Introduction

Recently, the academic community as well as the popular media has focused attention on the somewhat dramatic reversal of a previous trend in college matriculation; specifically, beginning in the 1980s, women’s rates of college attendance have eclipsed those of men’s. By 2004, women comprised approximately 57% of students at postsecondary institutions nationwide (Knapp et al. 2006). While females of all racial/ethnic groups now attend and complete college at rates higher than their co-ethnic male peers (Garcia and Bayer 2005; NCES 2005), recent research has called into question whether there is a one-size-fits-all explanation behind the higher postsecondary attendance rates of females from different racial/ethnic backgrounds (Barajas and Pierce 2001; Perna 2000).

On average, minority and majority youth differ greatly with regard to educational opportunities and attainment (Morales and Saenz 2007; Schneider et al. 2006; Stevens et al. 2006); it likely follows that minority females endure different obstacles and employ different strategies in their path to college than their white female peers. While some previous research has examined this possibility with regard to black–white differences (Buchmann and Diprete 2006), there has been scarce attention paid to the factors associated with the female postsecondary advantage among Hispanic youth.Footnote 1 Hispanics make up the fastest growing ethnic group in the United States, having increased by 267% over the past three decades (Suro and Passel 2003) and subsequently comprising a greatly increasing share of K-12 students. This study therefore contributes to the emerging research literature on the female postsecondary advantage by focusing explicit attention on the educational paths of Hispanic youth, with an eye towards understanding whether the higher college attendance rates of Hispanic and white females relative to their male peers are attributable to similar factors.

Specifically, the objective of this study is to examine whether gender differences in social capital in high school, in addition to gender differences in academic performance and orientation, function similarly to help explain the female postsecondary advantage in matriculation for both white and Hispanic youth. As girls’ higher academic performance and orientation during high school, including their higher grades and general efforts and engagement centered around school, have been found to be a key determinant of the greater aggregate matriculation rates of females in general (Buchman and DiPrete 2006; Downey and Yuan 2005; Jacob 2002), I therefore consider the specific contribution of such factors to the female college advantage for Hispanic as well as white youth. An additional major contribution of this study is that it considers the contribution of social capital to girls’ advantage in college-going. I examine whether high schools girls are more involved in the kinds of interpersonal relationships that are most beneficial for college-going, including those with friends, parents, and institutional agents such as high school counselors, and subsequently explore whether such forms of social capital contribute in similar ways to the female advantage in matriculation among Hispanic youth and among white youth.

To address this topic, I utilize data from the Texas Higher Education Opportunity Project (THEOP), which surveyed a recent statewide cohort of high school seniors. Texas has aggregate and gender-specific college matriculation rates that are highly comparable to the national average, but differs importantly from many other states due to its large representation of individuals of Hispanic origin (Knapp et al. 2006). As evidenced by the limitations of previous studies, the relatively small representation of Hispanic youth in current national datasets deters a thorough examination of the gendered paths to higher education among Hispanic youth.Footnote 2 Therefore, the THEOP data provides an exceptional opportunity to examine this important and timely topic.

Academic Performance and Orientation

Previous studies seeking to explain the contemporary female postsecondary advantage have focused on two key factors: girls’ greater levels of academic performance and their stronger orientation towards academically-related behaviors and activities both inside and outside of school (Buchmann and Diprete 2006; Jacob 2002). Before and during the high school years, girls are less likely to be placed in remedial courses and to drop out of school, earn higher grades across subjects, and take comparable or even larger numbers of college-preparatory courses (Bae et al. 2000; Freeman 2004; NCES 2001). And while the idea that girls often outperform boys in school is not necessarily new, it appears that their scope of advantage in high school is increasing. For example, girls now take Calculus, a traditionally male-dominated course, at comparable rates to boys (Correll 2001; Downey and Yuan 2005; Riegle-Crumb et al. 2006).

Additionally, previous studies have focused on the extent to which the contemporary college gap favoring females is resultant of girls’ greater academic orientation, such that they tend to be more academically engaged than boys, prioritizing school work and fulfilling the role of a ‘good student’ that is expected by school personnel (Downey and Yuan 2005; Johnson et al. 2001; Mickelson 1989). On average, girls also spend more time on homework and more time reading outside of school (Buchmann and Diprete 2006). In contrast, boys are more likely to get in trouble at school for behavioral problems, come to class without materials, and spend time out of school on non-academic endeavors (Jacob 2002). While correlated with academic performance, girls’ greater academic orientation likely has an independent effect on their college matriculation, as it may lead them to invest greater time and effort on college applications and decisions, as well as to prioritize continued participation in an academic world where they have previously been successful.

