Two Statistical Tests for the Detection of Item Compromise

Two independent statistical tests of item compromise are presented, one based on the test takers’ responses and other their response times (RTs) same items. The can be used to monitor an in real time during online continuous testing but also applicable as part post hoc forensic analysis. two statistics simple intuitive quantities sum RTs observed for takers item. Common features ease interpretation computational simplicity. Both uniformly most powerful under assumption known ability speed parameters takers. Examples power functions items with realistic parameter values suggest maximum 20–30 preknowledge response-based 10–20 RT-based test.