Precision, speed, and distraction in time-limit tests

Abstract

Three models are presented that account for the effects of speed and precision in mental test scores as measured by time-limit intelligence tests. To estimate precision, it is shown that the Rasch model- and, as a special case, the binomial error model -can be used, focusing on the number of items correct and the number of items attempted (usually considered as estimators of these effects). To estimate speed, a model called the Poisson-Erlang model is presented. It is shown that speed can be analyzed into two components: processing time and distraction time. Parameters of the Poisson-Erlang model are used to estimate these components. An outline is given of a new method for analyzing time-limit tests, combining the different models. In addition, it is shown that the Poisson-Erlang model can be used to account for response latencies commonly observed in mental concentration tests.