Yet as stated earlier, while research points to girls’ greater academic performance and orientation as key to understanding the aggregate female postsecondary advantage, we currently have little knowledge of the extent to which such factors contribute equally to the matriculation patterns observed among the Hispanic student population as among the majority white population. Compared to their co-ethnic male peers, Hispanic girls do exhibit higher levels of academic preparation in high school, and are also more likely than boys to articulate the importance of education as a way of getting ahead and voice higher educational aspirations (Crosnoe et al. 2004; Gándara 2005; López 2003; Mead 2006; Saenz and Ponjuan 2009). However, it does not necessarily follow that Hispanic females are similarly able to convert their higher relative levels of academic performance and orientation to greater rates of postsecondary matriculation as white female peers do. For example, greater financial constraints may prevent high-achieving Latina students from realizing their postsecondary goals, as could strong obligations to the family of origin (Sy 2006). Some have also suggested that the academic credentials of minority youth are viewed with greater skepticism by college admission counselors who may doubt that the achievement of students in a high-minority school is sufficiently comparable to students from other schools (Zarate and Gallimore 2005). On the other hand, gender differences in academic preparation might be even more important in explaining Hispanic girls’ higher relative rates of matriculation if their performance represents not just mastering skills and content, but signals perseverance and determination in the face of many obstacles to their opportunities to learn (Barajas and Pierce 2001). Therefore, it remains an empirical question whether academic performance and orientation contribute equally to the female postsecondary advantage among both Hispanic and white youth.

Social Capital

While the contributions of academic performance and orientation have been previously examined in the literature, there is a potential factor in women’s college matriculation advantage that has received far less attention. In short, girls may have greater resources of social capital that promote matriculation to college. Previous research has shown that enrollment in college is related to the resources that young people can access through their relationships with others, such as parents, teachers, school staff, and peers (McDonough 1997; Perna and Titus 2005). Such relationships have the potential to increase college matriculation in a myriad of ways, including providing psychological encouragement, emotional support, academic assistance, and relevant information and guidance in the college application process (Carbonaro 1998; Kim and Schneider 2005; Stanton-Salazar and Dornbusch 1995). A focus on relationships with friends, parents, and school staff as potential sources of social capital broadens the scope of consideration of factors that increase a student’s chance of college attendance beyond that student’s own individual choices and actions.

Previous research on gender differences in social capital has focused on adult women’s lack of access to the kinds of social networks that facilitate occupational and economic success (Burt 1998; Lin 2000). In contrast to this evidence of women’s social capital deficit in the labor force, I suggest that compared to boys, high school girls may have greater amounts of the kinds of social capital that are conducive to successful college matriculation. For example, adolescent girls typically have personal relationships with friends that involve significant amounts of discussion and sharing of information, not only about social concerns but also pertaining to school and academic issues (Felmlee 1999; Giordano 2003; Leslie et al. 1998). In contrast, boys’ friendships are more often focused on competition and recreational activities, and less on academic support (Beutel and Marini 1995). Subsequently, several recent studies find stronger links between friendship ties and academic outcomes during high school for girls than boys (Frank et al. 2008; Riegle-Crumb et al. 2006).

Additionally, girls’ other-directedness extends beyond close relationships with friends to those with parents and other adults as well, providing them greater potential access to the social capital inherent in such ties (Lieberman et al. 1999). Some studies find that adolescent girls report more closeness and attachment to their relationships with parents than boys do, as well as a greater willingness to initiate conversations with them (Laursen and Collins 2004; Stattin and Kerr 2000). Similarly, López (2003) finds that minority girls report a stronger awareness of their parents’ educational expectations and generally strive harder to reach them. Within the walls of school, there is also evidence suggesting that girls are more likely than boys to forge close ties with teachers and staff, possibly as a result of their greater tendency to ask questions and work to meet and exceed school rules and expectations (Eccles and Blumenfeld 1985; Hughes et al. 2001). These relationships form possible conduits through which students can receive both concrete information and guidance, as well as access less tangible resources like emotional and psychological support.

