Res High Educ (2015) 56:1–28 DOI 10.1007/s11162-014-9340-0 FAFSA Filing Among First-Year College Students: Who Files on Time, Who Doesn’t, and Why Does it Matter? Lyle McKinney • Heather Novak Received: 12 June 2013 / Published online: 15 July 2014 Ó Springer Science+Business Media New York 2014 Abstract Students who do not file the free application for federal student aid (FAFSA), or who file after the priority application deadline, are at risk of not receiving grant aid that could help them persist and graduate from college. This study used data from the beginning postsecondary student study (BPS:04/06) to examine FAFSA filing behavior (i.e. early, late, did not file) among students attending community colleges, public 4-year, and private non-profit 4-year institutions. Results indicate that later filers, on average, receive less total state and institutional grant aid compared to students who filed earlier. Attending college part-time and delaying enrollment into college after high school were strongly associated with not filing a FAFSA and filing late. There were notable differences in FAFSA filing across institutional sectors as a function of students’ gender, race/ethnicity, income status, high school context, and pre-college academic experiences. These findings serve as the basis for recommendations aimed at increasing the rates of early FAFSA filing among students at the greatest risk of leaving money on the table. Keywords Free application for federal student aid (FAFSA) Student financial aid State grant aid Institutional grant aid Community colleges Public 4-year colleges Private 4-year colleges L. McKinney (&) Department of Educational Leadership and Policy Studies, University of Houston, 427 Farish Hall, Houston, TX 77204-5019, USA e-mail: llmckinney@uh.edu H. Novak Office of Institutional Research, Colorado State University, 1004 Campus Delivery, Fort Collins, CO 80523-1004, USA e-mail: heather.novak@colostate.edu 123 2 Res High Educ (2015) 56:1–28 Introduction In the current US student financial aid system, the free application for federal student aid (FAFSA) represents the gateway to the majority of financial assistance for college students. The FAFSA is used by the federal government to determine student eligibility for federal Pell Grants, Stafford Loans, and Work-Study opportunities. States utilize information from the FAFSA to allocate resources from state-funded student financial aid programs. In addition, most colleges and universities use the FAFSA to make decisions about the allocation of institutional financial aid to their students. The FAFSA is a central component of the current US financial aid system and filing this application is the critical first step in helping procure the financial assistance that so many students desperately need to enroll in and graduate from college. Unfortunately, every year millions of college students miss out on the opportunity to receive financial aid simply because they do not file a FAFSA (Kantrowitz 2009; King 2006). For the 2007–2008 academic year, Kantrowitz estimated that approximately 40 % (about 8.4 million students) of all undergraduates in the US did not file. The most common reason students said they did not file a FAFSA was because they assumed, incorrectly in many cases, they would not qualify for financial aid. Other reasons students reported not filing a FAFSA include: they could afford college without financial aid; they missed the application deadline; and they had concerns about privacy. Some students elect not to file because they receive financial aid that does not require submission of a FAFSA, such as aid from their employer (King 2006). Regardless of their reason for not filing a FAFSA, there is growing concern that so many of these non-filers come from lower-income families and would have qualified for needbased grant aid that does not need to be repaid. Approximately 2.3 million students who did not complete the FAFSA in 2007–2008 would have qualified for a Pell Grant and 1.1 million of these non-filers would have likely qualified for a full Pell Grant (Kantrowitz 2009), which was $4,310 for the 2007–2008 academic year. About 42 % of community college students who were Pell Grant eligible in 2007–2008 did not file a FAFSA. Studies have found that failing to file a FAFSA is associated with lower within-year persistence rates among first-year students attending community colleges and 4-year institutions (McKinney and Novak 2013; Novak and McKinney 2011). Increasing the number of students who file a FAFSA each year has been promoted as a way to help increase college enrollment and graduation rates (College Board Advocacy and Policy Center 2010; Novak and McKinney 2011). But simply whether or not a student files a FAFSA does not tell the whole story. The timing of when a student files a FAFSA can impact the type and amount of financial aid received. For example, LaManque (2009) points out that in California the Cal Grant Entitlement awards typically have an application deadline of March, and Extended Opportunity Programs and Services grants are limited and awarded early in the term to students with completed applications. He explains that, ‘‘a late financial aid application can result in less aid awarded for students with equal need because of fewer funds being available to later applicants. Early application and award are less likely to result in the need for working extra hours, asking friends and family for help, or applying for loans. The assumption is that an early FAFSA filer with a good financial aid package will be more likely to succeed and persist in college than later FAFSA filers with exactly the same need and less financial aid’’ (p. 6). Pell Grants are awarded to eligible students regardless of the date when their application is submitted. But many state and institutional financial aid programs operate on a first come, first serve basis and awards are made until the program funds have been depleted 123 Res High Educ (2015) 56:1–28 3 (see 2013–2014 FAFSA Deadlines at www.fafsa.gov). Consequently, qualified applicants who file a FAFSA after the priority deadline may not receive any funding from these programs, or receive smaller awards than students who filed early. Missing out on this grant aid can have severe consequences for late applicants, as multiple studies have found that receipt of state grant aid (Mendoza et al. 2009; St. John and Paulsen 2001) and institutional grant aid (Gansemer-Topf and Schuh 2005; Gross et al. 2007) is positively associated with student persistence. Late filers who miss out on this grant aid during their first year of enrollment may rely more heavily on loans in order to bridge the gap between their financial resources and the total costs of attending college. However, higher levels of debt can hinder persistence among students who are lower-income, academically underprepared, and pursuing degrees in fields with lower earning potential (St. John et al. 2003). The purpose of the present study was to better understand the factors that predict FAFSA filing behavior among first-year college students, and the relationship between the timing of FAFSA filing and a student’s grant aid award. Two research questions guided this study: 1. 2. How does the timing of FAFSA filing impact the total amount of state and institutional grant aid received by first-year college students? What factors influence late FAFSA filing and non-filing among first-year students attending community colleges, public 4-year, and private not-for-profit 4-year institutions? This present study represents a useful contribution to the extant literature in several ways. First, this study extends the work of LaManque (2009), which explored FAFSA filing among students at a single community college in California, by investigating the factors that influence FAFSA filing behaviors among a nationally-representative group of first-year college students attending 2- and 4-year postsecondary institutions. Second, our analyses employed multivariate inferential statistics to identify the characteristics of both non-filers and late FAFSA filers. Prior reports on FAFSA filing have primarily relied on descriptive statistics, considered a relatively small number of variables, and overlooked potential differences between students who filed their FAFSA early and those who filed late. Finally, our study offers a more careful examination of an issue that has received considerable attention from policymakers in recent years. Findings from this study can be used to shape policies and institutional practices that increase the utilization of grant aid among at-risk student groups who are the most dependent on financial aid to persist and graduate. Conceptual Framework and Literature Review The present study adapted Perna’s (2006a, b) multilayered conceptual model of student choice in an effort to better understand the wide range of factors that can influence a student’s FAFSA filing behavior. This conceptual model, ‘‘integrates aspects of the economic theory of human capital and sociological notions of social and cultural capital and recognizes that multiple layers of context influence an individual’s college-related decision making’’ (Perna 2006b, p. 1621). In addition to offering a multi-disciplinary approach to the study of student decision-making, an important contribution of Perna’s model to the current study is the attention given to exogenous variables (e.g. financial aid deadlines, institutional characteristics) that can shape student’s FAFSA filing behavior. Perna has demonstrated the relevance of this model for studying financial aid by applying it to 123 4 Res High Educ (2015) 56:1–28 understand how students’ obtain and use information about financial aid (2006b) and to identify the forces that shape high school students’ willingness to borrow for college (2008). Applied to the current study, the conceptual model suggests that each student’s knowledge of, and experiences with, the FASFA occurs within a situated context. That is, students’ experiences with the financial aid application process are influenced directly and indirectly by their family context, the characteristics of the high schools and postsecondary institutions they attend, and by the larger social, economic, and public policy context. Figure 1 depicts an adaptation of Perna’s conceptual model and the variables examined in this study that are hypothesized to potentially impact FAFSA filing. The following sections give attention to contextual layers of the model, focusing on the specific constructs within each layer that are relevant in explaining students’ experiences with the financial aid application process. Research literature related to college knowledge and financial aid utilization is highlighted within each section to corroborate the applicability of the conceptual model for this study. Layer 4 of Perna’s model (i.e. the social, economic, and policy context) was not explicitly addressed during data analysis, but was relied upon to frame the