---
name: comparative-study-appraisal
description: Compare research studies by their questions, methods, results, and limitations while preserving differences in context and evidence quality.
---

# Comparative study appraisal

Compare studies against a specified question, not against an assumed common result. Read the relevant sources where possible and make any access limitations visible.

## Establish comparability

Identify the user's comparison question and verify each study's identity, version, publication status, and corrections when feasible. Extract the features needed for this question: objective, population or system, setting, design, exposure or intervention and comparator, outcomes or quantities, observation period or parameter regime, methods, and results with uncertainty. For theoretical or computational work, capture assumptions, boundary conditions, approximations, and validation. Record source locations for consequential statements; mark missing information as not reported or not accessible rather than filling it in.

Compare definitions, measurement instruments, estimands, units, reference groups, analysis methods, and sample or model scope before interpreting differences. Identify overlapping cohorts, reused data, and multiple reports of one study so they are not counted as independent evidence. A larger sample, a statistically significant result, or a journal label alone does not establish stronger evidence.

## Interpret differences

Assess limitations proportionately to each design: selection, measurement, confounding, missing data, analytical choices, uncertainty, model assumptions, and generalizability as relevant. Report estimates and intervals in their original context; harmonize units or effect directions only when the transformation is justified and shown. Do not interpret one significant and one nonsignificant result as proof that their effects differ; compare estimates and uncertainty or an appropriate direct contrast.

Explain whether differences could reflect population, design, conditions, chance, bias, or competing mechanisms. Distinguish what the studies observed from the comparison's own inference. If they are too heterogeneous for a shared conclusion or pooling, say so. Quantitative pooling requires a separately justified synthesis design and compatible data; do not calculate a pooled estimate merely because the studies address similar topics.

## Deliver

Provide a compact comparison table with source links and a narrative addressing the original question, material disagreements, study limitations, and uncertainty. Identify source passages inspected, unresolved questions, and the next comparison or evidence that would reduce uncertainty. Do not present an abstract-only review as a full methods appraisal.