Girls’ greater willingness to seek out or engage in interpersonal contact with significant others may therefore be a key to their academic success. Previous studies confirm that Hispanic girls rely more on their friends for academic support than do their male co-ethnic peers, and that they are similarly more likely to have strong relationships with parents as well as tighter connections with school personnel (Barajas and Pierce 2001). Thus, to the extent that both white and Hispanic females participate more actively in such relationships compared to their male peers, girls’ greater levels of social capital could be a key factor in their postsecondary educational advantage.

Yet it is also important to point out that social capital may function differently in the path to college for Hispanic and white youth. Research by Perna (2000) found that social capital mattered more in the decision to attend college for Hispanic and black students than it did for white students. Given their comparatively low levels of achievement and relative lack of human capital resources at home (Kao and Rutherford 2007; Schmid 2001; White and Glick 2000), minority youth may be the most in need of social capital benefits. In other words, there is perhaps greater potential for social capital to operate and make a real difference in Hispanic students’ educational trajectories. This may be particularly true with regard to relationships with institutional agents such as teachers and counselors who can offer vital knowledge about the details of the college admission process and the benefits of college attendance (Tornatzky et al. 2002). As a high percentage of Hispanic youth will be first-generation college goers, they may benefit the most from relationships with those individuals that can supply the information they need to make college a reality (Gonzalez et al. 2003; Perez and McDonough 2008). If Hispanic girls are more likely than their male peers to engage and personally interact with school personnel, then such factors could be even more important in explaining the gender gap in college attendance among Hispanic youth than among white youth.

On the other hand, other studies find evidence that Hispanic youth are less likely to have academically productive and encouraging interactions with school personnel due to several factors, including language and cultural barriers, as well as a more subtle form of racism or patronization that comes with the damning of faint expectations (Ream 2003; Stanton-Salazar and Dornbusch 1995). For example, Gonzalez et al. (2003) found that while many Hispanic female students report having caring relationships with teachers, nevertheless these relationships did not focus on academic development or college opportunities. The capacity for social relationships and networks to function as social capital for college is necessarily dependent not only on the existence of personal ties and access, but also on the quality of information and support that is passed along (Gibson et al. 2004). Thus, while close ties to peers and parents have the potential to serve as social capital for Hispanic students, this potential may be severely curtailed by such individuals’ lack of college knowledge (Perez and McDonough 2008; Turley 2006). In other words, Hispanic youth may be less able than their white peers to convert some interpersonal connections and relationships into the social capital that will benefit their future academic pursuits.

In short, we know little about whether and how the social and academic factors underlying the postsecondary female advantage differ for Hispanic and white students. This study therefore represents an important foray into research on access to higher education as it rejects the notion that there is a singular gendered path to college, and instead recognizes the diversity of potential paths that lead to the presence of more women than men on contemporary college campuses.

Data

This study uses data from the Texas Higher Education Opportunity Project (THEOP). THEOP data is based on a stratified random sample of Texas public schools in the spring of 2002. After eliminating ineligible schools (schools with a senior class of less than ten students, charter schools, or special education schools), the sample frame contained 1,258 eligible public schools, of which 108 were selected from eleven principal strata based on metropolitan status, racial/ethnic composition, and size. THEOP achieved a school participation rate of 93.3% (n = 97 schools). The sample design for Wave 1 entailed a census of all students in participating schools, resulting in a total of 13,803 student surveys that were completed by seniors during the academic year 2001–2002 (mostly in-class but a small number were done by mail). This represents approximately a 78% response rate. In 2004, THEOP conducted a longitudinal follow-up study of a sub-sample of students, designed to be approximately half the size of the original sample. With the exception of African American and Asian students, who were over-sampled, the Wave 2 longitudinal sample was drawn by a proportionate sample design. Approximately 70% of those 2002 seniors selected for the Wave 2 sub-sample in 2004 were successfully contacted and completed full surveys (n = 5836). The main survey instrument was a computer-assisted-telephone interviewing (CATI) questionnaire. For more information on the THEOP data, see http://www.theop.princeton.edu).

As mentioned previously, the explicit focus of the paper is the examination of the female postsecondary advantage among Hispanic students in comparison to their white majority peers. Therefore, I restrict the sample to white and Hispanic students in the longitudinal sample who graduated from high school (N = 3641). These two groups comprise over 70% of the full sample, which is roughly consistent with the racial/ethnic distribution of the state of Texas, where white and Hispanic youth together account for well over 80% of the public school students (Texas Higher Education Coordinating Board 2004). In exploratory analyses I did consider trends among African American students in the sample. Although a higher number of African American females vs. males did attend college, the difference is almost 50% smaller than that observed and discussed below for both white and Hispanic youth and was also not statistically significant (p = 0.21), thereby making an examination of the factors behind the female postsecondary advantage unwarranted for this group. Thus while the THEOP data is not well-suited to an in-depth analyses of gendered college trends for all racial/ethnic groups, it is arguably the best-suited current educational dataset with which to compare the matriculation patterns of white and Hispanic youth, a topic that to date has not been adequately addressed in the literature.

