Can Re-Enrollment Campaigns Help Dropouts Return to College? Evidence From Florida Community Colleges

Abstract

In this study, we develop and implement a text messaging re-enrollment campaign to encourage former community college students to return to college. Former students are randomly assigned to one of two treatment groups that either receives information to simplify the re-enrollment process or receives both information and a one-course tuition waiver. We find that providing information and a one-course tuition waiver to former students increases their likelihood of re-enrollment by 1.5 percentage points (21.1%), enrolling in multiple courses by 0.9 percentage points (16.4%), full-time enrollment by 0.5 percentage points (21.7%), and persistence by 0.6 percentage points (23.1%). This study highlights the importance of targeted interventions that address informational and financial barriers facing former students.

Keywords

higher education community colleges educational policy experimental design

as open-access institutions with diverse student bodies, community colleges provide access to higher education for a disproportionate number of underserved students (Bailey et al., 2015). When compared with high school graduates, individuals who attend college earn significantly higher wages (Belfield & Bailey, 2011) and accrue noneconomic benefits, including improved health, longer life expectancy, and lower rates of incarceration (Trostel & Chase, 2015). The financial benefits of community college attendance are especially high for students who complete a larger number of credits or years of study, even in the absence of degree completion (Belfield & Bailey, 2011). Unfortunately, the majority of students who begin at a community college drop out of college before they are able to obtain the benefits of enrolling in additional courses and persisting in college (Snyder et al., 2018). Disparities in college dropout rates across demographic groups are especially worrisome, with Black and Hispanic students persisting in college at significantly lower rates than their White peers (Shapiro et al., 2017).¹

Despite the well-established benefits of persistence in college, 36 million Americans have obtained some postsecondary education but eventually dropped out of college. Roughly 10% of these former students have already made substantial academic progress toward earning a degree and been identified as the most likely to re-enroll, persist, and eventually graduate from college (Shapiro et al., 2019). In light of stagnant enrollments, declining state appropriations, and a renewed emphasis on student success, many colleges have launched re-enrollment campaigns targeted at former students who have made considerable progress toward obtaining a degree (Schwartz, 2019). To encourage re-enrollment, colleges reach out to former students by sending a combination of letters, emails, and text messages. In addition to using a variety of modalities of communication, colleges sometimes offer targeted scholarships to incentivize re-enrollment among former students. Although the number of re-enrollment campaigns in higher education has grown considerably in recent years, there is a lack of causal evidence on their effectiveness.

In this article, we conduct a randomized controlled trial to evaluate the effect of text message–based interventions on re-enrollment behavior among previously successful former students at five high-enrollment community colleges. In partnership with participating community colleges, we sent multiple text messages to former students who had already made considerable progress toward their degree but were not currently enrolled at any postsecondary institution. More than 27,000 former students comprise our sample and are randomly assigned to receive one of two interventions through text messaging.

The first intervention consists of multiple text messages that provide information on how to re-enroll at the student’s prior institution and a direct link to do so. In essence, the texts simplify and streamline the re-enrollment process by providing embedded links to a website that informs the student how to re-enroll and apply for financial aid. Furthermore, the texts inform students of critical enrollment deadlines. The second intervention is similar to the first intervention but also offers a tuition waiver that covers the actual cost of enrolling in their first course. Former students who are randomly assigned to a control group do not receive any texts and, thus, are not offered a one-course tuition waiver or information that might facilitate re-enrollment.

The rationale for developing these interventions is motivated by the specific challenges that community college students face that may affect their decision to drop out and potentially re-enroll. The central motivation behind the information-only treatment rests on the assumption that community college students who are no longer enrolled have limited information and would benefit from a simplified, easily navigable path to re-enrollment (Institute for Higher Education Policy, 2011). According to models of behavioral economics, the decision making of individuals with limited information can vary on the margins when decisions are complex, risky, or uncertain (Karlan et al., 2016; Meyer & Rosinger, 2019; Thaler & Benartzi, 2004). Due to the perceived challenges associated with making complex decisions, former students may be inattentive to critical deadlines or deterred in carefully evaluating the costs and benefits of re-enrolling. Accordingly, they may avoid taking any action altogether. Because our interventions provide timely reminders and information that simplifies the re-enrollment process, treated individuals may be more likely to re-enroll.

The most prevalent contributing factor to explain why former students drop out of college is their inability to finance their postsecondary education (Bers & Schuetz, 2014). Among former students who made substantial progress toward earning a degree before leaving college, financial incentives in the form of scholarships or tuition waivers have been identified as critical to re-enrollment efforts designed to address the reasons former students dropped out of college in the first place (Institute for Higher Education Policy, 2011). Therefore, offering students one-course tuition waivers lowers near-term costs of financing a college education and may incentivize re-enrollment.

Overall, we find no relationship between providing former students informational nudges, alone, and their likelihood of re-enrollment. However, we do find positive effects on the likelihood of former students re-enrolling when informational nudges are combined with a one-course tuition waiver. More specifically, former students who received both the information and one-course waiver are 1.5 percentage points (21.1%) more likely to return to college. The effects of the informational nudges combined with a course waiver are especially strong among former students who have lower grade point averages (GPAs), earned a substantial number of credit hours when previously enrolled, and who are older than traditional undergraduates.

Our study makes two important contributions. Most obviously, this research sheds light on the efficacy of types of outreach campaigns aimed at fostering re-enrollment among former students. Despite the growing popularity of these targeted outreach programs (Schwartz, 2019), little is known regarding whether these targeted outreach programs are actually effective. As such, our study provides the first causal evidence that outreach campaigns can help former students return to college. Second, there is a nascent body of evidence that shows informational nudges alone do not significantly improve college enrollment or students’ academic outcomes (Bettinger et al., 2012; Bird et al., 2019; Oreopoulos & Petronijevic, 2019). Our experimental findings align with this prior work by showing that informational nudges alone do not improve the likelihood of former students re-enrolling in college. However, our study provides strong evidence that combining informational nudges with a financial incentive improves the likelihood that former students return to college.

Literature Review

Community colleges play a democratizing role for millions of underserved students (Belfield & Bailey, 2011; Rouse, 1995), but many community college students are unable to persist in college (Snyder et al., 2018). Approximately 47% of students who begin at a community college are no longer enrolled at any college or university, and only 38% of community college students are able to remain enrolled, persist, and eventually graduate (Shapiro et al., 2017).

The Contributing Factors of Student Attrition

Prior literature has shown that many students drop out of college due to informational and financial barriers that are unrelated to their academic ability or prior performance in coursework (Long, 2007); however, most research examining why students fail to persist in college is correlational. In a survey of community college students who were close to completing their degree (e.g., 45 or more credits) before dropping out, previous researchers reported that former students were confused regarding how to be able to take additional courses after departing college and expressed a general dissatisfaction with the quality of student services and advising while they were enrolled (Bers & Schuetz, 2014). Additional work shows that students often leave college due to many financial issues unrelated to their ability to succeed in the classroom (e.g., Stinebrickner & Stinebrickner, 2008). More specifically, former students may drop out of college due to a need to work additional hours, an inability to pay the required tuition and fees, and a financial disruption caused by family obligations (Cox et al., 2016). The contributing factors affecting community college students’ likelihood of dropping out of college can be exacerbated among specific subgroups of students.

