Maintaining and Attaining Educational Expectations: A Two-Cohort Longitudinal Study of Hispanic Youth

Abstract

Hispanic adolescents have high expectations for their own educational attainment, but educational attainment outcomes for Hispanic young adults are relatively low on average. Limited scholarship links Hispanic adolescents’ educational expectations with attainment in adulthood. Using a longitudinal within-group approach, the authors examined changes in Hispanic adolescents’ educational expectations from 10^th to 12^th grade as well as the discrepancy between educational expectations and later attainment in two cohorts (N = 1,372 and N = 1,521). Based on national surveys approximately 12 years apart (NELS:88 and ELS:02), expectations of a college degree became more common for Hispanic adolescents. Most Hispanic young adults attained less education than they expected, and this expectations-attainment gap grew over time, consistent with U.S. national trends. Parent education was positively associated with attaining expectations, and family income became more important for attaining expectations over time. Consistent with the immigrant paradox and immigrant optimism hypothesis, second-generation youth came closer to attaining their expectations than third- or higher-generation youth of equal positioning. Educational expectations and attainment patterns also varied by cohort: for instance, immigrants in the more recent cohort had an absolute disadvantage in terms of attaining their expectations compared to their other-generation counterparts. Finally, parental involvement in education (e.g. parents’ educational aspirations) was associated with educational expectations and attainment outcomes. This study advances knowledge of factors that contribute to maintaining and attaining educational expectations for Hispanic youth and how these factors have changed over time. Study results inform efforts to support Hispanic adolescents’ and young adults’ educational potential.

Educational disparities exist between Hispanic¹ youth and youth of other ethnicities. The high school dropout rate among Hispanic students is considerably higher than that of their non-Hispanic counterparts (McFarland et al., 2017), and Hispanic young adults earn bachelor’s degrees at roughly half the rate of other young adults (18.7% and 36.1% respectively; U.S. Department of Education, 2016a). Although college enrollment is rising among Hispanic high school graduates, this trend is partially accounted for by disproportionately high enrollment at two-year rather than four-year institutions (Fry, 2002; Ma & Baum, 2016). These disparities, coupled with the reality that Hispanics are one of the fastest-growing U.S. populations, evidence the need for a better understanding of factors that shape Hispanic youth’s educational outcomes to ensure the continued development of human capital in the U.S. (Flores, 2017).

Hispanic youth’s educational expectations (i.e. the level of education they expect to attain) are higher than or comparable to those of non-Hispanic youth (Mello, 2009). Most Hispanic young adults (89%) and parents (88%) say that earning a college degree is important for success (Lopez, 2009; Stepler, 2016). High educational expectations, however, do not always translate to later educational attainment (Oyserman & Lewis, 2017). Given that Hispanic youth have high early expectations yet relatively low attainment outcomes on average, we need a better understanding of factors that promote fulfilment of educational expectations for these youth.

The current study examined change in Hispanic youth’s educational expectations from 10^th to 12^th grade and the discrepancy between expectations and actual attainment 10 years later. Social position factors (e.g. immigrant generation status), family characteristics (e.g. parental involvement in education), and individual characteristics (e.g. academic achievement) were examined as predictors of these outcomes in two large cohorts of Hispanic youth. Utilizing data from two national surveys (National Education Longitudinal Survey of 1988 – NELS:88 and Education Longitudinal Survey of 2002 – ELS:02), this study also includes historical cohort comparisons. Educational attainment is not the only means of achieving one’s goals, and college is not necessarily a suitable path for every student. Thus, this study approached educational outcomes not by comparing students’ expectations and attainment to those of others’, but by examining changes in and attainment of students’ own educational expectations.

Educational Expectations and Attainment: Theoretical Frameworks and Change

Educational expectations have been characterized as “a student’s best estimate of attainment using available information” (Jacob & Wilder, 2010, p. 17). Expectations are typically distinguished from aspirations (Bohon, Johnson, & Gorman, 2006; Reynolds & Pemberton, 2001). Whereas aspirations refer to the level of education that a student hopes or desires to attain, expectations are more realistic appraisals of attainment (Hanson, 1994; Park, Wells, & Bills, 2015). According to the status attainment model (Sewell, Haller, & Portes, 1969), youth’s expectations of how far they will go in school are shaped by socioeconomic status (SES) and parents’ beliefs. Academic achievement is also highlighted as a source of influence on expectations, as this information shapes beliefs about realistic educational possibilities.

Expectations are implicated in agentic behaviors, including educational persistence and performance (Eccles et al., 1983; Messersmith & Schulenberg, 2008). Expectancy-value theory of achievement motivation states that an individual’s expectations for success and the degree to which they value a given activity are factors in determining persistence and performance in that activity (Wigfield & Eccles, 2000). Decades of research have shown educational expectations to be one of the most important predictors of later educational attainment among diverse populations (Driscoll, 1999; Mello, Anton-Stang, Monaghan, Roberts, & Worrell, 2012). Changes in expectations across development are also indicative of attainment potential, particularly during the critical period of 10^th to 12^th grade (Hanson, 1994). Earlier in adolescence, expectations may be based on hopes and ambitions, whereas expectations later in adolescence are informed by concrete possibilities (Kao & Tienda, 1998). Expectations can even reflect plans that are already in place (Messersmith & Schulenberg, 2008). Between 10^th and 12^th grade, students may prepare for post-high school plans (e.g., taking admissions tests, applying to college and for financial aid). Because expectations in 12^th grade are thought to better reflect postsecondary plans, expectations that decline from 10^th to 12^th grade may be indicative of lost potential (Hanson, 1994). Research shows that youth whose expectations start high and remain high have a greater likelihood of postsecondary enrollment than youth with low or decreasing expectations over high school (Bozick, Alexander, Entwisle, Dauber, & Kerr, 2010).

Educational expectations have also changed historically across cohorts. Educational expectations in the U.S. rose steadily from the 1970s into the 2000s, as expectations of a college degree became increasingly common (Goyette, 2008; Reynolds, Stewart, MacDonald, & Sischo, 2006). One explanation for this shift is that the earnings gap between those who held a bachelor’s degree and those who did not widened steadily during this time and still exists today (Snyder, de Brey, & Dillow, 2018; Wirt et al., 2003). As postsecondary degrees have been reinforced by the labor market, students have been compelled to attain higher levels of education (Fry, 2002). Changing demographics may also contribute to rising expectations, as students tend to expect a level of education that will maintain a social positioning that is at least as high as their parents’ (Goyette, 2008; Park et al., 2015). Also, students’ expectations have become more loosely tied to their academic and socioeconomic background, suggesting that in more recent cohorts, pursuing postsecondary education is viewed more as a typical developmental progression for young adults (Goyette, 2008). As educational expectations have risen over time, the proportion of U.S. students who expect a college or graduate degree have become far greater than the proportion of the population who actually attain these levels (Reynolds et al., 2006). In other words, even though educational attainment has risen overall, expectations have risen faster such that they outpaced attainment (Reynolds et al., 2006).

Consistent with trends in the general population, the educational expectations of Hispanic youth have risen historically. National data from NELS:88 and ELS:02 show that almost half of Hispanic high school sophomores in 1990 expected a college degree or higher, and by 2002, approximately two thirds expected at least a college degree (Ingels, Burns, Chen, Cataldi, & Charleston, 2005; Ingels, Schneider, Scott, & Plank, 1994). Educational attainment of Hispanic students has also risen, with high school dropout rates decreasing and college enrollment rates increasing from the 1980s and into the 2000s (Gramlich, 2017; McFarland et al., 2017; Tienda & Mitchell, 2006). However, empirical work linking Hispanic adolescents’ educational expectations to their attainment in adulthood is quite limited (Mello, 2009; Roche et al., 2017). Further, few studies have examined factors that promote fulfillment of Hispanic students’ educational expectations. Understanding shifts over time in the factors that promote educational outcomes for Hispanic students is important for informing educational policy and practice.

