Abbreviated Judgements and Asymmetries

In many circumstances people choose whether to obtain additional information prior to making a decision under uncertainty, e.g., an editor deciding whether to desk-reject an article or postponing the decision until after receiving recommendations from referees. In many cases, these decision rules are asymmetric; e.g., desk-rejections are common but desk-acceptances are rare in many fields. I question what types of informational, value-based, and frequency-based rationales may explain these asymmetries. To do so, I analyze a model in which a decision maker (DM) who receives a signal may invest in acquiring a second signal to increase the accuracy of her binary classification. It may be optimal for the DM to adopt a decision rule in which she may make one type of classification without obtaining further information, but where she always obtains further information before making the other type of classification. When the signals are symmetric, asymmetric priors as well as asymmetric error-avoidance preferences can cause these one-sided summary-classification rules to be optimal, but prior-asymetries have dominance over preference-asymmetries. Moreover, these one-sided classification rules can be optimal even when the DM has symmetric priors and error-costs, because the signals may be asymmetric. I discuss implications.