Dependent Variable

The outcome of interest in the following analyses measures students’ postsecondary matriculation in the year following high school graduation, self-reported from students’ responses in Wave 2. The measure has the following three categories: 0 ‘no college’; 1 ‘2-year college’; 2 ‘4-year college’.Footnote 3 It is important to note that Hispanic students are over-represented among 2-year college matriculates (Kurlaender 2006), so this represents a likely college trajectory for many students in Texas.

Independent Variables

I consider the roles of academic performance and orientation, as well as social capital in accounting for the female advantage in postsecondary attendance. All of these factors are measured by students’ responses to survey questions in the Wave 1 senior year survey. Academic performance for college is measured by three variables. The first variable is the average of students’ self-reported grades on the typical four point scale (where 4 represents an ‘A’ average) in the following subjects: math, science, English and history. The second variable is a dichotomous indicator of whether or not students had completed or were currently taking college-preparatory courses in both math (Algebra II) and science (Chemistry), as these course subjects are well-known predictors of college matriculation (Adelman 1994). The final variable is also dichotomous and measures whether students have ever taken or are currently taking an Advanced Placement (AP) or International Baccalaureate (IB) class. Although the availability of such courses has increased dramatically in the past decade (US Department of Education, National Center for Education Statistics 2007), it appears that 12 schools in the sample did not offer such courses, and a dummy variable was therefore added to the analyses to control for availability (not shown in tables). Additionally, I imputed values for the AP/IB course-taking variable for those students attending schools where it was not offered.

Academic orientation is measured by two variables. The first is a three-category ordinal variable indicating whether students feel that grades are very important, where 1 = ‘disagree’, 2 = ‘agree’, and 3 = ‘strongly agree’. The second variable measures the amount of time spent on homework outside of school each day, with the following values: 1 = 0 h, 2 = less than 1 h, 3 = 1–2 h, and 4 = 3 or more hours.

Social capital is measured by three indicators of the interpersonal relationships that students have which may facilitate their successful matriculation to college. The measures capture three different potential sources of social capital: friends, parents, and counselors. The friendship measure summarizes the academic focus of students’ friends. I created it by summing three separate indicators of whether or not students reported having three or more friends who did the following: do well in school, work hard in school, and plan to go to college. A ‘yes’ answer to each single question was coded as 1 while a ‘no’ was coded as 0. These three questions had an alpha reliability score of 0.73. For parents, a measure that summarized the closeness of the parent–child relationship was created by averaging students’ responses to the following four questions: my parents accept me as I am; I like to get my parents point of view; my parents consider my point of view; I tell my parents about problems. Response categories ranged from 1 = strongly disagree to 4 = strongly agree. These four items had an alpha score of 0.83. The social capital potential of interactions with high school counselors was captured by the students’ average across the following five items: frequency of discussing educational plans with counselors in the last year; frequency of discussions about choosing a college; frequency of discussing college applications; frequency of discussing financial aid; and whether counselors encouraged the student to go to college. Response categories on all items were: 1 = never, 2 = once, 3 = twice, and 4 = 3 or more times. These five items have an alpha score of 0.83.

Students’ gender is coded as 1 for female and 0 for male. All models include a set of background variables to capture students’ family and school characteristics. Parents’ level of education is coded as the highest level attained by either the mother or father with a series of dichotomous variables. Specifically, I include separate variables to distinguish between those who did not graduate from high school, those who received a high school degree (reference category), those who attended some college but did not graduate, those who attained a 2-year college degree, those who attained a 4-year college degree, and those who received a professional or advanced degree. Occupational prestige is also coded as the higher level of either the mother’s or father’s score based on their US Census occupational category and ranges from 0 to 100. Students’ immigrant generational status is coded 1 (not born in the US) or 0 (born in the US).Footnote 4

Measures of school context are also included, taken from the school administrative data that THEOP collected. The first measures the proportion of the student body comprised of minority students. The second is the percentage of the student body that is considered economically disadvantaged, defined by the Texas Education Agency as being eligible for free or reduced lunch, or living with a family whose annual income is at or below the official poverty line or who is eligible to receive food stamps. Both of these measures are included to capture differences in the school context, particularly between white and Hispanic students, which may have implications for students’ postsecondary opportunities.

