Education Qualitative Coding Analysis
Goal
Turn qualitative education data into a transparent coding process, evidence-backed themes, and a findings section.
Use After
Use after data sources and qualitative design are defined:
education-qualitative-study-design
education-mixed-methods-design
education-action-research-design
education-design-based-research
education-policy-comparative-analysis
education-program-evaluation
Do not expose the skill name to users. Present it as "质性材料编码分析" or "访谈/观察/文本分析".
Inputs
- Research questions
- Data type: interview, focus group, observation, document, policy, student work, AI feedback text, open-ended survey
- Corpus description and file list
- Chosen qualitative tradition if any: thematic analysis, grounded theory, case study, content analysis, discourse analysis
- Whether multiple coders are involved
- Language and anonymization requirements
Workflow
- Prepare corpus:
- transcribe audio/video
- clean obvious transcription errors
- anonymize names, schools, locations, accounts
- assign document and participant IDs
- Choose analysis approach:
- thematic analysis for patterns of meaning
- grounded theory for model/theory building
- content analysis for categories/frequency
- case analysis for bounded cases
- discourse-oriented analysis for language/interaction patterns
- Create initial coding unit:
- sentence
- paragraph
- speaking turn
- classroom episode
- document section
- artifact segment
- First-cycle coding:
- open coding
- descriptive coding
- in vivo coding
- process coding
- values/emotion coding where relevant
- Build codebook:
- code name
- definition
- inclusion/exclusion rules
- example quote
- Second-cycle coding:
- axial coding
- pattern coding
- focused coding
- category aggregation
- Theme/model development:
- theme names
- theme definitions
- relationships among themes
- negative cases/contradictions
- Write analytic memos throughout.
- Build evidence table linking claims to excerpts.
- Draft findings structure and discussion implications.
Tool Calls
Transcription
Whisper: https://github.com/openai/whisper
Otter: https://otter.ai/
讯飞听见: https://www.iflyrec.com/
Qualitative Coding Software
NVivo: https://lumivero.com/products/nvivo/
MAXQDA: https://www.maxqda.com/
ATLAS.ti: https://atlasti.com/
Dedoose: https://www.dedoose.com/
Taguette: https://www.taguette.org/
QualCoder: https://github.com/ccbogel/QualCoder
Lightweight Text Analysis
R:
install.packages(c("tidyverse", "tidytext", "quanteda", "readtext"))
Python:
pip install pandas nltk spacy scikit-learn
Use NLP only as support. Human interpretation remains required for qualitative analysis.
Output Format
1. Corpus Inventory
| Document ID |
Data Type |
Participant/Source ID |
Date |
Length |
Anonymized |
Notes |
2. Coding Plan
| Decision |
Choice |
| Analysis approach |
|
| Coding unit |
|
| First-cycle coding |
|
| Second-cycle coding |
|
| Software/tool |
|
| Multiple coders |
Yes/No |
| Reliability check |
Yes/No/Not applicable |
3. Codebook
| Code |
Definition |
Include When |
Exclude When |
Example Excerpt |
Notes |
4. Category / Theme Table
| Theme |
Related Codes |
Meaning |
Evidence Count |
Representative Excerpts |
Negative Cases |
5. Evidence Table
| Finding Claim |
Supporting Excerpt ID |
Excerpt Summary |
Data Source |
Strength |
Caveat |
6. Findings Section Outline
Finding 1: [Theme name]
- Main claim
- Representative evidence
- Variation across participants/cases
- Negative or boundary case
Finding 2: [Theme name]
...
Approach-Specific Guidance
Thematic Analysis
Use when the goal is to identify patterns of experience, perception, practice, or meaning.
Steps:
- Familiarize with data.
- Generate initial codes.
- Search for themes.
- Review themes against data.
- Define and name themes.
- Write findings with evidence.
Grounded Theory
Use when the goal is to build a process model or explanatory theory from data.
Steps:
- Open coding.
- Constant comparison.
- Axial coding.
- Theoretical sampling if possible.
- Selective coding.
- Theoretical saturation judgment.
- Model construction.
Output model:
Conditions -> Actions/Interactions -> Consequences
Content Analysis
Use when categories and frequencies matter, such as policy texts, textbooks, AI feedback comments, classroom discourse moves, or open-ended survey responses.
Steps:
- Define coding categories.
- Define coding unit.
- Code sample and refine categories.
- Code full corpus.
- Count and interpret category patterns.
- Combine frequency with qualitative examples.
Quality Rules
- Do not treat AI-generated codes as final findings.
- Every major theme needs multiple pieces of evidence unless it is explicitly a rare/negative case.
- Keep raw excerpt, code, category, and theme traceable.
- Distinguish participant claim from researcher interpretation.
- Include negative cases or contradictions when they matter.
- For classroom data, preserve context: lesson phase, activity, teacher/student role, task.
- For minor students, anonymize names and identifiable writing details.
User-Facing Closure
End with the next practical decision:
现在可以进入正式编码。你希望我先帮你做一版初始代码本,还是先根据 2-3 份材料做试编码,再修订代码本?
