paper-types

name	paper-types
description	Use when selecting or learning about paper types. Given research questions or a study design, identifies the matching paper type (11 types: IMRaD empirical, systematic/meta-analysis, narrative review, scoping review, qualitative, case report, RCT report, observational, theoretical, CS conference, short paper) and points to its reference guide and rubric.

Paper Types — Selection & Guidance

Maintenance note: reporting-guidelines/SKILL.md's "Step 1: Identify Your Study Type" flowchart asks the same primary-data/design/scope questions as the tree below, for a different purpose (guideline/checklist selection vs. reference/rubric file selection). If you change the classification logic here (add/remove a type, change a routing question), update that flowchart too — they are two independent copies, not one shared source.

Quick Decision Tree

1. Did you collect primary data (your own)?

   • Yes → Go to step 2
   • No, you reviewed literature → Go to step 5

2. Is your study randomized with treatment assignment?

   • Yes → RCT Report (Section 7)
   • No → Go to step 3

3. What's your design?

   • Cohort, case-control, or cross-sectional → Observational Study (Section 8)
   • Single case with detailed timeline → Case Report (Section 6)
   • Interviews/focus groups/ethnography → Qualitative Study (Section 5)
   • Experiment/lab study → Empirical Research (IMRaD) (Section 1)

4. For empirical studies: Does your research question test a hypothesis or explore mechanistically?

   • Tests hypothesis/comparison → Empirical Research (IMRaD)
   • Describes/explores without pre-registered hypothesis → Qualitative Study or Case Report

5. You reviewed literature. What scope?

   • Systematic search of all databases, explicit criteria, ≥2 screeners, flow diagram → Systematic Review / Meta-analysis (Section 2)
   • Mapping the field (breadth over depth) → Scoping Review (Section 4)
   • Narrative synthesis of known themes, selective scope → Narrative Review (Section 3)

6. Publishing venue?

   • Academic journal, any discipline → empirical/review/qualitative/case/observational per above
   • CS/engineering conference (IEEE/ACM/ICML/NeurIPS) → Conference Paper (Section 9)
   • Conference with strict page limit (2–4 pages) → Short Paper / Letter (Section 10)
   • Theory, no empirical data, building argument architecture → Theoretical/Conceptual Paper (Section 11)

---

Type Summaries (click reference links for full guidance)

1. Empirical Research Article (IMRaD)

When: Testing a hypothesis or comparing conditions with experimental/quasi-experimental design. Primary data collection (lab, field, quasi-experiment). Typical venue: peer-reviewed journal in natural/social sciences, engineering.
Structure: Introduction (RQ/hypothesis) → Methods (design, participants, procedures, measures) → Results (statistical/descriptive findings) → Discussion (interpretation, limitations, implications) → Conclusion.
Word budget: ~6000–8000 words; typical proportions I:15% M:25% R:25% D:30% C:5%.
Key discipline: Hypothesis ↔ findings tracing; statistical reporting standards (STROBE for observational, CONSORT for RCT).
Reference: imrad.md
Rubric: rubrics/imrad.md

2. Systematic Review / Meta-analysis

When: Answering a question by synthesizing ALL evidence from a systematic search; optionally pooling results quantitatively (meta-analysis). Rigorous protocol (ideally pre-registered). Typical venue: health sciences, social sciences, systematic review journals.
Structure: Introduction (RQ, rationale) → Methods (PRISMA protocol, search strategy, inclusion/exclusion, screening process, data extraction) → Results (flow diagram, characteristics table, synthesis/meta-analysis results) → Discussion (certainty of evidence, implications) → Conclusion.
Special requirements: PRISMA 2020 27-item checklist; flow diagram (identified → deduplicated → screened → excluded+reason → included); search-strategy reproducibility (databases, dates, keywords, hit counts per database).
Word budget: ~8000–12000 words; detailed Methods & Results.
Key discipline: Search reproducibility (every query logged), inclusion/exclusion criteria applied consistently by ≥2 reviewers.
Reference: reference/systematic-review.md
Rubric: rubrics/systematic-review.md

3. Narrative Review

When: Synthesizing known themes in a topic; NOT exhaustive search, but expert-curated synthesis. Conceptual framework builds the narrative. Typical venue: review journals, book chapters, journals seeking "state of the art" pieces.
Structure: Introduction (topic scope, why now) → thematic sections (each theme addressed via 2+ sources in conversation, not list) → Synthesis & gaps (conceptual framework, what's missing, future directions) → Conclusion.
Word budget: ~6000–10000 words; varies by venue.
Key discipline: Synthesis ≠ annotated bibliography; each paragraph weaves ≥2 sources together to build a point.
Reference: reference/systematic-review.md (search strategy basics apply; narrative reviews use selective scope, documented openly)
Rubric: rubrics/systematic-review.md (adapt "systematic" requirements: scope is selective, not exhaustive; document it clearly)