Design and Analyses

All analyses utilize the sample weights constructed by THEOP to adjust for unequal probabilities of selection and sample attrition. Additionally, as the THEOP data has a multi-stage survey design where schools and then students were selected into the sample, the standard errors of all estimates are adjusted to take into account the correlations between individuals within the same cluster (i.e., schools). Missing values on independent variables were imputed, with the exception of race and gender. After discussing descriptive statistics of key variables, the results of multinomial logistic regression analyses predicting college attendance (either 4-year or 2-year attendance vsersu no college attendance) are presented. To facilitate interpretation of results, I present odds ratios, which indicate the change in the odds of college attendance that is associated with a unit change in a given independent variable. Odds ratios greater than one indicate an increase in the likelihood of college attendance, while ratios less than one indicate a decrease. Separate analyses were conducted for Hispanic and white students. This provides the opportunity to see how accounting for academic performance, orientation, and social capital decreases the female matriculation advantage for each ethnic group, as indicated by the reduction of the gender odds ratio that occurs with the addition of each set of indicators.

Results

Descriptive Results

Table 1 displays the proportion of Hispanic and white students of each gender who attended 4-year college, 2-year college, or no college. Among Hispanic students, the female advantage in 4-year college going is small (31–29%), while the disparity for 2-year college attendance is approximately 8% (39% females versus 31% males). Among white students, 54% of women go to 4-year colleges compared to 44% of men. A similar proportion of white students of both genders attend 2 year colleges (32 and 31%). Only 14% of white females do not attend college of any type, compared to 25% of white men. A much higher proportion of all Hispanic students do not attend any college in comparison to white students, but Hispanic females have lower rates of non-attendance than their male peers (29% of Hispanic women versus 40% of Hispanic men attend no college).

Table 1 Descriptive statistics (means with SD in parentheses)
Full size table

Table 1 also displays means for all independent variables subsequently included in the multivariate analyses. Beginning with background characteristics, it is clear that in general, Hispanic youth come from less socio-economically privileged homes. For example, while almost one quarter of white students in the sample had a parent with an advanced degree, only 10% of Hispanic males and 6% of Hispanic females reported a similarly high level of parental education. Conversely, 28% of Hispanic students indicated that the highest level of education among their parents was less than a high school degree; the percentage reported by whites was less than 5%. Strong patterns of school segregation in Texas are also apparent. Hispanic students attend schools that are approximately 70% minority, compared to white students who attend schools that are only 30% minority.

In considering differences in academic performance, both Hispanic and white girls report significantly higher grades than their male counterparts. Hispanic girls are significantly more likely than their male peers to have taken both Algebra 2 and Chemistry. A significant gender difference is also apparent in the proportion of both Hispanic and white youth that report taking an AP/IB course. Similarly, for the academic orientation indicators, both white and Hispanic girls report spending significantly more time out of school doing homework each day compared to their male peers, and place a higher value on receiving good grades.

With regard to the social capital indicators, both Hispanic and white girls report having friendships that are significantly more academic in their emphasis than those of their male peers. Additionally, girls in both groups report engaging in significantly more dialogue about college with their counselors. Finally, compared to Hispanic and white boys, girls in each group report having significantly closer relationships with their parents.

Examining the Female Postsecondary Advantage

In sum, the descriptive statistics displayed in Table 1 reveal gender differences in academic performance and orientation, as well as social capital. Next, I turn to a series of multinomial logistic regression analyses to examine how accounting for these factors reduces the gender gap in college attendance for Hispanic youth and for white youth. I begin with a baseline model (model 1) that includes only student and school background characteristics. In the next three models (models 2–4) I add separately the indicators of academic preparation, orientation, and social capital. Finally, the last model (model 5) includes all indicators together, and allows us to see, for example, whether or not the estimated effects of social capital are independent of academic performance. In all cases, the primary focus is whether and to what the degree the introduction of each set of indicators reduces the female-to-male odds ratio representing females’ relative advantage in matriculation for each ethnic group. I begin with the analyses of Hispanic students, followed by the analyses of white students.