For example, Crosta (2013) highlights student attributes associated with the decision to drop out and finds community college dropouts are more likely to be older and less likely to receive financial aid. The financial barriers associated with student attrition are worse among adult students who typically have a greater burden caused by external constraints, such as the financial challenges described previously and higher levels of attrition than their younger peers (M. Bergman et al., 2014). Black and Hispanic students are also significantly more likely to drop out of college when compared with their White peers (Shapiro et al., 2017).

Prior work has also shown that low-income students are more likely to drop out, even when they receive need-based financial aid, because the amount of aid low-income students receive is typically inadequate to cover their cost of attendance beyond tuition and fees, such as transportation costs, the opportunity cost of lost wages, and financial assistance to low-income family members to cover rent and groceries (Joo et al., 2008; Peters et al., 2019). In addition to the above demographic and socioeconomic characteristics associated with dropping out of college, several academic characteristics may also help to explain why students fail to persist at a community college. College dropouts are significantly more likely to identify their poor academic performance or relatively low grades as a major reason why they left college early (Hoyt & Winn, 2004).

Institutional Responses to Student Attrition

Given widespread concerns that not enough qualified students reap the benefits associated with completing college, the federal government invests more than $120 billion each year to various initiatives designed to encourage college enrollment and persistence in college (Scott-Clayton, 2017). Many public colleges and universities, especially those whose funding is tied to institutional performance measures, invest considerable resources toward improving student persistence. In recent years, targeted initiatives have expanded beyond improving the outcomes of current students by seeking to help former students in good academic standing return to college and succeed academically.

Former students who left college before earning their degree have reported that they did not understand what was required to be able to re-enroll and lacked a clear list of next steps and specific deadlines related to the re-enrollment process (InsideTrack, 2019). Behavioral nudges can be used as a mechanism to remove informational barriers for individuals who may not be attentive to or aware of critical information or deadlines related to how to return to college (Bird et al., 2019). In an effort to encourage recent noncompleters to return to college, numerous colleges have enacted re-enrollment campaigns to increase enrollment and completion numbers (Schwartz, 2019).

As one example, the University of Memphis created a re-enrollment initiative targeted at students who stopped enrolling in courses at least a semester ago, have a 2.0 GPA, and earned 90 or more credit hours through their previous coursework (University of Memphis Finish Line Program, 2019). Waubonsee Community College (WCC) recently implemented a re-enrollment campaign in which students received a postcard or phone call offering information about re-enrolling, but the former students who were contacted via postcards or phone calls did not re-enroll at a higher rate than those students who were not contacted through the campaign (Lashure et al., 2019). In addition, Project Win-Win, a national re-enrollment campaign that took place between 2009 and 2013, included 41,710 former students from 61 institutions across nine states. Project Win-Win represents one of the most expansive re-enrollment campaigns to date. The campaign focused primarily on degree audits to award degrees to former students who were eligible to earn a degree but had not submitted paperwork to complete their associate degree, but the final step in the Project Win-Win sequence of tasks was to attempt to contact potential completers to persuade them to return to college. Although there were considerable variations in how the participating institutions implemented the steps of the campaign and causal inference was not a goal of the project, project personnel reported that future re-enrollment campaigns should focus specifically on subgroups of students who had already made considerable progress toward their degree (fewer than 12 credits remaining) and recently failed to persist (Adelman, 2013).

Prior Evidence on Interventions to Increase Enrollment

Numerous studies have employed informational interventions to determine the effect of nudges on the behaviors of both prospective and current college students. Castleman and Page (2015) found that their text messaging campaign had a positive impact on whether recent high school graduates enrolled in college. Additional work has shown positive effects associated with the implementation of text messaging programs on financial aid renewal among college freshmen (Castleman & Page, 2016).

Some prior research has revealed positive effects of informational nudges on prospective students’ likelihood of enrolling in college (Barr & Turner, 2018; Castleman et al., 2014; Castleman & Page, 2015; Page & Gehlbach, 2017). Castleman et al. (2014) found that contacting high school graduates via phone, email, and text messaging increased the likelihood of college enrollment among Hispanic males. Hyman (2020) mailed letters encouraging high-achieving high school seniors to consider college and engage with a custom website containing additional information. The author reported that low-income students were more likely to attend college after receiving the informational letters, but the results appeared to be driven by low-income student enrollment at 4-year institutions. Additional research has revealed that, among community college students, the positive impact of a localized text messaging campaign on college enrollment was stronger for individuals with lower high school GPAs (Castleman & Page, 2016).

Several researchers have sent emails or mailed letters to describe the benefits of college to prospective college students and found no effect on the likelihood of enrollment (P. Bergman et al., 2019; Gurantz et al., 2019). Bettinger et al. (2012) found no change in college enrollment in response to a treatment offering financial aid information, but an additional treatment offering both financial aid information and professional assistance completing the Free Application for Federal Student Aid (FAFSA) led to an increase in enrollment among dependent students and independent students with no prior college experience.

In a recent study examining the efficacy of nudge interventions at scale, Bird et al. (2019) employed a state-level and nationwide campaign that reached more than 800,000 students, finding no impact of their information-based nudging campaign on the likelihood of college enrollment for any student subgroups. Oreopoulos and Petronijevic (2019) designed online and text message interventions to improve student achievement in college and found no effect on college students’ academic outcomes—even among those students who were identified as more likely to drop out of college. According to a systematic review of experimental and quasi-experimental literature on the efficacy of outreach by colleges, outreach programs can increase access to higher education among disadvantaged students when the program offers supplemental counseling or simplifies the college application process, but outreach programs are not effective when they merely provide general information (Herbaut & Geven, 2020).

Previous literature has also examined the effect of financial aid on the likelihood of college enrollment. Castleman and Long (2016) employed a regression discontinuity design and found that receiving the Florida Student Access Grant had a positive impact on students’ likelihood of enrollment, which aligns with prior work showing the positive relationship between financial aid receipt and college enrollment (e.g., Fack & Grenet, 2015; Goldrick-Rab et al., 2016). Bettinger (2015) also found that students who received need-based financial aid through the Ohio College Opportunity Grant initiative were less likely to drop out of college. However, recent experimental evidence of the Wisconsin Scholars Grant, a privately funded need-based financial aid program, examined multiple cohorts and reported that the financial aid program had no effect on longer term outcomes of degree completion and graduate school enrollment (Anderson et al., 2020). Additional work focused specifically on nontraditional students found that eligibility to receive tuition vouchers and cash payments via a state financial aid program—California’s Competitive Cal Grant program—had virtually no impact on students’ likelihood of college attendance, degree completion, or labor market outcomes (Gurantz, 2020).

In a quasi-experimental study examining the impact of community college tuition discounts on student enrollment among recent high school graduates, Denning (2017) reported that community college enrollment increased by 5.1 percentage points for each $1,000 reduction in tuition, with Black students responding more strongly to tuition discounts than White students. Although prior work has shown that the Pell Grant receipt was not associated with an increase in college enrollment (Denning et al., 2019; Kane, 1995; Rubin, 2011), grants that supplement the Pell grant are more likely to positively influence college enrollment (e.g., Bettinger, 2015; Castleman & Long, 2016). Richburg-Hayes et al. (2015) conducted a randomized controlled trial to examine the impact of a one-time, performance-based scholarship on students’ likelihood of enrolling in college, finding that providing a $1,000 performance-based scholarship increased students’ likelihood of enrolling in college by 5 percentage points. The authors reported that the performance-based scholarship was particularly effective in inducing matriculation among students who were academically underprepared with lower GPAs, but the positive impact of the scholarship was generally consistent across student subgroups.

