Determining differential changes in outcomes across years for particular groups of students involved a test of interaction effects. These interaction effects were tested with a two-way Analysis of Variance (ANOVA). For example, in comparing the change in total degree completion between BPS participants in 1989–90 and 1995–96 by income level, a test was conducted on the interaction between income level and a variable representing year. An interaction effect significant at the 0.05 level indicated that the amount of change in completion between the two cohorts was different for students from different income levels.

In creating the two-way Analysis of Variance, the squares of the standard errors, the variance between the means, and the unweighted sample sizes were used to partition total sums of squares into within- and between-group sums of squares. These were used to create mean squares for the within- and between-group variance components and their corresponding F statistics. The F statistics were then compared with F values associated with a significance level of 0.05. Significant values of both the overall F and the F associated with the interaction term were required as evidence of a relationship between year and the row variable of interest. Means and standard errors were calculated by the DAS. Unweighted sample sizes are not available from the DAS and were provided by NCES through a restricted-use data license agreement.