Social Position and Educational Outcomes for Hispanic Youth

Cultural-ecological models suggest that social position is fundamental to understanding developmental competencies (García Coll et al., 1996). For Hispanic youth, generation status is an important social position factor in the study of educational outcomes (Marks, Ejesi, & García Coll, 2014). Theoretical and empirical work in this area have been guided by the immigrant optimism hypothesis, which states that children in immigrant families (i.e. first- and second-generation youth) benefit from their parents’ optimism about their social, economic, and educational opportunities (Kao & Tienda, 1995). However, because first-generation youth may be more likely to face challenges such as limited English proficiency and low SES, the immigrant optimism hypothesis posits that second-generation youth are best positioned to excel academically (Kao & Tienda, 1995; Kaufman, Chavez, & Lauen, 1998). Another phenomenon is the immigrant paradox, in which later generations are observed to have poorer outcomes than more recently-immigrated generations, even though later-generation youth live with families who have higher income, more highly educated parents, and greater English proficiency on average (Kao, 2004; Rodríguez, 2002; Tienda & Mitchell, 2006). These two hypotheses are similar in that they both suggest second-generation youth will fare better than their third- or higher-generation² counterparts, but only the immigrant optimism hypothesis suggests that second-generation youth are better equipped educationally than first-generation youth.

Empirical work on educational expectations shows mixed findings on the role of generation status. Among Hispanic 8^th graders in a national survey (NELS:88), first- and second-generation youth were more likely to expect a college degree than third-generation youth, consistent with the immigrant optimism hypothesis and the immigrant paradox (Kao & Tienda, 1995). However, other studies show no generational differences in Hispanic adolescents’ educational expectations (Hernández, Robins, Widaman, & Conger, 2016; Ibañez, Kuperminc, Jurkovic, & Perilla, 2004). Mixed findings for generation status may be due in part to the developmental period in which educational expectations were assessed. Studies have shown that the expectations of first-generation Hispanic youth are relatively less stable, such that they start high and dip in middle adolescence (i.e. 10^th grade) before rising again in late adolescence, whereas their native-born counterparts’ expectations are more stable (Kaufman et al., 1998; Perez-Brena, Delgado, Rodríguez De Jesus, Updegraff, & Umaña-Taylor, 2017). Given historical changes in educational expectations, historical period may also contribute to inconsistent findings (Hernández et al., 2016). These issues highlight the need for longitudinal, multi-cohort studies that assess generational differences in expectations.

Hispanic youth’s educational attainment is generally consistent with the immigrant optimism hypothesis. Second-generation Hispanic students are more likely to persist in high school, enroll in four-year institutions, and go on to graduate school than their other-generation counterparts (Driscoll, 1999; Fry, 2002; Hagy & Staniec, 2002). Conversely, first-generation Hispanic students generally fare worse; in 2013, 73% of first-generation Hispanic young adults had a high school credential compared to approximately 90% of their native-born counterparts (Musu-Gillette et al., 2016). First-generation Hispanic young adults are also less likely to enroll in and complete college (Fry, 2002; Lopez, 2009). As a caveat, the association of generation status with attainment depends on what other factors are accounted for statistically. For example, accounting for SES and language use, first-generation Hispanic youth are no less likely to enroll in postsecondary education (Jasinski, 2000). Other studies have found a third-generation disadvantage in enrollment after accounting for multiple factors, consistent with the immigrant paradox (Hagy & Staniec, 2002; Perreira, Harris, & Lee, 2006). As such, this study examines generation status differences in outcomes with and without the effects of other factors.

In addition to immigrant generation status, another relevant social position factor is sex. Extant research has shown that male adolescents have higher educational expectations (Hanson, 1994), but more recent studies show that female adolescents have higher expectations. Young men may be more likely to lower their educational expectations and to have unrealized expectations compared to young women, but these differences are explained in part by academic achievement (Feliciano & Rumbaut, 2005; Hanson, 1994). Traditional gender roles at home may promote girls’ academic achievement (Feliciano & Rumbaut, 2005). Consistent with gender patterns in other racial/ethnic groups, Hispanic female students enroll in college at higher rates than their male counterparts (Feliciano & Rumbaut, 2005; Fry, 2002). Although sex differences are not a central focus, the current study accounts for the effects of sex on the outcomes.

Socioeconomic Predictors of Educational Expectations and Attainment

The effects of SES on educational outcomes are well-documented (Dubow, Boxer, & Huesmann, 2009; Reynolds et al., 2006). Consistent with the status-attainment model, SES is positively linked with adolescents’ educational expectations in national and Hispanic samples (Hernández et al., 2016; Jacob & Wilder, 2010; Kao & Tienda, 1998). SES is also associated with change in expectations among diverse adolescents: higher SES predicts maintaining or increasing expectations (Mello, 2009; Park et al., 2015), while lower SES predicts lowered expectations (Hanson, 1994; Trusty & Harris, 1999). However, one study found that SES was not associated with changes in expectations for Mexican-origin adolescents (Perez-Brena et al., 2017). SES also predicts postsecondary enrollment and attainment outcomes for diverse and Hispanic young adults (Fuligni & Witkow, 2004; Jasinski, 2000; Roche et al., 2017). Over time, SES became a stronger predictor of educational attainment based on multiple nationally representative studies (Reynolds et al., 2006). However, whether socioeconomic factors have become a more important predictor of attainment outcomes for Hispanic students is unclear.

Existing work on expectations and attainment has often considered SES a composite measure, including parent education and family income (e.g. Mello, 2009; Perez-Brena et al., 2017). However, parent education and family income may have differing effects on educational outcomes. Parent education has a slightly greater effect than family income on changes in expectations for diverse adolescents (Kao & Tienda, 1998; Park et al., 2015). Parent education is also likely important for attaining expectations, as parents with higher education may be better equipped to help youth navigate the details of attaining their expectations. National surveys show that youth are more likely to attend college if their parents attended at least some college, and students whose parents have a bachelor’s degree are more likely to persist in college (Goyette, 2008; Nunez, Cuccaro-Alamin, & Carroll, 1998). Family income may also play a role in attaining expectations by way of affording educational opportunities, even though some evidence suggests that its effects are smaller than parent education on attainment (Fuligni & Witkow, 2004; Ward, 2006). The effects of parent education versus family income for Hispanic youth’s expectations and attainment outcomes are still unclear due to the use of composite measures or single indicators of SES. As such, the current study examines the effects of parent education and family income separately. Beyond providing resources and capital, parents also serve as sources of socialization for adolescents (Kao & Tienda, 1998).

Parental Involvement in Education and its Role in Expectations and Attainment

The status attainment model and other theoretical models (i.e. social cognitive models; Bandura, Barbanelli, Caprara, & Pastorelli, 1996) cite parents’ education-related beliefs and behaviors as sources of influence on youth’s educational expectations. Parental involvement in education may be particularly important for Hispanic youth, who may give weight to parents’ educational values given the prominence of familism in Hispanic cultures (Trusty, Plata, & Salazar, 2003). Researchers have theoretically conceptualized parental involvement in education as a multifaceted construct including school-based³ and home-based strategies, as well as academic socialization (Bhargava & Witherspoon, 2015; Hill & Tyson, 2009). However, empirical evidence on different forms of parental involvement and their effects on educational outcomes in Hispanic families is still lacking (Bhargava, Bámaca-Colbert, Witherspoon, Pomerantz, & Robins, 2017). Additionally, long-term effects of parental involvement in education during adolescence are not often examined (Benner, Boyle, & Sadler, 2016). The current study examines longitudinal effects of multiple types of parental involvement.

