BM3100 Research

Methods and Data

Analytics

CONTENT CHECKLIST:

Week 6: Research Paradigms and

Week 1: Introduction and Suggestions
Qualitative Data
Week 2: Research Question and Literature
Week 7: Quantitative data collection and
Review
creation
Week 3: Case Study
Week 8: Data cleaning and Description
Week 4: The messy process of coding
Week 9: Quantitative data analysis
qualitative data

Week 10: Data Visualisation

Week 5: research Ethics
 Research is always connected to theories
Week 1: Introduction and
Suggestions Theories are needed to explain results.
- What do we know? What can we explain? Where
are gaps? How can we design the research? How
Purpose of Module: to be able to develop research to collect data? How do we generate theory?
skills - Can be explanatory and exploratory

Why do we care about research methods?: there

could be different versions of truth, to get answers we
need questions

What are they?

- How do we know what is true?
- How do you assess what is good data?
- Which information can we trust?
- How do we analyse data?
- What is the best way of obtaining useful
information?
- How do we ask the right questions?
- Develop anakystical skills that never become
outdated?
Stage 1: Research Question, Literature Review
Stage 2: Formulating research Design
RESEARCH METHODS = accepted rules and
procedures to construct good questions, collect and Stage 3: Data Collection
analyze reliable and valid data, and find reliable and Stage 4: Data Analysis
valid answers to our questions! Stage 5: Interpretation of Findings
 Type of research question influences the design, the
Week 2: Research Questions and alignment of data.
Literature Review
What’s the most reliable data: depends on the research
question and how you access the data
Understanding alignment between Stage 1 & 2
The process is circular
How to come up with a good RQ?
Why need a RQ?
- To get answers we need questions - What are you curious about?
- We need good questions through appropriate - Who are you speaking to?
research - What conversation are you part of?

Types of RQs: Generating ideas: rational & creative techniques

EXPLORATORY - How? What?
DESCRIPTIVE - Who? When? What? Criteria for good RQ: RELEVANCE (what is the purpose of the
EXPLANATORY - How? Why? research?), FEASIBILITY(time available, skills, expertise, data
EVALUATIVE - How much? To what extent? needed), INTEREST(is it something you and the target are
interested in?)
Three of these engage with theory, descriptive doesn’t.
Practical Checklist for good RQ:
To assess quality of RQ: appropriateness and - Is there evidence this is a ‘real’ practical problem? Are
capability framework: there ppl. Talking about this
- large/small problem?
Appropriateness: Is it interesting? Has it been done
- Does it apply widely or to a narrow range of contexts,
already? Is it appropriate and within the scope of
orgs. Or settings?
question?
- Will answering the RQ help solve the problem?
Capability: do you have the right resources ie. time? - Does answering help: identify mechanisms underlying
Access to data? Skill sets access? the problem, identify causes of the problem
Academic Checklist for good RQ: Write down a temporary RQ in general terms, based on
Interest and Topic
● What evidence is there that this question is Read Literature
academically important or interesting?
● What did previous research show are the answers Literature Review: It’s not reading to find an answer, it
to the RQ? won’t answer your question
● What are the strengths and weaknesses of this You can reformulate questions
research?
● Is your RQ a replication of previous research? What is the connection between the research
Why? Is the context new? How will replicating help question and the literature review?
specifically?
● In what way is your RQ different from, or similar to - To ask the right question, you need to know what
RQs already asked in the academic literature? others have already said
● What theoretical mechanisms are presumed to - Key assumptions about the topic identified
underlie the phenomena? What is the quality of
this theory? So what is literature review?

- Report of relevant literature: discussion of what

has been established and agreed on the topic
- Clarification of you reasoning and logic for your
research
- Where you establish the framework for your
analysis: you define the terms you use
- Outline of strengths and weaknesses of the
existing arguments and explanations: the “gap” to
be addressed
The Inverted Pyramid Approach:
Strat with an Interest which is important for you and put
this in the context of a General Topic
It isn’t:
- Descriptive list Principles for a literature review:
- Article by article summary
- Survey of everything ever written on the topic Making sense of existing knowledge: what concepts,
theories and labels are used in this research area?
Literature: Original work that researches your relevant Are there any significant controversies, are there
area of interest, situates your project in a particular area inconsistencies, is there a significant gap?