4. Scoping Review

When: Mapping the field to identify scope, gaps, and research types (not deep synthesis). Useful before launching a full systematic review. Follows PRISMA-ScR. Typical venue: specialized journals, health/social sciences.
Structure: Introduction (topic, PEO/PIO framing) → Methods (PRISMA-ScR protocol, search strategy, inclusion criteria, screening, extraction) → Results (flow diagram, characteristics table, mapping table showing themes/gaps) → Discussion (implications, limitations) → Conclusion.
Special requirements: PRISMA-ScR 22-item checklist; mapping table (papers × dimensions like methodology, population, outcome).
Word budget: ~7000–10000 words.
Key discipline: Breadth over depth; documenting what types of research exist in the field.
Reference: reference/systematic-review.md (PRISMA-ScR variant)
Rubric: rubrics/systematic-review.md

5. Qualitative Study

When: Exploring lived experience, perceptions, meanings, processes (interviews, focus groups, ethnography, document analysis). Rigor via member checking, reflexivity, saturation. Typical venue: social sciences, nursing, education, psychology journals.
Structure: Introduction (research question, context, rationale) → Methods (design, participants, data collection procedures, analysis procedure, reflexivity/researcher positionality) → Findings (themes with supporting quotes, page numbers) → Discussion (interpretation, theoretical implications, limitations) → Conclusion.
Special requirements: COREQ or SRQR 32–32-item checklist; theme–quote–source chain (every quote includes source ID and page; no orphan themes); saturation criterion documented.
Word budget: ~6000–8000 words.
Key discipline: Quote fidelity (verbatim, page numbers, context) and theme rigor (saturated, not forced).
Reference: reference/qualitative.md
Rubric: rubrics/qualitative.md

6. Case Report

When: Describing a rare, unusual, or instructive single case (medical, clinical, organizational). CARE checklist required. Typical venue: medical journals, case-report journals, clinical practice.
Structure: Introduction (case significance, rationale) → Case presentation (patient/case demographics, history, timeline of events, intervention, outcome) → Discussion (what this case teaches, positioning vs literature, limitations) → Conclusion.
Special requirements: CARE 13-item checklist; detailed timeline (dates and sequence matter); explicit consent statement.
Word budget: ~2000–3000 words.
Key discipline: Timeline precision; medical/legal transparency (consent, privacy).
Reference: reference/case-report.md
Rubric: rubrics/imrad.md (adapt: single case = different Results structure)

7. RCT Report

When: Testing a randomized treatment assignment in a controlled trial (typically health, education, behavior). Highest evidence level for causal claims. Typical venue: medical/health/education journals.
Structure: Introduction → Methods (design, randomization procedure, blinding, participants, intervention, outcome measures, statistical analysis plan) → Results (flow diagram with participant attrition, baseline characteristics table, outcome results by group, statistical tests, adverse events) → Discussion → Conclusion.
Special requirements: CONSORT 25-item checklist; flow diagram (enrolled → randomized → allocated → followed-up → analyzed, with exclusion counts at each stage); pre-registration (trial ID, date); analysis plan stated before results.
Word budget: ~6000–8000 words.
Key discipline: Blinding specification, statistical precision (means ± SE, p-values, CIs), participant flow transparency.
Reference: reference/imrad.md (structure); CONSORT checklist lives in the reporting-guidelines skill
Rubric: rubrics/imrad.md

8. Observational Study

When: Cohort, case-control, or cross-sectional design; no randomization. Describing patterns, associations, prevalence. Typical venue: epidemiology, social sciences, psychology journals.
Structure: Introduction (research question, context) → Methods (study design specification, participants/cases, exposure/outcome measurement, data collection, confounders, statistical analysis plan) → Results (participant flow, characteristics tables, main and sensitivity/subgroup analyses) → Discussion → Conclusion.
Special requirements: STROBE 22-item checklist (variant by design type: cohort/case-control/cross-sectional); confounder specification; flow diagram.
Word budget: ~6000–8000 words.
Key discipline: Design-specific reporting (cohort: follow-up; case-control: matching; cross-sectional: point estimate + CI).
Reference: reference/imrad.md (structure); STROBE checklist lives in the reporting-guidelines skill
Rubric: rubrics/imrad.md

9. Theoretical / Conceptual Paper

When: Building argument, no empirical data. Proposing theory, framework, model, or analyzing concepts. Typical venue: theory journals, philosophy, conceptual papers in specialized journals.
Structure: Introduction (problem being addressed, thesis statement) → argument architecture (proposition → evidence/warrant → implication, repeated across sections) → Implications & future directions → Conclusion.
Word budget: ~5000–7000 words.
Key discipline: Claim–evidence–warrant chain (Toulmin model); explicit proposition–evidence mapping (avoid unsupported claims).
Reference: reference/theoretical.md
Rubric: rubrics/theoretical.md

