Educational expectation trajectories and attainment in the transition to adulthood

Social Science Research 42 (2013) 818–835

Contents lists available at SciVerse ScienceDirect

Social Science Research journal homepage: www.elsevier.com/locate/ssresearch

Educational expectation trajectories and attainment in the transition to adulthood Monica Kirkpatrick Johnson a,⇑, John R. Reynolds b a b

Washington State University, United States Florida State University, United States

a r t i c l e

i n f o

Article history: Received 17 January 2012 Revised 26 November 2012 Accepted 6 December 2012 Available online 19 December 2012 Keywords: Educational expectations Educational attainment Life course Status attainment

a b s t r a c t How consequential is family socioeconomic status for maintaining plans to get a bachelor’s degree during the transition to adulthood? This article examines persistence and change in educational expectations, focusing on the extent to which family socioeconomic status shapes overtime trajectories of bachelor’s degree expectations, how the influence involves the timing of family formation and full-time work vs. college attendance, and how persistence in expectations is consequential for getting a 4-year degree. The findings, based on the high school senior classes of 1987–1990, demonstrate that adolescents from higher socioeconomic status families are much more likely to hold onto their expectations to earn 4-year degrees, both in the early years after high school and, for those who do not earn degrees within that period, on through their 20s. These more persistent expectations in young adulthood, more so than adolescent expectations, help explain the greater success of young people from higher socioeconomic status backgrounds in earning a 4-year degree. Persistence of expectations to earn a bachelor’s degree in the years after high school is shaped by stratified pathways of school, work, and family roles in the transition to adulthood. Ó 2012 Elsevier Inc. All rights reserved.

1. Introduction Sociologists have focused much attention on adolescents’ ambitions, both as important correlates of educational and occupational attainment, but also because they provide an important window through which to see how adolescents view their opportunities and themselves. Expectations are often measured by adolescents’ reports of how far they really think they will go in school, and recent estimates from the Educational Longitudinal Study indicate that 75% of 12th graders in 2004 expected to earn a bachelor’s degree or more (Ingels and Dalton, 2008). In what became known as the ‘‘Wisconsin model’’ of status attainment, adolescents’ expectations about their future educational attainment and occupational destinations were thought to be a key link between parental socioeconomic status (SES) and offspring’s attainment (e.g., Haller and Portes, 1973). Recent refinements to the theory emphasize that children from higher SES families are more able to maintain high educational expectations from elementary school through high school, and that stable high expectations are better predictors of educational attainment than are expectations measured toward the end of high school (Bozick et al., 2010). We know much less about how processes linking socioeconomic background and expectations play out in the years beyond high school, and their ultimate implications for attainment (Morgan, 2004). What happens to educational ambitions as adolescents from disparate socioeconomic backgrounds leave high school and embark on futures that may or may not be what they expected? Despite growing attention to rising expectations across ⇑ Corresponding author. Address: Department of Sociology, Washington State University, Pullman, WA 99164-4020, United States. E-mail address: [email protected] (M.K. Johnson). 0049-089X/$ - see front matter Ó 2012 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.ssresearch.2012.12.003