The Gender Gap Among Hispanic Students

Beginning with model 1, net of family and school characteristics, there is a significant female advantage in postsecondary attendance, such that Hispanic females are 1.7 times more likely (or 70% more likely) to attend a 4-year college versus no college compared to Hispanic males, and almost two times more likely to attend a 2-year college versus no college (Table 2). Also apparent in model 1 is the positive and significant association between parental education level and matriculation. Additionally, first-generation students are approximately 50% less likely to go to either a 4-year (OR = .464) or 2-year college (OR = .556) compared to their non-immigrant Hispanic peers.

Table 2 Multinomial logistic regression analyses predicting college attendance for Hispanic students (odds ratios with standard errors in parentheses)
Full size table

Model 2 adds indicators of academic performance and shows that Hispanic students with higher GPA’s have a significantly higher likelihood of going to either a 4-year or 2-year college (vs. no college). Both course-taking indicators significantly predict only 4-year college attendance. For example, students in AP/IB courses are slightly over four times more likely to attend a 4-year college. Importantly, with the inclusion of these measures of academic performance, Hispanic females’ advantage in 4-year matriculation is no longer statistically significant; the female-to-male odds ratio decreases to just slightly over 1 and the corresponding gender coefficient (not shown) is reduced by almost 75%. For 2-year college attendance, the female advantage among Hispanic students is reduced only slightly and remains significant.

In model 3, variables measuring students’ academic orientation are introduced. While placing a high value on grades is not a significant predictor of matriculation, the number of hours spent daily on homework outside of school is significantly associated with an increase in the odds of Hispanic students’ college attendance at both types of college. Further, the inclusion of this indicator somewhat attenuates the female advantage in both types of attendance, although noticeably less so than for academic performance.

The inclusion of social capital indicators in model 4 also decreases the Hispanic female advantage in matriculation from baseline models. Two measures of social capital are significant predictors of both 4-year and 2-year attendance: having more academically-focused friends and engaging in more interactions with counselors regarding college-related issues. Results in model 4 indicate that close relationships with parents do not significantly increase the likelihood of college attendance.

Finally, in Model 5, all academic performance, orientation, and social capital indicators are included in the model. In this final model, the collective contribution of these indicators leads to the largest reduction in the female-to-male odds ratio from the baseline models for both 4- and 2-year colleges. Further, the social capital measure of friends’ academic focus remains a significant predictor of 2-year matriculation, while engaging in discussions with counselors about college continues to be associated with an increase in the likelihood that students attend both types of college, independent of their academic performance and orientation.

To better understand the potential for counselors to influence students’ chances of matriculation, I calculated the predicted probabilities of 4-year college matriculation for a Hispanic student who is average on all other indicators in the model, varying only the student’s frequency of interactions with counselors. A Hispanic student who on average reported never discussing college plans, selection, or financial aid with a counselor has a predicted probability of .23 for attending a 4-year college. In contrast, a Hispanic student who reported on average discussing such issues once with a counselor has a predicted probability of 0.33, and a student who reported having on average two discussions has a probability of 0.45. Finally, a student who reported 3 or more discussions with a counselor has a predicted probability of 0.57 for attending a 4-year college.

The Gender Gap Among White Students

As seen in model 1, net of family and school background characteristics, white females are over two and a half times more likely to attend 4-year college rather than no college compared to their male peers (Table 3). White females are also nearly two times as likely to attend 2-year college (vs. no college) than their male peers. Among the background characteristics included, only parental education level is significantly associated with the odds of college attendance for both 2- and 4-year matriculation. For example, a student with at least one parent who graduated from a 4-year college is almost 12 times more likely to attend a 4-year college rather than no college, compared to a student whose parents only graduated from high school.

Table 3 Multinomial logistic regression analyses predicting college attendance for white students (odds ratios with standard errors in parentheses)
Full size table

Model 2 introduces indicators of academic performance. Similar to the results for Hispanic students shown in Table 2, grade point average is a significant predictor of both types of college attendance, while course-taking only significantly increases the odds of attending a 4-year college. For example, white students who take both Algebra 2 and Chemistry are almost five times more likely to attend a 4-year college than their peers who did not take both college preparatory courses. After including these measures of academic performance, the odds ratio representing the female advantage in 4-year college attendance is reduced substantially from model 1, so that white females are now just over two times more likely to attend 4-year college versus none. However, unlike the analyses of Hispanic students, even after accounting for differences in academic performance, white female students remain at a much higher likelihood of attending a 4-year college than their male peers. Yet similar to the results in Table 2, there is less of a reduction in the female-to-male odds ratio in 2-year college attendance.

