---
name: research-results-interpretation
description: Use when interpreting research results, statistical output, model estimates, or qualitative findings supplied by the user.
---

# Interpret research results

Inspect the question, design, outcome definitions, sample or data, analysis method, and actual output. Check units, group direction, reference levels, transformations, uncertainty measures, and what each statistic represents before interpreting it. If necessary output is absent, identify it and provide only a conditional interpretation.

Explain the magnitude and direction of findings with the reported uncertainty and practical context, separating evidence for an effect from evidence about its size. Distinguish statistical uncertainty from bias and discuss assumptions, missingness, multiplicity, and selection where relevant. Do not turn a nonsignificant result into proof of no effect, a correlation into a causal effect, or model fit into external validation. For qualitative material, ground interpretations in supplied excerpts and describe the scope and alternative readings.

Compare findings with verified literature only when requested or needed, preserving citation-to-claim relationships. Clearly label exploratory, post hoc, and proposed interpretations. Never fabricate analyses, data, confidence intervals, p-values, quotes, or sources. Use available computation to check arithmetic and state what was actually run.

Deliver an accessible interpretation tied to the evidence, important limitations, and any clarification or follow-up analysis needed. Outputs do not establish clinical, legal, or policy advice without relevant expertise and context.