discussion of results. Student and Family Context The foundational layer of Perna’s conceptual model assumes that an individual’s demographic and familial characteristics, as well as his/her access to social and cultural capital, exert an influence on college-related decisions (Perna 2006a). This layer is characterized by a concept referred to as the student’s habitus, which has been defined as, ‘‘a common set of subjective, internalized, class-based perceptions that shape an individual’s expectations, attitudes, and aspirations’’ (McDonough and Calderone 2006, p. 1704). Prior research suggests that a student’s habitus can exert both direct and indirect effects on their access to information about the FAFSA and influence their decisions about applying for financial aid (Perna 2006b, 2008). With regards to student background characteristics, Kantrowitz (2009) found that nonfilers who would have qualified for a Pell Grant were more likely to be male, independent without dependents other than a spouse, and 24 years or older. Several reports indicate that lower-income students are disproportionally represented among students who do not submit a FAFSA (King 2006; Kantrowitz 2009, 2011). Students from less affluent families often lack the information, resources, and support necessary to successfully navigate the complex financial aid process (Kane and Avery 2004; McDonough and Calderone 2006; Perna 2006a, b, Perna 2008; Perna and Titus 2005). This lack of familiarity with the FAFSA, and with financial aid in general, helps explain why so many lower-income students do not apply for financial aid (Bettinger et al. 2009; Kane and Avery 2004). Many of these students do not file a FAFSA because they are unaware that financial aid is available to them, or because they do not know that filing a FAFSA is the first step towards receiving financial assistance for college. Theories of social and cultural capital are particularly useful to develop a better understanding of the non-monetary factors that influence how students navigate the financial aid application process. Teachers, counselors, and peers that can share information and provide guidance represent valuable forms of social capital that students can access when exploring how to pay for college (McDonough 1997). In addition, students with college-educated parents often benefit from a home environment that offers a wealth of information about college, to include advice about financial aid, that students can access 123 Res High Educ (2015) 56:1–28 5 Fig. 1 Conceptual model of factors influencing FAFSA filing behavior among first-year college students (McDonough 1997; Perna and Titus 2005). Conversely, first-generation college students are more likely to have parents that do not know how to help them navigate the financial aid application process (Feeney and Heroff 2013; Vargas 2004). Human Capital Theory The rational human capital investment model is the underpinning of Perna’s model. From an economic perspective, an individual is motivated to pursue higher education when they believe the long-term benefits outweigh the immediate costs (Becker 1997; Manski and Wise 1983). While human capital theory claims that individuals perform this cost-benefit 123 6 Res High Educ (2015) 56:1–28 comparison based upon the information that is available to them, it does not presume that individuals have all of the pertinent information when making decisions (DesJardins and Toutkoushian 2005). This suggests that knowledge regarding how to navigate collegegoing processes, including applying for financial aid, can vary greatly across different student groups. This is corroborated by studies that have found many lower-income and minority high school students are less informed about financial aid than their more affluent peers (Hossler et al. 1999; McDonough 1997; Perna 2008). Relying on incomplete information, or inaccurate assumptions, can help explain why so many students who are eligible for grant aid do not file a FAFSA or file long after the priority application deadline (Feeney and Heroff 2013). As depicted in the conceptual model, a student’s demand for higher education is influenced by his/her pre-college academic preparation and performance. Students with greater academic preparation and who excel academically in high school (measured by factors such as the highest level of math completed, cumulative high school GPA, and SAT/ACT test scores) are more likely to pursue postsecondary education. Those students with a high demand for college are more likely to proactively seek out information and guidance about filing a FAFSA, as college application and FAFSA completion are closely linked. Conversely, high school students who do not view themselves as college-bound, or who intend to delay enrollment into college after high school, may be less likely to seek advice on applying for financial aid. The conceptual model hypothesizes that a student’s supply of financial resources contributes to their decisions about applying for financial aid to help pay for college. Students from high-income families that can afford college without the receipt of financial aid may consciously decide not to file a FAFSA. Conversely, students from lower-income families are often dependent upon the receipt of financial aid in order to enroll and persist in college (Christou and Haliassos 2006). A growing number of today’s college students work while they pursue higher education (Perna 2010) and many of these students may not file a FAFSA because they intend to pay for college using the income received from employment. Working students who attend college part-time may incorrectly assume they do not qualify for aid and work additional hours to pay expenses that financial aid would have covered (Ziskin et al. 2014). With regards to the expected benefits of higher education, high school students with more clearly defined career plans and aspirations tend to commence their college search and planning efforts earlier (Hossler et al. 1999; Somers et al. 2001). For a high school student who begins college planning early and knows what they wish to study, they may complete college applications earlier than their undecided peers and file a FAFSA at this time as well. The highest level of postsecondary credential a student expects to earn may also impact their decision to apply for financial aid. For example, a student pursuing a 1-year vocational certificate at a community college may consciously decide not to apply for financial aid because they believe they can afford this credential without financial assistance or they may receive employer-provided tuition assistance. Conversely, a firstyear college student who intends to eventually earn a graduate or professional degree may recognize the need for ongoing financial support to achieve this goal and therefore be motivated to file a FAFSA. Institutional Contexts The conceptual model gives attention to students’ high school context and reflects what McDonough (1997) refers to as organizational habitus. The model also considers the 123 Res High Educ (2015) 56:1–28 7 higher education context and recognizes the major role that postsecondary institutions play in shaping students’ behaviors and choices (Perna 2006a, b, Perna 2008). Applied to the present study, these contextual layers give attention to the types of resources and support systems (or lack thereof) high schools and colleges make available to students that can influence their decisions about applying for financial aid. The high school context often reflects the amount of social capital that is available to a particular student. High-income or private high schools, in comparison to low-income or urban public schools, typically provide students with greater resources and support that can help them make informed decisions about paying for college (Perna 2008). Students attending private schools are more likely to have immediate access to knowledgeable financial aid counselors, teachers, and peers who can provide valuable information about the financial aid process (Perna 2008; Stanton-Salazar 1997; Tierney and Venegas 2006). Conversely, lower-income high schools may exhibit structural constraints that make it difficult for students to make informed decisions about paying for college. For example, the least informed counselors about college-related issues are typically found in lowerincome schools (McDonough 1997) and disbursing information about financial aid may not be a top priority for these counselors. Postsecondary institutions also play a role in providing students and their families with information about financial aid (Perna 2006b), which can influence students’ FAFSA filing behavior. The present study examined FAFSA filing among students attending public 2-year (i.e. community colleges) and public and private 4-year colleges. Unlike most 4-year institutions, the open access mission of the community college allows students to register for classes as late as the first day of a given semester. Many of these late enrollees apply for financial aid during the registration process, and filing this late can result in missing out on a grant award from state and/or institutional aid programs whose funds have already been depleted (LaManque 2009). Furthermore, financial aid offices at many community colleges are sorely understaffed and these financial aid counselors may not be able to provide every student with the information and guidance they need to successfully complete the FAFSA (College Board Advocacy and Policy Center 2010; McKinney and Roberts 2012; McKinney et al. 2013). As discussed previously, students attending community colleges who are Pell Grant eligible are much less likely to file a FAFSA than their peers attending 4-year institutions (Kantrowitz 2009). The variation in FAFSA filing rates across institutional sectors may be attributed to aforementioned factors such as differences in students’ social and cultural capital, income status, the type of high school attended, and college attendance patterns. For example, compared to community colleges, private 4-year institutions enroll larger proportions of higher-income students and students with high academic ability, which are factors that can increase the likelihood of early FAFSA filing among students attending these institutions. Methodology Data Source and Sample The data analyzed in this study were derived from the Beginning Postsecondary Students (BPS:04/06) survey conducted by National Center for Education Statistics (NCES). The samples used for these analyses consisted of first-year undergraduates who were enrolled in 123 8 Res High Educ (2015) 56:1–28 community colleges (unweighted n = 3,990),1 public 4-year institutions (unweighted n = 4,050), and private 4-year not-for-profit institutions (unweighted n = 3,230) during the 2003–2004 academic year and were eligible to receive federal financial aid (i.e. US citizens and resident aliens). Variables Perna’s conceptual model and existing research literature on financial aid