Recipe

Research hypothesis writer

Generate testable, falsifiable hypotheses from raw research questions using structured prompt engineering.

Ingredients

A research question or domain of inquiry
Known variables (independent / dependent)
Population or sample constraints
Confidence threshold (default: p < 0.05)

Prompt template

You are a senior research methodologist. Given the research question below, produce a single testable hypothesis that includes:

1. Null hypothesis (H₀)
2. Alternative hypothesis (H₁)
3. Independent and dependent variables
4. Proposed statistical test
5. Required sample size estimate

Research question: {{QUESTION}}
Population: {{POPULATION}}
Alpha: {{ALPHA}}

Output format

{
  "null_hypothesis": "...",
  "alternative_hypothesis": "...",
  "independent_variable": "...",
  "dependent_variable": "...",
  "statistical_test": "...",
  "sample_size_estimate": 0
}

Notes

Always request structured JSON output. If the model hedges, append: “Respond with valid JSON only. No commentary.” For multi-group designs, specify ANOVA or MANOVA in the prompt override.

This recipe pairs well with the Literature gap finder recipe for identifying under-explored research areas before hypothesis formulation.