Why is 'Phase 1: Hypothesis Formulation' mandatory?

Formulating a hypothesis before looking at the data is the cornerstone of the scientific method. It ensures that you aren't simply finding patterns that support a pre-existing belief (confirmation bias).

Do I need a specific database to use this skill?

Yes, this skill is designed to work with SQLite databases. It assumes you have already imported and cleaned your data using the complementary datapeeker skills.

What is p-hacking and how does this skill prevent it?

P-hacking is the practice of manipulating data or searching for patterns until something statistically significant emerges. This skill prevents it by forcing you to define your hypothesis and test design before you are allowed to query the data.

Can I use this for complex statistical modeling?

While this skill provides the framework for rigorous testing, the actual statistical complexity depends on the SQL queries you write. It is ideal for A/B testing, segment comparisons, and trend analysis.

Hypothesis Testing Framework

Name: Hypothesis Testing Framework
Author: tilmon-engineering

bytilmon-engineering

•

데이터 과학 및 ML

Guides users through a scientifically rigorous five-phase hypothesis testing process to eliminate bias and p-hacking in data analysis.

The Hypothesis Testing Framework is a specialized skill designed to bring scientific rigor to data analysis within Claude Code. It enforces a strict 'hypothesis-first' methodology, requiring users to formulate and document their null and alternative hypotheses before exploring data. This prevents common analytical pitfalls like confirmation bias and p-hacking. The skill provides a structured path from formulation and test design to SQL-based analysis and skeptical interpretation, ensuring that every conclusion is defensible, reproducible, and based on objective evidence.

주요 기능

01Enforced hypothesis formulation prior to data exploration to prevent bias

02Structured five-phase workflow tracked via TodoWrite for full auditability

03Mandatory data quality and schema validation checkpoints

041 GitHub stars

05Integrated SQLite query management with systematic documentation of results

06Standardized templates for test design, analysis queries, and result interpretation

사용 사례

01Testing the impact of specific variables on KPIs while controlling for confounding factors

02Conducting reproducible research and deep-dive data investigations with high integrity

03Validating business claims such as comparing weekend vs. weekday sales performance

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add tilmon-engineering/claude-skills hypothesis-testing

For use in Claude.ai and ChatGPT

Download Skill