Recommendation #1: Use the Central Limit Theorem Evaluation scores are usually mean scores across a large number of questions. This is a perfect fit for the Central Limit Theorem. In the paper, we suggest reporting the CLT-based standard errors alongside mean scores.