were used as guides to select the independent variables for the logistic regression models. The variables were organized into four categories: background characteristics and social/cultural capital (gender, race/ethnicity, dependency status, parent’s level of education); demand for higher education (pre-college academic performance, delayed enrollment, declared a major); supply of financial resources (expected family contribution, hours worked per week); and expected benefits and costs of pursuing higher education (highest level of education ever expected, attendance intensity). Figure 1 depicts how these variables fit within the conceptual model. In recognition of the differences between community college and 4-year students, there were variations in several of the measures included in the regression models for each sector. ACT/SAT composite score and high school GPA were used as measures of precollege academic performance for the samples of 4-year students. High school type (public or private) was also included in the models for 4-year students. For the community college sample, the highest level of high school math completed was used as a measure of precollege academic achievement because many of these students do not take the ACT or SAT. Additionally, having earned a high school diploma (compared to a GED or not completing high school) was used for the community college sample in recognition that some of this student population did not earn a high school diploma. Parent’s level of education was included in the 4- and 2-year models, but was coded differently based on each samples’ proportions. Parental education level was included in the 2-year model as ‘‘associates or higher’’ and ‘‘less than associates’’ because 51 % of parents in the 2-year sample had earned an associate’s degree or higher. But in the 4-year models this variable was coded as ‘‘bachelor’s or higher’’ and ‘‘less than a bachelors’’ because 57 % of the public 4-year and 63 % of the private 4-year sample had earned a bachelor’s degree or higher. The outcome of interest for the first research question was the average total amount of state and institutional grant aid received as a function of the month in which a student filed the FAFSA. The outcomes for the second research question were whether a student ever filed a FAFSA (coded 0 = filed a FAFSA, 1 = did not file a FAFSA), and subsequently, whether a student who filed a FAFSA submitted their application early or late (coded 0 = filed a FAFSA early, 1 = filed a FAFSA after the priority application deadline). Filing a FAFSA and filing early were coded as the reference levels so the results are interpreted in terms of the increased odds of the negative outcome (i.e. not filing, filing late) because these were the behaviors of primary interest in this study. In the relatively few cases when a predictor variable was negatively associated with the reference level, an inverse odds ratio (see Desjardins 2001) was used so the results are more easily interpreted in terms of the increased odds of filing or filing early. 1 All unweighted n’s reported in this study are rounded to the nearest ten per NCES data security guidelines. 123 Res High Educ (2015) 56:1–28 9 Defining ‘Early’ and ‘Late’ FAFSA Filing For the purposes of the logistic regressions predicting ‘late’ FAFSA filing, we wanted to understand the potential consequences of filing a FAFSA after the priority application deadline and identify students who were more likely to file after this deadline had passed. The application deadlines vary across state and institutional contexts, but colleges typically encourage their students to file a FAFSA before the state aid deadline in order to maximize the amount of aid for which they may be eligible (Feeney and Heroff 2013; King 2004). Using the 2003–2004 FAFSA form, we reviewed the state financial aid deadlines for the time period under investigation in our study (i.e. 2003–2004 academic year) and defined ‘late’ filing based on students’ state of residency when possible. A total of 30 states had identified priority application deadlines for the 2003–2004 academic year: six states had an application deadline by March, seven states by April, eight states by May, three states by June, four states by July, and two states by October. The BPS:04/06 dataset indicates the month in which the student filed a FAFSA, but does not provide the specific date within that month. Therefore, we adopted the more conservative approach and states with an early- or mid-month financial aid application deadline had their late deadline set for the subsequent month. For example, in 2003–2004 Kentucky had a state priority application deadline of March 15, so Kentucky residents were considered late filers if they filed a FAFSA in April 2003 or later. Arkansas had a priority financial aid application deadline of April 1, so we used the more conservative approach (i.e. students who filed on the April 1 deadline were not classified as late filers) and students in this state who filed their FAFSA in April 2003 or earlier were considered early filers and students who filed in May or later were considered late filers. This study aimed to provide a national perspective on FAFSA filing behaviors among first-year college students. Therefore, the next empirical decision was to determine how to operationally define early versus late FAFSA filing for students in the 20 remaining states that did not specify a priority application deadline during the 2003–2004 academic year. These states often just encourage students to file the FAFSA as early as possible. For these 20 states, we relied on prior research (see King 2004) that defined April 1 as the benchmark for early versus late FAFSA filing. King argued that April 1 was an appropriate benchmark for ‘late’ FAFSA filing because many institutions advise their students to complete the FAFSA before April in order to maximize the amount of aid for which they may be eligible. In addition, the month of March is the priority federal financial aid application deadline and many high school seniors who intend to matriculate into college file their FAFSA on or before the April 1 deadline. Therefore, the decision to use April 1 as the benchmark for late filing in these 20 states that did not set priority application deadlines in 2003–2004 was based on prior research and common practice related to FAFSA filing. Again we adopted the more conservative approach and students in these 20 states who filed in April 2003 or earlier were considered early filers and those filing in May 2003 or later were classified as late filers. It is important to note that in defining what constituted ‘early’ and ‘late’ submission of the FAFSA, consideration was also given to the differences in the college search and application process between 4-year and community college students. For students attending 4-year institutions, the process of applying to college and filing a FAFSA often occurs during their senior year of high school. But because of the open access mission of the community college, many incoming students register for classes and apply for financial aid as late as the first week of the semester (College Board Advocacy and Policy Center 2010; LaManque 2009). Many of these students have made a last-minute decision to 123 10 Res High Educ (2015) 56:1–28 pursue higher education and have therefore given limited prior consideration to completing a financial aid application. Therefore, we initially explored using August 1, 2003 as the operational definition of late filing for the sample of community college students. But the magnitude and direction of the regression coefficients did not vary significantly when the models were run with August 1 as the priority application deadline. Moreover, community college students who file near the time of Fall registration have still missed the priority application deadlines for most state grant aid programs and therefore remain at higher risk than early FAFSA filers of not receiving grant aid or receiving a smaller award amount. For these reasons, we made the decision to use the same approach to identifying the priority application deadlines for students in both the 4-year and community college sectors. Analytic Methods Descriptive statistics were used to examine sample characteristics for students as a function of their institutional sector. These descriptive statistics are shown for the full samples across each sector and also for the samples limited to only FAFSA filers. The first research question was addressed by plotting the average state and/or institutional grant aid amount students received as a function of the month in which they filed the FAFSA. This information is presented for students attending community colleges, public 4-years, and private not-for-profit 4-year institutions. Logistic regression was used to determine factors associated with whether or not a student files a FAFSA and then with the timing (i.e. early or late) of FAFSA filing for those who did file. The outcome variables were coded with not filing and filing late as the comparison group, so the results from the models are interpreted in regards to the odds of not filing and the odds of filing late. Since the outcome variables were binary (i.e. filed or did not file; filed early or filed late), a logistic regression was the appropriate method for the second research question (Hosmer and Lemeshow 2000). A benefit of logistic regressions is that the coefficients are easily interpreted as odds ratios. Odds ratios are obtained directly from the logistic regression model by exponentiating an independent variable’s coefficient. An odds ratio greater than one indicates that the comparison group has higher odds of not filing a FAFSA (or filing a FAFSA late) compared to the reference group, after controlling for all other predictors included in the model. An odds ratio less than one indicates lower odds for the comparison group (Long 1997). Inverse odds ratios are used to discuss the results that have an odds ratio less than one. An inverse odds ratio is equal to one divided by the variable’s odds ratio and is interpreted in terms of higher odds of the reference event occurring (e.g. filing) rather than lower odds of the comparison event occurring (e.g. not filing) (DesJardins 2001). Since the outcome variables are coded with the negative behavior (not filed or filed late) equal to one, using the inverse odds ratio in the discussion of the results avoids the double negative of having lower odds of not filing or filing late. The regression coefficients from the logistic regression models are presented in tables as odds ratios with their corresponding 95 % confidence interval. When inverse odds ratios are used in the discussion the odds ratio from the table is referenced. As recommended with all complex and/or multi-stage cluster sample datasets, weights and design effects were used to account for the oversampling of certain groups and clusters of homogeneity within sampling levels (Thomas and Heck 2001). Survey design effects were accounted for using the survey commands in Stata as recommended for logistic regression analyses with large-scale, secondary datasets (Dowd and Duggan 2001). All statistical analyses were conducted using Stata 10. 