The EDS is a randomized experiment, and if we examine gains for subgroups that participate a lot or a little, we lose the advantages offered by randomization and open up the findings to competing interpretations. Although cautious about the potential pitfalls of this approach, the previous report from this evaluation (St.Pierre et al, 2003) included analyses of the relationship between amount of participation and outcomes for Even Start children and parents based on pretest and posttest outcome data. Findings from those previous analyses are summarized below (these analyses were not replicated with the follow-up data).
One analysis eliminated all Even Start children whose parents reported that they did not receive early childhood services, as well as all control group children whose parents reported that they did receive early childhood services. The data showed that segmenting the sample in this fashion did not change the evaluation findings.
A second analysis investigated the relationship between child and parent literacy outcomes and a host of variables including monthly hours of child participation in early childhood education, monthly hours of parent participation in adult education and parenting education, and family background factors. We found (1) no relationship between the amount that parents participated in adult education and their scores on literacy outcomes; (2) children who participated more intensively in early childhood education scored higher on literacy outcomes; (3) parents who participated more intensively in parenting education had children who score higher on literacy outcomes; and (4) more intensive participation in adult education is associated with lower scores on some child outcomes. It is important to remember that amount of participation was not manipulated experimentally, and so factors other than participation in Even Start may be responsible for the observed relationships.