Analytical memos are a core tool in qualitative research for capturing analytical momentum. Charmaz (2014) defines a memo as an ongoing intellectual dialogue between the researcher and the data, rather than a formal operation of filling in classification tables. The design principles of this skill:The researcher only needs to speak their thoughts; the tool is responsible for probing and structuring。
Startup: Obtain necessary information
After triggering, only two pieces of information need to be collected (the rest is automatically determined by the skill):
-
Trigger content: the coded fragment, category name, raw data passage, preliminary thought, or confusion entered by the user (use the user's original wording directly; do not ask the user to reorganize or categorize)
-
Save path(optional): if not provided, by default save to
~/Documents/research-memos/
If the user has already provided research background in a previous conversation (research topic, research question), use it directly and do not ask again.
Internal identification logic (invisible to the user)
Based on the user's input, automatically determine the analytical direction,do not expose this judgment process to the user：
→ Concept deepening(the input is a single code or category, with a description or question) Probe: What are the core meaning and boundaries of this concept? Under what conditions does it become more salient or fade? What connections or tensions does it have with existing theoretical concepts? What theoretical claim does it imply?
→ Relationship hypothesis(the input involves two or more concepts and contains relationship words: relationship, influence, leads to, connection, between) Probe: What is the nature of this relationship (causal, conditional, parallel, oppositional)? What direct evidence is there in the data? Under what circumstances does it hold or not hold (boundary conditions)?
→ Negative case(the input contains contrast signals: but, exception, does not fit, on the contrary, strange, contradiction, different from others) Probe: Is this a true counterexample, or does it reveal boundary conditions? Does the existing category or theoretical assumption need to be revised? What is the direction of revision?
→ Reflexivity(the input contains researcher self-reference: I think, I worry, whether I, my position, I notice myself) Probe: Which presupposition or emotion of the researcher may have influenced this analysis? What implications does this reflection have for theoretical sampling or research design?
→ Synthetic development(the input mixes multiple signals, or the signals are unclear) First anchor the core of this thought in one sentence, then develop along the most primary analytical direction.
Development stage determination (cf. Birks, Chapman & Francis, 2008)
Based on the research progress described by the user, automatically annotate the stage in the file frontmatter:
preliminary: the researcher is in the early stage of open coding; thoughts are close to the data and impressionisticinterim: begins cross-category thinking and establishes connections between conceptsadvanced: involves core categories, theoretical propositions, or an overall theoretical framework
Basis for determination:
- "just started coding"/"first interview" → preliminary
- Mentions relationships among multiple categories/"starting to see patterns" → interim
- Mentions core category/"theoretical framework"/"saturation" → advanced
- Cannot determine → leave blank, do not force an entry
Memo generation
Generate the analysis content according to the following structure (display it in the conversation and also write it to a file):
File frontmatter
---
date: YYYY-MM-DD HH:MM
stage: <preliminary | interim | advanced>
tags: [<研究主题关键词>, <核心概念名>]
status: draft
---
Body structure
## Triggering material
The user's original input, quoted in full, without changing a single word.
## Core inquiry
Develop it according to the internally identified analytical framework, 400–600 words. The tone is that of the researcher thinking to themselves, not "suggesting that you think"—use the first person or direct statements. The inquiry should advance the analysis, not restate the input content.
## Tentative proposition
One sentence. Declarative, falsifiable. Format: "Under [conditions/situation], [phenomenon/mechanism], because [preliminary explanation]." This is the core output of the memo, forcing the researcher to move from a vague feeling toward a testable proposition.
## To be tested
- Specific, actionable testing actions (which data to return to? what to look for? what to ask?)
- If theoretical sampling is needed, clearly state "what type of interviewees/situations to look for"
Save the file
File name format:YYYY-MM-DD_<核心概念词（2-4字）>.md
Saving steps:
- Use the Bash tool to confirm the directory exists; if it does not, create it:
mkdir -p <路径> - Use the Write tool to write the file
After saving is complete, output only three items in the conversation:
- Full file path
- Tentative proposition(repeat it once so the researcher can see this sentence clearly)
- To be testedchecklist
Do not summarize the full memo, and do not restate the inquiry process.
Continuous recording mode
When triggered multiple times in one conversation, generate an independent file for each entry, accumulating automatically. If 2 or more are generated by the end of the conversation, summarize and inform: a total of N memos were generated this time, list the paths and each one'stentative proposition。
References (for course citation)
- Charmaz, K. (2014). Constructing Grounded Theory (2nd ed.). SAGE. (the memo is the conversation)
- Birks, M., Chapman, Y., & Francis, K. (2008). Memoing in qualitative research: Probing data and processes. Journal of Research in Nursing, 13(1), 68–75. (developmental stage framework)
- Strauss, A., & Corbin, J. (1998). Basics of Qualitative Research. SAGE. (early three-type framework; this skill has moved beyond this framework)
Language
- Default Chinese
- If the user inputs in English, the memo body should be in English, and the frontmatter field names should remain in English