Y2-Y4: listed in order of relative scatter about the regression line with 0.994 the least, 0.960, next and 0.848 exhibiting the greatest scatter.]
Question 2: Highly skewed observations can lead to artificially high correlations. The Spearman correlation coefficient uses the ranks of the
data, not the raw data values and, as such, are not unduly influenced by extreme values.
Activity 7.3
Question 1: Late attendance = 20.33 – 0.37 Age –3.68 Employment
Question 2: Yes. This can only be true for an unrealistically restricted range due to the high negative gradient of Years of employment. For
Appendix: Model Answers to Activities
Bradford MBA 155
example, if Age = 30 years; Years of service = 10 years, Late attendance would be, on average, 20.33 –0.37 30 –3.68 Employment = –27.57!!
Activity 7.6
Question 1: So that you do not unconsciously set the significance level such that the test would be deemed significant. This could happen if you are aware of the corresponding critical value for the distribution (e.g. ±1.96 for a 5% significance level for a two-tailed test using the Normal distribution or 1.645 for a one-tailed test for the Normal distribution)
Question 2: The outlier (an observation that is some distance from the
bulk of the remaining observations) is not ‘consistent’ with the other observations and has an undue influence. This leads to the regression equation being distorted resulting in the negative gradient. If the outlier was removed, the equation would be quite different and so the outlier is also an INFLUENTIAL observation.
Study Book: Business Research
156 Bradford MBA
Unit 8
Activity 8.1
Question 1: Highest for non-card holders = D; Lowest for card holders = E.
Question 2: Because 50% of the distribution is contained in the central box, 50% MUST be contained in the two tails.
Question 3: Yes. The central box contains the inter quartile range (IQR). The median is the second quartile (Q2) and as this value is always present in here, it will always be present in the box.
Question 4: It is most unlikely that the population mean values will be the same. The median values between card and non-card holders for each store are likely to be the same, as are the mean values. However, for example, the median values for store D are equivalent to values near the positive extreme for several of the remaining stores and are greater than the maximum values for store C.
Activity 8.2
Question 1: If present, you may be able to see patterns in the data that would not be apparent when the data are presented in a tabular form. You can also see how the data are distributed over the range of values and if there is any skewness present.
Question 2: a) Interval and ratio data; b) Nominal and ordinal data, integer valued interval and ratio data (Likert-type measures); c) Nominal and ordinal data.