Promoting College Enrollment among Disadvantaged Students: A Randomized Controlled Trial of Two Low-Cost Interventions
Last registered on March 06, 2020

Pre-Trial

Trial Information
General Information
Title
Promoting College Enrollment among Disadvantaged Students: A Randomized Controlled Trial of Two Low-Cost Interventions
RCT ID
AEARCTR-0005530
Initial registration date
March 05, 2020
Last updated
March 06, 2020 3:05 PM EST
Location(s)
Primary Investigator
Affiliation
UCLA Luskin School of Public Affairs
Other Primary Investigator(s)
PI Affiliation
UCLA Luskin School of Public Affairs
Additional Trial Information
Status
Completed
Start date
2011-10-03
End date
2019-11-25
Secondary IDs
Abstract
This research project is a random assignment evaluation of the effects of two variants of a virtual college counseling intervention designed to reduce informational and social support barriers to college application and enrollment among socioeconomically disadvantaged students. The 15-month intervention was implemented by EdBoost Education Corporation, a Los Angeles non-profit education services provider. The intervention began in the spring of students’ junior year and involved students from low-income, predominantly poor, non-white high schools in Southern and Central California. Students were randomly assigned to one of two nested variants of the Virtual Student Outreach for College Enrollment (V-SOURCE) intervention. Students in the Milestones variant received access to a comprehensive website; emails and text messages sent several times a month with information tailored to the timing of particular college access activities and reminders about important deadlines, as well as links to relevant content on the website; and $20 electronic gift card rewards for completing four key milestones in the college application process. Students assigned to the Complete variant received everything in the Milestones variant plus access to a personal advisor who communicated with the students through emails, texts, phone, and social media.
External Link(s)
Registration Citation
Citation
Phillips, Meredith and Sarah Reber. 2020. "Promoting College Enrollment among Disadvantaged Students: A Randomized Controlled Trial of Two Low-Cost Interventions." AEA RCT Registry. March 06. https://doi.org/10.1257/rct.5530-1.0.
Sponsors & Partners
Sponsor(s)
Partner(s)
Type
ngo
Experimental Details
Interventions
Intervention(s)
We study two variants of Virtual College Outreach for College Enrollment (V-SOURCE), a 15-month intervention designed to help disadvantaged students with all aspects of the college and financial aid application process. Students in the Milestones variant received access to a comprehensive website; emails and text messages sent several times a month with information tailored to the timing of particular college access activities and reminders about important deadlines, as well as links to relevant content on the website; and $20 electronic gift card rewards for completing four key milestones in the college application process. Students assigned to the Complete variant received everything in the Milestones variant plus access to a personal advisor who communicated with the students through emails, texts, phone, and social media.
Intervention Start Date
2012-03-21
Intervention End Date
2014-08-18
Primary Outcomes
Primary Outcomes (end points)
College enrollment in the first fall after projected high school graduation, by type of college. Persistence in college, by type of college.
Primary Outcomes (explanation)
We obtained data from the National Student Clearinghouse (NSC), matched to study participants using name and DOB, to construct the enrollment and persistence outcomes. The NSC data were merged with college characteristics from IPEDS and Barrons. We construct indicators for the enrollment outcomes for the same categories as applications and acceptances: (1) any four-year college, (2) any selective 4-year college, (3) any CSU, and (4) any UC. Selective colleges are those with Barron's ratings of very competitive plus to most competitive. The enrollment measure is equal to 1 if a student enrolled in the specified category in the first fall after expected on-time graduation from high school. The measures of persistence are equal to one if a student met the criteria in both the first and second falls after expected on-time high school graduation. For example, the two-year persistence measure for attending a 4-year college is equal to one if the student attended a 4-year college in both the first fall and the second fall, even if it was not the same institution or enrollment was not continuous.
Secondary Outcomes
Secondary Outcomes (end points)
1) Measures of students' experience with the process of applying to college and for financial aid: Sought Information, Had Information, Had Support.
2) Completion of key steps in the application process: Registered for the SAT/ACT, took the SAT/ACT, submitted college applications to two separate systems, submitted FAFSA on time.
3) Indicators for applying to at least one four-year college in each of the categories described above for the enrollment and persistence outcomes.
4) Indicators for having been accepted to at least one four-year college in each of the categories described above for the enrollment and persistence measures.
Secondary Outcomes (explanation)
For on-time FAFSA completion, we used data obtained from the California Student Aid Commission (CSAC), matched to participants in our study using name and DOB. All other outcomes were constructed from self-reports on the Follow-up survey.
Experimental Design
Experimental Design
The V-SOURCE Milestones treatment was less expensive compared to V-SOURCE Complete, so Milestones would be cost effective with smaller treatment effects. Thus, to improve power to detect small treatment effects in the Milestones treatment and stay within our budget, we chose to assign fewer students to V-SOURCE Complete than to the other two treatment arms. We planned to assign students to Complete, Milestones, and Control in a 2:3:3 ratio. However, we over-recruited slightly in cohort 2 and divided the extra students evenly between Milestones and Control. Because earlier randomized studies of college access programs found heterogeneous effects by gender, parental education, and/or race/ethnicity and home language, we randomly assigned students to each treatment within blocks created by fully interacting gender (2 categories: male and female), parental education (2 categories: at least one parent attended college, excluding vocational; and no parent attended college), and a race/ethnicity-home language composite (3 categories: Hispanic and speaks Spanish at home, Hispanic and does not speak Spanish at home, and all other students). The interaction of these categorical variables generated 12 blocks; we put students who had missing data on any of these variables in a separate block.
Experimental Design Details
Randomization Method
Random assignment was done on a computer using STATA.
Randomization Unit
Random assignment was at the student level.
Was the treatment clustered?
No
Experiment Characteristics
Sample size: planned number of clusters
Students from 84 high schools participated in the research, but random assignment was at the student level.
Sample size: planned number of observations
6640 students
Sample size (or number of clusters) by treatment arms
2536 in control, 2553 in Milestones intervention, 1551 in Complete intervention
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
In our grant application, we calculated a Minimum Detectable Effect Size on four-year college enrollment of .038 for comparing Milestones to Control and .043 for comparing Complete to Control and Complete to Milestones (two-tailed test, 80% power, alpha = .05), assuming we met our recruitment targets.
Supporting Documents and Materials
Documents
Document Name
Background
Document Type
other
Document Description
This document provides a brief description of the research and how we handled the choice of sub-groups for analyzing heterogeneous treatment effects.
File
Background

MD5: d4e9f587fd81ee731807dfadb6001f65

SHA1: 77ea229e7440413e1ba7a226c464ac8de1251df0

Uploaded At: March 04, 2020

IRB
INSTITUTIONAL REVIEW BOARDS (IRBs)
IRB Name
UCLA Institutional Review Board
IRB Approval Date
2011-10-03
IRB Approval Number
11-002465
Post-Trial
Post Trial Information
Study Withdrawal
Intervention
Is the intervention completed?
No
Is data collection complete?
Data Publication
Data Publication
Is public data available?
No
Program Files
Program Files
Reports, Papers & Other Materials
Relevant Paper(s)
REPORTS & OTHER MATERIALS