Thematic Analysis Flashcards
When to use thematic analysis?
Identifying a pattern
What are different types of qualitative analysis?
Interpretive phenomenological analysis- exploring experience of a phenomenon
Grounded theory- developing a new theory from the ground up (based on PPTS data)
Discourse analysis- how language is used
What are the two overreaching types of Thematic analysis?
Deductive and inductive
Characteristics of deductive thematic analysis
Deductive- we are using an existing theory to inform our data analysis
Focused question- code and group answers from different perspectives - explanation 1 2 and 3
Characteristics of inductive thematic analysis
Inductive- we analyse the data as is, no theory/ framework exists to code the data (bottom up)
Vast amount of data from multiple sources general question of the phenomenon - conceptual analysis by coding categorisation theming- overreaching theme theory, model
Deductive TA example
Behavioural, normative and control beliefs about earthquake preparedness- Earthquake preparedness in Tehran is low due to beliefs of Tehran inhibitions about ep- aimed at elicit beliefs about earthquake preparedness among citizens
Theory of planned behaviour
1- beliefs about the likely outcomes of the behaviour and the evaluation of these outcomes (positive or negative thoughts about the behaviour)
2- belief about the normative explanations ( how are people in your group behaving)
3- belief about the presence of factors that may facilitate or impede performance of the behaviour and the perceived power of these factors (what can prevent or facilitate me carrying out the behaviour )
Thematic analysis steps (Braun and Clarke 2006)
Braun and Clarke 2006
Transcription, familiarisation with data, coding, searching for themes, reviewing themes, defining and naming themes, writing- finalising analysis.
Stage 1- transcription
- Data analysis can only begin once all data has been collected in SPSS (in qualitative it isn’t essential to have all data before analysis)
- there is no distinct separation between the end of data collection and starting the analysis (in quant, you need all numerical data before analysis)
- you can begin transcribing and analysing before all interviews/ focus groups have been conducted
- can collect part of data review, refine and reorient subsequent data collection
Stage 2- data familiarisation
- immerse yourself in the data, become familiar, notice things that might be relevent to your research question
- repeated reading/listening of data - gain general impression, note any thoughts in a research journal (notes help with data analysis process)- memory aid that triggers when developing analysis later
- notes reflect what we bring to data- highlighting importance of reflective practice
- not passive process- reading data actively, analytically and critically
Stage 3 - coding- What is it?
- identifying aspects of the data that relate to research question
Essentially labeling important features of the transcript
Stage 3- selective coding
Identifying instances of the phenomenon that you’re interested in and selecting these out
- example of data reduction- bowl of m and ms - picking out the blue ones (certain type)
Stage 3- Complete coding
Identify anything and everything of interest or relevance to answering your research question within your entire dataset- becomes more selective later in the analytic process
Stage 3- code
Code can be a word of brief phrase that captures the essence of why you think a particular bit of data might be useful in relation to research question
- not exclusive process- coded in multiple ways as fits purpose
- code whatever extracts are relevant- may be large chunk of text, or a few words
- if you have a broad research question- code wisely and comprehensively then refine later
- specific question- maybe large chunks of text where nothing needs to be coded
Types of code- semantic (data derived)
-Reflect semantic content of data
- succinct summary of the explicit content of data
- mirror participants language and concepts
- explicit- themes are identified within surface meanings of the data and the analyst is not looking for anything beyond what the participant has said or what has been written
Types of code- Latent (researcher derived)
- conceptual or theoretical interpretations of the data
- codes go beyond the explicit content of the data (read between the lines)
- invoke researchers conceptual and theoretical frameworks to identify implicit meanings in the data- never expresses a particular sentiment
- interpretive- identifies or examines the underlying ideas, assumptions and conceptualisations and ideologies theorised as shaping or informing the semantic content of the data
What does the coding process look like?
- by hand/Microsoft word/ Nvivo software
- codes should be conscice as possible but captures essence of what is about the data that interests you
- Process same way for rest of data- are existing codes applicable? Do you need to make a new one?
Group 4 - developing candidate themes
- review codes to help identify patterns across data set
- pattern analysis presumes ideas which occur across data set capture something socially or psychologically meaningful
- focus on capturing most meaningful pattern
(Saliency analysis- Buetow 2010- something in the data can be important without appearing frequently - theme captures something important about data in relation to dataset and goes some way to answer research questions
Stage 4- Important note on language- themes do not emerge
- articles ‘ 4 themes emerged from data’ - statements suggest analysis is passive where you identify something that already exist
- however TA is active- actively make choices about how to shape and craft raw data
Stage 5- develop candidate themes
- review all codes and the data related to each code. Aim is to identify and overlap between codes
- some codes, if they are large, rich and complex enough can be promoted to themes ( Charmaz, 2006)
- want to identify themes that capture most salient pattens in data, relevent to answering research question
Important things to remember at stage 5
- candidate themes are provisional and will be revised and refined through developing analysis
- themes are not identified in a quantitative fashion- they need to be identified in quantitative fashion- they need to be identified across a proportion of the dataset
- themes dont have to cover everything - just address research question
- be prepared to let go of themes if they don’t fit, or use miscellaneous category
- Ta is selective- telling a particular story that answers research questions and not represent everything that was said in the data
Stage 5- how to ensure you have developed a good theme
- consider each theme on its own and in relation to other themes
- themes should be distinctive but still fit with others to form overall analysis
- 3 levels of themes (don’t need to have a all 3)
Overreaching themes (capture and idea encapsulated in several themes
Themes
Sub themes (develop a certain aspect of a theme further)
Stage 5- how many themes should I have?
- no magic number
- bigger dataset = likely to find more themes= but they have to be good quality so not always the case
- one reported analysis will not tell you everything about data, analyse multiple times to explore different strands
Stage 5- thematic maps
-See the relationship between themes
- Explore relationships and refine