To explain why receiving informational nudges and one-course tuition waivers may affect a former student’s likelihood of re-enrolling in college, this study considers the logic of behavioral economics, which rests on the assumption that students’ decision making can vary on the margins when decisions are complex, risky, and uncertain. In general, studies incorporating behavioral economics are designed to identify barriers to making an optimal choice, classify the typical responses to the previously identified barriers, and develop solutions to overcome the typical responses (Meyer & Rosinger, 2019). Seminal work by Kahneman and Tversky (1979) examines how the framing and context of choices presented to individuals can affect how those individuals make complicated decisions under some type of constraint or uncertainty. Given the challenges associated with making complicated decisions with limited information, individuals may be deterred by the challenges associated with carefully evaluating the costs and benefits of a complex decision and subsequently delay or avoid taking any action altogether (Karlan et al., 2016; Thaler & Benartzi, 2004).

In response to these dynamics, researchers have developed and implemented a host of behavioral nudges and interventions designed to alter individuals’ decision making in ways that encourage what has been identified as an optimal choice (Thaler & Sunstein, 2008). After the identification of the behaviorally rooted challenges that individuals encounter, colleges can make structural programmatic changes and offer support to simplify and improve the likelihood of re-enrollment. For example, Meyer and Rosinger (2019) highlight that colleges can simplify processes by removing unnecessary barriers to entry and support decision-making processes by sending timely, salient, and actionable reminders that can lead to active choices by the individuals receiving the behavioral nudge or intervention. In alignment with prior literature and the logic of behavioral economics, we examine the efficacy of informational nudges and one-course tuition waivers on the likelihood of former students re-enrolling in college.

Research Design

This experimental study was conducted in collaboration with five high-enrollment community colleges located in the state of Florida. Each community college was selected due to its large enrollment size (roughly 64,000 students per community college) and diverse student body. We collaborated with each institution during the 2018–2019 academic year to conduct the text message–based re-enrollment campaign.

Data and Sample

The text message–based re-enrollment campaign focused solely on former students who were identified as previously successful academically and eligible to re-enroll at their prior community college. To be included in our sample and eligible for treatment, former students were required to meet our agreed-upon criteria of 30 accrued credit hours, a 2.0 GPA or better, and no behavioral or financial holds that would prevent the former student from being allowed to re-enroll in college. The minimum sample criteria of 30 credit hours and 2.0 GPA were developed to include former students who made significant progress toward degree completion and would be eligible to graduate based on their prior GPA. We worked directly with the participating community colleges to further restrict our sample to former students who were enrolled previously in degree-seeking programs but stopped enrolling in courses within the past 3 years. Each participating community college provided administrative data files including all former students’ demographic characteristics, academic information, and cell phone number.

Because the intervention from this study included a text messaging component, any former students who did not have a cell phone number on file were excluded from our sample. We also partnered with a third-party vendor to ensure that all former students in our sample had an active cell phone number by removing any students who only had an inactive cell phone number on file. Taken together, 21.9% of former students were excluded from the sample due to not having a cell phone number on file or having a cell phone number on file that was later determined to be inactive. To ensure that we identified and removed any former students who had already re-enrolled or graduated from a different college or university, we matched student-level data obtained from each community college with data from the National Student Clearinghouse at the beginning of the project period and immediately before sending text messages.² The residual sample of the study includes 27,028 former community college students. Before partnering with a text messaging platform to employ the interventions (described in detail directly below), we randomly assigned eligible students to a control group or one of two treatment groups.

To better understand the distinction between our control and treatment groups, we begin by outlining the types of re-enrollment efforts that former students in the control group may have experienced. Specifically, Palm Beach State College and Hillsborough Community College did not have active re-enrollment campaigns, Broward College implemented a phone call re-enrollment campaign for a small subset of students who were close to graduation, and Valencia College would send occasional postcards or letters to encourage former students to re-enroll. Miami Dade College had the most sophisticated re-enrollment campaign, as it would send emails and text messages to students who were close to graduation (fewer than 15 credits remaining) and stopped enrolling up to 1 year prior. None of the participating community colleges previously used custom websites designed to simplify the re-enrollment process or one-course tuition waivers for former students who returned to college.

Control Group

Students within the control group received no information through this experiment.

Information-Only Treatment

Students within the “information-only” treatment group received 10 text messages directing them to re-enroll in college and visit our custom website to be able to streamline the re-enrollment process, apply for financial aid, or contact an assigned advisor for assistance. Within this targeted correspondence, former students received additional (and streamlined) information pertaining to how to re-enroll, timely reminders to re-enroll before critical deadlines, links to a previously nonexistent website designed specifically to simplify the re-enrollment process, the name and contact information of an assigned advisor, and encouraging appeals to return to college (e.g., the college “wants you back” or “you are so close to finishing your degree”).

Information + One-Course Waiver Treatment

Students within the “information + one-course waiver” treatment group received the same information described in the above treatment and a one-course tuition waiver. This one-course tuition waiver could be a conditional waiver (additional aid received only if the student did not already have all coursework covered by financial aid) or unconditional waiver (refundable credit received even if all coursework is already fully covered by financial aid). Although we recommended that all participating community colleges offer an unconditional waiver, only two community colleges elected to offer the unconditional waiver and the remaining three offered the conditional waiver.

Measures

To examine the impact of each treatment on former students’ likelihood of re-enrollment, we measure re-enrollment as the primary outcome of interest and consider a host of demographic and academic characteristics to better understand our main and heterogeneous effects. First, we measure re-enrollment as a binary indicator capturing whether former students in our sample enrolled in at least one for-credit course at their prior institution during Summer 2018, Fall 2018, or Spring 2019. To further explore this outcome, we consider not only our main effects including the whole sample but also heterogeneous effects for the following subgroups of interest: individuals with lower GPAs (below 3.0), near completers (more than the median of 42 credit hours), low-income students, underrepresented minority students, and adult students (25 years of age or older). Although the motivation to examine heterogeneous effects is exploratory in nature, the decision to focus on these specific subgroups is motivated by prior literature suggesting that certain subgroups of students are more likely to drop out of college and may respond differently to targeted interventions designed to foster re-enrollment.

To improve the precision of our empirical models, we include numerous covariates related to students’ background characteristics, as outlined in Table 1. Finally, we include college fixed effects and specify additional models to consider whether each institution used a conditional or unconditional one-course tuition waiver. As described above, a conditional one-course tuition waiver allows students to waive tuition and fees for one course if at least one of their courses was not already covered by financial aid, whereas an unconditional one-course tuition waiver allows students to waive tuition and fees for one course regardless of the amount of financial aid they receive.