M.K. Johnson, J.R. Reynolds / Social Science Research 42 (2013) 818–835

819

cohorts of adolescents (Goyette, 2008; Reynolds et al., 2006; Schneider and Stevenson, 1999) and what that has meant for educational attainment (Reynolds and Johnson, 2011), we have only scattered evidence of whether these expectations persist or change as young people move through their 20s, when ambitions often go unfulfilled and other life events like parenthood or full-time employment may encourage goal revision or postponement. Further, an ongoing debate is whether increasingly widespread ambitions to obtain a 4-year college degree or more are on balance beneficial or harmful to youth, particularly those from socioeconomically disadvantaged backgrounds (Domina et al., 2011a,b; Rosenbaum, 2011). Though hard to interpret as all positive or negative, a reality is that poorly prepared and/or economically disadvantaged adolescents face losing odds to realize their expectations to get a bachelor’s degree (Alexander et al., 2008). Many questions in this area remain unanswered. Are young people who expect to earn bachelor’s degrees able to hold on to these expectations in the years after high school, particularly those from lower SES backgrounds? Do SES-related differences in the persistence of bachelor’s degree expectations help explain stratification in attainment? Finally, what role do competing commitments that arise in adulthood play in SES-related patterns of persistence and change in expectations? In this study, we draw on panel data from a national sample of high school seniors from the Monitoring the Future Study (the 1987–1990 senior year cohorts) to examine patterns of persistence and change in expectations to earn a 4-year degree in the early and later part of the decade following high school. Our first objective is to examine how SES is related to the way bachelor’s degree expectations come and go during the transition to adulthood. Our second objective is to examine the impact of these expectation trajectories on educational attainment and evaluate the extent to which they explain the higher attainments of higher SES youth. Our third objective is to examine the ongoing life events that shape expectations as young people make the transition to adulthood for their part in SES-differentiated expectations and attainments. Our focus with respect to educational expectations is on young people’s subjective probabilities of obtaining a bachelor’s degree. That is, we seek to better understand the way educational advantage is passed on through the tendency for young adults to perceive with greater certainty and constancy they will get a 4-year degree. Our focus on the bachelor’s degree is grounded in the growing prominence of such plans among adolescents, who are often encouraged to pursue such plans based on average and projected payoffs to higher education, despite rapidly rising tuition and fees at 4-year institutions (e.g., Baum et al., 2010). Focusing on expectations to obtain a bachelor’s degree also is consistent with the small but growing literature on what happens to educational expectations beyond adolescence (Alexander et al., 2008; Hanson, 1994; Trusty, 2001; Trusty and Harris, 1999; Uno et al., 2010). It should be noted however that this differs from others who examine educational expectations in terms of how far adolescents and young adults think they’ll ultimately go in school (e.g., Andrew and Hauser, 2011; Morgan, 2004). Building on recent work by Bozick et al. (2010) and Alexander et al. (2008) and drawing on the life course perspective, we argue that higher SES enables the maintenance of expectations to earn a BA in the years after high school. Stable high expectations facilitate investments in higher education, resulting in higher educational attainment. That is, part of the reason youth from higher SES families are more successful and expedient at completing bachelor’s degrees is due to their more steadfast plans to do so. We also argue that the lower or more unstable bachelor’s degree expectations of lower socioeconomic youth result in part from different life course trajectories of work, schooling, and family formation during the transition to adulthood. We examine these dynamics in two phases of young adulthood: the first 5–6 years after high school and then another 5–6 years beyond that. This enables us to assess whether the benefits of stable high bachelor’s degree expectations or the links between SES and persistence extend beyond the normative period of college-going. The percent of high school graduates in the 2010 Census who were enrolled in school dropped steeply between 18 and 25, after which enrollment declines taper only slightly with advancing age through age 30 (Ruggles et al., 2010). We label these two phases the ‘‘early phase’’ and ‘‘late phase’’ of young adulthood, but also use the ‘‘early twenties’’ and ‘‘late twenties’’ for ease of expression. 1.1. Socioeconomic status and persistence vs. change in educational expectations Adolescent educational ambition is widespread in the US. Turner (1960) long ago characterized US educational ideology as one that emphasizes an open-ended competition, encouraging widespread high expectations into late stages of schooling. In the aggregate, we know young adults lower their educational ambitions in the years after high school (Jacob and Linkow, 2011). Underneath this aggregate downgrading of ambitions, however, lies a variety of trajectories. Young adults may persist in their expectations, give them up for more modest achievements, or aim even higher than before—trajectories both influenced by socioeconomic origins and potentially consequential for subsequent attainments. A possible educational advantage of youth from higher SES families is that they are more likely to maintain expectations for a bachelor’s degree and less likely to let go of those plans than lower SES youth (Alexander et al., 2008; Hanson, 1994; Trusty, 2000; Trusty and Harris, 1999; Uno et al., 2010). Bozick et al.’s (2010) recent work provides a possible explanation for why young people from higher status backgrounds may sustain their expectations over time, and suggests that the persistence of high expectations plays a large role in the link between socioeconomic origins and post-secondary educational attainment. Bozick and colleagues demonstrate that expectations to attend college fluctuate from the elementary school years through the high school years and the resulting trajectories of expectations are more strongly related to later college enrollment than expectations measured once in 11th grade. Those with stable high expectations were much more likely to attend college than others who also expected to attend college as juniors in high school but had not held consistent expectations. In other words, later formed expectations, even those at a