Home-based involvement (e.g., supervising homework and providing structure to reinforce learning) generally has weaker effects on educational outcomes than other forms of parental involvement (Fan & Chen, 2001; Hill & Tyson, 2009). Limited research has examined the role of home-based involvement for Hispanic youth, particularly as it relates to students’ long-term educational development. One study showed that parents’ home-based involvement was associated with Hispanic youth’s effort in school but not with educational expectations (Ceballo, Maurizi, Suarez, & Aretakis, 2014). As students progress through school, parents may reduce direct home-based involvement in favor of strategies that provide adolescents with more autonomy (Bhargava et al., 2017). Academic socialization includes communicating the utility of education, discussing learning strategies and goals, and making plans for the future with youth (Hill & Tyson, 2009) and is particularly well-suited for the developmental needs of teenagers (Wang & Sheikh-Khalil, 2014). Studies show that academic socialization is positively associated with Hispanic youth’s school engagement and beliefs about the utility of education (Ceballo et al., 2014; Rivas-Drake & Marchand, 2016). A specific form of academic socialization is parents’ aspirations for their youth’s education (i.e., how far in school they would like their child to go; Bhargava & Witherspoon, 2015; Sewell et al., 1993), which are consistently one of the strongest predictors of youth’s educational expectations (Bozick et al., 2010). Youth’s perceptions of their parents’ aspirations contribute to the development and maintenance of their own beliefs (Sewell et al., 1993). Limited work shows that parents’ aspirations are linked with Hispanic adolescents’ educational beliefs (Ceballo et al., 2014; Hernandez, Vargas-Lew, & Martinez, 1994).

Individual Academic Indicators: Achievement and English Proficiency

Based on expectancy value theory and the status attainment model, academic achievement serves as a source of feedback for students that influences their expectations for success in future educational endeavors (Sewell, Haller, & Ohlendorf, 1970; Wigfield & Eccles, 2000). Academic achievement predicts the development of high expectations (Jacob & Wilder, 2000), maintenance of expectations (Bozick et al., 2010), postsecondary enrollment (Perna, 2000), persistence towards degree completion (Fuligni & Witkow, 2004), and eventual educational attainment (Benner et al., 2016). For many Hispanic youth, academic achievement and educational attainment are tied to English proficiency (Lopez, 2009). Limited English proficiency is associated with lower levels of achievement and higher levels of high school dropout among Hispanic youth (Lutz, 2007; Warren, 1996). The current study accounts for the effects of academic achievement and English proficiency on other educational outcomes.

Current Study

The first goal of the current study was to examine predictors of change in educational expectations from 10^th to 12^th grade in two large cohorts of Hispanic youth (NELS:88 and ELS:02). The second goal was to examine predictors of the discrepancy between educational expectations in 10^th grade and educational attainment 10 years later, when most students will have finished their schooling (Lowell & Suro, 2002). Guided by the status attainment model and a cultural-ecological framework, social position factors, family characteristics, and individual characteristics were examined as predictors of these outcomes. Limited research has linked educational expectations in adolescence with ultimate educational attainment for Hispanic students (Mello, 2008), and existing research on educational expectations and attainment is based primarily on older surveys (NELS:88 and older). The current study assesses attainment outcomes in 2012 using one of the most recent national surveys available (ELS:02) in comparison with NELS:88, thereby contributing valuable knowledge on factors that promote Hispanic students’ educational outcomes and how these factors have changed over time, which has implications for educational practices and policies targeting Hispanic students’ educational success.

Study hypotheses are as follows. As expectations have risen faster than attainment (Reynolds et al., 2006), it was hypothesized that educational expectations would be higher in ELS:02 and that the ELS:02 cohort would be further from attaining their expectations than the NELS:88 cohort. Consistent with the immigrant optimism hypothesis (Kao & Tienda, 1995), second-generation status would be positively associated with attaining expectations whereas first-generation status would be negatively associated with attaining expectations, as immigrant youth may be more optimistic despite having lower educational persistence (Fry, 2002; Musu-Gillette et al., 2016). It was hypothesized that being female would be positively associated with change in expectations and attaining expectations, given higher college enrollment among Hispanic female students than male students (Fry, 2002). Consistent with research in the status-attainment tradition (Trusty, 2000), parent education and family income were hypothesized to be positively associated with outcomes. Socioeconomic factors were hypothesized to have a greater effect on attaining expectations in the more recent ELS:02 cohort (Reynolds et al., 2006). Parents’ aspirations and academic socialization were hypothesized to be positively associated with change in expectations and attainment of expectations. Academic achievement and English proficiency were hypothesized to be positively associated with outcomes; change in expectations would be positively associated with attaining expectations.

Method

Data Source and Analytic Samples

The data were drawn from two national surveys conducted by the National Center for Education Statistics (NCES): the National Education Longitudinal Study of 1988 (NELS:88) and the Education Longitudinal Study of 2002 (ELS:02). The similarities in sampling and study designs of NELS:88 and ELS:02 allow for comparisons between the two cohorts (Ingels et al., 2014). This study was approved by The Pennsylvania State University Institutional Review Board (STUDY00009119: Educational Expectations and Attainment).

NELS:88 survey and sample selection

Data collection began in 1988 when students were in 8^th grade. Follow-up surveys were conducted at two-year intervals, and a final follow-up survey was conducted in 2000, when participants were modal age 26. The full procedures for NELS:88, including sampling, surveying, and attrition, have been detailed elsewhere (Curtin, Ingels, Wu, & Heuer, 2002). The selection of the current study sample began with the N = 12,144 participants who responded to the final follow-up survey. Participants were selected if they were members of the 1990 10^th grade cohort (N = 10,982). Participants who identified as Hispanic were selected (12.5%), resulting in the final NELS:88 study sample of N = 1,371. Most of these participants were enrolled in 12^th grade in 1992 (n = 1,164, 84.9%), but participants who had dropped out of school (n = 166, 12.1%) or who were enrolled in a grade other than 12^th (n = 38, 2.8%) at the time of the 1992 survey were still included in the current study. The current study primarily utilized data from the NELS:88 first follow-up survey in 1990, when participants were in 10^th grade. Educational expectations in 12^th grade were obtained from the 1992 survey, and ultimate educational attainment was obtained from the final follow-up survey in 2000. Sociodemographic variables were obtained from parent surveys. Because there was no parent survey in 10^th grade, youth generation status and parents’ highest education were obtained from the base year parent survey when participants were in 8^th grade.⁴ Family income in 10^th grade was retroactively obtained from the 12^th grade parent survey.

ELS:02 survey and sample selection

The surveying schedule for ELS:02 resembled that of NELS:88, minus the 8^th grade survey. Data collection began in 2002 when students were in 10^th grade. Subsequent follow-up surveys were conducted at two-year intervals, and a final follow-up survey occurred in 2012, when the modal age of participants was 26. Complete study procedures are available elsewhere (Ingels et al., 2014). Selection of the current study sample began with the N = 13,250 participants who completed the final follow-up survey. Of these participants, N = 13,133 were members of the 10^th grade cohort. Participants who identified as Hispanic were selected (N = 1,727, 13.2%). The final ELS:02 study sample was N = 1,521, after excluding participants who responded “don’t know” to the educational expectations survey item in 10^th grade (n = 206, 11.93%).⁵ The majority of the sample was enrolled in 12^th grade in 2004 (n = 1,196, 78.6%), but as with the NELS:88 cohort, participants with alternative circumstances at the time of the 12^th grade survey were still included in the current study, including those who dropped out of school (n = 96, 6.3%), graduated early (n = 64, 4.2%), or were enrolled in a grade other than 12^th (n = 51, 3.4%). The current study primarily utilized data from the ELS:02 base year 2002 survey, when participants were in 10^th grade. Educational expectations in 12^th grade were obtained from the first follow-up survey in 2004, and ultimate educational attainment was obtained from the final follow-up survey in 2012. Sociodemographic variables (i.e. family income, parent education, generation status) were drawn from the base year parent survey.

Participant Demographic Information

NELS:88 participant demographics

The average age in 10^th grade was 16.15 years (SD = 0.51). The median level of parents’ highest education was a high school degree but less than a four-year college degree; for 13.4% of participants, parents’ highest education was at least a four-year college degree. The median annual family income was $20,001-$25,000 (compared to a national median of $29,943 in 1990; U.S. Census Bureau, 2018). Participants were offered only four options to describe their ethnicity/nationality of descent in the NELS:88 survey: Mexican/Chicano (n = 798, 58.2%), Puerto Rican (n = 114, 8.3%), Cuban (n = 57, 4.2%), or “Other Hispanic” (n = 262, 19.1%). Ethnicity/nationality of descent was missing for n = 140 (10.2%). Estimates of generation status proportions were 14.3% first generation, 44.2% second generation, and 41.4% third or higher generation (estimation procedures are described in Data Analysis section). Generation status could not be calculated for n = 276 (20.1%).

