Influencing policy (training slides from Fast Track Impact)
The analysis of qualitative data 22nd Oct 2015
1. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
Qualitative Research II:
The Analysis of Qualitative Data
Dr Matt Maycock
MRC/CSO Social and Public Health Sciences Unit
23rd October 2015
2. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
Lecture Aim and Objectives
• Aim
To introduce students to the analysis of
qualitative data
• Objectives
By the end you will have an appreciation of:
1 The principles of analysing qualitative data
2 The Qualitative Analytical Process
3 Qualitative Data Management Tools (Nvivo)
3. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
PhD research far-west Nepal
Masculinity, Modernity and Bonded Labour: Continuity and Change
amongst the Kamaiya of Kailali District, far-west Nepal (School of
International Development, UEA, Norwich)
Background to my research…
Yearlong fieldwork in
Nepal:
Three month language
training and key
informant interviews
Nine months in two
fieldsites
4. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
PhD – Ethnographic Methods/ Analysis
• Household survey
• Life History interviews
• Participant observation
• Photographs
• Participatory Development Tools (such as resource mapping)
• Data analysis took as long as data collection
• I used Nvivo extensively to support and shape my data analysis
5. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
Post-doc - prison masculinities in
Scotland
Data collection
• Observation of
sessions
• Interviews with
participants who
completed programme
• Interviews with
participants who did
not complete
• Focus group with
prison staff
Ongoing Data Analysis
Grounded theory
I have used Nvivo
extensively
Coding framework developed
collaboratively
6. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
• Non-numerical – converse of quantitative data
• Typically word based – but may include
imagery, video, etc.
• Can record attitudes, behaviours, experiences,
motivations, etc.
• Descriptive – describing events/opinions etc.
• Explanatory – explaining events/opinions etc.
What is Qualitative Data?
7. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
Examples of Qualitative Data Sources
• Interviews
• Focus groups
• Speeches
• Questionnaires
• Journals/diaries
• Documents
• Observation
• Audio/visual
materials
• Websites
• Social media
8. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
Identify
similarities
Extract themes
Identify
relationships
Highlight
differences
Create
generalisations
(?)
Analysing Qualitative Data
9. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
To draw conclusions
To develop theories
To test hypotheses
Objectives of Qualitative Analysis
10. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
Principles of Analysing Qualitative Data
1 Proceed systematically and rigorously
(minimise human error)
2 Record process, memos, journals, etc.
3 Focus on responding to research questions
4 Appropriate level of interpretation appropriate
for situation
5 Time (process of inquiry and analysis are often
simultaneous)
6 Seek to explain or enlighten
7 Evolutionary/emerging
11. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
Qualitative Research:
Common Features of Analytic Methods
(Miles and Huberman, 1994)
1 Affixing codes to a set of field notes drawn from data
collection
2 Noting reflections or other remarks in margin
3 Sorting or shifting through the materials to identify
similar phrases, relationships between themes, distinct
differences between subgroups and common sequences
4 Isolating patterns and processes, commonalties and
differences, and taking them out to the filed in the next
wave of data collection
5 Gradually elaborating a small set of generalisations that
cover the consistencies discerned in the data base
6 Confronting those generalisations with a formalised
body of knowledge in the from of constructs or theories
12. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
The Analysis Continuum
Raw Data
Descriptive
Statements Interpretation
13. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
The Credibility of Qualitative Analysis
The credibility for qualitative inquiry depends on three
distinct but related inquiry elements:
1 Rigorous techniques and methods for gathering high-
quality data that is carefully analysed, with attention to
issues of validity, reliability, and triangulation
2 The credibility of the researcher, which is dependent on
training, experience, track record, status, and
presentation of self
3 Philosophical belief in the phenomenological paradigm,
that is, a fundamental appreciation of naturalistic
inquiry, qualitative methods, inductive analysis and
holistic thinking
14. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
Data Analysis During Collection
• Analysis often not left until the end
• To avoid collecting data that are not important the
researcher must ask:
• How am I going to make sense of this data?
