Improving the identification of high-risk infants.

In previous works, information contained on standard U.S. birth certificates was linked to intelligence, academic achievement, and social adjustment in first grade for a statewide random sample. The present screening effort was concentrated on a sample that was already considered to be at high but still variable levels of risk. The families of the 52 black subjects were characterized by low levels of formal education, a fairly low level of maternal intelligence, and very low incomes. By focusing on a group at relatively high risk, we hoped to increase the precision and economy of early identification by using more process-oriented information about the children's early development and environment. As the children attained 6 months of age, characteristics of the mother, the child, and the home were assessed. At 2 years of age, the children were given a Stanford-Binet and divided into a nonretarded group (IQ greater than 85, n = 27) and a high-risk group (IQ less than 85, n = 25). We used a step-wise discriminant analysis to derive a predictor set consisting of mother's democratic attitudes, child's temperament, and the amount of time the child spent outside the home that allowed for the correct prediction of 75 percent of the children, with an overall miss rate of 20 percent and a false positive rate of 29.6 percent.