Sources of Literature: Making order out of existing knowledge:

How can studies in this are be grouped and ordered?
Academic Knowledge: Academic Journals, Monographs What research methods, or design have been used to
and Edited Books, Theses and dissertations, Working study this area?
Papers - Use these

Empirical Knowledge (not original works): Newspapers, Research Design / Methodology

BUsiness Magazines, Market reports, Web PAges,
What is it?
Company Publications - “These pointed me”
The decisions the researcher needs to make about:
Types of Literauture:
- Overall methodological approach chose
- Data Collection - sources, techniques
Integrated Review: brings together main ideas on a topic
- Data Anlaysis
to provide an overview and generate a new framework or
perspective
These decisions need to be explained and justified:
Critical review: What are the most valuable works on this
- How each step aligned to your RQ
topic and why? Where does this lead us?
- Citing others doing similar research who took
Scoping Review: size and scope of available literature
similar approaches
Systematic Review: a specific search strategy
- How you managed in the process
Meta- Analysis: pools and analyses the results from
multiple similar studies
Deductive vs Inductive Approaches:
Descriptive:
Deductive [Explanatory] Questions ■Accurate representation of events, situations
■Survey, sampling, interviews, re-analysis of secondary
■Enough literature on the topic exists to identify which data
concepts to include in your RQ ■Often forerunner to explanatory study
■Examine the relationship between concepts already
specified in the literature Explanatory:
■Theory and hypothesis first > Data second ■Study problem/situation to explain relationship between
■Often, but not always, Quantitative Data variables
■Surveys, experiments
Inductive [Exploratory] Questions
Criteria to determine quality of Research Design:
■Not enough literature on the topic already exists to
identify which concepts to include in your RQ Internal Validity: the extent to which the answer given to
■Explore / discover factors not yet clear or specified in the RQ takes into account alternative explanations
the literature
■Data > Theory and hypothesis generation External Validitiy: the extent to which the answer to the
■Often, but not always, Qualitative Data RQ is generalizable beyond the specific data (sample,
case) considered in the study
Three main types of Research Design:
Use this order Accuracy: The extent to which the answer to the RQ
Exploratory: takes into account the details and complexities of the
■new insights, new questions phenomenon under study
■Searching literature, interviews
■Broad focus, becomes narrower Reliability: The extent to which the answer to the RQ
■Flexible & Adaptable would be the same if your study was repeated by another
■Caution in drawing definitive conclusions researcher
 When useful:
Week 3: Case Study Research - Examine specific contemporary events
- Generating new ideas and explaining new
Characteristics: phenomena
- Illustrate theories
Small Sample - an organisation, individual, supply chain - Showing processes and relations
market - Often dictated by access
In-depth analysis using multiple sources of - usually
textual - data: interviews, documents, observations, A good case study provides a “full picture” of the situation
websites under investigation, involves triangulation, which helps
Focused on how and why questions about complex, make the research more robust
real-world situations
Assumptions of Case Study Research:
Case Study Research isn’t [Link], case example - Interpretivist: no single truth, the lived experience
of participants matters
Purpose of Case Study Research: - Participants are knowledgeable: their
- Theory building, testing (less frequently) interpretations inform the explanations, answer
- Develop new concepts, theories or theoretical developed by researchers
explanations grounded in the experience of people - Researchers’ subjective interpretations of data:
who “live and breathe’ the phenomenon: systematically and transparently recorded,
- Use time and space to build an understanding of logically ordered into a model or explanation that
the topic ‘makes sense’

Case Study Research: criticisms and challenges

Understanding of Case Study Approach: confusing case
study research with case study teaching. Inability to
generalize?
How they are conducted: selecting and accessing a case,
lack of rigor, time taken
Conducting Case Study Research profitable in the face of values opposed to their own
values?
Type of question: How - qualitative, case study