Table 1

Descriptive Statistics and Balance Tests

Student background characteristics	ControlM	Information vs. control	Information and one-course waiver vs. control	F-statistic(All = control)	Observations
Student background characteristics	(1)	(2)	(3)	(4)	(5)
Male	0.414	0.004	−0.001	0.248	27,028
Male		(0.007)	(0.007)	(.780)	27,028
Age	31.515	0.002	0.104	0.352	27,028
Age	(9.594)	(0.142)	(0.142)	(.703)	27,028
Black	0.237	0.004	0.004	0.276	27,028
Black		(0.006)	(0.006)	(.759)	27,028
Hispanic	0.165	−0.003	−0.002	0.188	27,028
Hispanic		(0.005)	(0.005)	(.829)	27,028
Multiracial	0.294	−0.002	−0.005	0.310	27,028
Multiracial		(0.006)	(0.006)	(.733)	27,028
Other	0.029	−0.002	−0.003	0.752	27,028
Other		(0.002)	(0.002)	(.471)	27,028
Limited English	0.127	−0.003	−0.005	0.494	27,028
Limited English		(0.005)	(0.005)	(.610)	27,028
Grade point average	2.774	0.015*	0.002	1.793	26,551
Grade point average	(0.623)	(0.009)	(0.009)	(.166)	26,551
Accumulated credits	44.700	0.318	0.310	0.572	27,028
Accumulated credits	(24.332)	(0.339)	(0.339)	(.564)	27,028
Full-time	0.121	0.007	−0.004	2.397	27,028
Full-time		(0.005)	(0.005)	(.091)	27,028
Transfer student	0.203	−0.006	−0.008	1.296	27,028
Transfer student		(0.005)	(0.005)	(.274)	27,028
Need-based aid recipient	0.485	0.006	−0.005	1.205	27,028
Need-based aid recipient		(0.007)	(0.007)	(.300)	27,028
Grade point average missing	0.018	−0.001	−0.001	0.080	27,028
Grade point average missing		0.002	(0.002)	(.923)	27,028

Note. Standard deviations are reported in parentheses in column (1). Standard errors are shown in parentheses in columns (2) and (3). p values for F tests are reported in parentheses in column (4). The number of nonmissing observations are reported in column (5).

Significant at *10% level, **5% level, and ***1% level.

Intervention Design

Former students were randomly assigned to the control group, the information-only treatment, or the information and one-course waiver treatment. In May 2018, we began our text message–based re-enrollment campaign, which included targeted messaging to treated individuals immediately before matriculation deadlines for the Summer 2018, Fall 2018, and Spring 2019 semesters. The final text message was sent in December of 2018. The extent to which these former students had received any type of outreach prior to the intervention varied across colleges. Former students assigned to the control group did not receive any text messages, but they would be granted the same opportunities for advising and financial aid assistance as any other student if they initiated the original point of contact with the institution. After the introductory text message for treated individuals, text messages were sent every 2 to 3 weeks, with a total of 10 text messages being sent to former students in either treatment group. A small number of text messages had slight differences in the language of the text message due to character limits that would not allow a single text to note that the information and one-course waiver treatment simplified the re-enrollment process and provided a one-course tuition waiver. The majority of text messages sent to former students were nearly identical. Supplementary Appendix A in the online version of the journal provides the content of each text message, which varied according to whether former students were assigned to the information-only or information and one-course waiver treatment group. A total of 1,192 former students replied to one of our text messages with “stop” to opt out of receiving future text messages. If a former student replied to any of the text messages with a question, we shared the reply and the student’s contact information with the appropriate institutional personnel.

We reviewed the “business as usual” approach associated with re-enrolling at each community college and discovered that the standard process of re-enrolling was complex, challenging to navigate, and inconsistent across all five participating community colleges. To address these potential information barriers, we coordinated across all participating community colleges and created custom websites for each intervention at each community college to simplify the re-enrollment process (10 unique websites in total). For the information-only intervention, the custom website provided a single button to re-enroll immediately, a single button to apply for financial aid immediately, and the contact information of an assigned advisor—including the advisor’s name, email, and phone number—for each former student. For the information and one-course waiver intervention, former students received the same information designed to simplify the re-enrollment process, but both the text messages and custom website clearly noted that they would receive a one-course tuition waiver upon re-enrolling.

The one-course tuition waiver covered the equivalent of three credit hours of in-state tuition at the participating community college. Each text message, regardless of the intervention, prompted former students to visit their assigned custom website. If former students replied to any of our text messages, they received an automated reply directing them to their custom website. Over the duration of the project period, 6,480 former students visited our custom websites a total of 12,348 times, with 68.7% of visits going to the information and one-course waiver website and 31.3% of visits going to the information-only website. Supplementary Appendix B in the online version of the journal shows example website for the information-only intervention and the information and one-course waiver intervention, respectively.

Randomization and Baseline Equivalence

Our study is based on a randomized block design, where we randomly assigned treatment status to former students within each of the five participating community colleges. Among the 27,028 former students in our sample, 9,010 were randomly assigned to the control group, 9,009 were randomly assigned to the information-only treatment group, and 9,009 were randomly assigned to the information and course waiver treatment group. In Table 1, we report basic descriptive statistics on students’ demographics and prior academic circumstances. More specifically, we report means for the control group and differences between each treatment group (alongside standard errors). Unsurprisingly, we find little evidence of significant differences in observed means between the control group and treatment groups following the randomization, providing reassurance that we achieved baseline equivalence and the randomization was successful.

Among former community college students, only 41.4% of the control group is male, with equal proportions of men across the treatment groups. On average, individuals in our sample are older than traditional undergraduate students. The average age for individuals in the control group slightly exceeds 31 years. Our sample also demonstrates the racial diversity of American community colleges. Almost 24% of individuals in the control group are Black while Hispanics comprise 16.5% of the control group. Also, in the control group, more than 29% of individuals identify as multiracial. About 13% of former students in the control group have limited English skills. Across every demographic attribute, we do not find significant differences in observed means between participants in the control group and those in either the information-only group or information and one-course tuition-waiver group.

We also collected data on individual’s academic performance and circumstances based on the last academic year they attended community college. For the control group, the average GPA earned is nearly 2.8. The average number of credits accumulated is 45. In general, we do not find any statistically significant differences between these measures of academic performance in the control group and the treatment groups, although there is one exception. Participants in the information-only group have marginally higher GPAs than those individuals in the control group. Nonetheless, this difference is very small and does not appear to be meaningful. As for the circumstances surrounding their most recent academic enrollment, for the control group, we find that only 12.1% of individuals enrolled full-time and that 20.3% of these individuals transferred to a community college from a different postsecondary institution. Finally, it is notable that many individuals in our sample received some form of financial assistance when last attending college. In the control group, 48.5% of individuals received need-based aid. Broadly, we report no statistically significant differences between any academic circumstances in the control group and treatment groups.

Empirical Strategy

As we are utilizing random assignment, we can measure the impact of the interventions by making simple comparisons of the outcomes of individuals assigned to either the information-only treatment or the combined information and tuition-waiver treatment with those individuals in our sample who were not treated. We can estimate the intent-to-treat (ITT) effects by using a linear probability model:

\begin{array}{l} y_{i j} = β_{0} + β_{1} I N F O_{i j} + β_{2} I N F O_W A I V E R_{i j} \\ + β_{3} X_{i j} + θ_{j} + ε_{i j}, \end{array}

(1)

where $y_{i j}$ is an indicator equaling one or zero for whether individual i who was formerly enrolled in college j decides to re-enroll in community college. The variable $I N F O_{i j}$ is a treatment indicator that equals one if an individual is offered information on how to re-enroll at college j and takes the value zero otherwise. The variable $I N F O_W A I V E R_{i j}$ is also a treatment indicator that takes a value of one if an individual is offered both information on how to re-enroll and a financial incentive in the form of a one-course tuition waiver by college j. $X_{i j}$ represents a vector of demographic variables and information on the academic performance and circumstances of individual i who formerly attended community college j. We also include dummy variables for the last academic term in which each student enrolled. All covariates included in the model are represented in Table 1. Also, $θ_{j}$ is a fixed effect for college j and signifies the randomization was conducted within each community college. The error term is represented by $ε_{i j}$ . Finally, we use robust standard errors to account for heteroscedasticity induced from using linear probability models.

