Endnotes
1. Many other states would be appropriate for this purpose. Florida, Georgia, Nebraska, and South Carolina were chosen because the larger project through which this research was funded focused on the southeastern United States. Because of regional variations in both crime and the structural correlates specified in social disorganization theory, this study includes a midwestern plains state to assess, for a second region, the generalizability of the findings.2. For a detailed discussion of these statistical problems and their resolution, see Osgood, 2000.
3. This is a simple means of conveying the information in the regression coefficients of the negative binomial analysis. That statistical model assumes a logarithmic relationship between the explanatory variable and the outcome, which implies that unit differences on the social disorganization variables are associated with proportional differences on delinquency rates.
4. Although the analysis included the unemployment rate as a second index of economic status, the results for this variable were not very informative because the rates varied so little within each state that the estimates were too imprecise to be meaningful. Although unemployment was associated with higher rates of most of the offenses examined, none of those relationships approached statistical significance (p > .35 in all cases). For a complete presentation of these analyses, see Osgood and Chambers, 2000.
5. This implies that the relationship was curvilinear. In technical terms, the analyses allowed for curvilinearity by adding the square and cube of population size as additional terms in the regression model. There was significant evidence of a curvilinear relationship for the Violent Crime Index, aggravated assault, and simple assault. For rape, the deviation from linearity was of borderline significance, as was the overall relationship of population size to offending (p < .10).
| Previous | Contents | Next |
 |
|
| Community Correlates of Rural Youth Violence | |