ABSTRACT
This chapter looks at some of the problems that are particularly significant in multivariate statistical analysis. It provides some problems their users have encountered or are likely to encounter. Bernard Lander's Towards an Understanding of Juvenile Delinquency has been justly honored as the first work in delinquency to make significant use of multivariate statistical techniques. In explanation, the new variable is causally prior to the original independent variable; the statistical finding that it accounts for the original relation thus leads to discarding the original relation as causally spurious. Lander's interpretation of the results of his factor analysis resembles those of his partial correlation analysis. He rejects a correlation of.684 between his anomic and socioeconomic factors as merely statistical. The key to demonstrating that socioeconomic conditions are a cause of delinquency is precisely the statistical relation that Lander had invoked to get rid of socioeconomic conditions.
