PREDICTION OF SPECIAL-EDUCATION PLACEMENT FROM BIRTH CERTIFICATE DATA

The overall goal of this research effort was to develop procedures for accurately identifying children at high risk for special education pl acement, based on information available at the time of birth. A file c ontaining information on all births in New York City between 1976 and 1986 was matched against the 1992 BIOFILE, which contains information on all children enrolled in the New York City public school system in 1992. A matched file containing birth and school information on 471,16 5 children resulted from this process. Three sets of risk factors were derived from birth certificate data: parental, pregnancy-related, and child-related. Using these risk factors as independent variables, a s urvival analysis model was developed predicting special education plac ement for each of three major disability categories: learning disabili ty, emotional disorder, and mental retardation. A model combining all disability categories was also developed. The significant predictors o f special education placement were Medicaid payment for birth (a pover ty indicator), unmarried status of mother, large family size, low pare ntal education, a mother born in the United States, a low level of pre natal care, male gender, low birthweight, and a low Apgar score. Male gender was the strongest risk factor in all models. Examination of sel ected survival curves indicated that the predictive power of the model s is substantial. The methodology described in this article can be use d to identify at-risk children for whom screening and other early inte rventions, including preschool programs, may be appropriate. In additi on, these methods can be adapted to conduct long-term tracking of at-r isk children, to conduct cost-effective evaluation of early interventi ons, and to contribute to the development of long-term enrollment proj ection models.