• As they collect data the researcher must ask
• Why do the participants act as they do?
• What does this focus mean?
• What else do I want to know?
• What new ideas have emerged?
• Is this new information?
15. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
Data Analysis After Collection
• One way is to follow three iterative steps
1. Become familiar with the data through
1. Reading
2. Memoing
2. Exam the data in depth to provide detailed
descriptions of the setting, participants, and
activities.
3. Categorizing and coding pieces of data and
grouping them into themes.
16. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
Data Analysis Strategies
• Identifying themes
• Begin with big picture and list “themes” that emerge.
• Events that keep repeating themselves
• Coding qualitative data
• Reduce data to a manageable form
• Often done by writing notes on note cards or in
coding software (such as Nvivo) and sorting into
themes.
• Predetermined categories (A priori) vs.
emerging categories (In vivo)
17. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
A common approach for analyzing qualitative data
is called content analysis. It involves 5 steps:
1. Get to know your data
2. Focus the analysis on your research
questions
3. Categorize the information
• Identify themes or pattern
• Organize them into coherent categories
4. Identify patterns and connections within
and between categories
5. Interpretation – bring it all together
18. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
What is the Process of Data Analysis?
Codes the Text for
Description to be Used
in the Research Report/ thesis
Codes the Text for
Themes to be Used
in the Research Report / thesis
The Researcher Codes the Data (i.e., locates text
segments and assigns a code to label them)
The Researcher Prepares Data for analysis
( e.g., transcribes fieldnotes)
The Researcher Collects Data (i.e., a text file, such as
fieldnotes, transcriptions, optically scanned material)
The Researcher Reads Through Data
( i.e., obtains general sense of material)
SimultaneousInteractive
19. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
Step 1. Get to know your data. Good qualitative data analysis
depends upon understanding your data. Spend time getting to “know” your data.
• Read and re-read the text
• Listen to tape recordings if you have them; transcribe data
• Check the quality of the data. Is it complete and
understandable. It it likely to add meaning and value? Was it
collected in an unbiased way?
20. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
Step 2. Focus the analysis
• Review the purpose of the evaluation and what you wanted to
find out (keep referring to your research questions).
• Based on your ‘getting to know your data’, think about a few
questions that you want your analysis to answer and write them
down.
• You might focus your analysis by question, topic, time period,
event, individual or group.
21. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
Step 3. Categorize information
Some people call this process ‘coding’ the data.
It involves reading the data and giving labels or codes to the
themes and ideas that you find.
You may have themes or ideas you search for (pre-set categories)
and/or create categories (emergent categories) as you work with
the data.
I will discuss using Nvivo to do this in more detail shortly
22. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
How do You Prepare and Organize the Data?
• Develop a matrix or table of sources that can be used to
organize the material
• Organize material by type
• Keep duplicate copies of materials
• Transcribe data
• Prepare data for hand (if you don’t like computers) or
computer analysis (and select computer program)
23. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
A Visual Model of the Coding Process in
Qualitative Research
Reduce Codes to
5-7 Themes
Initially read
through data
Divide text
into segments
of information
Label
segments of
information
with codes
Reduce
Overlap and
redundancy
of codes
Collapse
codes into
themes
Many
Pages
of Text
Many
Segments
of Text
30-40
codes
Codes
reduced
to 20
24. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
Approaches to coding
• A priori codes are developed before
examining the data
• In vivo codes are derived from the data
• Co-occurring codes partially or completely
overlap
25. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
In vivo codes
• Treat data as answers to open-ended questions
• ask data specific questions
• assign codes for answers
• record theoretical notes
Strauss and Corbin, 1998, Ron Wardell, EVDS 617 course notes
26. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
A priori codes
• Categories are created ahead of time
• from existing literature
• from previous open coding
• Code the data just like open coding
Ron Wardell, EVDS 617 course notes
27. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
Step 4. Identify patterns within and between
categories
• Once you have identified the categories, you might:
• Sort and assemble all data by theme
• Sort and assemble data into larger categories
• Count the number of times certain themes arise to show
relative importance (not suitable for statistical analysis)
• Show relationships among categories
28. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
Working with others (key stakeholders, other program staff,
participants) in the coding and interpretation process is helpful.