123 Res High Educ (2015) 56:1–28 11 Limitations There are several limitations that should be considered when interpreting the results from this study. First, the sample under investigation consists of students who did enroll in higher education and therefore this is not a study about college access. The focus of this paper was on providing a better understanding of FAFSA filing behaviors among first-year college students. It was beyond the scope of this research to examine how the timing of FAFSA filing, and average state and/or institutional grant aid awards, impacted students’ decisions to enroll, or not enroll, in college. A different dataset would be required to determine how FAFSA filing behaviors affect students’ likelihood of matriculating into higher education. Similarly, it is possible that being offered a smaller grant aid award, as a result of filing a FAFSA after the priority application deadline, influenced a students’ decision to attend college part-time rather than full-time. This could introduce the potential for endogeneity bias if there is a causal relationship between FAFSA filing timing and enrollment intensity during the first year of college. These are issues we were unable to determine with the BPS dataset, but are worthy of investigation in future research. Second, a common challenge that exists when using secondary datasets like BPS is that the analyses are limited to the existing variables in the dataset. This introduces the possibility of omitted variable bias if criterion variables that may potentially influence the outcome of interest are unavailable for inclusion in the analyses (Cellini 2008). For example, accounting for a student’s interactions with a high school counselor about financial aid would have been a useful variable to include in the regression models. But the BPS:04/06 dataset does not include this information or direct measures of other pre-college activities (e.g. attending a financial aid workshop) that could influence a student’s FAFSA filing behavior. A different NCES dataset, National Postsecondary Student Aid Survey (NPSAS), has these types of measures but does not include the FAFSA filing date in the most recent version (NPSAS:08). Thus, despite lacking some potentially relevant predictor variables, BPS was the most appropriate national dataset for the present study given our focus on the timing of FAFSA filing among first-year students. Lastly, the present study used a nationally representative sample of first-year college students to examine differences in FAFSA filing and the average total amount of state and institutional grant aid received by late and early filers. Almost every US state has some type of financial aid program for college students, but these state-funded programs can vary greatly in their scope, purpose, funding levels, and eligibility criteria. Similarly, there is often considerable variation across colleges in terms of how institutional aid is awarded to students. Findings from our study are intended to provide a national perspective on FAFSA filing behaviors and show the consequences of late filing among first-year college students. But future studies should examine these issues within a particular state or institutional context using more recent data that accounts for the characteristics of specific financial aid programs. Our study offers an important baseline for future research on this topic and an analytic framework that can be used to carry out these studies. Results Descriptive statistics are presented in Tables 1, 2, and 3 for each of the three institutional sectors. Proportions are used to describe the categorical variables and means are provided for the continuous measures. Across all three institutional sectors, higher proportions of the sample were female, White, and categorized as dependent for financial aid 123 12 Res High Educ (2015) 56:1–28 purposes. Community colleges enrolled a greater proportion of minority (i.e. non-White) students (35 %) overall compared to public (26 %) and private (24 %) 4-year institutions. Compared to the 4-year sectors, higher proportions of community college students were enrolled part-time (39 %), delayed enrollment into college after high school (28 %), and had not declared a major during their first year of college (39 %). The average expected family contribution (EFC) during the 2003–2004 academic year was approximately $4,000 less for community college students and they worked an average of 10 more hours per week than students attending 4-year public institutions. With regards to FAFSA filing behaviors, 44 % of first-year community college students did not file a FAFSA compared to 26 % of the public 4-year and 18 % of the private 4-year students. Community college students were also much more likely than the 4-year students to file their FAFSA late. Among students that did file a FAFSA, approximately 24 % of public 4-year and 17 % of private 4-year students filed their FAFSA late, while 54 % of community college students filed their FAFSA late. Does Timing of FAFSA Filing Impact Amount of State and Institutional Grant Aid Received? Figure 2 graphs the average total amount of state and/or institutional grant aid a student receives as a function of the month in which they filed their FAFSA. For all three sectors, the average amount of aid decreased the closer the FAFSA filing date was to the beginning of the Fall 2003 semester. The absolute dollar differences in average award amounts between January and August vary greatly by sector; however, students in all three sectors who filed their FAFSA in August had significantly lower awards compared to students who filed in January, February, or March. For example, community college students who filed in August had an average award of $474, which was 60 % less than their peers who filed in January and had an average award of $1,165. Students at public and private 4-year institutions who filed in August had a state and/or institutional grant award that was considerably lower, on average, than their peers who filed in January (73 % lower at publics and 71 % lower at privates). In sum, there was a clear negative association between later FAFSA filing and the average total state and/or institutional aid received. Factors Associated with Not Filing a FAFSA and Filing Late Logistic regression models were used to identify factors that are associated with students’ FAFSA filing behavior. The results from these models are presented in Table 4 (community colleges), Table 5 (public 4-year institutions), and Table 6 (private 4-year institutions). The results from two regression models examining different outcomes (i.e. filed versus did not file; filed early versus filed late) are presented within each table. Filing a FAFSA and filing early were the reference categories used in these models (i.e. coded as zero). The odds ratios are interpreted in terms of the comparison categories of not filing and filing late; however, when an odds ratio is less than one the inverse odds ratio is used in this discussion so the interpretation flips to the odds of filing or filing early. With regards to the community college sample, the first three columns of data in Table 4 display the results for the logistic regression that examined factors associated with not filing a FAFSA. The overall model is statistically significant (f = 16.28, p \ .000) and has a Cragg–Uhler pseudo r squared of .181 and correctly classifies 69 % of the 3,990 cases. Male students had 61 % higher odds of not filing compared to females, while 123 Res High Educ (2015) 56:1–28 13 Table 1 Descriptive statistics for community college students Filers and non-filers % of sample Filers only % of sample Gender Femalea 53.12 59.48 Male 46.88 40.52 Whitea 64.60 61.17 African American 14.05 18.08 Hispanic 16.26 15.67 5.10 5.08 Race/ethnicity Asian Dependency status Dependenta 87.57 87.56 Independent 12.43 12.44 Associates or highera 50.95 43.95 Less than an associates 49.05 56.05 92.73 93.42 7.27 6.58 Parental education HS degree Has a degreea No high school degree or equivalent Enrollment intensity Full-timea 61.13 67.37 Part-time 38.87 32.63 High school math Algebra2 and highera 76.07 77.84 Below algebra 2 23.93 22.16 Delay enrollment Did not delaya 71.83 77.30 Did delay for a year or more 28.17 22.70 Major Declareda 61.36 67.06 Undeclared 38.64 32.94 Highest expected level of education BA or highera 86.84 86.88 AA or less 13.16 13.12 Expected family contribution (divided by 1,000)b 9.22 5.96 Hours worked (divided by 10) per weekb 2.15 2.05 Filed a FAFSA (rounded unweighted n) Yesa 55.87 (2520) No 44.13 (1470) Filed early (rounded unweighted n) Yesa 45.96 (1200) No 54.04 (1320) a Indicates reference category for each variable b mean is shown for continuous variables 123 14 Res High Educ (2015) 56:1–28 Table 2 Descriptive statistics for students attending public 4-year institutions Filers and non-filers % of sample Filers only % of sample Femalea 55.50 56.13 Male 44.50 43.87 74.52 70.93 Gender Race/ethnicity Whitea African American 9.22 11.38 10.16 11.52 6.10 6.17 Dependenta 97.16 97.36 Independent 2.84 2.64 Hispanic Asian Dependency status Parental education Bachelors or highera 56.62 51.38 Less than a bachelors 43.38 48.62 Publica 89.20 90.21 Private 10.80 9.79 Full-timea 93.77 95.12 Part-time 6.23 4.88 [3.5a 46.36 46.37 3.0–3.5 36.82 37.31 0–2.999 16.82 16.32 2.22 2.21 93.22 95.17 6.78 4.83 HS type Enrollment intensity High school GPA ACT test score (divided by 10)b Delay enrollment Did not delaya Did delay for a year or more Major Declareda 73.76 76.13 Undeclared 26.24 23.87 MA or highera 72.74 72.66 BA or less 27.26 27.34 13.22 12.18 1.01 1.01 Highest expected level of education Expected family contribution (divided by 1,000)b Hours worked (divided by 10) per weekb Filed a FAFSA (rounded unweighted n) Yesa 73.87 (3040) No 26.13 (1020) Filed early (rounded unweighted n) Yesa 76.32 (2370) No 23.68 (670) a Indicates reference category for each variable b Mean is shown for continuous variables 123 Res High Educ (2015) 56:1–28 15 Table 3 Descriptive statistics for students attending private non-profit 4-year institutions Filers and non-filers % of sample Filers only % of sample Femalea 56.41 56.51 Male 43.59 43.49 Gender Race/ethnicity Whitea 76.02 74.06 9.49 10.79 10.28 10.93 4.21 4.21 Dependenta 98.17 98.09 Independent 1.83 1.91 Bachelors or highera 62.61 58.27 Less than a bachelors 37.39 41.73 Publica 79.42 82.63 Private 20.58 17.37 Full-timea 97.16 97.90 Part-time 2.84 2.10 [3.5a 53.07 52.88 3.0–3.5 30.97 31.04 0–2.999 15.95 16.08 2.28 2.25 93.72 94.54 6.28 5.46 Declareda 67.50 69.51 Undeclared 32.50 30.49 MA or highera 77.35 76.53 BA or less 22.65 23.47 15.75 14.48 0.72 0.75 African American Hispanic Asian Dependency status Parental education