# Validation Report Summary

Conjoint Survey validates its analysis workflow with known-effects synthetic data. The validation task is simple: generate data where the true positive, negative, and null effects are known in advance, then verify that the same analysis pathway used by the app recovers those effects.

## Result

Passed.

- Synthetic respondents: 5,000
- Profile-level observations: 80,000
- Respondent clusters: 5,000
- True effects tested: positive, negative, and null
- Estimator: profile-level linear probability model
- Standard errors: clustered by respondent

## Recovery Table

| Attribute level | True effect | Estimate | SE | 95% CI | Absolute error | Pass |
|---|---:|---:|---:|---|---:|---|
| Positive Signal: Present | 0.100 | 0.099 | 0.003 | 0.093 to 0.106 | 0.001 | Yes |
| Negative Signal: Present | -0.100 | -0.102 | 0.003 | -0.109 to -0.096 | 0.002 | Yes |
| Null Signal: Present | 0.000 | -0.001 | 0.004 | -0.008 to 0.006 | 0.001 | Yes |

## Interpretation

This is a software and estimator validation, not a guarantee that every user-created study design is substantively valid. Passing the known-effects check means that when randomization is clean and the true effect is known, the app's analysis path recovers the expected values within the stated tolerance.

Full validation page: https://conjointsurvey.com/validation