10. CS / Engineering Conference Paper

When: Publishing at IEEE/ACM/ICML/NeurIPS/AAAI/etc. Possibly with strict page limits and camera-ready rules. Emphasis on novelty, reproducibility, comparison rigor.
Structure: Intro (problem, contribution) → Related work (explicit positioning vs prior art) → Method/Approach → Experiments (baselines, datasets, metrics, ablations, results tables/figures) → Discussion/Conclusion.
Special requirements: Page limit (usually 6–8 pages + references); camera-ready formatting (IEEE/ACM style); code/data availability statement (NeurIPS reproducibility checklist); comparison fairness (equal experimental conditions, multiple seeds, statistical significance).
Word budget: ~5000–6000 words (strict).
Key discipline: Baseline selection (fair, well-known, recent); ablation design; seed protocol & significance testing; positioning vs related work (not just citations, but explicit comparison).
Reference: reference/conference-cs.md
Rubric: rubrics/conference-cs.md

11. Short Paper / Letter

When: Preliminary findings, negative results, rapid communication, letters to the editor, conference short papers. Venue-specific format. Typical venue: any journal's "Short Communication" or "Letter" section; some conferences.
Structure: Varies by venue; typically Introduction + Methods + Results + Discussion compressed into 2–4 pages; or Introduction + key point + conclusion.
Word budget: 1500–2500 words max (strict).
Key discipline: Every word counts; prioritize the novel/surprising finding; constraints are part of the art.
Reference: reference/conference-cs.md (page-limit discipline applies; adapt proportions)
Rubric: rubrics/conference-cs.md (adapt for brevity)

---

Canonical Type → File Resolution

PROJECT.md's - **Type:** bullet holds one of 15 enum keys (see the template's own comment). Consumers (submit-check, reviewer-critic, critique) MUST look up both the reference file and the rubric file in this table — never substitute reference/<type>.md or rubrics/<type>.md naively. Several type keys deliberately share a file (e.g. narrative/scoping review reuse the systematic-review reference; case-report/RCT/observational share the IMRaD rubric), and the two columns are NOT the same mapping — reference and rubric diverge for several types.

Type key (PROJECT.md)	Reference file	Rubric file
empirical-imrad	reference/imrad.md	rubrics/imrad.md
systematic-review	reference/systematic-review.md	rubrics/systematic-review.md
narrative-review	reference/systematic-review.md	rubrics/systematic-review.md
scoping-review	reference/systematic-review.md	rubrics/systematic-review.md
qualitative	reference/qualitative.md	rubrics/qualitative.md
case-report	reference/case-report.md	rubrics/imrad.md
rct	reference/imrad.md	rubrics/imrad.md
observational	reference/imrad.md	rubrics/imrad.md
theoretical	reference/theoretical.md	rubrics/theoretical.md
conference-cs	reference/conference-cs.md	rubrics/conference-cs.md
short-paper	reference/conference-cs.md	rubrics/conference-cs.md
thesis-bsc	(see thesis-structure skill, not a reference/ file)	rubrics/thesis.md
thesis-msc	(see thesis-structure skill, not a reference/ file)	rubrics/thesis.md

RCT and observational studies also carry their own reporting-guideline checklist (CONSORT / STROBE) — that lookup is separate and already handled by submit-check's Step 2 (type + subtype → guideline table), not by this table. All paths are relative to .claude/skills/paper-types/.

Cross-Cutting Rules

Hypothesis ↔ Findings Tracing (empirical only):
In Introduction, state the hypothesis/RQ clearly. In Results, use the same terminology. In Discussion, directly address what you found vs what you hypothesized. word_budget.py --rq-trace checks this automatically.

Terminology Consistency:
Choose one term per concept (e.g., "CNN-based fault diagnosis" not also "deep learning anomaly detection" for the same thing). EN term + TR term on first mention (bilingual output).

Limitations Honesty:
Every paper must include limitations (Methods or Discussion, depending on type). Mark assumptions; don't hide them.

Figure / Table Sufficiency:
At least one table or figure per major result; captions are self-contained (reader shouldn't need to read text to understand the figure).

---

Next Steps

1. Choose your type using the decision tree above.
2. Read the reference file linked in your type's summary.
3. Check the rubric (reviewer-critic will read it; you can too for self-review).
4. Use /draft to start writing; academic-writing skill covers register + tense + hedging.
5. Use /verify-citations before finalization (applies the reporting-guidelines checklist).