Testing the Significance of Comparisons in Sociological Data

Abstract

To explain the difference in parole violation rates of lone and group offenders, a large group of parole cases were successively subdivided as native or foreign born, as having or not having a previous criminal record, and finally, as group or lone offenders. Among the foreign-born first offenders the difference in violation rates was significant; it was not significant, however, for recidivists. But a comparison between these two differences indicated that the absence of a previous criminal record was definitely associated with the lower violation rate of group offenders. A similar treatment of native cases showed that there was no such association between criminal record and presence or absence of associates. However, the difference in these differences as between native and foreign-born cases was significant and therefore would warrant further subdivision with an additional variable. This process of successive subdivision is suggested as a fruitful method of sociological analysis.