In learning potential (LP) tests, intelligence is measured by repeated administrations of reasoning tasks, with interpolated training in problem-relevant strategies. In a comparison of the differential sensitivity and validity of LP and IQ measures, subjects were divided into three IQ groups: bright normal, dull-to-average, and subnormal (EMR). In the first study, both low IQ groups gained more from the coaching than did the bright group. In the second study, the concurrent validity of the LP test was equal to that of IQ in the entire sample and in the bright group, but superior to IQ in the low, disadvantaged groups. (Author)