There are several limitations to the study related to attrition of the most severe cases and generalizeability, which have been detailed in prior papers using this sample (Delucchi et al., 2004; Kaskutas et al., 2005; Matzger and Weisner, in press; Weisner et al., 2003a). Concerning attrition, although the study had high follow-up rates, attrition was higher for those whose problems were more severe and thus may produce some bias in our findings. This is a constraint with the use of any longitudinal data. Regarding generalizeability, the county chosen was selected on the basis of diversity in its population characteristics and mix of rural and urban areas (Greenfield and Weisner, 1995; Weisner and Schmidt, 1992).
