An Analysis and Evaluation of the Internal Validity of the Remote Associates Test: What Does it Measure?

It is shown that items constructed like those of Mednick's Remote Associates Test (1962) fall into distinct categories. These categories of items are shown to interact significantly with subjects and with subject groups determined on the basis of RAT performance. Furthermore, RAT performance is shown to be more related to performance on items containing idiomatic relationships between stimuli and targets than to performance on items in which such relationships are semantic in nature. It is suggested either that the RAT is measuring a loosely related group of subtraits, or that its item population is poorly specified. Some doubts are cast on the validity of the RAT.