HS type Enrollment intensity High school GPA ACT test score (divided by 10)b Delay enrollment Did not delaya Did delay for a year or more Major Highest expected level of education Expected family contribution (divided by 1,000)b Hours worked (divided by 10) per weekb Filed a FAFSA (rounded unweighted n) Yesa 81.94 (2640) No 18.06 (590) Filed early (rounded unweighted n) Yesa 82.60 (2240) No 17.40 (400) a Indicates reference category for each variable b Mean is shown for continuous variables 123 16 Res High Educ (2015) 56:1–28 $12,000 $10,000 2003-04 Average State and Ins tu onal Grant Awards by Sector and 2003 FAFSA Filing Date $9,775 $8,000 $6,000 $4,000 $2,858 $2,663 $2,000 $- $1,165 January February March Private 4-Year $721 $474 April May Public 4-Year June July August Community College Fig. 2 Average amount of state and/or institutional aid by sector awarded by FAFSA filing month prior to Fall 2003 African Americans had 80 % (1.799 = 1/.556) higher odds of filing compared to White students. Community college students whose parents had earned less than associate’s degree had 68 % (1.675 = 1/.597) higher odds of filing a FAFSA compared to those with higher levels of educational attainment. Part-time students, compared to those attending full-time, had 75 % higher odds of not filing. Students who delayed enrollment had 75 % higher odds of not filing a FAFSA compared to peers who enrolled in the community college immediately after high school. Not declaring an academic major upon initial enrollment in the community college was associated with 43 % higher odds of not filing a FAFSA. As income status (measured by EFC) increased, the odds of not filing a FAFSA increased. With every $1,000 increase in EFC, the odds of not filing a FAFSA increased by 5 %. The last three columns of Table 4 include the results from the logistic regression that examined the factors associated with late filing among community college students who did file a FAFSA. The overall model is statistically significant (f = 7.32, p \ .000) and has a Cragg–Uhler pseudo r squared of .067 and correctly classifies 63 % of the 2,520 cases. Among FAFSA filers, male students have 27 % higher odds of filing late compared to females. Additionally, community college students attending part-time (73 % higher odds) and those who delayed enrollment into college after high school (181 % higher odds) had a greater probability of filing their FAFSA late. Prior academic performance and achievement in high school appear to impact the timing of FAFSA filing among community college students. Compared to students who earned a high school diploma, students who did not finish high school or earned their GED had considerably higher odds of filing late (130 % higher odds). Students who did not complete Algebra II or higher in high school, compared to those who did complete at least Algebra II, had 43 % higher odds of filing their FAFSA late. Table 5 presents results from the logistic regression models for students attending public 4-year institutions. The first three columns of data display the odds ratios and confidence intervals for factors associated with not filing a FAFSA. The overall model is statistically significant (f = 14.74, p \ .000) and has a Cragg–Uhler pseudo r squared of .121 and correctly classifies 76 % of the 4,050 cases. Compared to White students, African American (276 % higher odds; 3.759 = 1/.266) and Hispanic students (95 % higher odds; 1.953 = 1/.512) were more likely to file a FAFSA. Students whose parents have earned less than a bachelor’s degree, compared to their peers whose parents had earned a bachelor’s degree or higher, were more likely (116 % higher odds; 2.155 = 1/.464) to file. Graduates of private high schools have 35 % higher odds of not filing a FAFSA compared 123 Res High Educ (2015) 56:1–28 17 Table 4 Logistic regression models examining non-filing and late FAFSA filing: community college students Variable Factors associated with not filing a FAFSA Factors associated with filing a FAFSA late Odds ratio Odds ratio 95 % CI for odds ratio Lower Upper 95 % CI for odds ratio Lower Upper Ethnicity African American 0.556*** 0.401 0.772 1.502 0.987 2.286 Hispanic 1.212 0.864 1.700 1.268 0.859 1.869 Asian 0.989 0.641 1.526 1.064 0.595 1.900 1.607*** 1.331 1.941 1.273* 1.022 1.585 1.148 0.854 1.543 1.402 0.899 2.185 0.597*** 0.497 0.717 1.030 0.806 1.316 1.005 0.712 1.420 2.300** 1.384 3.821 1.745*** 1.393 2.187 1.731** 1.240 2.415 1.206 0.985 1.476 1.430** 1.113 1.838 1.752*** 1.390 2.208 2.811*** 2.060 3.835 1.425** 1.135 1.789 0.843 0.645 1.103 1.303 Gender Male Dependency status Independent Parental education Less than an associates High school degree No degree or GED Enrollment intensity Part time High school math Less than Algebra II Delay enrollment Yes (1 year or more) Major Undeclared Highest level of education expected 0.953 0.688 1.321 0.891 0.609 EFC (divided by 1,000) AA or less 1.045*** 1.032 1.059 0.989 0.977 1.002 Hours worked (divided by 10) 1.036 0.977 1.099 0.913 0.833 1.001 Model fit statistics Unweighted N 3990 2520 Correctly classified (classified ? if predicted Pr(D) [ = .5) 68.84 % 63.29 % Cragg–Uhler (Nagelkerke) R2: 0.181 0.067 BIC -2 Log likelihood (df) 4810.879 -2343.255 (15) 3339.124 -1610.831 (15) * p B .05; ** p B .01; *** p B .001 to graduates of public high schools. Similar to the community college sample, 4-year students who attended part-time (125 % higher odds), delayed enrollment into college (193 % higher odds), and did not declare an academic major in their first year (44 % higher 123 18 Res High Educ (2015) 56:1–28 Table 5 Logistic regression models examining non-filing and late FAFSA filing: public 4-year students Variable Factors associated with not filing a FAFSA Factors associated with filing a FAFSA late Odds ratio Odds ratio 95 % CI for odds ratio Lower Upper 95 % CI for odds ratio Lower Upper Ethnicity African American 0.266*** 0.173 0.407 1.034 0.686 1.558 Hispanic 0.512*** 0.359 0.728 0.953 0.625 1.451 Asian 0.821 0.536 1.260 0.591 0.348 1.003 0.968 0.796 1.178 1.195 0.954 1.495 1.109 0.542 2.270 1.949 0.896 4.239 0.464*** 0.382 0.562 0.911 0.713 1.163 1.349* 1.016 1.791 1.147 0.772 1.705 2.250*** 1.498 3.380 2.030** 1.292 3.191 3.0–3.5 1.047 0.844 1.300 1.256 0.970 1.628 0–2.999 1.274 0.957 1.698 1.393 0.921 2.109 0.990 0.796 1.233 0.574*** 0.459 0.718 2.933*** 1.938 4.438 3.874*** 2.146 6.993 1.440*** 1.175 1.764 1.228 0.900 1.676 1.482 Gender Male Dependency status Independent Parental education Less than a bachelors High school type Private Enrollment intensity Part time High school GPA ACT test score (divided by 10) Delay enrollment Yes (1 year or more) Major Undeclared Highest level of education expected 0.960 0.764 1.206 1.160 0.908 EFC (divided by 1,000) BA or less 1.008** 1.002 1.013 1.003 0.996 1.010 Hours worked (divided by 10) 0.987 0.909 1.072 1.175*** 1.074 1.285 Model fit statistics Unweighted N 4,050 3,040 Correctly classified (classified ? if predicted Pr(D) [ = .5) 75.46 % 79.37 % Cragg–Uhler (Nagelkerke) R2: 0.121 0.111 BIC 4,346.789 3,105.138 -2 Log likelihood (df) -2102.798 (17) -1484.419 (17) * p B .05; ** p B .01; *** p B .001 odds) were much more likely to not file a FAFSA. As income status increased the odds of not filing a FAFSA increased. With every $1,000 dollar increase in EFC, the odds of not filing a FAFSA increased by 1 %. 123 Res High Educ (2015) 56:1–28 19 Table 6 Logistic regression models examining non-filing and late FAFSA filing: private 4-year students Variable Factors associated with not filing a FAFSA Factors associated with filing a FAFSA late Odds ratio Odds ratio 95 % CI for odds ratio Lower Upper 95 % CI for odds ratio Lower Upper Ethnicity African American 0.429** 0.232 0.792 1.199 0.753 1.910 Hispanic 0.739 0.454 1.202 1.584 0.727 3.455 Asian 0.789 0.476 1.308 0.828 0.338 2.026 0.938 0.735 1.197 1.009 0.751 1.357 1.378 0.404 4.706 1.036 0.338 3.174 0.390*** 0.283 0.540 0.807 0.559 1.165 2.062*** 1.618 2.627 1.084 0.710 1.656 3.602*** 1.819 7.132 1.875 0.552 6.371 3.0–3.5 1.416* 1.059 1.892 1.517** 1.126 2.046 0–2.999 1.560* 1.022 2.382 2.181*** 1.486 3.201 1.349 0.970 0.605*** 0.482 0.761 3.052*** 1.664 5.598 6.786*** 3.563 12.923 1.377** 1.084 1.750 0.831 0.586 1.179 1.970 Gender Male Dependency status Independent Parental education Less than a bachelors High school type Private Enrollment intensity Part time High school GPA ACT Test Score (divided by 10) Delay enrollment Yes (1 year or more) Major Undeclared Highest level of education expected 0.870 0.616 1.230 1.465* 1.090 EFC (divided by 1,000) BA or less 1.006 0.999 1.012 1.004 0.995 1.014 Hours worked (divided by 10) 0.836* 0.724 0.965 1.173* 1.012 1.358 Model fit statistics Unweighted N 3,230 2,640 Correctly classified (classified ? if predicted Pr(D) [ = .5) 82.04 % 86.25 % Cragg–Uhler (Nagelkerke) R2: 0.131 0.191 BIC 2,934.901 2,061.777 -2 Log likelihood (df) -1,398.787 (17) -963.9433 (17) * p B .05; ** p B .01; *** p B .001 The last three columns in Table 5 display the odds ratios and confidence intervals for factors associated with filing a FAFSA late. The overall model is statistically significant (f = 8.16, p \ .000) and has a Cragg–Uhler pseudo r squared of .111 and correctly 123 20 Res High Educ (2015) 56:1–28 classifies 79 % of the 3,040 cases. Again, part-time attendance (103 % higher odds) and delaying enrollment into college (287 % higher odds) were very strong predictors of late filing. The other statistically significant variables in this model where SAT/ACT score and the number of hours worked off campus each week. Every 10 point increase in SAT/ACT test score resulted in 74 % higher odds (1.742 = 1/.574) of filing early, while there was 18 % higher odds of filing late for every 10 h a week a student worked for wages. Table 6 presents results from the logistic regression model for factors associated with not filing a FAFSA among students attending private 4-year institutions. The overall model is statistically significant (f = 10.85, p \ .000) and has a Cragg–Uhler pseudo r squared of .131 and correctly classifies 82 % of the 3,230 cases. Results were very similar to the public 4-year students. For instance, compared to White students, African American (133 % higher odds; 2.331 = 1/.429) were more likely to file a FAFSA. Students whose parents have earned less than a bachelor’s degree, compared to their peers whose parents had earned a bachelor’s degree or higher, were more likely (156 % higher odds; 2.564 = 1/.390) to file. Graduates of private high schools have 106 % higher odds of not filing a FAFSA compared to graduates of public high schools. Similar to both the community college sample and the public 4-year sample, private 4-year students who attended part-time (260 % higher odds), delayed enrollment into college (205 % higher odds), and did not declare an academic major in their first year (38 % higher odds) were much more likely to not file a FAFSA. Unlike the public 4-year students, hours worked is positively associated with filing