Study 1
Barr and Castleman (2017a) examined the effects of the Bottom Line program through a two-cohort randomized controlled trial in three cities: Boston, Mass. (treatment n = 510; control n = 118), New York, N.Y. (treatment n = 350; control n = 262), and Worcester, Mass. (treatment n = 135; control n = 54). In the first of the two cohorts (Cohort 1), eligibility criteria to participate included having a GPA of at least 2.5, having a family income below 200 percent of the federal poverty line, and being the first in their family to go to college. High school students (n = 1,429) who applied and met eligibility requirements were randomized through a lottery to receive either an offer to participate in Bottom Line services or no services. Students who were randomized into the treatment group met with college access advisors at Bottom Line’s local offices every 3–4 weeks throughout their senior year.
The study authors did not provide specific information on the Bottom Line treatment group’s baseline characteristics. Rather, they reported results from models in which they had regressed student-level baseline characteristics on the treatment indicator and site-by-cohort fixed effects, in addition to the means of the control group. In the Cohort 1 control group, 70.3 percent of youth were female, and 78.8 percent were U.S. citizens. In regard to race, 33.4 percent were Hispanic, 29.5 percent were Black, 25.1 percent were Asian, and 9.2 percent identified as other race. The students’ average GPA was 3.3. In terms of family background, parents had an adjusted gross income of $21,424; 64.0 percent had a mother who was employed, and 68.3 percent had a father who was employed. Eighty-two percent were first-generation students, and 18.9 percent reported participating in other college access programs. There were no statistically significant differences between the treatment and control groups on baseline characteristics, except for the following: 1) treatment group youth were less likely to be U.S. citizens, and 2) treatment group youth were less likely to have fathers who were employed.
Enrollment data were collected through the National Student Clearinghouse each semester for 2 years (for a total of four semesters, or through students’ second year of college). Variables were created to indicate enrollment in a 4-year college, enrollment in a 2-year college, and enrollment in any college. Variables were also calculated to indicate whether the student was continuously enrolled (all four semesters) and the total number of semesters in which they were enrolled (ranging from 0 to 4). Findings were assessed using regression analyses that controlled for each of the reported demographic variables and site-by-cohort fixed effects, to account for the fact that the proportion of youth assigned to the treatment versus the control group varied by site and cohort.
Subgroup analyses were conducted to examine separate treatment effects for female, Black, and Hispanic youth; those who reported not participating in any other programs focused on college enrollment at study baseline; those with low GPAs; and those from low-income households.
Study 2
Barr and Castleman (2017b) used the same randomized control design to study the impact of the Bottom Line program on students in Cohort 2 (n = 993). In Cohort 2, participants were also from Boston (treatment n = 350; control n = 75), New York (treatment n = 232; control n = 188), and Worcester (treatment n = 110; control n = 38); however, data were collected only through students’ first year of college (for a total of two semesters). National Student Clearinghouse data were used to assess enrollment in a 4-year college, enrollment in a 2-year college, and enrollment in any college.
The study authors did not provide specific information on the Bottom Line treatment group’s baseline characteristics. Rather, they reported results from models in which they had regressed student-level baseline characteristics on the treatment indicator and site-by-cohort fixed effects, in addition to the means of the control group. In the Cohort 2 control group, 68.8 percent were female, and 78.5 percent were U.S. citizens. In terms of race, 31.2 percent were Black, 31.2 percent were Hispanic, 23.9 percent were Asian, and 9.6 percent reported other race. The students’ average GPA was 3.3. In terms of family background, parents had an adjusted gross income of $24,112; 64.2 percent had a mother who was employed, and 70.5 percent had a father who was employed. Almost 80.0 percent were first-generation college students, and 41.5 reported participating in other programs. There were no statistically significant differences between the treatment and control groups on baseline characteristics,
Findings were assessed using regression analyses that controlled for each of the reported demographic variables, and site-by-cohort fixed effects, to account for the fact that the proportion of youth assigned to the treatment versus the control group varied by site and cohort. No subgroup analysis was conducted.