A group of 105 lower-lower middle class kindergarten children were tested on a set of single variable classification tasks and related-component tasks dealing with color and number. Children who failed to reach criterion on the classification test were randomly assigned to one of five experimental groups, stratified on the basis of performance on both the classification and Component Test scores. Each group was given instruction on classification tasks, component tasks and/or control tasks, followed by Classification and Component Posttests. Children receiving classification instruction performed significantly better on the Classification Posttest than control groups who did not. Children who received component instruction performed better than the comparison group that did not receive it, although the difference was statistically significant only for one of three levels of pretest performance. There were no apparent effects of classification instruction on Component Posttest performance or of component instruction on Classification Posttest performance. (Author)