What Can Qualitative Software Do for My Research? November 5, 2013 APHA Conference
Dec 25, 2015
Abt Associates | pg 2
Outline of Session
Overall Goal: Understand how qualitative data analysis software can improve the rigor of your public health research
Short introduction to qualitative analysis and computer-assisted qualitative data analysis software (CAQDAS)
Introduction to coding and coding exercise
NVivo demonstration and exploration
Abt Associates | pg 3
A Brief History of CAQDAS
1981: Lyn and Tom Richards develop NUD*IST, the precursor to NVivo
1994: Miles and Huberman discuss the use of software in qualitative analysis in their widely cited text
2007: National Science Foundation publishes guidelines for the use of software in qualitative data analysis
2013: NVivo, AtlasTi, EZ-Text, ANSWR, MaxQDA, HyperResearch and Dedoose are among the most commonly used tools today
Today at APHA 2013: Over 30 presentations mention using NVivo in their abstracts (see handout)
Abt Associates | pg 4
How can Software Help Improve the Rigor of Qualitative Analysis?
GATHER
Existing Research
Current Information
Data Sources
ORGANIZE
Data Sources
Coding Structure
Demographics
Literature Review
ANALYZECoding
Queries
Memos
Visuals
VISUALIZE RESULTSReports
Journal Articles
Dissertations
Study PresentationsX Y
Z
• Transparency• Saturation
Methodology
Abt Associates | pg 5
Considerations in Choosing to Use Software
Sample size and multiplicity of data sources
Emphasis on replicability, rigor and transparency
Likelihood that there will be future opportunities to perform secondary analyses on the same dataset
Desire to publish in peer-reviewed journals
Interest in merging close-ended attributes into the qualitative dataset
Building capacity of analysis team including training time and costs
Budgetary parameters and software investment
Abt Associates | pg 6
Promoting Reliability and Validity in Analysis Document the process of analysis including what is the
statement of a respondent and what is interpretation by a
researcher
Involve multiple analysts to check biases
Document in detail the process by which analytical themes or
codes are developed
Train coders or analysts on coding structure and create well-
defined themes. Refine again and again.
In analysis, check inter-rater reliability
Develop conventions for transcribing data so that transcripts
are comparable across data source
Develop saturation guidelines
Abt Associates | pg 7
Future Directions in Qualitative Analysis
Mixed methods tools, such as the capacity to work with datasets containing both fixed response and open-ended material
Web-based data, including social media and online discussion boards
Capacity for larger samples, especially large quantities of text (qualitative studies are no longer small)
More tools for comparing coding by researcher, theme, and participant group
Abt Associates | pg 8
A Future Look at Mixed Methods with NVivo
Descriptive Statistics Inferential Statistics Meta-Analysis
Coding of TextAudio, Video andImage Data Open-Ended Survey and Interview Responses Transcribing tools for Audio and Video
Import/Export from Excel, text and database files Open-Ended and Fixed Response Questions Within and Between Group Analysis of Coding Text Analysis Kappa Coefficient Cluster Analysis of Word and Coding Similarity
Qualitative QuantitativeNVivo
Abt Associates | pg 10
Brief Coding Exercise: Purpose
We will code a short transcript manually
– Given the short timeframe we will not do this thoroughly
We will talk about what we found and how we might begin to refine our codes
We will discuss how we would use qualitative software to accomplish the same steps
Abt Associates | pg 11
Coding
Any researcher who wishes to become proficient at doing qualitative analysis must learn to code well and easily. The excellence of the research rests in large part on the excellence of the coding.
(Anselm L. Strauss, Qualitative Analysis for Social Scientists, 1987, p. 27)
Abt Associates | pg 12
What is Coding?
Codes are short words or phrases that symbolize the essence of a piece of text, visual image, or other qualitative data.
Codes reduce a large quantity of data into more manageable “themes.”
Interpret qualitative data into meaningful themes (meaningful depends on the lens of the analyst)
Abt Associates | pg 13
EXAMPLE: Focus Group on Treating Chronic Fatigue Patients
Physician Participant
“There are also trends over times. When I was in training, everybody who we now consider chronic fatigue or even chronic fibromyalgia was largely looked into a group that they called the hypochondriacal patients. Now you hardly ever hear the diagnosis hypochondriasis anymore.”
Code Trends in Diagnoses OR Physician training OR Chronic fatigue syndrome
familiarity
Abt Associates | pg 14
Coding process
Initial codes will be defined, redefined, collapsed as more and more data are coded
Analysts will develop definitions, inclusion and exclusion criteria for each code
Abt Associates | pg 15
Analysis of Codes/Themes
Codes/themes are analyzed for patterns, e.g.
– frequency,
– similarity and differences across respondent types,
– meaning,
– sequence,
– associations with other codes,
– causation, etc.
Abt Associates | pg 16
Brief Coding Exercise - Instructions
Read the interview with Thomas – 10 minutes
Remember the purpose of the research. (Description in packet.)
Look for themes. Put parentheses around the text and write the word/phrase that summarizes the text next to it.
Note questions or ideas that occur to you as you read the interview.
Report out and discussion – 10 minutes
Abt Associates | pg 18
Brief Coding Exercise – Discussion
What codes are similar to each other?
How will we capture the changing definitions over time?
How will we merge codes?
How will we assess inter-rater reliability?
What questions or comments occurred to you as you read the interview?
Abt Associates | pg 19
Brief Coding Exercise – Summary
What works manually with a small amount of qualitative research becomes more complicated as the number of qualitative sources increase
Defining, redefining, merging, and separating codes is easier to do and easier to keep track of with qualitative software
Documenting the analysis process systematically is a benefit of the software; replicability is possible
Abt Associates | pg 21
Let’s explore NVivo in practice
How does NVivo store data? Sources may be text, audio, video, pictures, categorical or social media data
How do I code in NVivo? Select data and assign to one or more Nodes
How can I see my coding in NVivo? Turn on Coding Stripes, and open Nodes
Isn’t there more the software can do?
Text Search and Word Frequency Queries can help you code.
Matrix Coding Queries can reveal patterns within and across themes or groups of participants.
Abt Associates | pg 22
Systematic, Rigorous, Quick
Increase accessibility of data and transparency of analysis
o Node content, memos, annotations, coding stripes, event log
Identify and test ideas about emerging patterns and themes
o Text Search Query, Matrix Coding Query
Utilize open-ended text and non-text data
o Transcripts, audio, video, pictures, survey and other categorical data, social media
Generate output for reporting
o Query results, node exports, visualizations
Support team-based analysis
o View team members’ coding, calculate Kappa coefficient
Abt Associates | pg 23
Utilizing Non-text Data
Classification (First Cell)
Person Gender Age Range Years in Down East
Primary Residence
Barbara Female 40-49 40 Yes
Charles Male 60-69 72 Yes
Dorthy Female 20-39 40 Yes
Helen Male 70-79 N/A Yes
Attributes (First Row)
Nodes (First Column)
Abt Associates | pg 25
Contact Information
Lisa LeRoy, MBA, PhD
617-349-2723
Ilana Ron, MSc
301-347-5339
Cynthia Jacobs, EdD
617-491-1850