The instability of the delinquency types also underscores the need for multiwave longitudinal data that are collected over several contiguous time periods to study delinquency (and perhaps other behavior problems). Classification of individuals into types on the basis of cross-sectional data is likely to misclassify the nature of many individuals, and the identification and strength of risk or protective factors derived from such data may be in error. Similarly, longitudinal designs involving a single early measure and a measure much later in life may also suffer similar problems (unless infallible memories are presumed, and the intermediate data are obtained).
Finally, it should be noted that the process presented here is largely descriptive in nature and not based on complex statistical models. On the one hand, this is a very real limitation -- the statistical procedures developed and being developed for longitudinal data are not being utilized. On the other hand, interest in typological approaches also appears to be increasing once again, and an emphasis on longitudinal taxonomic description may provide valuable insights, thus following Cattell's ( 1965) dictum that "nosology precedes etiology."
Parental monitoring. This is a measure of parents' knowledge of a child's whereabouts during the day, fixed times to be home, and knowledge of child's friends. (Alpha inappropriate, and test-retest correlation not available.)
Youth involvement in various activities. This measure indicates the amount of time a youth spends in various activities, such as school activities or religious activities. (Alpha inappropriate, and test-retest reliability not available.)
Attitudes toward delinquency. This is a 12-item measure that indicates how wrong a respondent believes it is to engage in different delinquent acts. (α = .91)
Neutralization. This 11-item measure indicates the willingness of a respondent to use a variety of excuses for engaging in delinquent behavior. (α = .83)
Guilt. This is a 7-item measure that indicates how bad or guilty a respondent would feel if he or she engaged in a variety of delinquent behaviors. (α = .85)
Empathy. This is an 11-item measure developed by Eysenck and Eysenck ( 1978) to measure empathetic responses in a number of situations. (α = .67)