Troubleshooting

Reliability statistics need at least three valid items and at least 30 complete-case respondents. Below those thresholds, ReliCheck shows the count and a sample-size note instead of an estimate.

Check three things: are reverse-scored items flagged, do all items measure the same construct, and is the sample large enough? Item-total correlations point at weak contributors. The reliability guide pitfalls covers the most common causes.

The mapping wizard requires column headers in row 1 and at least one Likert, single-choice, or open-ended column. Files over 10MB or with more than 50,000 rows need to be split. CSVs with mixed delimiters (comma plus semicolon) need cleaning first.

Check three things: is the survey published (not just saved), is the access window open, and is the response cap not exceeded. Settings under the survey’s Distribute tab show all three at a glance.

AI features need an API key configured server-side. If you self-host or your admin has not set anthropic_api_key in the platform config, AI features return a clear "AI features are not configured" message. Contact your admin or support.