For example, several people might review the data independently
to identify categories. Then, you can compare categories and
resolve any discrepancies.
How else might you involve others in your qualitative data
analysis?
Collaboration with coding
29. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
Step 5. Interpretation
• Stand back and think about what you’ve learned. What
do these categories and patterns mean? What is really
important
• What did you learn?
• Interpretation is not neutral
• Reflect about the personal meaning of the data
• Compare and contrast personal viewpoints with the
literature
• Address limitations of the study
• Make suggestions for future research
30. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
How do You Validate the Accuracy of Your
Findings?
• Member/participant checking: Members/participants
check the accuracy of the account
• Triangulation: Using corroborating evidence
• Triangulation involves gathering data on the same
theme from a variety of sources. Mixed methods
approaches are useful here. Triangulation can be
useful in data analysis whether or not there are
correspondences or discrepancies.
• External: Hiring the services of an individual outside
the study to review the study. Or you can ask a friend…
31. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
Qualitative analysis is not as easy as it
looks…
• Qualitative analysis is ‘as much a test of the enquirer as it
is a test of the data:
• first and foremost, analysis is a test of the...ability to think
- to process information in a meaningful and useful
manner...qualitative analysis remains much closer to
codified common-sense than the complexities of statistical
analysis of quantitative data’ (Robson 1995: 374).
• As a result it is important to recognise some common
failings...
32. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
Common failings part one…
(Robson, 1995: 374-5)
• data overload - too much
to process and remember
• information availability -
information which is
difficult to get hold of
(wrongly) gets less
attention
• positive instances - a
tendency to ignore
evidence which conflicts
with hypotheses
• uneven reliability - that
some sources are more
reliable than others
sometimes gets ignored
• fictional base - tendency
to compare with a base or
average when no base
data is available
• confidence in judgement -
excessive confidence once
a judgement is made
• co-occurrence -
interpreted as strong
evidence of correlation
• inconsistency - repeated
evaluations of the same
data which differ
33. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
Common failings part two…
• Listing all narrative comments without doing any analysis
• Including information that makes it possible to identify
the respondent.
• Generalizing from comments to the whole group.
Qualitative information seeks to provide unique insights,
understanding and explanation – it is not to be
generalized.
• Using quotes to provide a positive spin. Consider your
purpose for including quotes.
• If you have a large amount of data analysis is
challenging
• No standard processes for coding or extracting themes
• Time constraints
34. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
Practical exercise
Read the text (which is an extract from my fieldnotes from
my post-doc research) on the following page and assign
each line of the text to one of the codes below:
• CODE A - The prison context
• CODE B - Researcher reflections
• CODE C – Health in prison
35. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
1. The walk from the prison entrance to the prison gym passes through
2. old and new parts of the prison, fenced outside spaces, gardens and
3. various noticeably non-descript rooms, but does not include direct
4. passage through any of the prison accommodation areas (halls). The
5. sound of rattling keys is the soundtrack to this walk. There are a
6. various motivational messages located on the walk up the two flights
7. of stairs to get to the prison gym, including: “we live longer healthier
8. lives”. Going through the nine barriers and gates to get to the gym
9. becomes a kind of routine through which one becomes enveloped
10. within the prison, it encloses around you.
“Entering the gym” (Field note extract, December
2014, Session one, Prison B)
CODE A - The prison context
CODE B - Researcher
reflections
CODE C – Health in prison
36. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
1. The walk from the prison entrance to the prison gym passes through
2. old and new parts of the prison, fenced outside spaces, gardens and
3. various noticeably non-descript rooms, but does not include direct
4. passage through any of the prison accommodation areas (halls). The
5. sound of rattling keys is the soundtrack to this walk. There are a
6. various motivational messages located on the walk up the two flights
7. of stairs to get to the prison gym, including: “we live longer healthier
8. lives”. Going through the nine barriers and gates to get to the gym
9. becomes a kind of routine through which one becomes enveloped
10. within the prison, it encloses around you.