Results

Estimated Effects of the Re-Enrollment Campaign

Table 2 reports our main results for the estimated effects of the information-only treatment and the estimated effects of the bundled information nudge with financial incentives on community college re-enrollment. We report the findings in separate columns based on the full sample of community colleges and whether a community college offered an unconditional or conditional tuition waiver. In the first column, we report results based on pooling individuals attending each of the community colleges. In the second column, we report results for the two community colleges that offered unconditional one-course tuition waivers. In the third column, we report results for the three community colleges that offered conditional one-course tuition waivers.

Table 2

Estimated Effects of Re-Enrollment Campaign on Likelihood of Re-Enrollment

	Institutions	Waivers	Waivers	Institutions
Information	0.005	−0.002	0.007*	0.0065
Information	(0.004)	(0.010)	(0.004)	(0.0040)
Information and one-course Waiver	0.015***	0.025**	0.011***	0.011***
Information and one-course Waiver	(0.004)	(0.010)	(0.004)	(0.003)
Information × Unconditional Waiver Offered				−0.008
Information × Unconditional Waiver Offered				(0.011)
Information and One-Course Waiver × Unconditional Waiver				0.014
Information and One-Course Waiver × Unconditional Waiver				(0.011)
Control mean	0.071	0.119	0.057	0.071
R ²	.052	.071	.029	.052
p value (F test of equality of treatment effects)	.012
p value (F test of joint significance of interaction effects)				.116
n	27,028	6,130	20,898	27,028

Note. Coefficients reported are from linear probability models of the estimated effects of information and financial nudges on re-enrollment of college dropouts, controlling for baseline covariates and college-level fixed effects. Baseline covariates are those variables included in Table 1. Observations with missing grade point average are coded with the median values and we include an indicator for covariate missingness. Robust standard errors are reported in parentheses.

Significant at *10% level, **5% level, and ***1% level.

The first column reports findings for the full sample of all community colleges. As shown, the re-enrollment rate for individuals in the control group is slightly more than 7%. For this sample, we find that the information-only treatment does not have an effect on re-enrollment but that the information nudge combined with the course tuition waiver has a significant impact on re-enrollment. Compared with the control group, individuals offered both information and one-course waiver are 21.1% more likely to re-enroll. This estimated effect is statistically significant at the 1% level. The effect of the information-only treatment on re-enrollment is 6.5% and is not statistically significant. An F test of the joint hypothesis for whether the estimated re-enrollment effect from receiving the information and one-course tuition-waiver treatments is equal to the effect of receiving the information-only treatment can be rejected at the 5% level of significance.

As noted previously, two community colleges offered former students unconditional one-course tuition waivers if they re-enrolled, and the remaining three community colleges provided conditional one-course tuition waivers. Students from the community colleges offering conditional one-course tuition waivers make up 77% of the sample. Although we report main effects separately for community colleges that offer conditional or unconditional tuition waivers (see columns 2 and 3 of Table 2), we also assess whether the treatment effects are in fact different depending on the type of one-course tuition waiver offered by the participating community colleges. To inform this hypothesis, we pool both samples of community colleges and include interaction terms that capture whether the treatment effects vary based upon whether a community college offered unconditional waivers. We find that each of these estimated interaction effects is not statistically significant. These findings suggest that the format of the one-course tuition waiver does not appear to matter and that we can focus on results from the pooled sample. As an exploratory exercise, we also compare the two treatment groups to estimate the impact of the one-course tuition waiver, as both treatment groups would receive the information aspect of the treatment, and the only meaningful difference is the one-course tuition waiver. We provide suggestive evidence of the causal effect of the one-course tuition waiver and find that offering a one-course tuition waiver increases a former student’s likelihood of re-enrollment by 13% (see Supplementary Appendix E in the online version of the journal).

Finally, we examine the effects of each treatment on former students’ likelihood to re-enroll in multiple courses, re-enroll as a full-time student, and persist to later semesters upon re-enrollment (see Table 3). Although the information-only treatment has no impact on former students’ likelihood of re-enrolling in more than one course or as a full-time student, the combined intervention including information and a one-course tuition waiver had a statistically significant positive effect on all outcomes of interest. Specifically, the information and one-course waiver treatment increases the likelihood of re-enrolling in multiple courses by 16.4%. In addition, the combined intervention increases former students’ likelihood of re-enrolling as a full-time student by 21.7%. Finally, the information and one-course waiver treatment increases the likelihood of persisting to later semesters, for returning students, by 23.1%.

Table 3

Estimated Effects of Re-Enrollment Campaign on Likelihood of Enrolling in Multiple Courses and Persisting

	Re-enroll > 3 credits	Re-enroll ≥ 12 credits	Persistence
Information	0.005	0.002	0.003
Information	(0.003)	(0.002)	(0.002)
Information and one-course tuition waiver	0.009***	0.005**	0.006**
Information and one-course tuition waiver	(0.003)	(0.002)	(0.002)
Control mean	0.055	0.023	0.026
R ²	.045	.029	.028
p value (F test of equality of treatment effects)	.301	.181	.258
N	27,028	27,028	27,028

Significant at *10% level, **5% level, and ***1% level.

Heterogeneous Effects

The effects of the information-only treatment and combined information and one-course tuition-waiver treatment may vary based upon certain observable characteristics of individuals and their academic circumstances. For all subgroups that we examined, we found little evidence that the information-only treatment has a statistically significant effect on re-enrollment. However, our findings do point to some evidence of heterogeneity for the effectiveness of the information and one-course tuition-waiver treatment (see Table 4). For example, we find that the estimated effects of the information and course waiver treatment are concentrated among individuals with lower GPAs. More specifically, for students holding a GPA less than 3.0, the intervention increases community college re-enrollment by 25.4%, while the estimated effect on individuals with higher GPAs is qualitatively smaller and not statistically significant.³

Table 4

Estimated Effects of Re-Enrollment Campaign on Likelihood of Re-Enrollment by Academic Performance and Demographics

	GPA		Accumulated credits		Semesters since dropped out
	GPA < 3.0	GPA ≥ 3.0	Accumulated credits ≤ 42	Accumulated credits > 42	Dropout ≤ Fall 2015	Dropout ≥ Spring 2016
Information	0.006	0.004	0.002	0.008	0.001	0.010
Information	(0.005)	(0.007)	(0.005)	(0.005)	(0.004)	(0.020)
Information and one-course waiver	0.017***	0.010	0.008	0.021***	0.011***	0.020***
Information and one-course waiver	(0.005)	(0.007)	(0.006)	(0.005)	(0.004)	(0.008)
Control mean	0.067	0.081	0.077	0.065	0.042	0.112
n	16,899	9,651	13,671	13,357	16,050	10,978
	Age		Race/ethnicity		Socioeconomic status
	Age ≤ 24	Age ≥ 25	Black or Hispanic	Non-Hispanic White	Low income	Not low income
Information	0.011	0.002	0.003	0.011	0.005	0.004
Information	(0.009)	(0.004)	(0.006)	(0.008)	(0.005)	(0.005)
Information and one-course waiver	0.006	0.017***	0.017***	0.021***	0.012**	0.017***
Information and one-course waiver	(0.009)	(0.004)	(0.006)	(0.008)	(0.006)	(0.005)
Control mean	0.113	0.057	0.072	0.070	0.070	0.072
n	6,752	20,276	10,883	6,257	13,121	13,907

Note. Coefficients reported are from linear probability models of the estimated effects of information and financial nudges on re-enrollment of college dropouts, controlling for baseline covariates and college-level fixed effects. Baseline covariates are those variables included in Table 1. Observations with missing GPA are coded with median values and we include an indicator for missing GPA. Robust standard errors are reported in parentheses. GPA = grade point average.