820

M.K. Johnson, J.R. Reynolds / Social Science Research 42 (2013) 818–835

critical juncture such as nearing the end of high school, mattered less than did the pattern of expectations held over a number of years. Importantly, expectation trajectories were strongly patterned by social class (see also Jacob and Linkow, 2011). High SES youth were more likely to hold stable expectations to attend college as they moved through school. Kao and Tienda (1998) similarly find greater stability in the educational expectations of white and Asian students throughout high school compared to black and Hispanic students, and attribute those differences to socioeconomic background. Bozick and colleagues argue that this is because the feedback high SES youth receive from their parents and through their academic performance is more consistently supportive of the idea that they are ‘‘college material.’’ Those who hold stable high educational expectations experience consistent feedback from parents early on and consistently earn grades that reinforce an orientation toward college. Like Bozick and colleagues, this paper uses a dynamic, life course, approach to examining educational expectations. The life course perspective facilitates taking a long view of lives (Elder, 1994), in this case mapping how orientations and achievement at any moment are embedded in ongoing trajectories shaped by systems of stratification that have meaningful consequences for lives. The configurations of responsibilities and resources at those moments affect school decisions largely because they shape the costs of attending, the probability of success, and the utility of completing a degree (Astone et al., 2000). We argue the same processes Bozick and colleagues identify in childhood and adolescence occur in the post-high school years as well. Higher SES young people are more likely to continue to receive feedback about their greater likelihood of completing bachelor’s degrees based on their higher academic achievement and better preparation for college studies while in secondary school. Higher achieving students are more likely to maintain expectations for a bachelor’s degree than their lower achieving peers both in the earliest years after high school as well as into the late 20s (Alexander et al., 2008; Trusty, 2000; Trusty and Harris, 1999). Young people who earn poorer grades as they start college or are directed toward remedial coursework, may be more likely to lower their expectations. This is evident in Clark’s (1960) classic work on the disjuncture between high aspirations and limited opportunities for success, where he argued that 2-year schools serve a ‘‘cooling out function’’ to lower aspirations and to direct blame toward the self and away from social institutions. But there are additional mechanisms that likely link SES and expectations in the post-high school years as well, beyond feedback about college worthiness. These mechanisms include nonacademic resources as well as competing risks for role changes that may detract from college pursuits. For those with greater socioeconomic resources, college simply remains more affordable and college-educated relatives provide important first-hand knowledge (Plank and Jordan, 2001; Roderick et al., 2008). Enrollment in 4-year schools, and in 2-year schools as well if for most of the year, is associated with maintaining expectations to earn a 4-year degree during the transition to adulthood, and helps explain why higher SES youth have more stably high expectations (Alexander et al., 2008). Continuous vs. interrupted college attendance also may play a role (Goldrick-Rab, 2006). Though a variety of education-related experiences are likely involved, the current study examines how enrollment in 2 and 4-year post-secondary institutions shapes expectations to earn a bachelor’s degree. We expect enrollment in both types of institution, but especially enrollment in 4-year institutions, facilitates the maintenance of expectations net of high school achievements. We further expect that enrollment patterns will help explain the relationship between SES and expectations to earn a bachelor’s degree. A life course approach also directs us to consider the multiple domains of life woven together over time (Alexander et al., 2008; Elder et al., 2003). Socioeconomic status strongly distinguishes other aspects of the life course pathways of young people, with those from lower SES backgrounds generally transitioning into adult roles earlier. That is, young people from lower socioeconomic status families marry, become parents and work full-time at earlier ages than their more advantaged counterparts (Dariotis et al., 2011; Sandefur et al., 2005). Lower socioeconomic status young adults may hold more volatile expectations or more frequently drop their plans through assuming adult roles that may conflict with that of student.1 Recent work by Andrew and Hauser (2011) shows that although educational expectations are not completely stable across the 8th, 10th, and 12th grades, expectations do not tend to change in response to test scores and are only responsive to very large changes in grades. They suspect that notable shifts in expectations are more likely to occur after other major life transitions that dramatically alter ‘‘information about and/or perceptions of future academic success’’ (p. 514). Roksa and Velez (2012) find that the post-secondary educational careers of young adults in the 1997 National Longitudinal Survey of Youth were thus affected by the timing of full-time employment and family formation after high school. Socioeconomically disadvantaged youth were more likely to delay their entry into college and had lower odds of completing a bachelor’s degree due to the combination of school, work, and family role responsibilities. These struggles related to major life course transitions may also be evident in diminished expectations to complete a bachelor’s degree. We expect marriage, parenthood, and full-time work to be negatively related, and part-time employment to be positively related, to holding expectations to earn a bachelor’s degree, and that differences in the rate of holding these adult social roles will help explain the link between SES and bachelor’s degree expectations. The life course perspective also draws attention to the significance that timing has for college pursuits, specifically whether those who postpone attempts to get a college degree are similarly constrained by socioeconomic origins as are those just out of high school. Elman and O’Rand (2004, 2007) document a pattern of cumulative (dis)advantage among adults 1 There is a limited literature on links between educational expectations and school attendance, family formation and work roles. Findings regarding these roles are mixed, perhaps because studies tend to examine one or two of these in isolation of the others, or vary in the age-range for which they capture these transitions (Alexander et al., 2008; Hanson, 1994; Jacob and Linkow, 2011; Uno et al., 2010). None has evaluated the extent to which socioeconomic background operates through employment and family formation role pathways to shape expectation trajectories.

