Agent Skills
› brycewang-stanford/Awesome-Journal-Skills
› revedres-organizing-framework
revedres-organizing-framework
GitHub用于将已编码的文献语料转化为RER论文的分析框架，避免罗列式综述。通过构建分类、模型等骨架整合研究发现，确保类别互斥且穷尽，解释矛盾并生成理论或政策贡献。
Trigger Scenarios
语料已完成编码但文章像流水账
无法概括领域整体意义
缺乏明确的章节架构
Install
npx skills add brycewang-stanford/Awesome-Journal-Skills --skill revedres-organizing-framework -g -y
SKILL.md
Frontmatter
{
"name": "revedres-organizing-framework",
"description": "Use when turning a screened, coded corpus into an analytic spine that advances theory or policy for a Review of Educational Research (RER) manuscript. Builds the conceptual contribution beyond a tally; it does not run the search (revedres-literature-synthesis) or appraise robustness\/risk-of-bias (revedres-comprehensiveness-and-balance)."
}
Organizing Framework (revedres-organizing-framework)
When to trigger
- The corpus is complete and coded, but the draft reads as study-after-study
- You can describe what each study found but not what the field, taken together, means
- A reviewer is likely to say "this is a competent summary, but what is the contribution?"
- You need section architecture and cannot decide how to carve the literature
The RER contribution is the framework, not the tally
The cardinal RER failure mode is the annotated bibliography: a competent, even exhaustive, study-by-study summary with no argument about the field. RER wants a comprehensive, critical, integrative review — integrative means the studies are reorganized under a frame that the reader could not have built from the abstracts alone, and that advances theory or policy. A meta-analysis is not exempt: a pooled number with no conceptual account of why effects vary is a calculation, not a synthesis.
A good framework is the lens through which scattered findings become a coherent claim. Candidate spines:
| Spine type | When it fits | Example shape |
|---|---|---|
| Taxonomy / typology | many studies, no shared map | classify studies by a 2–3 axis scheme so every study has a home |
| Theoretical model | a mechanism organizes the field | a model whose links the literature does/doesn't support |
| Moderator architecture | a meta-analysis with real heterogeneity | the framework is the explanation of variation |
| Developmental / stage account | the field has moved through phases | what each phase established and left open |
| Tensions / competing perspectives | live theoretical or methodological debate | steelman each camp; locate the evidence between them |
| Construct clarification | the field conflates distinct ideas | disentangle the constructs, then re-read the evidence |
Three tests a framework must pass
- MECE-ish. Categories are reasonably mutually exclusive and collectively exhaustive — every included study has exactly one principled home, and there are no empty "misc" bins doing the real work.
- Reconciling. The frame explains apparent contradictions — studies that looked to disagree turn out to sit in different cells (different populations, constructs, designs), or genuinely conflict in a way the frame names.
- Generative. The frame's thin or empty cells become the review's agenda for theory, research, or policy — the contribution that makes the review worth citing.
Then make the section architecture mirror the framework's cells, so the structure of the paper enacts the argument rather than marching through databases or chronology.
Building the frame from the coded corpus (not before it)
The frame must be earned from the evidence, not imposed on it. Work bottom-up from the coding dataset (revedres-literature-synthesis):
- Cluster the coded studies by what actually differentiates their findings — population, construct, design, mechanism, or context — and look for the axis along which the literature naturally separates.
- Name the axis (or axes) that does the most explanatory work; a 2- or 3-axis scheme usually beats a long flat list of themes.
- Test the frame against the hard cases — the studies that seemed to contradict each other. If the frame puts them in different cells and explains why, it is reconciling; if it just relabels them, it is not.
- Read the empty cells as the agenda. A cell with no studies is either an untested combination (a research gap) or an impossible one (a boundary condition) — both are contributions if you say which.
A frame discovered this way is defensible because a reviewer can trace it back to the coded studies; a frame asserted in the introduction and never grounded reads as the author's prior, not the field's structure.
Checklist
- A one-sentence claim about the field stated up front (not withheld to the conclusion)
- A named spine (taxonomy / model / moderator architecture / stages / tensions / constructs)
- Every included study slots into the frame; no junk-drawer "other" category
- The frame reconciles at least one apparent contradiction in the literature
- Thin/empty cells are turned into a concrete theory/research/policy agenda
- Section architecture mirrors the frame, not the search or the calendar
- The framework advances theory or policy — it is not just a sorting convenience
Anti-patterns
- An annotated bibliography dressed up with section headings (the named RER sin)
- A "framework" that is really chronology ("early work / recent work") or by-database order
- A meta-analysis that reports a pooled effect and moderators with no conceptual story of why
- A taxonomy with a giant "miscellaneous" cell that hides the studies that do not fit
- Centering the frame on the author's own program so it functions as self-promotion
- A frame that sorts but never reconciles conflict or generates an agenda (sorting ≠ integrating)
Output format
【Field claim】<one sentence: what the literature, integrated, shows>
【Spine】<taxonomy | model | moderator architecture | stages | tensions | constructs>
【MECE check】every study has one home; no junk-drawer cell? Y/N
【Reconciles】<an apparent contradiction the frame resolves>
【Generative】<the theory/research/policy agenda the thin cells imply>
【Section map】<how sections mirror the frame's cells>
【Next step】→ revedres-comprehensiveness-and-balance (test coverage, risk-of-bias, robustness)
Version History
- 1839142 Current 2026-07-05 14:22