ELS:02 participant demographics

Participants in the ELS:02 cohort averaged 16.19 years of age in 10^th grade (SD = 0.64). The median level of parents’ highest education was a high school degree but less than a four-year college degree; for 21.5% of participants, parents’ highest level of education was at least a four-year college degree. The median annual family income was $25,001-$35,000 (compared to a national median of $42,409 in 2002; U.S. Census Bureau, 2018). Participants identified as Mexican/Chicano (n = 970, 63.78%), Puerto Rican (n = 190, 12.49%), Cuban (n = 60, 3.94%), or a range of options that were new to the ELS:02 survey, including a South American nationality (n = 110, 7.23%), a Central American nationality (n = 100, 6.57%), and Dominican (n = 50, 3.29%). Ethnicity/nationality of descent was missing for n = 40 (2.63%)⁶. Estimates of generation status proportions were 25.6% first generation, 40.9% second generation, and 33.5% third or higher generation (estimation procedures are described in Data Analysis section). Generation status could not be calculated for n = 186 (12.2%).

Measures

Social position

Sex

Participants’ identified as 1: “male” or 2: “female” in 10^th grade.

Generation status

Parents/guardians reported birthplace of the student and the student’s biological parents. Generation status was coded as follows: first generation = youth who were born outside of the 50 U.S. states, second generation = U.S.-born youth with one or both parents born outside of the 50 U.S. states, and third generation = U.S.-born youth with U.S.-born parents.

Socioeconomic status

Parents’ education

Parents/guardians reported their own educational attainment and, if applicable, the educational attainment of their cohabitating spouse or partner. The highest value was used to assess level of education attained by parents.

Family income

Parents/guardians reported their family’s income. NELS:88 participants had two more response options than ELS:02 participants, which were combined with other categories to create a 13-point scale, ranging from “none” to “$200,000 or more.”

Parental involvement in education

Parents’ aspirations for youth education

Youth reported how far in school they thought their mother and father each wanted them to go. Although the response options for the NELS:88 and ELS:02 cohorts were similar, they differed slightly⁷ and were recoded so that parent aspirations were measured on the same scale in both cohorts: 1) less than high school graduation, 2) high school graduation or GED, 3) attend or complete a two-year program, 4) attend college but not complete a four-year degree, 5) graduate from college, or 6) master’s /Ph.D./M.D./other advanced degree. Response options “Does not apply,” “Don’t know,” and “Parent doesn’t care” were treated as missing.⁸ Mothers’ and fathers’ aspirations were averaged if both were provided; correlations between mothers’ and fathers’ aspirations was .84 for NELS:88 and .78 for ELS:02.

Academic socialization

Youth were asked how often in the first half of the school year either or both of their parents engaged in discussions of school courses or programs, school activities, topics studied in class, grades, preparation for the SAT or ACT, and going to college. Items were rated on a 3-point scale ranging from “never” to “often.” The internal consistency of the six academic socialization items⁹ was good for bothNELS:88 (α = .80) and ELS:02 (α = .83).

Home-based involvement

Youth rated how often their parents engaged in activities such as checking homework and limiting time spent with friends on school nights. Items were rated on a 4-point scale ranging from “never” to “often.” Internal consistency for the six home-based involvement items¹⁰ was acceptable for NELS:88 (α = .71) and ELS:02 (α = .73).

Individual academic indicators

English proficiency

In 10^th grade, youth responded to four items about how well they understood spoken English, spoke English, read English, and wrote English on a 4-point Likert scale ranging from 1: “very well” to 4: “not at all.” English proficiency was computed as the average of these items and re-coded such that higher scores reflect greater proficiency. Internal consistency was excellent for NELS:88 (α = .93) and ELS:02 (α = .92) cohorts. Respondents skipped these items if English was the only language spoken in their home (NELS:88, n = 246) or if English was their native language (ELS:02, n = 733). Because these youth were assumed to be English-proficient, they were assigned a value of 4, the highest proficiency score.

Academic achievement

Achievement was based on the average of reading and math scores on NCES-administered tests, standardized as t-scores with a national mean of 50.

Outcomes

Change in educational expectations from 10^th to 12^th grade

Educational expectations were assessed with one item: “As things stand now, how far in school do you think you will get?” first in 10^th grade and again in 12^th grade. Response options, which differed slightly between the NELS:88 and ELS:02 surveys, are displayed in Table 1. NELS:88 responses were recoded to match the ELS:02 response options. For the change in expectations outcome, expectations in 10^th grade were subtracted from expectations in 12^th grade, so positive scores indicate an increase in expectations, and negative scores indicate a decrease in expectations.

Table 1.

Response options for educational expectations item: “As things stand now, how far in school do you think you will get?”

NELS:88 Responses		ELS:02/Current Study Responses
Less than high school graduation		(1) Less than high school graduation
High school graduation only		(2) High school graduation or GED only
< 2 years vocation/trade/business school 2+ years vocation/trade/business school *Degree from vocation/trade/business school		(3) Attend or complete two-year program
< 2 years of college 2+ years college		(4) Attend college but no four-year degree
Finish college (4- or 5-year degree)		(5) Graduate from college
Master’s degree or equivalent		(6) Master’s degree or equivalent
Ph.D., M.D., or other advanced degree		(7) Ph.D., M.D., or other advanced degree

Discrepancy between educational expectations and attainment

At the final follow-up survey, 10 years after the 10^th grade survey, respondents reported whether they had received a high school diploma or equivalent, whether they had ever attended a four-year institution, and the highest level of education they had attained. These three items were used to determine educational attainment, coded on the same 7-point scale as expectations (see Table 1). To calculate the discrepancy between expectations and attainment, expectations in 10^th grade were subtracted from attainment, such that negative scores reflect attaining less education than expected, and positive scores indicate attaining a higher level of education than expected.

Data Analysis

All analyses were conducted in Mplus version 7 (Muthén & Muthén, 2012), except for missing value analyses, which were conducted in SPSS version 25. For the NELS:88 and ELS:02 cohorts respectively, 12.29% and 12.78% of values were missing. To handle missingness, all analyses were conducted using full-information maximum likelihood (FIML) estimation with robust standard errors. FIML is a well-established procedure that produces parameter estimates based on available data (Schafer & Graham, 2002). All analyses were conducted using sample weights to adjust for probabilities of selection and participation rates; see supplemental materials for more information. For all analyses involving the change in expectations outcome, participants who responded “don’t know” to the educational expectations item in 12^th grade were excluded listwise (n = 77, 5.6% for NELS:88; n = 155, 10.2% for ELS:02). To include these participants would have treated their outcome data as missing when in fact they were not.

Predictor analyses

To examine predictors of each study outcome, multiple-group regression analyses were conducted with cohort as the grouping variable. Baseline educational expectations (i.e. in 10^th grade) were controlled in each model, and all predictors were entered simultaneously. The models were first freely estimated in both cohorts. Equality constraints were then imposed, holding all regression coefficients except baseline expectations equal between cohorts. Constraints were released individually, and a chi-square difference test was used to determine whether model fit improved. Given that the effects of socioeconomic variables were hypothesized to differ between cohorts, family income and parent education were tested first for cohort differences. The other equality constraints were then tested sequentially, beginning with social position variables, then individual academic indicators, and finally parental involvement variables. Because covariances among predictors also contribute to overall model fit, these covariances were allowed to vary if they differed between groups. In the final reported models, coefficients that did not vary between cohorts were constrained for parsimony. Unstandardized coefficients are reported because the standardized coefficients cannot be compared with equality constraints. The comparative fit index (CFI) and RMSEA were used to assess model fit, with CFI greater than .95 and RMSEA less than .06 indicating good fit (Hu & Bentler, 1999).

