Problems of generalizing from a sample of a population called deviant: The case of methadone-maintained women and their infants

The difficulties of obtaining classical experimental and control groups when sampling populations called deviant often lead researchers to analyze data from index and comparison groups (for example, drug users and non-users, respectively). When one is dealing with such samples, classical hypothesis testing and its associated inferential statistics are inappropriate. Yet, social scientists can generalize from their samples on the basis of facet theory, with its associated tools of the mapping sentence, smallest space analysis (SSA), and multidimensional scalogram analysis (MSA). These are illustrated and discussed with data taken from a large research project concerned with infants born to women who were using methadone.