Summary Of Findings

Eighteen Even Start projects and 463 families participated in the EDS - 309 families were randomly assigned to participate in Even Start and 154 were assigned to a control group that could avail themselves of any educational services to which they were entitled, but they could not participate in Even Start. Pretest data on child and adult literacy skills were collected in the fall, posttest data were collected in the spring/summer, and follow-up data were collected in the following spring.

Although analysis of pretest compared with posttest data did not show that Even Start children and adults performed better than control group children and adults (see St.Pierre, Ricciuti, Tao, et al, 2003), it was hypothesized that follow-up data might show positive effects of Even Start due to either, or both, of two factors. First, families that were assigned to Even Start had the opportunity to participate in the program for approximately one school year between pretesting and posttesting. Collecting follow-up data nine months after posttesting gave those families the opportunity to participate in Even Start for a second school year, possibly leading to positive effects under the assumption that a greater amount of exposure to Even Start would lead to larger literacy gains and to statistically significant program impacts.

A second hypothesized reason that positive effects might show up in follow-up data is that change in some outcome measures is more long-term in nature than others, and impacts might be evident after two school years, even if there were no impacts after one year. An example is household income, which might not change until parents have increased their literacy skills or education level and found a job, or acquired a better job. Another example is child literacy levels, which are hypothesized to change, in part, as a result of temporally prior changes in parenting skills and parent literacy. It takes time for the latter to occur, so it is possible that changes in child literacy that are not apparent after one school year of Even Start might be seen after two school years, due to changes that occur in parents.

The follow-up data do not support either of these hypotheses. Findings from analyses of literacy gains based on a comparison of pretest data with follow-up data (collected about 18 months after pretest) are almost identical to the findings based on a comparison of pretest data with posttest data (about nine months after pretest). In short, at follow-up, while Even Start children and parents made gains on literacy assessments and other measures, children and parents in the 18 Even Start programs that participated in the EDS did not gain more than children and parents in the control group.

We hypothesized that positive effects might be observed by the time of the follow-up data collection if families assigned to Even Start participated in instructional services for a substantially longer period of time. However, data from the Even Start Performance Information Reporting System (ESPIRS) show that Even Start EDS families participated in instructional services for an average of about 8 months between pretest and posttest data collection, and for an average of 10 months between pretest and follow-up data collection. Thus, providing the opportunity for an additional year of exposure to Even Start resulted in little additional participation on the part of families (an average of two additional months).

Another way in which we sought to understand participation in Even Start was to ask parents about the instructional services in which they participated during the prior year. At the posttest, significantly more Even Start parents than control parents reported that their children participated in early childhood education (72% vs. 32%), and that they participated in adult education (59% vs. 29%) and in parenting education (28% vs. 17%). By the follow-up assessment, more Even Start parents than control parents reported that their children participated in early childhood education (57% vs. 44%), and roughly equal percentages of Even Start and control parents reported participating in adult education (40% vs. 32%) and parenting education (17% vs. 20%). Although these data show that significantly more Even Start families than control families participated in instructional services, the participation rates for the two groups are not large in an absolute sense and are not nearly as different as might be expected, indicating both that Even Start families did not participate fully in the program, and that control families were able to find competing instructional services. Thus, in order for Even Start families to perform better than control families on literacy-related outcome measures, the instructional services in which Even Start families participated would have to be much more effective than the instructional services in which control group families participated. This is unlikely since many Even Start projects build on instructional services that already exist in the community - the same services in which control group parents and children may have been participating.

Finally, it was hypothesized that positive effects on child literacy might be observed in the follow-up data collection because of the additional time that parents have to improve their parenting and literacy skills. These improvements would then be hypothesized to translate into changes in their interactions with their children, and additional family resources. However, the data do not show significant impacts of Even Start on parents' literacy skills or parenting behaviors, so the subsequent lack of impacts on children is not surprising.

Top