scientific-critical-thinking

What it does

Reviews scientific claims by evaluating methodology, experimental design, statistical validity, biases, confounding factors, and evidence quality using GRADE and Cochrane risk-of-bias frameworks. It guides critical analysis of research papers and supports systematic reviews or meta-analyses. It also promotes adding schematics to documents and provides steps for generating diagrams via a dedicated schematic tool.

How it works

• Core capabilities include: Methodology Critique, Bias Detection, Statistical Analysis Evaluation, Evidence Quality Assessment, Logical Fallacy Identification, and Research Design Guidance.
• Each capability provides explicit criteria and checklists. For example, Methodology Critique asks to assess study design suitability, validity (internal, external, construct), and statistical conclusion validity, then evaluate randomization, blinding, control conditions, and measurement quality.
• Bias Detection covers cognitive biases, selection biases, measurement biases, analysis biases, and confounding, with prompts to review preregistration, sampling, attrition, measurement objectivity, and reporting practices.
• Statistical Analysis Evaluation outlines sample size/power, appropriate tests, multiple comparisons corrections, p-value interpretation, effect sizes with confidence intervals, missing data handling, regression/model considerations, and common pitfalls.
• Evidence Quality Assessment follows a hierarchy (design type, risk of bias, GRADE considerations, convergence of evidence, contextual factors) and cites quality signals like replication, consistency, and indirectness.
• Logical Fallacy Identification instructs naming specific fallacies across causation, generalization, authority, statistics, structural, and science-specific categories, with guidance on evidence needed.
• Research Design Guidance provides steps for refining questions, selecting designs, minimizing bias, planning samples, measurement, analysis, and transparency, including preregistration and reporting guidelines.
• Visual schematics are recommended by default; if diagrams are missing, it suggests using the scientific-schematics skill and provides a concrete command example for generating schematics.

When to use it

Use when evaluating research methodology and experimental design, assessing statistical validity and evidence quality, identifying biases and confounding, reviewing scientific claims, conducting systematic reviews or meta-analyses, applying GRADE or Cochrane assessments, or providing critical analysis of research papers.

What it can touch

• Tools declared for use: Read, Write, Edit, Bash
• Commands referenced include: python scripts/generate_schematic.py and the instruction to use the scientific-schematics skill for diagram generation. These are described as the means to generate and refine publication-quality diagrams and schematics, stored under a figures/ directory.

Caveats

• License: MIT
• Declared tools: claude-code (within frontmatter), with allowed-tools Read, Write, Edit, Bash
• The skill emphasizes adding schematics by default and cites external reference materials for deeper principles (e.g., GRADE, ROB, and common biases). No outcomes are promised beyond the explicit evaluation and guidance described above.