This report presents an analysis of the effectiveness of the Talent Search in Florida, Indiana, and Texas, drawing upon project, state, and federal administrative records to examine short-term program outcomes for program participants and a matched comparison group. The specific goals of the Talent Search program are to identify qualified youths with potential for postsecondary education programs, publicize the availability of student financial aide, and encourage secondary and postsecondary school dropouts to reenter an educational program.

KEY FINDINGS

The main research questions posed in this design and addressed in this report are (1) Is it possible to rely on administrative records to compile a complete, retrospective record of participation in Talent Search, characteristics of students in secondary school, and secondary and postsecondary outcomes? And (2) Can administrative data sources and quasi-experimental techniques be used to identify students similar to Talent Search students but who did not participate in the program, in order to generate valid estimates of the relationship of participation in Talent Search to secondary and postsecondary outcomes?

The Efficacy of Using Large State Databases to Inform Policy

The compilation of data from administrative data sources to study the effect of Talent Search on participants was a success in three states: Florida, Indiana, and Texas. Information was assembled on the characteristics and outcomes of the cohort of students who were in ninth grade during the 1995-1996 school year, along with participation by any of the cohort members in Talent Search. Because a large amount of data was compiled in each state, complex propensity score matching models were used to identify nonparticipating students who were most similar to Talent Search participants. However, analyzing the effect of Talent Search on participants in any one state suffered from serious limitations. Still, the variations in the data, as well as the strategies for drawing comparison students in each state, provided a good test of the sensitivity of the findings to using different data sources and comparison students.

Findings Across States

This research is limited in its ability to attribute participation in Talent Search to improved outcomes in any one state. However, findings that were consistent across all three states suggest greater robustness. These findings include:

• Financial Aid Applications. Talent Search Participants were much more likely than nonparticipants from similar backgrounds to be first-time applicants for financial aid in the 1999-2000 school year. The difference in financial aid application for Talent Search participants and nonparticipants was 14, 17, and 28 percentage points, respectively, for Indiana, Florida, and Texas (see Figure 1). The difference was smallest in Indiana, where we had the strongest measures of educational aspirations; even in Indiana, however, the gap represents applications levels for financial aid that are one-third higher for Talent Search participants.

• Postsecondary Enrollment. Talent Search participants were more likely than nonparticipants to enroll in a public college or university in their state by the 1999-2000 school year. Initial enrollment in a postsecondary institution was higher by 6, 14, and 18 percentage points, respectively, for Indiana, Florida, and Texas (see Figure 2).

• Enrollment in Two -Versus Four-Year Institutions. While the study reveals differences in enrollment rates for both two- four-year institutions, the differences are much larger and more robust for two-year enrollment. Enrollment by type of institution (two- or four-year) was linked to the type of institution hosting the Talent Search project. In general, projects did not increased enrollment by exposure to their type of institution, or to their specific institution.