psych-paper-deconstruction

心理学文献精读拆解助手

Overview

This skill provides a systematic workflow for deconstructing psychology research papers into structured analysis reports. Given a psychology paper (PDF, Word, or plain text), produce a comprehensive close-reading report that covers research hypotheses, experimental paradigms, independent/dependent variables, statistical methods, limitations, and practical implications — formatted for direct use in applied psychology course assignments.

Input Handling

Accept papers in the following formats:

1. PDF files — Use the Read tool to extract text. If the PDF is scanned or image-based, use OCR if available; otherwise inform the user and request a text version.
2. Word documents (.docx) — Use the Read tool or appropriate document processing tools to extract content.
3. Plain text / pasted content — Process directly.
4. Web links to papers — Use WebFetch to retrieve the abstract and available content; note that full-text access may be limited.

After extracting the paper content, confirm the paper title and authors with the user before proceeding to analysis.

Analysis Workflow

Execute the following steps in sequence to produce the deconstruction report.

Step 1: Extract Basic Information

Identify and record:

• Paper title (original language + Chinese translation if applicable)
• Authors and affiliations
• Journal name, volume, issue, year
• DOI or other identifiers
• Keywords listed by the authors

Step 2: Locate Research Background & Question

Extract:

• The broader research context the paper situates itself in
• The specific research gap or problem being addressed
• The main research question(s)

Distinguish between what the authors explicitly state and what can be inferred. Label inferences clearly with "[推断]".

Step 3: Identify Research Hypotheses

Extract all stated hypotheses (H1, H2, H3, ...). For each hypothesis:

• Quote the original statement
• Identify the direction (directional vs. non-directional)
• Map which variables it connects
• Note whether each hypothesis is a priori (pre-registered) or appears post-hoc

If hypotheses are not explicitly stated, reconstruct them from the research questions and predictions, labeling them as "[重构假设]".

Step 4: Analyze Research Design

Determine and document:

• Design type: between-subjects, within-subjects, mixed, correlational, longitudinal, case study, meta-analysis, etc.
• Experimental vs. observational: Is this a true experiment, quasi-experiment, or observational study?
• Number of conditions/groups
• Random assignment: Was random assignment used? What was the unit of randomization?
• Control conditions: What serves as the control/baseline?

Consult references/psychology_research_methods.md for design-type definitions and decision criteria.

Step 5: Identify Experimental Paradigm

Identify the experimental paradigm or task used:

• Is it a well-established paradigm (e.g., Stroop, Simon, Flanker, N-back, Posner cueing, IAT, dot-probe, lexical decision, go/no-go)?
• Is it a novel/adapted paradigm? Describe the task structure.
• What are the trial-level events and timings?
• What stimuli are used?

Consult references/psychology_research_methods.md → "Common Experimental Paradigms" section for a catalog of established paradigms and their characteristics.

Step 6: Extract Variables

Create a structured variable table:

Independent Variables (IVs):

Dependent Variables (DVs):

Control / Covariate Variables: List all variables held constant or used as covariates (e.g., age, gender, education level, IQ, mood state).

If operationalization is ambiguous, note it with "[操作化不明确]" and suggest a clearer definition.

Step 7: Identify Statistical Methods

Document:

• Descriptive statistics used
• Inferential tests (t-test, ANOVA, ANCOVA, regression, mediation/moderation, SEM, etc.)
• Significance level (α)
• Effect size measures reported (Cohen's d, η², partial η², r, OR, etc.)
• Corrections for multiple comparisons (Bonferroni, FDR, etc.)
• Software used (SPSS, R, Python, etc.)

If advanced or unusual methods are used, briefly explain what they test.

Step 8: Summarize Main Results

For each hypothesis:

• State whether it was supported, partially supported, or rejected
• Report key statistics (test statistic, p-value, effect size)
• Describe the direction and magnitude of effects
• Note any unexpected or null findings

Organize results by hypothesis, not by analysis order, to maintain coherence.

Step 9: Analyze Limitations

Identify limitations from two sources:

1. Author-stated limitations — Quote or closely paraphrase from the Discussion section.
2. Critically identified limitations — Assess:
   • Internal validity: confounds, demand characteristics, experimenter effects, history/maturation/attrition threats
   • External validity: sample representativeness, ecological validity, generalizability
   • Construct validity: operationalization adequacy, face/construct validity of measures
   • Statistical validity: power adequacy, multiple comparison issues, assumption violations
   • Ethical considerations: informed consent, debriefing, potential for harm

Label author-stated limitations vs. critically identified ones.

Step 10: Extract Practical Implications

Document:

• Theoretical contributions: How does this study advance theory? Does it support, challenge, or extend existing models?
• Applied/practical implications: What real-world applications does the study suggest? (clinical, educational, organizational, consumer, etc.)
• Future research directions: What do the authors recommend? What additional questions does the analysis reveal?

Step 11: Critical Evaluation

Provide a balanced critical assessment:

• Strengths: methodological rigor, novelty, theoretical contribution
• Weaknesses: design flaws, measurement issues, generalizability concerns
• Overall assessment: Rate the study's contribution on a scale (moderate / substantial / significant) with justification

Step 12: Generate Output

Format the complete analysis using the template in assets/analysis_template.md. Produce the report in Markdown. Offer to also export as a Word document if the user needs it for assignment submission.

Output Format

The final report must follow this structure (see assets/analysis_template.md for the full template):

1. 论文基本信息
2. 研究背景与问题
3. 研究假设
4. 研究设计
5. 实验范式
6. 变量分析（自变量 / 因变量 / 控制变量）
7. 统计方法
8. 主要结果
9. 研究局限
10. 现实引申
11. 批判性评价
12. 参考格式（APA 格式引用）

Quality Standards

• Accuracy: All statistics and findings must be traceable to the original paper. Never fabricate or infer data points.
• Completeness: Cover all 12 sections. If a section is not applicable (e.g., no explicit hypotheses), state so explicitly rather than omitting.
• Clarity: Use plain language. Define technical terms on first use. Assume the reader is an undergraduate psychology student.
• Honesty: Clearly distinguish between what the authors stated and what the analysis infers. Use labels: [原文], [推断], [重构假设], [操作化不明确].
• APA formatting: Use APA 7th edition style for any in-text citations and the reference entry.

References

• references/psychology_research_methods.md — Catalog of common experimental paradigms, research designs, statistical methods, and validity threat checklists. Consult this when identifying design types, paradigms, or validity issues.
• assets/analysis_template.md — Output template for the final deconstruction report. Use this as the structural basis for the generated report.