When: March 11, at 1-2 pm

Where: Room B705, Department of Statistics, Stockholm University

Abstract

In large achievement tests like the Swedish Scholastic Assessment Test (högskoleprovet), questions are pretested for use in future tests. We recapitulate a method for optimal pretesting based on the abilities of the examinees. An adaptive algorithm for computing the optimal pretesting designs will be discussed. We compare then the optimal design’s theoretical efficiency with an efficiency based on simulation results. Plans to use data from previous Swedish Scholastic Assessment Tests to investigate the value of pretesting methods are outlined.