Significant at *10% level, **5% level, and ***1% level.

Next, we aim to understand whether the effects vary based on the number of credits an individual accumulated previously. The median number of accumulated credits for students in our sample is 42. We examine the effects on students who achieved more than 42 credits and conduct a separate analysis for individuals who achieved fewer credits. The effect of the information and one-course waiver treatment for individuals who accumulated fewer credits is very small and not statistically significant. However, the estimated effect on individuals who earned more than 42 credits is considerably larger. For these students, the combined intervention increases community college re-enrollment by 32.3%.⁴

The timing of when a student stops enrolling at a community college may play a critical role in determining whether that individual chooses to return. In fact, among former students in our control group, we observe that recent dropouts re-enroll at more than twice the rate of individuals who drop out earlier—11.2% versus 4.2%. We therefore examine whether the effects of the information and one-course tuition-waiver treatment vary based on when individuals left community college. Both groups of former students who left community college during or since the Spring 2016 semester and former students who dropped out during the Fall 2015 semester or earlier are significantly affected by the combined intervention of simplified information and a one-course tuition waiver. However, the magnitude of the treatment effect on earlier dropouts is stronger than the effect on most recent dropouts.

The subgroup analysis we conducted related to socioeconomic circumstances pertains to whether the individual was a low-income student (as indicated by receiving need-based aid). Almost half of all individuals in our sample received need-based aid when last enrolled. We examine whether the effect of the information and one-course tuition-waiver treatment vary based upon whether an individual received need-based aid. Our findings are robust to whether or not someone was awarded need-based aid. For low-income individuals who receive need-based aid, we find that the combined intervention of an information nudge and one-course tuition waiver significantly increases their likelihood of re-enrolling by 17.1% at the 5% level of significance. For individuals who did not receive need-based aid and may be more likely to be economically advantaged, we find that the information and one-course tuition-waiver treatment significantly increases the likelihood of re-enrolling by 23.6% at a significance level of 1%.

Our study also examines whether treatment effects may differ based upon select demographic attributes. For this study, we investigate whether the treatment effects are significant for both Black and Hispanic students and, separately, for non-Hispanic White students.⁵ For the subgroup of Black and Hispanic students and the subgroup of White students, the results are remarkably similar and both are statistically significant. The information and one-course tuition-waiver treatment increases the likelihood of re-enrollment among Black or Hispanic students by 23.6%. For individuals who are non-Hispanic White, we find that the estimated effect is an increase of 30%. We also examine whether our interventions may have differential effects based on whether someone is older or younger. For individuals 24 years and younger, we find that the information and one-course waiver treatment has a small and statistically insignificant effect of 5.3% on re-enrollment. On the contrary, for older students, we show that the combined intervention increases the likelihood of re-enrollment by 29.8% at the 1% level of significance.

Discussion

The majority of students who begin at a community college leave school before they are able to obtain the benefits of enrolling in additional courses and persisting in college. In response to high rates of student attrition, a growing number of community colleges have implemented re-enrollment campaigns designed to encourage and potentially incentivize former students to return to college. This experimental study represents the first attempt to examine the causal impact of a re-enrollment campaign and provides evidence to show that the combined information and one-course tuition-waiver treatment had a positive impact on former students’ probability of re-enrollment, enrolling in multiple courses, full-time enrollment, and persistence. Through this targeted re-enrollment campaign, we identify an intervention that increases former students’ likelihood of re-enrollment in the aggregate and across multiple student subgroups, including low-income students, underrepresented minority students, adult students, and students with low GPAs.

Previous research has reported positive effects when examining the efficacy of localized text messaging campaigns (Castleman et al., 2014, 2015; Castleman & Page, 2015), but a growing body of recent evidence has shown that informational nudges alone, when offered at a larger scale, do not appear to significantly improve the likelihood of college enrollment (e.g., Bird et al., 2019). Additional work has suggested that behavioral outreach programs can increase access to higher education among disadvantaged students when the program simplifies the college application process (Herbaut & Geven, 2020). However, we find little evidence suggesting that streamlining the re-enrollment process, without offering financial incentives, significantly affects former students’ likelihood of re-enrollment, enrolling in multiple courses, full-time enrollment, and persistence.

This study offers strong evidence of the positive impact of the information and one-course tuition-waiver treatment on the likelihood of former students re-enrolling in college, enrolling in multiple courses, enrolling as full-time students, and persisting upon re-enrollment. For community colleges that may be finding limited success through letter-writing campaigns or third-party call centers, we show that interventions including both streamlined re-enrollment information and a one-course tuition waiver have a positive impact on the probability of former students returning to college, enrolling in multiple courses, enrolling as a full-time student, and persisting upon re-enrollment. Given that a substantial number of postsecondary students leave college due to informational and financial barriers as opposed to their academic ability or classroom performance (e.g., Long, 2007), this study highlights the importance of interventions that not only simplify complex processes but also reduce the financial barriers facing community college students.

Prior work also highlights the importance of exploring heterogeneous effects when examining the impact of financial aid by showing that additional investments in financial aid programs can benefit traditionally disadvantaged subgroups of students, in particular, such as academically underprepared, underrepresented minority, and low-income students (e.g., Anderson et al., 2020; Goldrick-Rab et al., 2016). Additional studies have also shown that adult students are more likely to drop out due to financial reasons (M. Bergman et al., 2014; Crosta, 2013) and academically underprepared students with lower GPAs may benefit disproportionately from scholarships designed to induce college enrollment (Richburg-Hayes et al., 2015).

In this study, we explored the impact of the re-enrollment campaign on former students’ likelihood of returning to college across multiple academic and demographic characteristics, finding that the positive impact of the information and one-course tuition-waiver treatment held across socioeconomic and race/ethnicity subgroups and appeared to be stronger for older students and students with lower GPAs. Former students with lower GPAs, who were more responsive to the information and one-course waiver treatment than their peers with higher GPAs, were also more likely to have been enrolled previously in a STEM (science, technology, engineering, and mathematics) degree program. Community college students struggle disproportionately in the types of introductory math courses required in STEM degree programs and occasionally run out of financial aid due to the need to retake these challenging courses, and these dynamics may help to explain, at least in part, why former students with lower GPAs were more responsive to the information and one-course waiver treatment offering financial assistance. In addition, in alignment with recommendations by Adelman (2013), we found that students who completed more than the median number of credit hours were more likely to return to college.