“Entering the gym” (Field note extract, December
2014, Session one, Prison B)
CODE A - The prison context
CODE B - Researcher
reflections
CODE C – Health in prison
37. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
1. The walk from the prison entrance to the prison gym passes through
2. old and new parts of the prison, fenced outside spaces, gardens and
3. various noticeably non-descript rooms, but does not include direct
4. passage through any of the prison accommodation areas (halls). The
5. sound of rattling keys is the soundtrack to this walk. There are a
6. various motivational messages located on the walk up the two flights
7. of stairs to get to the prison gym, including: “we live longer healthier
8. lives”. Going through the nine barriers and gates to get to the gym
9. becomes a kind of routine through which one becomes enveloped
10. within the prison, it encloses around you.
“Entering the gym” (Field note extract, December
2014, Session one, Prison B)
CODE A - The prison context
CODE B - Researcher
reflections
CODE C – Health in prison
38. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
Qualitative Data Management Tools
• Enables efficient data management by supporting the
processes of indexing, searching and hence data theorising
• Creates an environment to store and explore data and ideas,
it does not determine the research approach.
• The major advantage of the package is that it enables an
efficient and flexible approach to rigorously and
systematically analysing qualitative data.
• Such tools include, Nvivo, ATLAS.ti, MAXQDA
• You can buy Nvivo for £30 from IT, or it is installed on some
computers around the campus
40. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
• Divide data into meaningful units
• Use words/phrases e.g. ‘physical environment’,
‘interpersonal relationships’
• Codes can be ‘data-driven’ or ‘theory-
driven’
• A priori codes are developed before examining the
data
• In vivo codes are derived from the data
• Co-occurring codes partially or completely overlap
• In NVivo, codes are stored within Nodes
• Keep a master list of codes used
Coding Data
41. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
Types of Code
42. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
Types of Code
• This took place at Head
Office
• This is about
discrimination against
women
• This is a reflection on
misogyny in the
workplace
Analytic
Descriptive
Thematic
43. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
Interpersonal
relationships
Family
Parent-child
Spouse/
partner
Sibling
Non-family
Friends
Work
colleagues
Tree (Hierarchical) Coding
44. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
• Create database for storing different
sources (text, audio, video, web
resources, etc.)
Manage data
• Annotate data
• Attach memos to filesManage ideas
• Identify commonly occurring words
• Collate data relating to a theme or
concept
Query data
• Illustrate relationships using models
• Report knowledge developed from
data
Model & report
How can NVivo Help?
45. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
Qualitative Analysis Using Nvivo
Import Code
Query &
Visualise
AnnotateSummarise
47. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
• Source – your data
• Documents, audio, video, images, etc.
• Memo – item in a project linked to a document or
node
• Node – a code or concept (theme node) or a
component of your project e.g. participant or location
(case node)
• Can be free or tree
• Classification – applied to a case or participant e.g.
person, organisation, etc.
• Attributes – data (demographics) known about a
case (participant) recorded separately from the case
Terminology
48. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
• Descriptive code
• Classification/attributeWhat is this?
• Thematic code
• Annotation/memo
Why is this
interesting?
• Analytic code
• Memo
Why is this relevant
to my research
question?