Model 3 adds measures of academic orientation to the baseline model. Time spent on homework does not significantly predict any matriculation, in contrast to the results for Hispanic students in Table 2. Yet among white students, valuing good grades is associated with a significant increase in the odds of attendance at both types of college. Subsequent statistical tests confirm that this association is significantly different than that observed for Hispanic students.Footnote 5 Further, it is interesting to note that among white students, the reduction in the female-to-male odds ratios from the baseline models predicting 4- and 2-year attendance is somewhat greater in model 3 with the addition of academic orientation variables than that observed in model 2 when academic performance indicators were included.

In model 4, measures of social capital are introduced. Similar to the results for Hispanic students, having friends who are academically-focused is associated with a significant increase in the likelihood of college attendance for white students, but only for 4-year colleges. However, the measure for discussions with high school counselors does not significantly predict matriculation at either type of college for white students, and subsequent tests confirm that the coefficients for counselor interactions are significantly different for Hispanic students than for white students. The measure of closeness with parents is not significantly associated with the likelihood of matriculation for white students. Thus, having academically-inclined friends is the only significant source of social capital for 4-year matriculation among white students. In this model, the female-to-male odds ratio is reduced to 2.2; although this is not as large a reduction as that observed in previous models with the introduction of measures of academic performance and orientation, it represents a decline in the gender coefficient (not shown) of 16%. Again, we see that including measures of social capital leads to a smaller reduction of females’ greater likelihood of attending a 2-year college than it does for 4-year college.

Finally, in the full model (5) that includes measures of performance, orientation, and social capital, we see the greatest reduction in the female advantage in 4-year college matriculation. Accounting for all of these factors collectively reduces the female-to-male odds ratio from the baseline model for 4-year matriculation to only 1.6, which translates to a reduction in the corresponding gender coefficient of almost one-half. Valuing good grades and social capital in the form of friends’ academic focus remain significantly associated with 4-year matriculation, despite the strong predictive power of grades and advanced course-taking. For example, converting the odds ratios to percents reveals that a one unit increase in the level of friends’ academic focus corresponds to an almost 50% increase in the likelihood of 4-year college attendance. In contrast, no measures of social capital or academic orientation significantly predict 2-year matriculation in the final model, and the odds ratio measuring the female advantage is not reduced much in magnitude compared to the baseline model.

Discussion and Conclusion

While there is a growing research literature on the recent trend of greater relative female rates of college matriculation, there is a clear need for more research that explores racial/ethnic variation in such patterns, particularly with regard to the growing Hispanic student population. Subsequently, the goal of this study was to address this gap in the literature by considering whether the factors underlying the female college advantage might differ for Hispanic and white students, using newly available longitudinal data from a representative sample of high school seniors in the state of Texas. Additionally, this study argued for the need to consider girls’ potentially greater access to the kinds of social capital that facilitate the transition to college as a key factor in the female postsecondary advantage beyond the more typically studied factors of academic performance and related attitudes and behaviors. The results indicate that while there are clearly common factors that underlie the female college-going advantage among both white and Hispanic groups, there are nevertheless important differences which underscore the importance of moving beyond a one-size-fits-all model.

First, while gender differences in academic performance in high school play a strong role in explaining the female postsecondary advantage among both white girls and Hispanic girls, particularly for 4-year college attendance, it appears to be more important in explaining the gender gap among Hispanics. As Hispanic male youth exhibit the lowest levels of academic performance in the analytic sample, the results serve as a reminder that minority males are the group most likely to exit high school without having acquired the skills and prerequisites necessary for college. Additionally, accounting for students’ academic orientation leads to a reduction in the female advantage in matriculation for both ethnic groups. Yet while an emphasis on good grades is a significant predictor of college-going only for whites, in contrast, time spent on homework is associated with a significant increase in college-going for Hispanic youth, perhaps suggesting the important role that effort can play in the face of other educational obstacles.

Further, as hypothesized, the results indicate that both Hispanic and white high school girls possess greater levels of certain kinds of social capital than their co-ethnic male peers, and this appears to contribute to their relatively higher likelihood of college matriculation. These results speak to the need to move beyond a focus on the grades earned by individuals, for example, to consider how girls may occupy a more advantageous position with regard to social relationships and networks within schools that has important implications for their future educational trajectories. For example, compared to their co-ethnic male peers, both Hispanic and white girls reported having friendship groups that are more academically-focused, a factor that is significantly associated with higher odds of college matriculation. Such findings stand in somewhat dramatic contrast to research on women in the labor force, which finds that females have less access to the kinds of social networks that increase salaries, and promotions, for example (Burt 1998; Lin 2000).