Generation status analyses

To test for differences in outcomes by generation status, three sets of orthogonal contrast codes were entered in linear regressions to provide pairwise comparisons (i.e. first vs. second, first vs. third, second vs. third generation) for each cohort. Generational differences in outcomes were first examined without covariates (except baseline expectations, to aid interpretation) ¹¹ and then with covariates (i.e. study predictor variables). Given differences in some ELS:02 models with and without covariates (MacKinnon, Krull, & Lockwood, 2000), post-hoc tests of mediation were conducted to examine the effects of first- versus other-generation statuses on the outcome via other study variables (e.g. family income). The variables that significantly differed for first- versus other-generation statuses were entered individually into the model; the final model with all mediators is reported.

Results

Descriptive statistics with cross-cohort comparisons and bivariate correlations are in Tables 2 and 3, respectively. Generation status differences in variables are available by request.

Table 2.

Measure descriptive statistics with cross-cohort comparisons of NELS:88 and ELS:02

	NELS:88 (N = 1,371)		ELS:02 (N = 1,521)
	%		%		Wald χ2
Social Position Factors
Sex, female	52.5		52.1		0.10
Generation status					18.91***
1st generation	14.3		25.6
2nd generation	44.2		40.9
3rd+ generation	41.4		33.5
	M	(SD)	M	(SD)	Wald χ2
Socioeconomic Status
Parents’ highest education	2.28	(1.21)	2.61	(1.22)	21.49***
Family income	6.87	(2.29)	7.85	(2.42)	41.21***
Parental Involvement
Parent aspirations	4.43	(1.27)	4.75	(1.45)	14.71***
Academic socialization	1.99	(0.46)	2.05	(0.51)	3.88*
Home-based involvement	2.56	(0.66)	2.59	(0.73)	0.32
Individual Academic Indicators
English proficiency	3.80	(0.42)	3.76	(0.48)	2.61
Academic achievement	45.24	(9.05)	45.34	(9.42)	0.02
Outcomes
Change in expectations1	.19	(1.37)	−0.21	(1.66)	11.78***
Expectations-attainment discrepancy2	−1.08	(1.36)	−1.46	(1.54)	12.42***

¹Negative values indicate a decrease in expectations, and positive values indicate an increase in expectations.

²Negative values indicate lower attainment than expected, and positive values indicate attainment that exceeded expectations.

^***p < .001

^**p < .01

^*p < .05

Table 3.

Bivariate correlations among study variables for NELS:88 (below diagonal) and ELS:02 (above diagonal)

	1.	2.	3.	4.	5.	6.	7.	8.	9.	10.	11.	12.
1. Sex	--	−.03	−.03	−.05	.−.02	.09*	.17***	.01	.02	.16***	.16***	.14***
2. Parent education	−.10***	--	.40***	.18***	.15***	.22***	.18***	.13***	.27***	.16***	.16***	.27***
3. Family income	.05	.50***	--	.31***	.30***	.20***	.10**	.10**	.30***	.11***	.12***	.25***
4. Generation status	.01	.22***	.22***	--	.41***	.06	.01	.01	.17***	−.04	.04	.08**
5. English proficiency	.06*	.19***	.29***	.36***	--	.10***	.12***	.09**	.34***	.10***	.09**	.19***
6. Parent aspirations	.06*	.23***	.22***	.01	.17***	--	.25***	.15***	.24***	.37***	.28***	.30***
7. Academic socialization	−.03	.16***	.17***	.03	−.01	.25***	--	.53***	.21***	.42***	.29***	.25***
8. Home-based involvement	.02	.06*	.09*	.02	.07*	.15***	.48***	--	.05	.20***	.13***	.08**
9. Academic achievement	−.02	.29***	.33***	.03	.27***	.27***	.15***	-.04	--	.37***	.32***	.47***
10. Expectations in 10th	−.02	.32***	.28***	−.01	.09***	.52***	.35***	.11***	.47***	--	.43***	.41***
11. Expectations in 12th	−.08***	.24***	.23***	−.10**	.07**	.34***	.24***	.08*	.49***	.61***	--	.48***
12. Attainment	−.09**	.33***	.19***	−.02	.16***	.26***	.21***	−.01	.50***	.53***	.58***	--

Note. The change in expectations outcome and the discrepancy between expectations and attainment outcome were not included in the bivariate correlations table. Without controlling for baseline expectations, the correlations of these outcomes with other study variables is difficult to interpret.

^***p < .001

^**p < .01

^*p < .05

Cross-Cohort Comparisons of Study Variables

The ELS:02 cohort was of a more recently immigrated generation than participants in the NELS:88 cohort on average (χ²(1) = 18.91, p < .001). The ELS:02 cohort had higher parent education (χ²(1) = 21.08, p < .001) and family income (χ²(1) = 40.21, p < .001) than NELS:88. However, this income difference (NELS:88 median of $20,001-$25,000; ELS:02 median of $25,001-$35,001) was not large considering inflation of 37.6% between 1990 and 2002 (Bureau of Labor Statistics, 2018). The ELS:02 cohort reported higher parent aspirations (χ²(1) = 14.71, p < .001) and academic socialization (χ²(1) = 3.88, p < .05) than NELS:88. Change in expectations from 10^th to 12^th grade differed between cohorts (χ²(1) = 11.78, p < .001), with expectations increasing slightly on average in NELS:88 and decreasing slightly on average in ELS:02. Although both cohorts attained less education than they expected on average, the ELS:02 cohort was further from attaining their expectations than the NELS:88 cohort (χ²(1) = 12.42, p < .001).

Change in Educational Expectations from 10^th to 12^th Grade

Table 4 shows levels of educational expectations in 10^th grade and 12^th grade by cohort.

Table 4.

Levels of Educational Expectations and Educational Attainment for NELS:88 and ELS:02

	NELS:88 (N = 1,371)			ELS:02 (N = 1,521)
	Expectations in 10th Grade	Expectations in 12th Grade	Educational Attainment	Expectations in 10th Grade	Expectations in 12th Grade	Educational Attainment
	%	%	%	%	%	%
(1) Less than high school graduation	2.0	4.2	11.0	2.1	0.4	6.4
(2) High school graduation or GED only	15.1	3.6	16.2	10.8	6.4	13.5
(3) Attend or complete two-year program	14.7	22.2	36.8	6.9	20.7	38.3
(4) Attend college but no four-year degree	25.0	16.1	22.7	7.8	4.6	22.5
(5) Graduate from college	25.3	25.0	12.5	38.8	27.4	16.2
(6) Master’s degree or equivalent	7.4	10.9	1.0	17.7	17.1	2.5
(7) Ph.D., M.D., or other advanced degree	10.5	12.5	0.1	15.8	13.2	0.6
“Don’t know”	N/A	5.6	--	N/A	10.2	--
M	4.20a	4.39b	3.12c	4.85a	4.64b	3.38c
(SD)	(1.55)	(1.57)	(1.19)	(1.57)	(1.57)	(1.20)

Note. The response option “Don’t know” was not offered on the NELS:88 10^th grade survey. For the ELS:02 cohort, participants who responded “Don’t know” in 10^th grade (n = 206) were previously excluded during the sample selection phase, as they had missing data for both outcome measures and could not be compared to NELS:88 participants.

^aExpectations in 10^th grade were significantly higher for ELS:02 than NELS:88 on average (χ² = 42.48, p < .001).

^bExpectations in 12^th grade were significantly higher for ELS:02 than NELS:88 on average (χ² = 4.39, p < .05).

^cAttainment was significantly higher for ELS:02 than NELS:88 on average (χ² = 6.27, p < .05).

Predictors of change in expectations

Constraining regression coefficients to be equal between cohorts marginally worsened fit (Δχ²(9) = 15.29, p = .08). Releasing constraints on family income (Δχ²(1) = .01, p = .93) and parent education (Δχ²(1) = .01, p = .96) did not improve model fit, indicating that these predictors of change in expectations did not differ between cohorts. The remaining constraints were released sequentially. Releasing constraints on sex (Δχ²(1) = 9.86, p < .01) and generation status (Δχ²(1) = 4.52, p < .05) improved model fit, indicating that these effects differed by cohort. Releasing the constraint on academic achievement (Δχ²(1) = 3.53, p = .06) also marginally improved fit. The final model had good fit (CFI = 1.00, RMSEA < .01). Results are displayed in Table 5.