Coding in NVivo
49. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
Descriptive coding
•Research design, project outline
•Folders, templates, case nodes
Thematic coding
•Finding obvious themes,
autocoding
•In Vivo coding
Analytic coding
•Creating node hierarchies
•Using queries, matrices
Developing an NVivo Project
Structuringphase
Creative/analyticphase
Optionalanalyticiterations
Source: Edhlund, B & Mcdougall, A (2013), NVivo 10 Essentials, p. 14
50. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
• Install NCapture for Internet Explorer/Chrome
• Capture content from:
• Web pages
• Online PDFs
• Facebook
• Twitter
• LinkedIn
• YouTube
• http://help-
ncapture.qsrinternational.com/desktop/welcome/welcome.htm
• http://nsmnss.blogspot.co.uk/2014/08/7-ways-nvivo-helps-
researchers-handle.html
Web Pages and Social Media Data
51. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
• Code data at multiple nodes
• Use descriptive, thematic and analytic codes
• Keep a record of your codes and the themes that
evolve
• Use a Word Frequency query to help you identify key
phrases
• Use Text Search queries to help you explore themes
• Take time to reflect on what you have found and
record ideas using memos
• Keep a journal of your analysis process
Tips for using Nvivo
52. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
• Complex package that can take time to
learn
• Can distance researcher from their data
• Researcher can get caught in ‘coding
trap’
• Can identify references to phrases but
cannot discern different contexts
• Will not compensate for poor data or
weak interpretive skills!
Nvivo - Limitations
53. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
• NVivo Toolkit:
http://explore.qsrinternational.com/nvivo-toolkit
• Getting Started Guide:
http://download.qsrinternational.com/Document/NVivo1
0/NVivo10-Getting-Started-Guide.pdf
• QSR website:
http://www.qsrinternational.com/support.aspx
• QSR Support - @QSRSup - on Twitter:
https://twitter.com/QSRSup
• QSR on Facebook:
http://www.facebook.com/qsrinternational
• QSR on YouTube:
https://www.youtube.com/user/QSRInternational
Online Resources
54. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
• Qualitative Data Analysis with Nvivo
• Bazeley, P & Jackson, K (2013)
• NVivo 10 Essentials
• Edhlund, B & Mcdougall, A (2013)
• Using QSR‐NVivo to facilitate the development of a grounded
theory project: an account of a worked example
• Andrew John Hutchison, Lynne Halley Johnston, Jeff David Breckon
• International Journal of Social Research Methodology
• Vol. 13, Iss. 4, 2010
• Using NVivo to Answer the Challenges of Qualitative Research in
Professional Communication: Benefits and Best Practices Tutorial
• Hoover, R.S.; Koerber, A.L.,
• Professional Communication, IEEE Transactions on
• Vol.54, no.1, pp.68,82, March 2011
doi: 10.1109/TPC.2009.2036896
URL: http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=533
7919&isnumber=5718246
Literature
55. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
Conclusions
• Admit limitations in data collection
• strength, not weakness
• contributes to the research validity
• demonstrates your reflexivity as researcher
• Use qualitative data analysis software (such as Nvivo) as a
tool to support your analysis. Such software is only as good
as your analytical approach.
• Finally ...Good qualitative research depends upon good
analysis not just description.
• Your analysis will only be as good as your data
56. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
• Bazeley, P .Jackson, K (2014) Qualitative data analysis with Nvivo. London: SAGE
• Bazeley, P (2013) Qualitative Data Analysis: Practical Strategies. London: SAGE
• Bryman, A. (2012). Social research methods (2nd ed.). Oxford: Oxford University
Press. Chapter 24
• Fairclough, N. (2003). Analysing discourse. Textual analysis for social research.
London: Routledge.
• Grbich, C. (2012). Qualitative data analysis: An introduction. London: Sage.
• Madison, D S (2011) Critical Ethnography: Method, Ethics, and Performance.
London: SAGE Chapters 8 & 9
• Miles, M (2013) Qualitative Data Analysis: A Methods Sourcebook. London: SAGE
• Ritchie, J. Lewis, J (2013) Qualitative Research Practice: A Guide for Social
Science Students and Researchers. London: Sage. Chapters 10 & 11
• Robson, C (2011) Real World Research (third edition). London: John Wiley & Sons.
Part V
• Saldana, J (2012) The Coding Manual for Qualitative Researchers. London: SAGE
• Silverman, D. (2015). Interpreting qualitative data: Methods for analysing talk,
text and interaction. London: Sage.
Further reading…
57. MRC/CSO Social and Public Health Sciences Unit, University of Glasgow.
matthew.maycock@glasgow.ac.uk
www.matthewmaycock.com
Contact details…