a FAFSA (20 % higher odds of filing for every 10 h worked; 1.196 = 1/.836), but EFC is not related to filing a FAFSA. Among private 4 year students, lower high school grade point averages are associated with not filing a FAFSA, compared to students who earned a high school grade point average above 3.5. The last three columns in Table 6 display the odds ratios and confidence intervals for factors associated with filing a FAFSA late. The overall model is statistically significant (f = 13.17, p \ .000) and has a Cragg–Uhler pseudo r squared of .191 and correctly classifies 86 % of the 2,640 cases. Analogous to both the community college sample and the public 4-year sample, delaying enrollment into college (579 % higher odds) is a very strong predictor of late filing. Comparable to the public 4-year model, every 10 point increase in SAT/ACT test score resulted in 65 % higher odds (1.652 = 1/.605) of filing early, while there were 17 % higher odds of filing late for every 10 h a week a student worked for wages. Unique to the private 4-year model were the associations between lower high school grade point averages and lower educational expectations and filing the FAFSA late, as well as the lack of association for part-time enrollment and filing the FAFSA late. Discussion and Implications Results from our study indicate several of the factors that traditionally place a student atrisk of nonsuccess in college (i.e. delaying enrollment after high school, attending parttime, not declaring an academic major) are strong predictors of not filing a FAFSA or filing late. Delaying enrollment and attending college part-time, in particular, produced dramatic differences in the odds of first-year students’ FAFSA filing behavior. Our findings also reveal that on average, late filers receive less total state and institutional grant aid than students who filed their FAFSA earlier. This finding was consistent across all three institutional sectors. Overall, community college students exhibited the greatest number of 123 Res High Educ (2015) 56:1–28 21 factors that place them at risk for not filing a FAFSA or filing late, while students attending private 4-year institutions were much more likely to file and file early. The additional grant aid received by an early FAFSA filer, compared to a late filer, could have made it possible to enroll in additional courses during their first year, which has important implications for students’ academic momentum and likelihood of degree completion. First, the ability to complete a greater number of credit hours during the first year of college is a strong predictor of persistence for students attending community colleges (Adelman 2005) and 4-year institutions (Adelman 2006). Consequently, a student who files their FAFSA late may be forced to take fewer classes during their first academic year due to financial limitations, thus hindering their academic momentum. Second, financial constraints are one of the primary reasons that students, particularly those who are lowerincome, dropout of college (Chen 2008; Walpole 2003). By filing a FAFSA late, a firstyear student can miss out on grant aid that could provide them with the ability to pay for college-related expenses (e.g. textbooks, computer, transportation) while reducing their dependency on loans. For example, students attending community colleges who filed their FAFSA early received, on average, about $700 more in total state and institutional grant aid than late filers. During the time period under investigation (i.e. the 2003–2004 academic year), the average tuition and fees charged by community colleges was $1,905 and the average cost of books and supplies was $745 (College Board 2003). Therefore, the amount of additional grant aid received as function of filing the FAFSA early would have been enough to cover the costs of books and supplies, or approximately 38 % of the total costs of tuition and fees. Given the lack of financial resources among many community college students, this additional grant could have had a considerable impact. The following sections situate the results of our study within the extant financial aid research and policy literature. The contextual layers and key constructs of Perna’s conceptual model are used to organize our discussion and the implications of our findings. Background Characteristics and Social and Cultural Capital Gender was not a significant predictor of FAFSA filing behavior for 4-year students, but males attending community colleges were much more likely than females not to file a FAFSA and file late. Policy reports that included a combined sample of 2- and 4-year students found that males were disproportionally represented among non-filers (Kantrowitz 2009) and among students who indicated they did not file because the FAFSA was too much work (Kantrowitz 2011). The greater likelihood of not filing a FAFSA among community college males could potentially be attributed to differences in the college search and planning process between males and females (Stage and Hossler 1989), but additional research is needed to better understand why male community college students are at high risk of not filing. Prior policy reports on FAFSA filing (Kantrowitz 2009, 2011; King 2004, 2006) have not found significant proportional differences in filing rates as a function of students’ race/ ethnicity. Our study examined FAFSA filing by race/ethnicity while controlling for a number of factors than were not included in these descriptive policy reports. Compared to their White peers, African American students across all three institutional sectors were more likely to file a FAFSA. Hispanic students enrolled at public 4-year colleges were more likely to file than White students. Considering that receipt of grant aid is often particularly important for promoting college persistence among racial/ethnic minorities (Chen 2008), we believe these findings are promising. While there are mixed results in the 123 22 Res High Educ (2015) 56:1–28 literature, several studies have found that when controlling for other relevant factors, African American (Hurtado et al. 1997) and Hispanic (Plank and Jordan 2001) high school students exhibit higher demand for higher education than their White peers. Another possible explanation for this finding is these student groups, which are disproportionally represented among those from lower-income households, file a FAFSA because they realize they will need financial assistance to pay for college. Perna’s model underscores the importance of social and cultural capital in helping students navigate the college search and planning process. In our study, students across all three institutional sectors were more likely to file a FAFSA if their parents had lower levels of educational attainment. These results seem counterintuitive in light of the conceptual model, but as stated previously, may very well reflect the recognition by students from less affluent households that financial aid will be necessary for them to pursue higher education. Human Capital Perspectives Findings from this study indicate that the human capital model is particularly important in explaining financial aid application behaviors among first-year college students. Individuals who exhibited lower demand for higher education were much more likely not to file a FAFSA or to file late. Delaying enrollment into postsecondary education for 1 year or more after high school was a strong predictor of not filing and late filing across all three institutional sectors. We contend this finding is attributed to the reality that students who delay enrollment are distanced from the high school context, and therefore do not have immediate access to important forms of social capital such as guidance counselors, teachers, and peers that could help them with college planning. An alternative explanation could be that a lower financial aid award led a student to delay postsecondary enrollment because of a lack of financial resources to pay for college, but this is not discernable with the BPS:04/06 dataset since FAFSA filing behavior was examined only among students who did enroll in the 2003–2004 academic year. The high rate of non-filing and late FAFSA filing among students who delay enrollment presents a perplexing challenge for policymakers, high schools, and colleges. These students are disconnected from the high school context where they could receive help filing a FAFSA, and often they do not communicate with the postsecondary institution where they enroll until late in the registration process. One potential strategy to address this issue is to encourage every high school senior complete a FAFSA before the priority application deadline because filing is free for students. This information could help some high school students who are considering delaying college enrollment due to financial concerns understand they can afford college with help from financial aid. In addition, community outreach initiatives aimed at increasing FAFSA filing and college-going may be particularly valuable in reaching students who did not immediately pursue higher education after high school. As hypothesized, the results showed that students across all three institutional sectors that entered college with an undeclared major were more likely not to file a FAFSA than their peers that declared a major. High school students with more clearly defined career aspirations and goals typically engage in the college search and planning process earlier than their undecided peers (Hossler et al. 1999). While unrealistic to expect every high school student to have confirmed their academic major before entering college, this finding suggests that early commitment to a particular career field could have the indirect benefit of increasing the number of students who apply for financial aid during their college planning process. 