Data Collection: unit of analysis

You can’t research all aspects of a phenomenon of

interest in one study. You need to define:
- Unit of analysis (of the case)
- Unit of data collection (of the evidence about the
Selecting Your Case: case)
- Single or multiple
- Rationale? Why is the case suited to address your These exist at different “social system” levels:
RQ?: critical, unusual, common, revelatory, - Individual, group, organisation, a set of temporarily
longitudinal ordered events, an industry, a country
- Boundary: time-period, in-scope activities, actors
- ACCESS Example: For org.: we’d need data from the level of
individuals like interviews from employees, managers
Yoga Paper: Good example for academic research and at the level of organisation policies like brand
management & employment
Research question and literature review [“Theoretic At the level of individual we’d need data from individual
Motivations’] and group levels
What is the broad area of interest?: To implement Yoga to Put into context: The evolution from corporate social
the American/ Western markets into lucrative industry responsibility to supply chain responsibility: the case of
What is the theoretic topic?: How the movement that is Waitrose
growing uses a mechanism that goes against its own
ideology in order to grow faster Unit of analysis of the study = Organisational Level
What is their research question?: How do Unit of analysis of the data collection = Individual and
anti-establishment movements grow and become Organisational Level
Data Collection: types of data Interviews
- Guided conversations: structured (limited to
Primary: questions)/ semi-structured (flexible, interview
- Interviews (surveys & experiments are unlikely to guide, harder ot analyse), informants (people
be used) being interviewed),
- Focus Groups - Open questions: allow informants to elaborate43e
- Participant observation
- Participant diaries

Secondary
- Documents (ie. reports, policies, newsletter,
records, presentations, emails, meeting agendas, Purpose of interviews:
project documents)
- Archival Documents - Generate ideas or develop hypotheses
- Visual Images - Gain insight into complex issues
- Seek expert opinion
Develop a case study framework - Understand individual’s experiences
- Unit of analysis, unit of data collection - Collect information
- Triangulation - use multiple sources of evidence
- Construct validity - how does the data collected What’s the right number of informants - as long as there’s
relate to concepts being studied theoretical saturation (the subsequent interview you
- Justify the selection of sources conduct doesn’t tell you anything you haven’t already
- Reliability - could my research design be repeated heard) there’s no need for more
and obtain similar findings Documents
- Manage your time efficiently - Access to secondary data - like documents - is
- ACCESS usually easier than primary data collection
- Private vs public domain
 - Background Information
- Recording Device, transcript

Application: Yoga paper (end of Week 3)

Analysis (applied to any research method related

to qualitative not just case study)
You need to be structured and transparent about: How to convince others that you subjective
- What you are collecting interpretations are credible and plausible:
- Why, How, How many / what time frame - Root your RQ in existing literature
- Use a table to record information on your sources - Proivde detailed accounts of data collection
- Be reflexive and critical: why were the source - Triangulation
documents created? Who was the author? What - Be reflexive about yourself and your subjectivity
was the purpose? What was the context? What - SHOW your steps in your analysis tables
limitations might they have? - Use your raw data to build a convincing argument
- Show, don’t tell
Access = CONVINCE YOU HAVE IT
Recording, Organizing, Reporting Data
- Record, organize and report systematically
For interviews, observation, non-public documents:
- Is HARD - be realistic in your plans
Most Case Studies Include:
- Be willing to cold call/ cold email
- Timeline of events: the dates of the material
- Use family, friends, other contacts
relevant events covered by the study
- Be prepared to travel
- Interview tables: number of, and job role (or other
relevant characteristics) of interviewees
For interviews:
- Documentary/Secondary Data table: number, type
- Dont’ ask for more than 40 minutes
and use of other secondary data
- Be flexible with their availability
Credibility of analysis:
- Research Ethics: University Consent, Research
Information Sheet and Consent Form
 - How do they demonstrate the depth of their case Why is case method appropriate?
knowledge? - Its a how q, they can understand and immerse
- Do they show as well as tell? themselves to how something happens
What is the selected case?
Explaining Findings - Yoga from 1975
How do they justify their case selection?
■ Report the meaning of the case; what has been learnt - Its understandable that yoga has changed from
■ Typically, explain your findings in a figure (a framework, religious to capitalist target
a process, a theoretic model) You can’t research all aspects of a phenomenon of
■ Discuss how your findings are different / similar to the interest in on ecase study
literature review
■What are you adding Data Sources:
■How might your case apply to other contexts - What are their data sources? Yoga magazine,
interviews, participant observations
- How do they justify use of these sources?
Yoga Paper: Magazine: the most read, been around a long
time, how it influenced the industry
What is their broad area of interest? - How do they demonstrate robustness of their data
- Yoga’s Remarkable Transformation from a Sacred collection? Triangulation, building up on their data
Movement into a Thriving Global Market
What is the theoretic topic?
- How does something religious become a
consumer obsession/ capitalist target
What is their research question? In the final paragraph of
literature review
- How these movements become syncretic w Week 4: Qualitative Analysis
markets is what we explore here
Coding & Thematic Analysis
Methodology:
 - Coding and thematic analysis is how we can
Coding Qualitative Data: identify the relevant important vasribalws and how
What is it and why do we do it? they are connected

