NCJ Number
208873
Journal
European Journal of Criminology Volume: 2 Issue: 1 Dated: January 2005 Pages: 93-97
Date Published
January 2005
Length
5 pages
Annotation
This research note corrects a multilevel regression analysis on the effects of neighborhood contextual variables on serious juvenile offending, previously reported in this journal.
Abstract
Multilevel regression analysis is an innovative methodology that has not been widely used outside of the United States. As a result of inexperience with these models, the statistical analysis presented in Volume 1 of this journal was partially flawed; the multilevel regression analysis is corrected here by the original author. The flaw occurred in models 3 and 4, reported in Table 8; the level-1 predictors were incorrectly group-mean centered, resulting in an inflation of the estimates of the coefficients of the neighborhood-level predictors. The corrected model lowers these estimates and, as a result, only the variable “intergenerational closure” survives as a significant predictor. Moreover, the strongest variable in the original model is rendered insignificant in the corrected analysis. The more widespread use of multilevel analyses will hopefully contribute to a better understanding of both its advantages and disadvantages. Table, reference