In educational research and evaluation, a sample of subjects usually received some type of programmatic treatment. Outcome scores for these students are then compared with outcome scores of a control or comparison group. M. Lewis and H. McGurk (1972) have pointed out that there are some implicit assumptions when this approach is applied to programs for infants and toddlers. These are: that infant intelligence is a general unitary capacity; that mental development can be enhanced by enriching the infant's experience in a few areas; and that infant scales can reflect the improvement that results from a specific enrichment experience. The contrasting view is that infants and toddlers undergo rapid, nonlinear growth and change along many lines of development. This digest examines these contrasting assumptions, first by examining the short-term and long-term consistency of test scores. Research has shown that test-retest reliability is extremely low for infants and increases moderately for toddlers, a finding that is consistent with the view of the child going through nonlinear growth. Classic studies of mental growth in infants and toddlers show inconsistent and unpredictable growth rates and demonstrate little predictive validity for early studies of mental growth. A close look at statistical implications suggests that it is not likely that the researcher or evaluator could find significant differences if they do exist. For these reasons, in lieu of control-comparison group testing, the use of case studies, the computation of effect sizes, and the examination of growth curves are suggested for evaluations of programs for infants and toddlers. (Contains 13 references.) (SLD)