Study
Bos and colleagues (2012) used an experimental design to evaluate the SOURCE program in a sample of high school juniors from schools in the Los Angeles Unified School District (LAUSD). Students in this district were predominantly Latino/a and economically disadvantaged, based on qualification for free or reduced lunch. Eligible students were recruited by letters sent to students and families, flyers distributed at college fairs, and information sent to high school counselors. After obtaining informed consent from students and their parents or guardians, students were randomly assigned to the intervention (SOURCE program) or a control group. Each student in the intervention group was matched with an advisor (either 1 of 67 college student advisors or 1 of 2 coordinators who also served as advisors). College advisors worked individually with each of their assigned students to guide them through a set sequence of college and financial-aid application steps. The control group received usual services for students through the school system.
Of approximately 15,000 students targeted for recruitment, 2,499 were verified as eligible and consented to participate. Of this group, 1,000 were randomly assigned to the SOURCE intervention condition, and 1,499 were assigned to the control group. A subsample of 150 students initially assigned to the control condition were randomly assigned to a waiting list to ensure that all 1,000 program slots would be filled with eligible recruited students. The remaining 1,349 students served as the control group. The study sample was approximately 70 percent female. The ethnicity of the sample was Latino/a (62.0 percent), Black (12.8 percent), Asian or Pacific Islander (12.4 percent), and white (10.0 percent). The primary student languages were English (47.2 percent), Spanish (45.3 percent), and other (7.5 percent).
Data sources included 1) demographic and educational aspiration information collected from students at baseline; 2) administrative program data from SOURCE; 3) LAUSD administrative data on high school completion, GPA, and SAT outcomes; 4) a control group follow-up survey conducted 1 year after expected graduation from high school; 5) a follow-up survey administered to the full sample 1 year after expected high school graduation; and 6) 3 years of college data from the National Student Clearinghouse (NSC), with beginning data collected 25 months after random assignment. Qualitative and quantitative implementation data was also collected via information interviews with program staff.
Outcome data on college attendance and persistence was collected during the students’ third year following high school graduation (approximately 36 months following program participation), which corresponded to October of the students’ expected junior year of college. NSC data was used to assess whether students were enrolled in any college, at California State University (CSU) and/or University of California (UC), any 2-year college, or any 4-year college. NSC data was also used to determine (during the 26 months following the month of expected high school graduation) the number of months students had been enrolled in any college, CSU and/or UC, any 2-year college, or any 4-year college. Student transfers from a 2-year to a 4-year college were also obtained from NSC data. At the time of expected graduation from high school, administrative data from LAUSD was obtained for SAT completions, SAT scores, and graduation outcomes (whether students had graduated, GPA at graduation, percentage with GPA of 3.0 or higher, and eligibility to attend CSU). In the year following expected high school graduation, participants reported whether they had applied to college, and NSC data was used to assess enrollment in any college, 2-year college, 4-year college, and at UC or CSU.
Effects of the program on outcomes were evaluated using regression analyses with imputation for missing data. Analyses controlled for the following baseline covariates: student reports of their age and race/ethnicity, likelihood of attending college, having a sibling in college, parent college graduation status, and whether Spanish was the student’s primary language. Differences at baseline on these covariates between the intervention and control groups were not statistically significant.