Main Steps: What is coding about?: starting point for qualitative

1. General Research Questions analysis, fragmenting our data into units for use in our
2. Selecting relevant site(s) and subjects analysis, codes come from you rtext, not from theory
3. Collection of relevant data
4. Interpretation of data Initial steps in coding: preparation
5. Conceptual and theoretical work = before coding: transcription, cleaning, saving, storing,
- 5a. Tighter specification of the research qs organizing
- 5b. Collection of further data
6. Writing up findings/ conclusions Transcription software: ie. happyscribe, trint, [Link]

Initial steps in coding:

What are qualitative data? Pieces of information gathered - Start process of coding as you go long
in a non-numeric form: - Read through transcripts, notes, documents:
- Written notes of observations, Transcripts of familiarize with data
interviews, Emails, documents, diary entries, - Read again, take notes/ observations and
recordings and write ups of focus groups, can be reflections
primary or secondary - Note everything relevant, keywords
- Generate an index of terms that will help you
Purpose of qualititaive analysis: finding conceptual interpret and theorise
mechanisms and meanings that explain a phenomenon - This could influence who you interview next, the
questions you ask , the documents you seek
You choose qualitative analysis to explore a Initial coding
phenomenon: ■ Start coding > identify ‘meaning units’
- You don’t know up front what are the relevant ■ Words, sentences, paragraphs
important variables to the hneomenn
 ■ Connected with your research question in some way. Thematic Analysis
You can code e.g.: events, properties, actions, meanings, = general, overarching framework for analysis
relationships qualitative data
■ any item of data can be coded in multiple ways
■ keep a code book
Purpose: search for themes or patterns that occur
across a dataset, helps with massive data sets, helps
identify discrepancies for a more nuanced approach
Example: in a study looking at why people choose to
leave their jobs as paramedics, initial codes might
Codes = “pithy labels identifying what is of interest in the
include:
data”
- Job expectation, aligned experiences, misaligned
Themes: “ a common recurring patterns across a dataset,
experiences, satisfaction, intent to leave,
clustered around organizing concept”
challenges
How does it fit with Research design choices?
- Use TA from an objectivist (positivist) or
Review your codes:
subjectivist (interpretivist) position = assumptions
● Combine, if describing the same thing
affect how you interpret data
● Do you codes resemble terms and concepts in the
- You can use TA for inductive or deductive
literature?
research
● Are there connections between codes?
Main steps:
1. Familiarize yourself with data
2. Create initial codes
3. Decide what to code; add new codes where
needed
4. Collate data with the same code
5. Group code into themes
6. Evaluate and revise themes
Finding Themes: some suggestions

- Repetition
- Indigenous typologies or categories (local
expressions)
- Metaphors and analogies
- Similarities and differences
- Theory related

The way to do it for coding: Goiai Method

Some specific qualitative analysis methods:

Grounded Theory = methodology to introduce more

rigour in showing “how do you know what you claim you
know” ( you’re being guided by the data and you have to
interpret it)
Instead of starting with a theory, start with gathering
information and then formulate a RQ
- Onus is on the researcher to be transparent in
analytical steps and provide rich contextual detail
- Relies on constant comparison between data and
theory
- Moves through increasing levels of abstraction
- 1sr order: participants interpretation
- 2nd order: researcher’s interpretation/ theoretic
explanation
- iterative
 The research “onion”: layers of research design

Week 6:Research Paradigms What are the dimensions across which research philosophies
differ? (outermost layer)
Ontology = assumptions about the nature of reality
How do we find out about the world? Epistemology = assumptions about how knowledge of reality is
- Personal experience produced and justified
- Wisdom of crowds (opinion) Axiology
- Scientific research (evidence) Ideological creation
- Following leaders (belief) Methodology

Research = process that is undertaken in a systematic way

with a clear purpose, to find things out

Why is it important that we get our research design right?

Why do we need to think about research philosophies?

