Ninety percent of participants completed some T2 measures, and some data were available for 86% of participants at T3. To reduce problems associated with listwise and pairwise deletion, we imputed missing values in SPSS using the expectation maximization algorithm after specifying a normal distribution. Little’s (1988) missing at random (MCAR) test was not significant, χ2(5368, N = 214) = 5448.97, ns, which supports the method of imputing data. The following analyses are based on the imputed data. See Table 2 for the number of children in each problem behavior group at T1, T2, and T3 prior to and after we imputed missing values.