Table 5.

Standardized coefficients, standard errors, and 95% confidence intervals for predictors of A) the change in expectations from 10^th to 12^th grade and B) the discrepancy between expectations in 10^th grade and attainment 10 years later

	Change in Expectations1						Expectations-Attainment Discrepancy2
	CFI = 1.00; RMSEA < .01						CFI = 1.00; RMSEA < .01
	NELS:88 (N = 1,294)			ELS:02 (N = 1,366)			NELS:88 (N = 1,371)			ELS:02 (N = 1,521)
	b	SE	95% CI	b	SE	95% CI	b	SE	95% CI	b	SE	95% CI
(Constant)	.54	.64	[−.72, 1.80]	.74	.43	[−.66, 2.59]	−.01	.42	[−.82, .81]	−.98**	.32	[−1.60, −.35]
Expectations in 10th grade	−.59***	.05	[−.69, −.48]	−.73***	.04	[−.81, −.65]	−.58***	.05	[−.68, −.49]	−.71***	.03	[−.76, −.65]
Social Position Factors
Sex	−.17	.15	[−.45, .12]	.30**	.12	[.07, .53]	−.10	.12	[−.34, .14]	.14*	.06	[.02, .27]
Generation status	−.21**	.07	[−.36, −.07]	.03	.08	[−.12, .18]	−.07	.04	[−.14, .03]	−.07	.04	[−.14, .03]
Socioeconomic Status
Parent education	.02	.04	[−.04, .09]	.02	.04	[−.04, .09]	.12***	.03	[.07, .17]	.12***	.03	[.07, .17]
Family income	.03	.02	[−.01, .07]	.03	.02	[−.01, .07]	−.04	.02	[−.08, .01]	.04*	.02	[.01, .07]
Parental Involvement
Parent aspirations	.12*	.05	[.03, .22]	.12*	.05	[.03, .22]	−.07	.08	[−.23, .09]	.08**	.03	[.03, .13]
Academic socialization	.24*	.12	[.01, .48]	.24*	.12	[.01, .48]	.12	.08	[−.04, .28]	.12	.08	[−.04, .28]
Home-based involvement	−.01	.09	[−.18, .15]	−.01	.04	[−.18, .15]	−.11*	.05	[−.21, −.02]	−.11*	.05	[−.21, −.02]
Individual Academic Indicators
English proficiency	−.13	.13	[−.38, .13]	−.13	.13	[−.38, .13]	.16*	.08	[.01, .31]	.16*	.08	[.01, .31]
Academic achievement	.04***	.01	[.03, .07]	.03***	.01	[.01, .04]	.03***	.01	[.02, .04]	.03***	.01	[.02, .04]
Change in expectations	--	--	--	--	--	--	.23***	.03	[.18, .29]	.23***	.03	[.18, .29]

Note. Italicized coefficients were allowed to vary; non-italicized coefficients were held constant.

¹For the change in expectations outcome, positive coefficients represent prediction of an increase in expectations (or less of a decrease for the ELS:02 cohort, whose average change was negative), while negative coefficients represent prediction of a decrease in expectations (or less of an increase for the NELS:88 cohort, whose average change was positive).

²For the expectations-attainment discrepancy outcome, positive coefficients represent prediction of exceeding expectations or approaching expectations, while negative coefficients represent prediction of attaining less education than expected.

^***p < .001

^**p < .01

^*p < .05

Note that positive coefficients represent prediction of an increase in expectations (or less of a decrease for the ELS:02 cohort, whose average change was negative). Academic achievement was positively associated with change in expectations for both cohorts, and this association was marginally stronger for NELS:88 (b = .04, SE = .01, p < .001) than ELS:02 (b = .03, SE = .01, p < .001). Parent aspirations (b = .12, SE = .05, p < .05) and academic socialization (b = .24, SE = .12, p < .05) were also positively associated with change in expectations for both cohorts. For NELS:88 only, higher generation status was associated with less increase in expectations (b = −.21, SE = .07, p < .01), which is explored further in the next section. For the ELS:02 cohort only, female sex was positively associated with change in expectations (b = .30, SE = .12, p < .01). Socioeconomic factors, English proficiency, and home-based involvement were not associated with change in expectations for either cohort.

Generation status and change in expectations

Changes in expectations by generation status are depicted in Figure 1 for NELS:88 and Figure 2 for ELS:02. See supplemental materials for figures with confidence intervals (Figures S1 and S2). For NELS:88, first-generation youth had a greater increase in expectations compared to their second- (b = −.17, SE = .08, p < .05) and third-generation counterparts (b = −.23, SE = .08, p < .01), and these differences remained after accounting for other study variables (b = −.16, SE = .07, p < .05; b = −.24, SE = .07, p < .01). For ELS:02, change in expectations did not differ by generation status, with or without covariates.

Figure 1.

NELS:88 cohort: Educational expectations in 10^th and 12^th grades and educational attainment by generation status, (a) without covariates and (b) adjusted for all other study variables

Note. Response options on Y-axis represent (2) complete high school or GED, (3) attend or complete a two-year school program, (4) attend college but not attain a four-year degree, and (5) graduate from college.

Figure 2.

ELS:02 cohort: Educational expectations in 10^th and 12^th grades and educational attainment by generation status, (a) without covariates and (b) adjusted for all other study variables

Note. Response options on Y-axis represent (2) complete high school or GED, (3) attend or complete a two-year school program, (4) attend college but not attain a four-year degree, and (5) graduate from college.

Discrepancy between Educational Expectations and Attainment

Attainment levels are presented in Table 4. Attainment was higher for the ELS:02 cohort than the NELS:88 cohort (χ² = 6.27, p < .05). However, 73.5% of the ELS:02 cohort attained less education than they expected, compared to only 64.8% of the NELS:88 cohort.

Predictors of expectations-attainment discrepancy

Constraining regression coefficients to be equal between cohorts worsened model fit (Δχ²(10) = 24.99, p < .01). Releasing the constraint on family income improved model fit (Δχ²(1) = .23.42, p < .001), indicating that the effect of family income on expectations-attainment discrepancy differed by cohort. This was not the case for parent education (Δχ²(1) = 1.39, p = .24). Releasing constraints on sex (Δχ²(1) = 4.40, p < .05) and parent aspirations (Δχ²(1) = 5.81, p = .05) individually improved model fit, indicating that the effects of these variables differed by cohort. The final model fit the data well (CFI = 1.00, RMSEA < .01). Results are displayed in Table 5.

Note that positive regression coefficients represent coming closer to attaining one’s own expectations or exceeding expectations. In both cohorts, all academic indicators were positively associated with expectations-attainment discrepancy (academic achievement b = .03, SE = .01, p < .001; English proficiency b = .16, SE = .08, p < .05; change in expectations b = .23, SE = .03, p < .001). Home-based involvement was negatively associated with expectations-attainment discrepancy in both cohorts (b = −.11, SE = .05, p < .05), indicating that higher home-based involvement was associated with attaining less education than expected. Parent education was positively associated with expectations-attainment discrepancy in both cohorts (b = .12, SE = .03, p < .001), but family income was positively associated with this outcome for ELS:02 only (b = .04, SE = .02, p < .05). Also for ELS:02 only, female sex (b = .14, SE = .06, p < .05) and parent aspirations (b = .08, SE = .03, p < .01) were positively associated with expectations-attainment discrepancy. Academic socialization was not associated with expectations-attainment discrepancy for either cohort, and the effects of generation status are explored below.

Generation status and expectations-attainment discrepancy

Generation status differences in the expectations-attainment discrepancy are depicted in Figure 1 for NELS:88 and Figure 2 for ELS:02. For figures with confidence intervals, see supplemental materials (Figures S1 and S2). For the NELS:88 cohort, the expectations-attainment discrepancy did not differ by generation status before accounting for covariates. However, after accounting for other study variables, third-generation youth were further from attaining their expected level of education than second-generation youth on average (b = .13, SE = .06, p < .05).