To clarify research design
To recognise which research designs will work Ontology:
To ensure that we are being internally consistent with our - What is truth? Our assumption
questions, design, methods, analysis and conclusions - realism, internal realism, relativism, nominalism
To understand your role in research and conclusions spectrum
Epistemollogy:
Tree as a metaphor for the research process; - OUr assumptions about the best way of enquiring into
Trunk: research design: ontology, epistemology, axiology, the nature of the world and establish truth
methodology, methods and techniques - Theory of knowledge: methods - validity, scope
Roots: research traditions within different disciplines - distinction between justified belief and opinion
Fruit: research outputs - Positivism <> social constructivism (interpretivism)
Branches and leaves: data collection and analysis spectrum
Key philosophical assumptions: objectivism/subjectivism Rsearch philosophy and Methodology
Objectivism:
- Reality is external to social actors
- Social entities exist irrespective of how we label them/
think about them or if we are aware of them
- Research aims to find “laws” that govern behaviour

Subjectivism:
- Social reality made up of the perceptions and
consequent actions of social actors
- Reality constructed through social interaction

Differences in Quantitative and Qualitative Research

Designs

Relating research Philosophy to Research Design

Approaches to theory development

When is Qualititaive Reseasrch useful?
- To build new theory when prior theory is absent
- To capture individuals’ lived experiences and
interpretations
- To understand complex process issues
- To examine anrratives, discourse or other linguistic
phenomena

Challenges:
- Lack of structure guidance
- Messy
 Process:
1. Understand the problem
2. Data Collection
Week 7: Quantitative methods & 3. Data Cleaning
Data Analytics - Data Creation/ 4. Data exploration and analysis
5. Interpret the Results
Collection
Data Analytics
Big data and digital economy
- Growth of data
- Development of IT tools
- Skills of people and robots

= process of exploring and analyzing large datasets to

make predictions and help data-driven decision making
= process of collecting, cleaning and transforming data

GDPR (General Data Protection Regulation) 3 main

types of people:
- Data subject (creator of data)
- Data controller (manage and store)
- Data processor (process and analyse)

Types:
- Descriptive analytics
- Predictive Analytics
- Prescriptive Analytics
 - Overfitting (model too complex, too much data) or
underfitting (overly simplistic model)
Data Creation and Collection
Sampling Bias:
Data can be structured in different ways - ie. index, Cognitive Biases: j
tabular, tree etc Coverage Bias: Highlighting certain stories or facts while
ignoring others, like a news channel focusing only on
It can also be unstructured - ie. social media, celebrity news while overlooking significant global
discussions, call centre events.

BIased vs Unbiased Data: Concision Bias: Preferring short, simplified explanations

over longer, more complex ones, such as using sound
Fund performance: bites instead of in-depth analysis to discuss policy
- Self-selection (big group of students, ask how issues.
everyone did in exam, only ones who do well
answer and they self select), so probably you And for the biases mentioned in your image:
collected only good data
- Backfilling ( adding past data to records as if they Declinism: The belief that the past was better and the
were always there) present is a decline, like older generations claiming life
- Survivorship (only the ones that survived come up was better in their youth.
as data) bias
Framing Effect: Drawing different conclusions from the
Other common bias: same information based on how it is presented, like
- Confirmation (only using data that supports your perceiving a glass as half empty or half full.
previous opinion)bias
- Selection (choosing non-random data analysis, Dunning-Kruger Effect: The less a person knows, the
like surveying only college students to understand more confident they are in their abilities, like an amateur
all youth trends)bias
giving expert advice on complex topics.
Authority Bias: The tendency to trust and be influenced by
the opinions of authority figures, such as a celebrity
endorsing a product.

Groupthink: Conforming to a widely held belief to avoid

conflict, like agreeing with a group decision even if you
have doubts.

Availability Cascade: An idea gaining credibility as it

becomes more widespread, like a rumor being accepted
as truth because it's widely shared.

False Consensus: Overestimating how many people

share your beliefs, like assuming all your friends have the
same political views as you.

Summary and Key Messages:

Data analytics is essential in the big data and digital
economy.
Week 9 Data Analysis and Modelling

- Data sampling, simulation, hypothesis testing,

model evaluation
- Data analysis and modelling

1. Probability distributions
2. Simple and multiple regression
3. Measures of association
4. Principal components analysis
5. Classification
6. Clustering
7. Time series analysis
8. Machine learning

More advanced: optimization, agent-based modelling

Process:
Data Collection, preparation, exploration, modelling,
result interpetation