Because the provision of a one-course tuition waiver costs between $303 and $354 for participating community colleges, one potential concern may be that a former student will only return for a single course and not allow the institution to recoup course waiver costs and generate additional tuition revenue. We ran additional specifications to examine whether providing the course waiver can make financial sense for the institutions. For example, Table 3 shows that the informational nudge alone is unrelated to the likelihood of former students re-enrolling in more than one course, but the combination of the informational nudge and one-course tuition waiver led to an increase in former students’ likelihood of re-enrolling in more than one course, an increase in the likelihood of former students returning as full-time students, and an increase in former students’ likelihood of persisting upon their return.

From the institutional perspective, the findings of our experimental study suggest that the information and one-course waiver treatment can represent an evidence-based mechanism to meet enrollment demands and generate additional tuition revenue for participating community colleges. However, administrators should also consider the potential for unintended consequences such that current students may be incentivized to stop enrolling in the hope that they would be included in a future re-enrollment intervention and receive a one-course tuition waiver. From a societal perspective, the extent to which re-enrollment benefits former students represents an important area of future inquiry.

The simplest way to examine whether the information and one-course waiver treatment is “worth it” from a societal perspective is to consider the lifetime impact of returning to college. Prior work has repeatedly shown that college enrollment and degree completion have a positive impact on an individual’s lifetime earnings (e.g., Hoekstra, 2009; Zimmerman, 2014). Even in the absence of degree completion, community college students who accrue a larger number of credits have been found to experience greater financial benefits when compared with community college students with fewer credits (Belfield & Bailey, 2011). In addition, investments in financial aid appear to make sense from a tax perspective, as previous research has reported that the positive impact of federal financial aid on students’ earnings alone is enough to recoup government expenditures within 10 years (Denning et al., 2019). For the present study, the positive impact of the information and one-course waiver treatment should be placed in its proper context—a modest increase of 1.5 percentage points in the likelihood of re-enrollment.

Former students who return to college may be more likely to forgo wages while they are re-enrolled when compared with other students. If former students re-enroll but do not eventually complete their degree, they may incur additional debt without obtaining the benefits associated with making considerable progress toward their degree (Belfield & Bailey, 2011). In such a scenario, former students could potentially be in a worse financial position than before re-enrollment. Importantly, the National Student Clearinghouse has reported that completion rates were nearly two times higher among former students who re-enrolled after making substantial progress toward a degree and within 3 years of initially stopping out of college (Shapiro et al., 2019). Despite the modest effect size of the information and one-course waiver intervention and the opportunity costs associated with returning to college, our findings suggest that the intervention appears to be a worthwhile investment when one considers the increase in enrollment and potentially degrees awarded alongside the low cost ($303–$354) to the institution for each returning student.

Although the combination of an informational nudge and one-course tuition waiver represents a relatively affordable policy lever to foster re-enrollment among former community college students, further research is needed to better understand whether former students who re-enroll are also more likely to continue to persist and ultimately complete their degree upon returning to college. Such analyses will offer greater insight into the extent to which re-enrollment campaigns benefit former students. Additional research is also needed to explore why community college dropouts decide to leave college in the first place and what can be done to prevent their initial departure and thereby enhance their likelihood of degree completion.

Supplemental Material

sj-docx-1-epa-10.3102_0162373720979177 – Supplemental material for Can Re-Enrollment Campaigns Help Dropouts Return to College? Evidence From Florida Community Colleges

Supplemental material, sj-docx-1-epa-10.3102_0162373720979177 for Can Re-Enrollment Campaigns Help Dropouts Return to College? Evidence From Florida Community Colleges by Justin C. Ortagus, Melvin Tanner and Isaac McFarlin in Educational Evaluation and Policy Analysis

Footnotes

Acknowledgements

We thank Peter Bergman, Angela Boatman, Joshua Hyman, and Dennis Kramer for their detailed feedback. We also thank conference participants at the Association for the Study of Higher Education for their helpful comments, and Maryanne Long and Jiayao Wu for outstanding research assistance. Any errors or omissions remain the sole responsibility of the authors.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by a research grant from the Helios Education Foundation.

ORCID iD

Justin C. Ortagus

Notes

Authors

JUSTIN C. ORTAGUS is an assistant professor of Higher Education Administration & Policy and the director of the Institute of Higher Education at the University of Florida. His research examines how online education, community colleges, and various state policies affect the opportunities and outcomes of underserved students.

MELVIN TANNER is a PhD student in the Higher Education Administration & Policy program at the University of Florida. His research interests include community colleges and the impact of online education in higher education.

ISAAC MCFARLIN is an assistant professor of Higher Education Administration & Policy in the College of Education at the University of Florida. His research focuses on the economics of education.

References

Adelman

(2013). Searching for our lost associate’s degrees: Project Win-Win at the finish line. Institute for Higher Education Policy.

Anderson

D. M.

Broton

K. M.

Goldrick-Rab

Kelchen

(2020). Experimental evidence on the impacts of need-based financial aid: Longitudinal assessment of the Wisconsin Scholars Grant. Journal of Policy Analysis and Management, 39(3), 720–739.

Bailey

Jaggars

S. S.

Jenkins

(2015). Redesigning America’s community colleges. Harvard University Press.

Barr

Turner

(2018). A letter and encouragement: Does information increase postsecondary enrollment of UI recipients? American Economic Journal: Economic Policy, 10(3), 42–68.

Bers

Schuetz

(2014). Nearbies: A missing piece of the college completion conundrum. Community College Review, 42(3), 167–183.

Belfield

C. R.

Bailey

(2011). The benefits of attending community college: A review of the evidence. Community College Review, 39(1), 46–68.

Bergman

Gross

J. P. K.

Berry

Shuck

(2014). If life happened but a degree didn’t: Examining factors that impact adult student persistence. Journal of Continuing Higher Education, 62(2), 90–101.

Bergman

Denning

J. T.

Manoli

(2019). Is information enough? The effect of information about education tax benefits on student outcomes. Journal of Policy Analysis and Management, 38(3), 706–731.

Bettinger

E. P.

(2015). Need-based aid and college persistence: The effects of the Ohio College Opportunity Grant. Educational Evaluation and Policy Analysis, 37(1 Suppl.), 102S–119S.

10.

Bettinger

E. P.

Long

B. T.

Oreopoulos

Sanbonmatsu

(2012). The role of application assistance and information in college decisions: Results from the H&R Block FAFSA Experiment. The Quarterly Journal of Economics, 127(3), 1205–1242.

11.

Bird

K. A.

Castleman

B. L.

Denning

J. T.

Goodman

Lamberton

Rosinger

K. O.

(2019). Nudging at scale: Experimental evidence from FAFSA completion campaigns (No. w26158). National Bureau of Economic Research.

12.

Castleman

B. L.

Long

B. T.

(2016). Looking beyond enrollment: The causal effect of need-based grants on college access, persistence, and graduation. Journal of Labor Economics, 34(4), 1023–1073.

13.

Castleman

B. L.

Owen

Page

L. C.

(2015). Stay late or start early? Experimental evidence on the benefits of college matriculation support from high schools versus colleges. Economics of Education Review, 47, 168–179.

14.

Castleman

B. L.

Page

L. C.

(2015). Summer nudging: Can personalized text messages and peer mentor outreach increase college going among low-income high school graduates? Journal of Economic Behavior & Organization, 115, 144–160.

