Compensating for deficiencies in perinatal data sets: parametric perspectives.

Considering the investments made in a research demonstration project— including the hours of work in research design, provision of services, data collection, data entry, money spent, and professional reputation—great care should be taken when testing hypotheses, but often it is not. A small amount of extra effort at the data analysis stage frequently can yield great benefits in terms of more accurate findings. The reason: Very few data sets are ideal, and among nonideal data sets, those of perinatal drug abuse treatment projects have few equals.