| Abstract: | Previous attempts to ground structural effects have averaged the responses of individuals in survey research and consequently equate properties of the respondent with attributes of the system itself. This technique (Blau, 1960; 1974) generates philosophical controversy which can only be avoided by measuring properties of criminal justice agencies having no counterpart at the individual level (offenders, police, correctional workers) of analysis. Sampling designs must subsume both organizations and individuals to formulate generalizations about interorganizational processes. Applying conventional multivariate statistical models, we are able to examine one type of characteristic while simultaneously adjusting for the effects of the other. |
