# Qualitative Data Analysis Psychology I · Statistics · https://tryals.app/learn/psychology-i/qualitative-data-analysis ## Finding Patterns in Words Analyzing qualitative data like interview transcripts requires systematic methods distinct from numbers. **Thematic analysis** (Braun and Clarke) proceeds through structured stages: familiarizing with data, generating codes, searching for themes, refining them, naming them, and writing up with quotes. **Coding** marks meaningful text segments. **Open coding** applies labels with minimal preconceived categories, staying close to the text. **Axial coding** connects those codes to build an integrated structure. | Concept | What it involves | |---|---| | Open coding | Labeling data without preconceived categories | | Axial coding | Finding relationships between codes | | Theme | A broader pattern uniting related codes | ## Theory and Rigor **Grounded theory** (Glaser and Strauss) builds theory **inductively**, letting it emerge **from** the data during coding rather than testing it deductively beforehand. This suits novel, under-studied phenomena. Qualitative research checks **inter-rater reliability** by having multiple independent researchers code the same material, guarding against idiosyncratic bias. **Common pitfall:** Assuming qualitative coding is a purely subjective free-for-all. Rigor is maintained through explicit coding schemes, inter-rater checks, and actively searching for contradictory evidence. ## Practice questions 7 of this lesson's 11 practice questions, with answers. The full set is in the app. ### 1. Two researchers independently code 40 transcript segments and agree on 34. Why is the resulting 85% agreement figure a weaker guarantee than it appears? A. Because agreement between two coders says nothing about reliability, which requires at least three. B. Because percentage agreement measures validity rather than reliability, and the two are unrelated. C. Because 85% is below the accepted threshold, and only figures above 95% indicate reliable coding. D. Because some agreement would occur by chance alone. **Answer:** D. Because some agreement would occur by chance alone. **Why:** Raw agreement has a floor that has nothing to do with skill: two coders assigning at random over two categories agree about half the time, so 85% is 85 against a baseline of 50. Kappa asks how much of the room above chance was actually captured, which is why the same 85% can be excellent or mediocre depending on the coding scheme's structure. Page: https://tryals.app/practice/psychology-i/qualitative-data-analysis/two-researchers-independently-code-40-transcript-segments-and-agree ### 2. Inductive coding builds categories up from the data; deductive coding applies a framework decided in advance. Why is neither simply the better method? A. Because deductive coding is objective and inductive coding is subjective, so the choice depends on how much rigour a journal demands. B. Because inductive coding is more rigorous but too slow for most projects, so deductive coding is a practical compromise. C. Because inductive coding suits large datasets and deductive coding suits small ones. D. Because they answer different questions: discovery versus test. **Answer:** D. Because they answer different questions: discovery versus test. **Why:** The trade is between discovery and test, and it is the exploratory/confirmatory distinction in another form. Neither label survives inspection: a pre-specified framework embodies theoretical choices rather than objectivity, and inductive coding is held to account by reliability checks, audit trails and negative-case analysis. Most projects run both, in sequence. Page: https://tryals.app/practice/psychology-i/qualitative-data-analysis/inductive-coding-builds-categories-up-from-the-data-deductive-coding ### 3. Put the stages of thematic analysis (Braun and Clarke) in order. **Answer:** 1. Familiarize yourself with the data. 2. Generate initial codes, labeling meaningful segments. 3. Search for broader themes among the codes. 4. Review and refine the candidate themes. 5. Define, name, and write up each theme with illustrative quotes. **Why:** Familiarize, code, search for themes, review them, then define and write up. Coding the fine-grained segments precedes grouping them into broad themes - building the analysis from the ground up. Page: https://tryals.app/practice/psychology-i/qualitative-data-analysis/put-the-stages-of-thematic-analysis-braun-and-clarke-in-order ### 4. Sort each activity under the coding approach it belongs to. **Answer:** - Open coding: Labeling data with minimal preconceived categories, Staying close to what is actually present in the text - Axial coding: Looking for relationships and connections between existing codes, Building toward a more integrated structure among codes **Why:** Open coding generates labels directly from the data with few preconceptions; axial coding then finds relationships among those codes, building toward an integrated structure. Open makes the pieces; axial connects them. Page: https://tryals.app/practice/psychology-i/qualitative-data-analysis/sort-each-activity-under-the-coding-approach-it-belongs-to ### 5. Match each concrete analyst action to the concept it illustrates. **Answer:** - You mark a transcript segment with the short label 'fear of judgment' → Coding - You group several related codes under 'social anxiety' → Theme - Two researchers code the same interviews and compare their labels → Inter-rater reliability - You let the theory grow out of the data instead of testing a preset one → Grounded theory **Why:** Labeling a segment is coding, grouping related codes is a theme, comparing independent coders is inter-rater reliability, and letting theory emerge from data is grounded theory. Each action pins down one concept. Page: https://tryals.app/practice/psychology-i/qualitative-data-analysis/match-each-concrete-analyst-action-to-the-concept-it-illustrates ### 6. Two coders agree on 90% of segments, but nearly every segment belongs to one dominant category. Why does the raw agreement flatter them? A. Because chance agreement is high when one category dominates B. Because 90% is below the accepted threshold in any coding scheme C. Because agreement should be computed only on the rare categories D. Because two coders are insufficient to establish reliability **Answer:** A. Because chance agreement is high when one category dominates **Why:** The baseline moves with the category distribution: if 85% of segments are one label, blind agreement is already 85% and the coders have added five points. Cohen's kappa rescales against that floor, which is why raw percentage agreement is uninterpretable without knowing the scheme's balance. Page: https://tryals.app/practice/psychology-i/qualitative-data-analysis/two-coders-agree-on-90-of-segments-but-nearly-every-segment-belongs ### 7. Select every statement that is TRUE about qualitative data analysis. A. Deductive coding applies a pre-existing framework to the data. B. Qualitative analysis requires no attention to reliability. C. Cohen's kappa corrects agreement figures for chance. D. Theory built from the data upward, without a framework decided in advance, is the grounded-theory approach. E. Thematic analysis moves from familiarisation through coding to defining themes. **Answer:** B. Qualitative analysis requires no attention to reliability.; D. Theory built from the data upward, without a framework decided in advance, is the grounded-theory approach.; E. Thematic analysis moves from familiarisation through coding to defining themes. **Why:** Axial coding's linking role, grounded theory's inductive logic, and inter-rater reliability are all true. Qualitative analysis does have systematic procedures, and a theme is a broad pattern uniting many codes, not a single labeled segment. Page: https://tryals.app/practice/psychology-i/qualitative-data-analysis/select-every-statement-that-is-true-about-qualitative-data-analysis