When should I use the conjoint-diagnostics skill?

Use it when you are reviewing a conjoint manuscript, auditing analysis code, checking for survey satisficing, or ensuring your experimental interpretation follows current methodological standards.

Does this skill work with specific programming languages?

Yes, it is designed to review analysis code (particularly R and Python) to verify clustering specifications, estimands, and data cleaning procedures.

How does it handle measurement error?

It provides a workflow to assess Intra-Respondent Reliability (IRR) and suggests bias-correction methods, such as those found in the projoint R package, to account for swapping error.

Does it support power analysis checks?

Yes, it prompts users to verify if the sample size is justified by power analysis and flags potential Type S (sign) and Type M (magnitude) errors.

Conjoint Experiment Diagnostics

Name: Conjoint Experiment Diagnostics
Author: scdenney

byscdenney

•

데이터 과학 및 ML

Evaluates and audits choice-based conjoint experiments using systematic social science standards and statistical best practices.

A comprehensive diagnostic tool for social scientists and data analysts performing conjoint analysis. It provides a structured framework to audit experimental designs, verify estimation scripts, assess measurement error through Intra-Respondent Reliability (IRR), and validate the interpretation of Average Marginal Component Effects (AMCEs) and Marginal Means (MMs). Whether reviewing a manuscript or auditing a codebase, this skill ensures that identifying assumptions—such as profile-order, carryover, and fatigue—are empirically tested and that subgroup analyses are statistically sound.

주요 기능

01Detection of measurement error and swapping bias using projoint methods

02Systematic checklist for design, estimation, and external validity

03Verification of identifying assumptions (profile-order, carryover, fatigue)

04Code-level auditing for clustering, estimands, and IRR computation

05Guidance on subgroup analysis using Marginal Means vs. conditional AMCEs

0615 GitHub stars

사용 사례

01Auditing R or Python scripts to ensure correct clustering and estimand specifications

02Assessing the external validity and behavioral benchmarks of experimental designs

03Reviewing academic manuscripts or working papers involving conjoint analysis

주요 기능

01Detection of measurement error and swapping bias using projoint methods

02Systematic checklist for design, estimation, and external validity

03Verification of identifying assumptions (profile-order, carryover, fatigue)

04Code-level auditing for clustering, estimands, and IRR computation

05Guidance on subgroup analysis using Marginal Means vs. conditional AMCEs

0615 GitHub stars

사용 사례

01Auditing R or Python scripts to ensure correct clustering and estimand specifications

02Assessing the external validity and behavioral benchmarks of experimental designs

03Reviewing academic manuscripts or working papers involving conjoint analysis