123 Res High Educ (2015) 56:1–28 23 A student’s income status is a common measure used to gauge their supply of financial resources to pay for college-related expenses. Lower-income students have naturally been a major focus of prior studies that examine students’ FAFSA filing behaviors (Feeney and Heroff 2013; Kantrowitz 2009, 2011; King 2004, 2006; McKinney and Novak 2013; Novak and McKinney 2011) because these students have the greatest need for financial assistance. In the present study, as the level of income increased for students attending community college and public 4-year institutions, their odds of not filing a FAFSA increased. This finding is in agreement with previous reports suggesting that higher income students are often less likely to file a FAFSA (Kantrowitz 2011; King 2006). Pursuing higher education on a part-time basis, compared to full-time enrollment, was a strong predictor of not filing a FAFSA across all three institutional sectors and late filing at community colleges and public 4-years. As previously discussed, the relationship between FAFSA filing behavior and enrollment intensity in college is complex. A student intending to enroll full-time, for example, may have filed their FAFSA late and therefore been offered a lower financial aid package. This lower award amount may have forced to the student to procure employment and/or reduce the number of courses attempted during their first-year because of a shortage of financial resources. Understanding the relationship between FAFSA filing behavior and students’ enrollment decisions is an important area for future research. But enrolling part-time during the first year of college may reflect a lower demand for higher education among high school students, which could have resulted in filing the FAFSA after the priority deadline. The ‘pay as you go’ college financing strategy utilized by a significant number of today’s part-time students who work (Perna 2010) may have led these students to believe that applying for financial aid was unnecessary. But as Ziskin and colleagues (2014) have found, this behavior can become a barrier to persistence when students work additional hours to pay for college expenses that could have covered by financial aid. Students often choose to pursue college part-time because they do not want to take out student loans and avoiding debt is the reason some students reported they did not file a FAFSA (Kantrowitz 2011). These findings indicate the need to help better educate part-time students about their eligibility for grant aid and the benefits of applying early for financial aid. Institutional Contexts The type of high school and postsecondary institution a student attends can shape their decision-making during the college planning process (McDonough 1997; Perna 2006a, b). The type of high school a student attends may serve as one proxy for their level of access to information and guidance (i.e. social capital) about applying for financial aid, with private high schools typically being more resource-rich than publics. In this study, public and private 4-year college students who attended private high schools were more likely not to file a FAFSA than their peers who attend public high schools. While this finding may first appear counterintuitive, it is well-aligned with prior analyses of FAFSA filing that indicate higher-income students are less likely to file than their lower-income peers (Kantrowitz 2011; King 2006). Higher-income students are overrepresented in private high schools and many of these students elect not to file a FAFSA because their families can afford (or perhaps the student just believes they can afford) the first year of college without the receipt of financial aid. Increasing the rates of FAFSA filing, and early filing, among community college students could help increase persistence and graduate rates at these institutions (McKinney 123 24 Res High Educ (2015) 56:1–28 and Novak 2013). Nearly half of all community college students who enroll in a given Fall term will dropout before their second Fall term begins (American Association of Community Colleges 2012). Financial concerns are a primary reason for this institutional departure, so increasing FAFSA filing and early filing among this population could provide grant aid that helps these students persist. Consistent with prior recommendations (Kantrowitz 2011; King 2004), our results strongly suggest that FAFSA completion initiatives should be targeted at high schools seniors who are the most likely to enroll in community colleges and late enrollees at community colleges. Compared to students at public and private 4-year institutions, community college students exhibit more of the characteristics (e.g. delaying enrollment, part-time attendance, having an undeclared major) that place them at high risk of not filing a FAFSA or filing late. Partnerships between high schools and local community colleges that help students complete the FAFSA before the priority application deadline may be a particularly useful strategy (College Board Advocacy and Policy Center 2010). Social, Economic, and Policy Context Layer 4 of Perna’s conceptual framework is useful in situating the results from this study within the broader landscape of American higher education and the current financial aid system. In particular, the public policy context gives attention to the policies and structures that may discourage, or encourage, college-going behaviors (Perna 2006a, b). The complexity of the FAFSA form itself has been identified as a formidable barrier to college access, particularly among student groups that have traditionally been underrepresented in higher education (see Asher 2007; Bettinger et al. 2009; Dynarski and Scott-Clayton 2006, 2007, 2008a; National Economic Council 2009). Surveys of non-filers reveal that the complexity of the FAFSA is a deterrent for a considerable number of students (Kantrowitz 2011). Dynarski and Scott-Clayton (2008b) concluded that from an economic perspective, the benefit of targeting federal aid is grossly outweighed by the cost of the FAFSA’s complexity. In fact, it is estimated that the financial cost associated with the FAFSA, measured by time spent completing the application and administrative burden to colleges, is at least $4 billion a year (Dynarski and Scott-Clayton 2008b). Widespread concern about the complexity of the FAFSA and the financial aid application process resulted in a revised and shortened version of the FAFSA that was unveiled in 2010. The FAFSA now allows students to seamlessly retrieve their relevant tax information from the IRS to make completing the online FAFSA simpler. However, the full potential of linking tax data with the FAFSA has yet to be realized because relatively few students utilize this option (Dynarski and Wiederspan 2012). Several community-based initiatives have been introduced in recent years to help increase FAFSA rates, particularly among lower-income students and those who are not traditionally viewed as college bound. These initiatives include College Goal Sunday, the FAFSA H&R Block Experiment (Bettinger et al. 2009), and other community partnerships that integrate tax preparation with FAFSA completion (Duan-Barnett and Mabry 2012). Understanding how these collective efforts have impacted FAFSA filing rates is of considerable value to policymakers, researchers, and postsecondary institutions. Driven by the seemingly perpetual shortages in the availability of grant aid for college students, and in recognition of major inefficiencies in the current financial aid system, financial aid reform has been a hot topic on the policy agenda in recent years. Recently, more than a dozen national postsecondary associations and institutes were tasked, as part of the reimagining aid design and delivery (RADD) project funded by the Bill & Melinda 123 Res High Educ (2015) 56:1–28 25 Gates Foundation, to identify innovative policies and practices that could improve the US student financial aid system. A theme across multiple RADD reports is that further streamlining of the application process is necessary (see Mullin 2013), but must be supported by efforts to provide students with detailed information about their aid eligibility much earlier in their college search and planning process. A clear message from these reports is that simplifying the financial aid application process can improve college access and completion by helping to ensure that higher education is financially possible for a greater number of qualified students. Conclusion A clear understanding of who does, and does not, apply for financial aid has major implications for the success of any financial aid program (King 2004). Considering the FAFSA is the critical gatekeeper to many types of financial aid for most college students, few empirical studies have investigated students’ FAFSA filing behavior. The present study applied an integrated conceptual model to explain FAFSA filing among first-year college students and found that early applicants receive, on average, more total state and institutional grant aid than those who apply after the priority deadline. There are some promising results from our study in that across all three institutional sectors, students who are African American, attended a public high school, and whose parents had lower levels of educational attainment were more likely to file a FAFSA. However, it is troubling that many student groups (i.e. those attending part-time, delaying college enrollment, and exhibiting lower pre-college academic performance) who would benefit most from the receipt of additional grant aid are the most likely to file their FAFSA late, and therefore often miss out on receiving the total amount of grant aid for which they are eligible. Findings from our study can be used to inform current policy discussions, communitybased initiatives aimed at increasing FAFSA filing, and future research. We believe the greatest potential for future research in this area includes studies that examine FAFSA filing behaviors, and the consequences of filing late, within a more narrowly-focused context. For example, this line of inquiry may be particularly insightful for institutional researchers working in states, and at institutions, whose financial aid programs are often depleted before all eligible applicants have received a grant award. This type of analysis would better account for differences in priority application deadlines, program eligibility criteria, and award amounts across state and institutional contexts. The conceptual and methodological approaches used in our study can serve as a guide for future work on this topic. Though considerable efforts have been made to increase awareness of existing financial aid programs and the utilization of grant aid among lower-income students, today we are still struggling to achieve these important goals. The ongoing trend that such a larger number of eligible students miss out on this financial assistance suggests that current policy discussions around financial aid reform are necessary and of critical importance (Dynarski and Wiederspan 2012). There is a real need to rethink, and reinvent, the processes involved in applying for and receiving financial aid. As King (2004) has articulately stated, ‘‘no student should miss the opportunity for vital financial assistance because he or she lacks necessary information, is misinformed about the nature of student aid programs, or is unable to navigate the financial aid application process’’ (p. 7–8). But as findings from our study and other recent research suggest, these circumstances are still too often the case for many student groups that have traditionally 123 26 Res High Educ (2015) 56:1–28 been underrepresented in higher education. It would be unjust to carry on with business as usual when we know the current financial aid application process serves as a barrier to college access and success for thousands of students each year. Acknowledgments An earlier version of this paper was presented at the 2013 Association for Institutional Research (AIR) Forum in Long Beach, California. The authors thank Tom Biedscheid, Catherine