For the ELS:02 cohort only, first-generation youth attained less education than they expected to a significantly greater extent than both second- and third-generation youth (b = .18, SE = .05, p < .001; b = .17, SE = .05, p < .001). However, these differences no longer remained after including covariates, indicating that the effects of first-generation status may have been mediated by these variables (MacKinnon et al., 2000). Post-hoc mediation tests showed that the effect of first- versus second-generation status was partially mediated by income (indirect b = .04, SE = .01, p < .01) and achievement (indirect b = .06, SE = .02, p < .001). The effect of first- versus third-generation status was fully mediated by these same factors plus parent education (income indirect b = .04, SE = .02, p < .05; achievement indirect b = .08, SE = .02, p < .001; parent education indirect b = .03, SE = .01, p < .01). Finally, like the NELS:88 cohort, when covariates were included, third-generation youth were further from attaining their expected level of education than second-generation youth in ELS:02 (b = .08, SE = .04, p < .05).

To summarize, findings showed that third-generation youth were at a relative disadvantage compared to their second-generation counterparts of otherwise equal positioning in both cohorts. In the more recent ELS:02 cohort only, first-generation youth had an absolute disadvantage in terms of attaining their expectations compared to other-generation counterparts, and this was most attributable to income, academic achievement, and parent education.

Discussion

Extant literature documents high early educational expectations of Hispanic adolescents but low attainment outcomes among Hispanic young adults on average (McFarland et al., 2017; Mello, 2009). Drawing on the status-attainment tradition and expectancy value theory, this study examined changes in educational expectations from 10^th to 12^th grade as well as discrepancies between educational expectations and attainment for Hispanic youth in two nationally representative cohorts. Using a cultural-ecological framework, social position factors, family characteristics, and individual factors were examined as predictors of these outcomes. The current study extends existing knowledge in four primary ways, first, by documenting changes in Hispanic youth’s educational expectations historically (across cohorts) and developmentally (within cohorts) as well as attainment of those expectations. Second, the current study demonstrated that parent education is important for attaining expectations and that income has become a more important factor in Hispanic youth’s attainment outcomes over time. Third, the results of this study showed that third-generation individuals have a relative attainment disadvantage compared to second-generation counterparts of otherwise similar social positioning, and that first-generation individuals have more recently had an absolute attainment disadvantage compared to native-born counterparts on average. Finally, the current study provides support for the importance of parents’ educational aspirations and academic socialization strategies for promoting educational expectations in high school and beyond.

Changes in Educational Expectations and Educational Attainment

As expected, results showed that educational expectations of Hispanic students rose across cohorts. This is consistent with previous research showing that educational expectations have risen over time nationally (Goyette, 2008; Park et al., 2015; Reynolds et al., 2006) as well as among Hispanic students specifically (Ingels et al., 2005, 1994). This trend in rising expectations has been attributed to multiple factors, including increased earnings gaps between individuals with and without a bachelor’s degree, higher education among parents, and changing norms about higher education as part of a typical life course (Goyette, 2008; Nunez et al., 1998).

Beyond rising expectations across cohorts, the results of the current study showed cohort differences in the change in expectations from 10^th to 12^th grade for Hispanic youth: expectations increased slightly on average in the NELS:88 cohort and decreased slightly on average in the ELS:02 cohort. These results are consistent with data from the NELS:88 and ELS:02 national samples showing the same pattern (Park et al., 2015). Within the more recent ELS:02 cohort, the decrease in expectations from 10^th to 12^th grade is likely attributable to very high expectations in 10^th grade; so many 10^th graders expected to attain a college degree or higher that expectations may have hit a ceiling. In 12^th grade, some youth may realize academic, socioeconomic, or other constraints and downgrade their expectations accordingly (Park et al., 2015). As supported by the current study, greater decreases in expectations (or smaller increases for NELS:88) were associated with attaining a lower level of education than initially expected.

The current study showed that youth attained less education than they expected in both cohorts. There are multiple pathways by which youth may attain less education than they expected. For instance, youth who expect a four-year college degree or higher may ultimately attain a two-year college degree. In the current study, expectations of attending or completing a two-year program were relatively uncommon (14.7% and 6.9% for NELS:88 and ELS:02 respectively), but this was the level of education that was ultimately attained by a sizeable proportion of each cohort (36.8% and 38.8% for NELS:88 and ELS:02 respectively). Many students enroll in two-year programs with the intention of transferring to a four-year college, but despite these intentions, two-year institutions do not typically serve as “gateways” to bachelor’s degrees (Arbona & Nora, 2007; Fry, 2005; Ma & Baum, 2016). National data have shown that of students who entered two-year degree programs in 2003 with the intentions of gaining a bachelor’s degree, only 21% had attained a bachelor’s degree six years later (Ma & Baum, 2016). Other ways in which expectations may exceed attainment for Hispanic youth is by expecting a four-year college degree and enrolling in a four-year program, but not completing the degree (Fry, 2002) due to socioeconomic, academic, cultural, or other barriers.

By comparing Hispanic students’ expectations in adolescence to their own attainment in young adulthood in two cohorts, this study provides evidence that Hispanic students’ expectations are outpacing attainment. Results showed that the proportion of Hispanic students attaining less education than they expected has grown over time: 64.8% attained less education than they expected in NELS:88 versus 73.5% in ELS:02. This finding is consistent with trends in the general population (Reynolds et al., 2006), suggesting that this is a normative shift not specific to Hispanic youth. As “college for all” has become a more common ideal in U.S. society, students who initially expected to attain a four-year degree may later realize that enrolling in a two-year institution is a more viable alternative (Fry, 2002; Goyette, 2008). This phenomenon may be particularly applicable for Hispanic students. Between 1990 and 2010, the enrollment of Hispanic students in two-year institutions grew by a substantial 241% (compared to 37% nationally; Snyder et al., 2018). Increasing enrollment of Hispanic students in two-year institutions, combined with increasingly common expectations of a college degree, is likely one way in which Hispanic youth’s expectations have outpaced their actual attainment over time.

Maintaining and Attaining Educational Expectations: Contributing Factors

Given the current findings on change in expectations from 10^th to 12^th grade and discrepancies between expectations and attainment, predictors of these outcomes were examined.

Generation status

Previous findings on generation status and Hispanic youth’s educational expectations and attainment are mixed, likely due to changes over time, the developmental period considered, and whether other factors are accounted for in analyses. These inconsistencies in the literature were addressed in the current study by the use of two cohorts, a longitudinal design, and by examining the effects of generation status before and after accounting for other factors. Results showed that first-generation youth had a greater increase in expectations than second- and third-generation youth in the NELS:88 cohort. This is consistent with previous research showing that Hispanic immigrants’ educational expectations dip from middle school to sophomore year and then increase again through senior year, unlike native-born youth, whose expectations are relatively more stable (Kaufman et al., 1998; Perez-Brena et al., 2017). This pattern did not hold for the ELS:02 cohort. The fact that change in expectations did not differ by generation status for the ELS:02 cohort may be due to the changing nature of expectations. As more and more students optimistically expect to attain a college degree, “immigrant optimism” may no longer set Hispanic immigrants’ educational expectations apart.

In support of the immigrant optimism hypothesis and the immigrant paradox, second-generation youth in both cohorts came closer to attaining their expectations than their third-generation counterparts, only after accounting for individual and family characteristics. These findings are consistent with patterns in previous educational research: the odds of dropping out of high school were similar for Hispanic students regardless of their generation status, but after accounting for demographic characteristics and various forms of capital, the odds of dropping out were highest for third-generation youth (Driscoll, 1999; Perreira et al., 2006). Taken together, these findings suggest that the immigrant paradox phenomenon is more complex than a later-generation disadvantage in educational outcomes; third and higher generations might have a disadvantage relative to earlier generations of otherwise equal social positioning.