1---2name: education-qualitative-coding-analysis3category: 开发工具4description: Use when analyzing qualitative education data such as interviews, focus groups, classroom observations, reflective journals, student work, policy documents, teaching artifacts, AI feedback texts, and open-ended survey responses. Covers open coding, axial coding, selective coding, thematic analysis, grounded theory, content analysis, discourse-oriented coding, memo writing, evidence extraction, codebook development, and qualitative findings structure. This is a second-layer execution skill for qualitative, mixed-methods, action research, DBR, policy/comparative analysis, program evaluation, and AI-assisted education research.5metadata:6 source: https://skillmd.com/skills/ilog3/education-qualitative-coding-analysis7 short-description: Execute qualitative coding and thematic analysis for education data8---910# Education Qualitative Coding Analysis1112## Goal1314Turn qualitative education data into a transparent coding process, evidence-backed themes, and a findings section.1516## Use After1718Use after data sources and qualitative design are defined:1920- `education-qualitative-study-design`21- `education-mixed-methods-design`22- `education-action-research-design`23- `education-design-based-research`24- `education-policy-comparative-analysis`25- `education-program-evaluation`2627Do not expose the skill name to users. Present it as "质性材料编码分析" or "访谈/观察/文本分析".2829## Inputs3031- Research questions32- Data type: interview, focus group, observation, document, policy, student work, AI feedback text, open-ended survey33- Corpus description and file list34- Chosen qualitative tradition if any: thematic analysis, grounded theory, case study, content analysis, discourse analysis35- Whether multiple coders are involved36- Language and anonymization requirements3738## Workflow39401. Prepare corpus:41 - transcribe audio/video42 - clean obvious transcription errors43 - anonymize names, schools, locations, accounts44 - assign document and participant IDs452. Choose analysis approach:46 - thematic analysis for patterns of meaning47 - grounded theory for model/theory building48 - content analysis for categories/frequency49 - case analysis for bounded cases50 - discourse-oriented analysis for language/interaction patterns513. Create initial coding unit:52 - sentence53 - paragraph54 - speaking turn55 - classroom episode56 - document section57 - artifact segment584. First-cycle coding:59 - open coding60 - descriptive coding61 - in vivo coding62 - process coding63 - values/emotion coding where relevant645. Build codebook:65 - code name66 - definition67 - inclusion/exclusion rules68 - example quote696. Second-cycle coding:70 - axial coding71 - pattern coding72 - focused coding73 - category aggregation747. Theme/model development:75 - theme names76 - theme definitions77 - relationships among themes78 - negative cases/contradictions798. Write analytic memos throughout.809. Build evidence table linking claims to excerpts.8110. Draft findings structure and discussion implications.8283## Tool Calls8485### Transcription8687```text88Whisper: https://github.com/openai/whisper89Otter: https://otter.ai/90讯飞听见: https://www.iflyrec.com/91```9293### Qualitative Coding Software9495```text96NVivo: https://lumivero.com/products/nvivo/97MAXQDA: https://www.maxqda.com/98ATLAS.ti: https://atlasti.com/99Dedoose: https://www.dedoose.com/100Taguette: https://www.taguette.org/101QualCoder: https://github.com/ccbogel/QualCoder102```103104### Lightweight Text Analysis105106R:107108```r109install.packages(c("tidyverse", "tidytext", "quanteda", "readtext"))110```111112Python:113114```bash115pip install pandas nltk spacy scikit-learn116```117118Use NLP only as support. Human interpretation remains required for qualitative analysis.119120## Output Format121122### 1. Corpus Inventory123124| Document ID | Data Type | Participant/Source ID | Date | Length | Anonymized | Notes |125|---|---|---|---|---|---|---|126127### 2. Coding Plan128129| Decision | Choice |130|---|---|131| Analysis approach | |132| Coding unit | |133| First-cycle coding | |134| Second-cycle coding | |135| Software/tool | |136| Multiple coders | Yes/No |137| Reliability check | Yes/No/Not applicable |138139### 3. Codebook140141| Code | Definition | Include When | Exclude When | Example Excerpt | Notes |142|---|---|---|---|---|---|143144### 4. Category / Theme Table145146| Theme | Related Codes | Meaning | Evidence Count | Representative Excerpts | Negative Cases |147|---|---|---|---|---|---|148149### 5. Evidence Table150151| Finding Claim | Supporting Excerpt ID | Excerpt Summary | Data Source | Strength | Caveat |152|---|---|---|---|---|---|153154### 6. Findings Section Outline155156```text157Finding 1: [Theme name]158 - Main claim159 - Representative evidence160 - Variation across participants/cases161 - Negative or boundary case162163Finding 2: [Theme name]164...165```166167## Approach-Specific Guidance168169### Thematic Analysis170171Use when the goal is to identify patterns of experience, perception, practice, or meaning.172173Steps:1741751. Familiarize with data.1762. Generate initial codes.1773. Search for themes.1784. Review themes against data.1795. Define and name themes.1806. Write findings with evidence.181182### Grounded Theory183184Use when the goal is to build a process model or explanatory theory from data.185186Steps:1871881. Open coding.1892. Constant comparison.1903. Axial coding.1914. Theoretical sampling if possible.1925. Selective coding.1936. Theoretical saturation judgment.1947. Model construction.195196Output model:197198```text199Conditions -> Actions/Interactions -> Consequences200```201202### Content Analysis203204Use when categories and frequencies matter, such as policy texts, textbooks, AI feedback comments, classroom discourse moves, or open-ended survey responses.205206Steps:2072081. Define coding categories.2092. Define coding unit.2103. Code sample and refine categories.2114. Code full corpus.2125. Count and interpret category patterns.2136. Combine frequency with qualitative examples.214215## Quality Rules216217- Do not treat AI-generated codes as final findings.218- Every major theme needs multiple pieces of evidence unless it is explicitly a rare/negative case.219- Keep raw excerpt, code, category, and theme traceable.220- Distinguish participant claim from researcher interpretation.221- Include negative cases or contradictions when they matter.222- For classroom data, preserve context: lesson phase, activity, teacher/student role, task.223- For minor students, anonymize names and identifiable writing details.224225## User-Facing Closure226227End with the next practical decision:228229```text230现在可以进入正式编码。你希望我先帮你做一版初始代码本,还是先根据 2-3 份材料做试编码,再修订代码本?231```