L. Horn, Christopher M. Mullin, and two anonymous reviewers for their helpful comments and feedback on previous drafts. References Adelman, C. (2005). Moving into town—and moving on: The community college in the lives of traditionalage students. Washington, DC: U.S. Department of Education. Adelman, C. (2006). The toolbox revisited: Paths to degree completion from high school through college. Washington, DC: U.S. Department of Education. American Association of Community Colleges. (2012, April). Reclaiming the American Dream: A report from the 21st-Century Commission on the Future of Community Colleges. Washington, DC: Author. Asher, L. (2007). Going to the source: A practical way to simplify the FAFSA. Oakland: Institute for College Access & Success. Becker, G. S. (1997). Human capital: A theoretical and empirical analysis, with special reference to education. New York: National Bureau of Economic Research. Bettinger, E., Long, B.T., Oreopoulos, P., & Sanbonmatsu, L. (2009). The role of simplification and information in college decisions: Results from the H&R Block FAFSA Experiment. NBER Working Papers, No. 15361. National Bureau of Economic Research, Inc. Cellini, S. R. (2008). Causal inference and omitted variable bias in financial aid research: Assessing solutions. The Review of Higher Education, 31(3), 329–354. Chen, R. (2008). Financial aid and student dropout in higher education: A heterogeneous research approach. Higher Education, 23, 209–239. Christou, C., & Haliassos, M. (2006). How do students finance human capital accumulation? The choice between borrowing and work. British Journal of Sociology in Education, 24, 621–636. College Board. (2003). Trends in college pricing 2003. Washington, DC: College Board. College Board Advocacy and Policy Center. (2010). The financial aid challenge: Successful practices that address the underutilization of financial aid in community colleges. Washington, DC: College Board Advocacy and Policy Center. DesJardins, S. L. (2001). A comment on interpreting odds-ratios when logistic regression coefficients are negative. AIR Professional File, 81. DesJardins, S. L., & Toutkoushian, R. K. (2005). Are students really rational? The development of rational thought and its application to student choice. In J. C. Smart (Ed.), Higher education: Handbook of theory and research (Vol. 20, pp. 191–240). New York: Springer. Dowd, A. C., & Duggan, M. B. (2001). Computing variances from data with complex sampling designs: A comparison of Stata and SPSS. In North American Stata Users’ Group Meetings 2001 (No. 3.1). Stata Users Group. Duan-Barnett, N., & Mabry, B. (2012). Integrating tax preparation with FAFSA completion: Three case models. Journal of Student Financial Aid, 42(3), 25–45. Dynarski, S., & Scott-Clayton, J. (2006). The cost of complexity in federal student aid: Lessons from optimal tax theory and behavioral economics. KSG Faculty Research Working Paper Series RWP06013 and NBER Working Paper 12227. Dynarski, S. & Scott-Clayton, J. (2007). College grants on a postcard: A proposal for simple and predictable federal student aid. The Brookings Institution. Policy Brief No. 2007-01. Washington, DC. Dynarski, S., & Scott-Clayton, J. E. (2008a). Complexity and targeting in federal student aid: A quantitative analysis. HKS Faculty Research Working Paper Series RWP08-005 and NBER Working Papers 13801. Dynarski, S. M., & Scott-Clayton, J. E. (2008b). Complexity and targeting in federal student aid: A quantitative analysis. In NBER/tax policy & the economy (Vol. 22, pp. 109–150). University Of Chicago Press. Dynarski, S., & Wiederspan, M. (2012). Student aid simplification: Looking back and looking ahead. National Tax Journal, 65(1), 211–234. Feeney, M., & Heroff, J. (2013). Barriers to need-based financial aid: Predictors of timely FAFSA completion among low-income students. Journal of Student Financial Aid, 43(2), 65–85. 123 Res High Educ (2015) 56:1–28 27 Gansemer-Topf, A., & Schuh, J. (2005). Institutional grants: Investigating student retention and graduation. Journal of Student Aid, 35(3), 5–20. Gross, J. P., Hossler, D., & Ziskin, M. (2007). Institutional aid and student persistence: An analysis of the effects of institutional financial aid at public 4-year institutions. Journal of Student Aid, 37(1), 28–39. Hosmer, D. W., & Lemeshow, S. (2000). Applied logistic regression. New York: Wiley. Hossler, D., Schmit, J., & Vesper, N. (1999). Going to college: How social, economic, and educational factors influence the decisions students make. Baltimore: John Hopkins University Press. Hurtado, S., Inkelsa, K. K., Briggs, C., & Rhee, B. S. (1997). Differences in college access and choice among different racial/ethnic groups. Research in Higher Education, 38(1), 43–75. Kane, T. J., & Avery, C. (2004). Student perceptions of college opportunities: The Boston COACH Program. In Caroline Hoxby (Ed.), College decisions: The new economics of choosing, attending and completing college. Chicago: University of Chicago Press. Kantrowitz, M. (2009). Analysis of why some students do not apply for financial aid. Student Aid Policy Analysis. Retrieved on April 20, 2011 from http://www.finaid.org/educators/studentaidpolicy.phtml. Kantrowitz, M. (2011). Reasons why students do not file the FAFSA. Student Aid Policy Analysis. Retrieved on April 20, 2011 from http://www.finaid.org/educators/studentaidpolicy.phtml. King, J. E. (2004). Missed opportunities: Students who do not apply for financial aid. Washington, DC: American Council on Education, Center for Policy Analysis. King, J. E. (2006). Missed opportunities revisited: New information on students who do not apply for financial aid. Washington, DC: American Council on Education, Center for Policy Analysis. LaManque, A. (2009). Factors associated with delayed submission of the free application for federal financial aid. Journal of Applied Research in the Community College, 17, 6–12. Long, J. S. (1997). Regression models for categorical and limited dependent variables (Vol. 7). Thousand Oaks: Sage Publications Inc. Manski, C. F., & Wise, D. A. (1983). College choice in America. Cambridge: Harvard University Press. McDonough, P. M. (1997). Choosing colleges: How social class and schools structure opportunity. Albany: State University of New York Press. McDonough, P. M., & Calderone, S. (2006). The meaning of money: Perceptual differences between college counselors and low-income families about college costs and financial aid. American Behavioral Scientist, 49(12), 1703–1718. McKinney, L., & Novak, H. (2013). The relationship between FAFSA filing and persistence among firstyear community college students. Community College Review, 41(1), 63–85. McKinney, L., & Roberts, T. (2012). The role of community college financial aid counselors in helping students understand and utilize financial aid. Community College Journal of Research and Practice, 36, 761–774. McKinney, L., Roberts, T., & Shefman, P. (2013). Financial aid counselors’ perspectives and experiences of loan borrowing among community college students. Journal of Student Financial Aid. Mendoza, P., Mendez, J., & Malcolm, Z. (2009). Financial aid and persistence in community colleges: Assessing the effectiveness of federal and state financial aid programs in Oklahoma. Community College Review, 37, 112–135. Mullin, C. M. (2013). An analysis of the reimagining aid design and delivery (RADD) white papers. Washington, DC: American Association of Community Colleges. National Economic Council (2009). Simplifying student aid: The case for an easier, faster, and more accurate FAFSA. Executive Office of the President, Council of Economic Advisors. Retrieved online on January 14, 2013 from http://www.whitehouse.gov/assets/documents/FAFSA_Report.pdf. Novak, H., & McKinney, L. (2011). The consequences of leaving money on the table: Examining persistence among students who do not file a FAFSA. Journal of Student Financial Aid, 41(3), 5–23. Perna, L. W. (2006a). Studying college access and choice: A proposed conceptual model. In J. C. Smart (Ed.), Higher education: Handbook of theory and research (Vol. 21, pp. 99–157). New York: Springer. Perna, L. W. (2006b). Understanding the relationship between information about college costs and financial aid and students’ college-related behaviors. American Behavioral Scientist, 49, 1620–1635. Perna, L. W. (2008). Understanding high school students’ willingness to borrow to pay college prices. Research in Higher Education, 49, 589–606. Perna, L. W. (2010). Understanding the working college student: New research and its implications for policy and practice. Sterling: Stylus Publishing. Perna, L. W., & Titus, M. (2005). The relationship between parental involvement as social capital and college enrollment: An examination of racial/ethnic group differences. The Journal of Higher Education, 76(5), 485–518. Plank, S. B., & Jordan, W. J. (2001). Effects of information, guidance, and actions on postsecondary destinations: A study of talent loss. American Educational Research Journal, 38(4), 947–979. 123 28 Res High Educ (2015) 56:1–28 Somers, P., Cofer, J., & Vander Putten, J. (2001). The early bird goes to college: The link between early college aspirations and postsecondary matriculation. Journal of College Student Development, 43(1), 93–97. St. John, E. P., Musoba, G. D., & Simmons, A. B. (2003). Keeping the promise: The impact of Indiana’s twenty-first century scholars program. Review of Higher Education, 27(1), 103–123. St. John, E. P., & Paulsen, M. B. (2001). The finance of higher education: Implications for theory, research, policy, and practice. In M. B. Paulsen & J. Smart (Eds.), The finance of higher education: Theory, research, policy, and practice (pp. 545–568). New York: Agathon Press. Stage, F. K., & Hossler, D. (1989). Differences in family on college attendance plans for male and female ninth graders. Research in Higher Education, 30(3), 301–315. Stanton-Salazar, R. D. (1997). A social capital framework for understanding the socialization of racial Minority children and youths. Harvard Education Review, 67(1), 1–40. Thomas, S. L., & Heck, R. H. (2001). Analysis of large-scale secondary data in higher education research: Potential perils associated with complex sampling designs. Research in Higher Education, 42(5), 517–540. Tierney, W. G., & Venegas, K. M. (2006). Fictive kin and social capital: The role of peer groups in applying and paying for college. The American Behavioral Scientist, 49(12), 1687–1702. Vargas, J. H. (2004). College knowledge: Addressing information barriers to college. Boston, MA: The Education Resources Institute (TERI). Walpole, M. (2003). Socioeconomic status and college: How SES affects college experiences and outcomes. The Review of Higher Education, 27(1), 45–73. Ziskin, M., Fischer, M. A., Torres, V., Pellicciotti, B., & Player-Sanders, J. (2014). Working students’ perceptions of paying for college: Understanding the connections between financial aid and work. The Review of Higher Education, 37(4), 429–467. 123