Measuring Norms: Eliciting normative expectations with coordination games allowing for neutral report

We examine the use of a five-item response scale in a widely used norm-elicitation procedure based on coordination games. The method introduced by Krupka and Weber (2013) measures the degrees of social (in)appropriateness using a four-category scale. The even-numbered scale may limit the ability to evaluate asymmetries in deviations from the prescribed action and excludes ratings that indicate no normative valence. To address these limitations, we conduct a norm elicitation experiment using dictator games and vignettes, testing a scale that allows subjects to report neutral normative valence: ``neither socially appropriate nor inappropriate''. While a central neutral category may risk constituting a focal point, biasing responses, our findings reveal no evidence of such distortions and minimal changes in distribution shapes. The neutral category is robust to including incentivized uncertainty responses (``I don't know''), confirming that it captures genuine neutrality. In addition, including the neutral category slightly reduces bias and noise in predictions of behavior in the dictator game. The method remains robust to false positives (detecting norms where they do not exist) and to competing focal points. Overall, incorporating a neutral category improves the method's precision and interpretability, allowing a more nuanced understanding of normative expectations without significant drawbacks.