# Conjoint Survey Research Artifact v1

This artifact provides public, citable materials for reviewing the Conjoint Survey research workflow. It is prepared under the public organizational identity of System Analytics LLC and the Conjoint Survey Research Team. It does not include personal author names.

## Contents

- `validation-report.md`: known-effects validation summary.
- `sample-conjoint-profile-data.csv`: profile-level long-format sample conjoint data.
- `codebook.md`: variable definitions for the sample dataset.
- `replication.R`: R example using clustered standard errors.
- `replication.py`: Python example using a cluster-robust sandwich estimator.
- `replication.do`: Stata example using respondent-clustered standard errors.
- `metadata.json`: machine-readable artifact metadata.
- `CITATION.cff`: citation metadata for GitHub, Zenodo, and software directories.
- `.zenodo.json`: release metadata for DOI archiving through Zenodo.

## Citation

Conjoint Survey Research Team. (2026). *Conjoint Survey research artifact v1: Validation report, sample conjoint dataset, codebook, and replication scripts*. System Analytics LLC. https://conjointsurvey.com/research/research-survey-validation-replication

## Purpose

The artifact is meant to help researchers inspect how Conjoint Survey represents conjoint data and how profile-level analysis can be reproduced outside the app. It supports directory submissions, OSF deposits, Zenodo records, GitHub releases, and methodology pages.

## What the Sample Data Represents

The sample dataset is profile-level long-format conjoint data. Each row is one shown profile, not one respondent. Respondents see multiple tasks, each task shows multiple profiles, and the `chosen` field records whether that profile was selected.

This structure is the core requirement for AMCE/ATE-style conjoint analysis because the analysis needs both selected and unselected profiles, task/profile identifiers, attribute levels, and respondent IDs for clustered uncertainty estimates.

## Reproducibility

Run the scripts from this directory:

```bash
Rscript replication.R
python replication.py sample-conjoint-profile-data.csv
stata -b do replication.do
```

The R and Stata examples require their respective software environments. The Python script requires `numpy` and `pandas`.

## Privacy

The artifact contains synthetic/sample data only. It contains no real respondent records, no personal identifiers, and no private customer data.

## Product Link

Conjoint Survey is available at https://conjointsurvey.com.

## Archiving

This directory is release-ready as a standalone public repository. After the repository is connected in Zenodo, publishing a GitHub release archives the bundle and creates a citable Zenodo record with a DOI. The release metadata identifies the creator as the Conjoint Survey Research Team and the affiliation as System Analytics LLC; no personal author name is required in the public artifact metadata.