The current study also demonstrated an absolute disadvantage among first-generation individuals in the more recent ELS:02 cohort: first-generation youth were significantly further from attaining their educational expectations compared to their later-generation counterparts. However, after accounting for other factors, this first-generation disadvantage vanished. These findings are consistent with population surveys showing that first-generation Hispanic youth have lower enrollment rates than later-generation counterparts (Lopez, 2009; Musu-Gillette et al., 2016). The current results are also consistent with the handful of studies showing that accounting for sociodemographic factors partially or completely closes these enrollment gaps (Jasinski, 2000; Warren, 1996). A limitation of previous work is that few studies have tested factors that may explain generational differences in educational outcomes for Hispanic youth. The current study addressed this limitation by examining factors that explained first generation youth’s greater discrepancy between expectations and attainment in the ELS:02 cohort. Findings showed that income, achievement, and parent education accounted for differences in outcomes between first-generation and later-generation youth. Thus, the linkage between generation status and economic and human capital for Hispanic families in the U.S. appears to play a role in attainment disadvantages among more recent first-generation Hispanic students. Immigration patterns may be one possible explanation for the cohort differences in the first-generation disadvantage. In the 1980s and 1990s, the U.S. experienced high rates of immigration, and consequently, Hispanic enrollment in public elementary schools increased by 53% between 1993 and 2003 (a period that corresponds to ELS:02 elementary school years; Fry, 2006). During this time, approximately one fifth of public schools absorbed most new Hispanic enrollment; these schools grew most in size and poverty rates (Fry, 2006; Lowell & Suro, 2002). Public schools serving Hispanic youth may have been less equipped for the ELS:02 cohort, and earlier-generation Hispanic youth may have been disproportionately affected by these changes.

Sex

Results showed that being female was associated with maintaining and attaining expectations in the ELS:02 cohort. This is consistent with previous studies showing that male adolescents may be more likely to lower their educational expectations and male young adults may be more likely to have unrealized expectations compared to their female counterparts (Feliciano & Rumbaut, 2005; Hanson, 1994). The fact that this finding emerged for the most recent cohort is consistent with patterns in other racial/ethnic groups (U.S. Department of Education, 2016b) and reflect trends in Hispanic young adults’ postsecondary enrollment that emerged in the 1990s and 2000s, with female young adults enrolling at higher rates.

Socioeconomic status

The hypothesis that socioeconomic factors would be associated positively with study outcomes was partly supported. In both cohorts, youth whose parents were more highly educated came closer to attaining their expected level of education on average. These findings are consistent with the status-attainment model and with research highlighting the importance of parent education for enrollment in higher education (Nunez et al., 1998; Sewell et al., 1969). In the ELS:02 cohort only, youth whose families had higher income came closer to fulfilling their educational expectations. The finding that income was positively associated with attaining expectations in the recent cohort only is consistent with existing research demonstrating that the association of SES with attainment has become stronger over time (Reynolds et al., 2006). Whereas previous work has been limited by composite measures of SES, the current findings highlight that family income in particular became a more important predictor of attaining educational expectations for Hispanic students. With a higher proportion of ELS:02 youth expecting to complete at least a four-year college degree, family financial resources and constraints made a bigger difference in terms of ability to attain those expectations. As “college for all” has become a more common ideal, and as two-year colleges are more affordable and accessible (Fry, 2002), family income has become a stronger predictor of attaining expectations.

Contrary to hypotheses, changes in expectations were not predicted by family income or by parent education for either cohort, accounting for all other factors. One reason may be that in theoretical models of the development of educational expectations, the effects of SES operate through more proximal factors, including parents (Sewell et al., 1969; Trusty, 2000). The current study did not test for indirect effects of SES; future studies should do so to provide a more holistic portrait of these processes. Additionally, although previous work has shown that SES was associated with change in expectations from 10^th to 12^th grade for the population in general (Park et al., 2015), SES may be less predictive of educational outcomes for Hispanic, immigrant, and some other racial/ethnic minority youth (Hernández et al., 2016; Jasinski, 2000).

Parental involvement

The current study expands on existing knowledge of the longitudinal effects of multiple types of parental involvement in Hispanic youth’s education. Parents’ aspirations were positively associated with change in expectations for both cohorts and with attaining expectations for the ELS:02 cohort only. These findings are consistent with existing research linking parents’ aspirations with Hispanic youth’s educational beliefs (Ceballo et al., 2014; Hernandez et al., 1994), but this study is among the first to suggest that this form of parental involvement may help transform Hispanic youth’s expectations into attainment. The fact that parent aspirations predicted attaining expectations for ELS:02 but not NELS:88 may be related to the fact that initial expectations were so high in the ELS:02 cohort. As expectations have risen historically, they have also become more “unrealistic” for many students given actual socioeconomic, academic, and other constraints and life circumstances (Goyette, 2008; Reynolds et al., 2006). When parents, who are also attuned to these constraints, have higher aspirations for their youth’s education, youth may be more realistically positioned to achieve their expectations.

Findings also showed that academic socialization was also positively associated with change in expectations in both cohorts as hypothesized. This is consistent with research suggesting that academic socialization is a developmentally relevant strategy for many adolescents (Hill & Tyson, 2009). However, academic socialization did not predict attaining expectations in either cohort. It is possible that the effect of academic socialization on attaining expectations takes place indirectly through more proximal effects, such as school engagement or academic self-efficacy (Bandura et al., 1996; Wang & Sheikh-Khalil, 2014).

The parental strategy of home-based involvement was associated with attaining less education than expected in both cohorts. This finding should be interpreted in light of the fact that many parents who provide a high level of supervision and structure at home do so to support adolescents who have academic and related difficulties (Fan & Chen, 2001). Therefore, it is possible that adolescents who receive high levels of home-based involvement still develop high expectations but may encounter difficulty attaining them. This interpretation is consistent with extant work showing that parents’ home-based involvement is associated with Latino adolescents’ greater school effort (Ceballo et al., 2014) and negatively associated with Mexican American adolescents’ grades (Altschul, 2011). More research is needed on the longitudinal effects of home-based involvement, accounting for prior achievement and related indicators.

Academic indicators

Finally, English proficiency and change in expectations were associated with attaining expectations in both cohorts. Academic achievement was positively associated with change in expectations and attaining expectations in both cohorts, as hypothesized. The effect of achievement on change in expectations was marginally stronger for the NELS:88 cohort than the ELS:02 cohort. This cohort difference may again be due to the fact that expectations were so high in the ELS:02 cohort. Adolescents in older cohorts may have placed more weight on achievement as a form of feedback that informs their expectations, whereas in more recent cohorts, adolescents’ expectations are more loosely tied to their academic background given the ideal of “college for all” has become more common (Goyette, 2008).

Limitations and Future Directions

Although the current study extends the knowledge base on Hispanic adolescents and young adults’ educational outcomes, there are inherent limitations. The current study did not examine whether processes differ by ethnic group or nationality of origin. Because educational outcomes may vary considerably for different Hispanic groups (e.g. Cuban-origin individuals tend to have higher educational attainment outcomes than Mexican-origin individuals across generation statuses), the processes that promote educational outcomes may differ among these groups (Bohon et al., 2006). Another limitation is that, beyond generation status and English proficiency, the current study did not examine cultural processes that may further elucidate the findings. For example, youth with a greater sense of family obligation may be more inclined to attend two-year rather than four-year institutions so that they may continue living at home (Fry, 2002; Fuligni & Witkow, 2004), which may contribute to generational differences in outcomes. Relevant ethnic and cultural factors should be explicitly tested rather than assumed (Phinney & Jolene, 1998). To address this limitation, future national surveys must include measures of specific cultural indicators to allow for examination of population-level trends.

Implications and Conclusions

Although pathways of academic achievement and attainment begin long before sophomore year in high school, 10^th to 12^th grade is a critical period of planning for postsecondary activities (Perez-Brena et al., 2017). Because mismatch of expectations with attainment is indicative of a lack of knowledge about the prerequisites of the expected level, it is important for high school students to receive information that supports their plans, including information about tools to prepare for college admissions tests and college applications, as well as information about financial aid and scholarships. Additionally, diversity enhancing educational and professional development programs for high school and incoming college students may help improve representation of Hispanic individuals in higher education (Santiago & Brown, 2004). This is the first study to examine how the factors that promote successful fulfillment of Hispanic youth’s educational expectations have changed over time. The importance of educational attainment for economic and occupational success in our society is greater now than ever before, and the current study’s findings and implications highlight the need for continued research in this area.