15.

Castleman

B. L.

Page

L. C.

(2016). Freshman year financial aid nudges: An experiment to increase FAFSA renewal and college persistence. Journal of Human Resources, 51(2), 389–415.

16.

Castleman

B. L.

Page

L. C.

Schooley

(2014). The forgotten summer: Mitigating summer attrition among college-intending, low-income high school graduates. Journal of Policy Analysis and Management, 33(2), 320–344.

17.

Cox

B. E.

Reason

R. D.

Nix

Gillman

(2016). Life happens (outside of college): Non-college life events and students’ likelihood of graduation. Research in Higher Education, 57(7), 823–844.

18.

Crosta

(2013). Characteristics of early community college dropouts. Community College Research Center, Columbia University Teachers College.

19.

Denning

J. T.

(2017). College on the cheap: Consequences of community college tuition reductions. American Economic Journal: Economic Policy, 9(2), 155–188.

20.

Denning

J. T.

Marx

B. M.

Turner

L. J.

(2019). ProPelled: The effects of grants on graduation, earnings, and welfare. American Economic Journal: Applied Economics, 11(3), 193–224.

21.

Fack

Grenet

(2015). Improving college access and success for low-income students: Evidence from a large need-based grant program. American Economic Journal: Applied Economics, 7(2), 1–34.

22.

Goldrick-Rab

Kelchen

Harris

D. N.

Benson

(2016). Reducing income inequality in educational attainment: Experimental evidence on the impact of financial aid on college completion. American Journal of Sociology, 121(6), 1762–1817.

23.

Gurantz

(2020). Impacts of state aid for non-traditional students on educational and labor market outcomes. Journal of Human Resources. 1–43. https://ogurantz.github.io/website/Gurantz_competitive_prepublication.pdf

24.

Gurantz

Howell

Hurwitz

Larson

Pender

White

(2019). Realizing your college potential? Impacts of College Board’s RYCP campaign on postsecondary enrollment (Ed Working Paper No. 19-40). Brown University.

25.

Herbaut

Geven

K. M.

(2020). What works to reduce inequalities in higher education? A systematic review of the (quasi-)experimental literature on outreach and financial aid. Research in Social Stratification and Mobility, 65, 100442.

26.

Hoekstra

(2009). The effect of attending the flagship state university on earnings: A discontinuity-based approach. The Review of Economics and Statistics, 91(4), 717–724.

27.

Hoyt

J. E.

Winn

B. A.

(2004). Understanding retention and college student bodies: Differences between drop-outs, stop-outs, opt-outs, and transfer-outs. NASPA Journal, 41(3), 395–417.

28.

Hyman

(2020). Can light-touch college-going interventions make a difference? Evidence from a statewide experiment in Michigan. Journal of Policy Analysis & Management, 39(1), 159–190.

29.

InsideTrack. (2019). Re-engage, re-enroll, and prepare former students to finish strong. http://info.insidetrack.com/re-engage_re-enroll_and_prepare_ps

30.

Institute for Higher Education Policy. (2011). Crossing the finish line: A national effort to address near completion. http://www.ihep.org/sites/default/files/uploads/docs/pubs/brief_crossing_the_line-near_completion.pdf

31.

Joo

S.-H.

Durband

D. B.

Grable

(2008). The academic impact of financial stress on college students. Journal of College Student Retention: Research, Theory & Practice, 10(3), 287–305.

32.

Kahneman

Tversky

(1979).). Prospect theory: An analysis of decision under risk. Econometrica, 47, 263–291.

33.

Kane

T. J.

(1995). Rising public college tuition and college entry: How well do public subsidies promote access to college? (No. w5164). National Bureau of Economic Research.

34.

Karlan

McConnell

Mullainathan

Zinman

(2016). Getting to the top of mind: How reminders increase saving. Management Science, 62, 3393–3672.

35.

Lashure

Randall

Hinkle

(2019). The retention project and the investigation of a non-returning new student re-enrollment strategy. Strategic Enrollment Management Quarterly, 7(1), 25–36.

36.

Long

B. T.

(2007). The contributions of economics to the study of college access and success. Teachers College Record, 109(10), 2367–2443.

37.

Meyer

Rosinger

K. O.

(2019). Applying behavioral insights to improve postsecondary education outcomes: A review of Obama administration efforts and next steps under the Trump administration. Journal of Policy Analysis and Management, 38(2), 481–499.

38.

Oreopoulos

Petronijevic

(2019). The remarkable unresponsiveness of college students to nudging and what we can learn from it (No. w26059). National Bureau of Economic Research.

39.

Page

L. C.

Gehlbach

(2017). How an artificially intelligent virtual assistant helps students navigate the road to college. AERA Open, 3(4), 1–12.

40.

Peters

E. E.

Roberson

A. J.

Voight

(2019). The cost of opportunity: Student stories of college affordability. http://www.ihep.org/sites/default/files/uploads/docs/pubs/ihep_student_voices_full_report_v6_web_pages.pdf

41.

Richburg-Hayes

Patel

Brock

de la Campa

Rudd

Valenzuela

(2015). Providing more cash for college: Interim findings from the performance-based scholarship demonstration in California. MDRC.

42.

Rouse

C. E.

(1995). Democratization or diversion? The effect of community colleges on educational attainment. Journal of Business & Economic Statistics, 13(2), 217–224.

43.

Rubin

R. B.

(2011). The Pell and the poor: A regression-discontinuity analysis of on-time college enrollment. Research in Higher Education, 52(7), 675–692.

44.

Schwartz

(2019, June 19). How colleges are bringing back stopped-out students. Education Dive. https://www.educationdive.com/news/how-colleges-are-bringing-back-stopped-out-students/556706/

45.

Scott-Clayton

J. E.

(2017). Undergraduate financial aid in the United States. American Academy of Arts & Sciences.

46.

Shapiro

Dundar

Huie

Wakhungu

P. K.

Yuan

Nathan

Bhimdiwali

(2017). Completing college: A national view of student completion rates. National Student Clearinghouse Research Center.

47.

Shapiro

Ryu

Huie

Liu

(2019). Some college, no degree: A 2019 snapshot for the nation and 50 states (Signature Report No. 17). National Student Clearinghouse Research Center.

48.

Snyder

T. D.

De Brey

Dillow

S. A.

(2018). Digest of Education Statistics 2016 (NCES 2017-094). National Center for Education Statistics.

49.

Stinebrickner

(2008). The effect of credit constraints on the college drop-out decision: A direct approach using a new panel study. American Economic Review, 98(5), 2163–2184.

50.

Thaler

R. H.

Benartzi

(2004). Save more tomorrow: Using behavioral economics to increase employee saving. Journal of Political Economy, 112, 164–187.

51.

Thaler

R. H.

Sunstein

C. R.

(2008). Nudge: Improving decisions about health, wealth, and happiness. Yale University Press.

52.

Trostel

Chase

(2015). It’s not just the money: The benefits of college education to individuals and to society. Lumina Issue Papers.

53.

University of Memphis Finish Line Program. (2019). Finish line. https://www.memphis.edu/innovation/finishline/

54.

Zimmerman

S. D.

(2014). The returns to college admission for academically marginal students. Journal of Labor Economics, 32(4), 711-754.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

3.55 MB