Yet the analyses also reveal an important distinction in the kinds of social capital that may contribute to explaining the female postsecondary advantage among Hispanic youth and white youth. Specifically, social capital in the form of interactions with high school counselors is a significant predictor of matriculation unique to Hispanic students. When considering the fact that many Hispanic students are first-generation college-goers and that many Hispanic parents have low levels of college knowledge, this indicates that counselors represent a critical source of relevant information and guidance to postsecondary access for which these students have few substitutes (Gonzalez et al. 2003; Tornatzky et al. 2002). For white students, counselors may be a source of information and encouragement that is somewhat redundant with the support they receive from parents, peers, and others. Unfortunately, recent research suggests that it is exactly those students who are the most in need of access to counselors that are the ones least likely to have it. The average counselor-student ratio is highest in low-income, high-minority schools, and conversely lowest in the most advantaged schools (Perez and McDonough 2008; Perna et al. 2008). This study therefore offers additional evidence of the need for improved access of Hispanic youth to high school counselors.

However, despite this limited access, the results presented here show that Hispanic girls somehow tap into this essential resource more so than their male peers, and this appears to contribute to their higher relative likelihood of college-going. Relevant questions such as whether Hispanic girls are more likely to seek out counselors because they are already interested in college, whether counselors seek out these girls because they are better students than their male peers, or whether Hispanic males interact with counselors at similar rates as their female peers but about issues other than college, are beyond the scope of this paper. Yet understanding how school personnel can provide crucial social capital for Hispanic students’ educational success, and why and how girls have comparatively more access to such capital, is an important topic that needs more research.

It is also worth noting that among 2-year colleges, the college sector where Hispanic students are over-represented compared to other groups (Kurlaender 2006), social capital, academic performance, and orientation indicators did little to explain females’ higher rates of attendance. A similar pattern was also observed for white students. The analyses presented here therefore underscore the complexity and diversity of the gendered patterns and predictors of college attendance, and suggest the need for more research on the potentially unique factors that may influence the paths to 2-year colleges. For example, recent work by Turley (2009) suggests that Hispanic youth’s greater proximity to 2-year colleges rather than 4-year colleges in Texas and elsewhere is an important factor to consider when exploring differences in college trajectories.

As with any analyses, this study has several limitations worth noting. First, as the data are from students in Texas, the results cannot necessarily be considered representative of patterns nationwide. For instance, Texas has one of the highest high school drop-out rates in the country (Day and Jamieson 2003), and therefore the college-going population considered here may be a more select group than in other states. Furthermore, another limitation of the study is its reliance on self-reported indicators of students’ academic performance, which may introduce some upward bias. However, it is not clear whether that would bias the results in favor of one group or another.

Finally, while this study’s focus on social capital as an important factor behind the female postsecondary advantage provides important new insight into these gendered trends, the measures included here are somewhat limited. For example, the survey included far fewer questions querying students about their interactions with teachers than with counselors. More detailed information about the kinds of activities and discussions that students have with parents, peers, and school personnel, as well as questions that tap students’ impetus for talking with counselors and measures of parents’ own college knowledge, would also allow researchers to gain a clearer picture of the causal link between social capital and college matriculation. For example, students who are already bound for college may seek out similarly-oriented friends and initiate conversations with counselors. By controlling on individual measures of grades, course-taking, and academic orientation, as well as background, the potential for selection bias to influence results should be reduced. Yet as with any study of friendships or relationships, it is not possible to say with certainty that such involvements actually cause a particular outcome, such as college-going (Schneider et al. 2007).

In conclusion, it is important to point out that while both Hispanic and white girls enjoy an advantage in matriculation rates compared to their male peers, nevertheless, racial/ethnic disparities in college attendance consistently trump any gender disparities (Knapp et al. 2006). Specifically, minority youth remain far under-represented in college campuses nationwide. When taking these two patterns together, the position of disadvantage occupied by minority males is brought into sharp relief. In the specific case of Hispanic males, their low matriculation rates ultimately foreshadow reduced employment and economic prospects over the life course. There is clearly the need for more research that considers the factors that impede or alternatively facilitate the academic pathways of minority male youth in particular, as well as more consideration of the complex processes through which gender and race/ethnicity intersect in shaping individuals’ paths to college and beyond.