M.K. Johnson, J.R. Reynolds / Social Science Research 42 (2013) 818–835

821

Fig. 1. Early and late phase of assessment, educational expectation trajectories and attainment.

who return to school at midlife. Their analyses of the National Survey of Families and Households show that SES-stratified resources, life course roles, and institutional sorting (i.e., earlier non-baccalaureate coursework) sustain the effects of SES on educational attainment in midlife. The ability of young adults to maintain expectations to complete a bachelor’s degree may be similarly conditioned by life course timing. The influence of SES on persistence may be even stronger in young adults’ late 20s as compared to just after high school, due to similar cumulative effects. Our study examines whether socioeconomic status is related to change and persistence in bachelor’s degree expectations during the transition to adulthood and the extent to which these expectation trajectories can then explain the higher attainments of higher socioeconomic status youth. This study also extends inquiry into how role pathways may prompt revision of educational expectations during the transition to adulthood by assessing the impact of school enrollment (at 2- and 4-year institutions), employment (part- and full-time), and family formation on bachelor’s degree expectations and the extent to which the more frequent and earlier transitions to these roles explains the impact of socioeconomic background on these expectations. In doing so, we address Alexander and colleagues’ call for greater attention to the ‘‘more immediate conditions and experiences’’ of life within, but also beyond, the educational arena (2008, p. 391). Our study offers two additional strengths. First, we draw on nationally representative samples of high school seniors to give a more complete picture of persistence and change in bachelor’s degree expectations. Past studies have either examined change generally rather than examining distinguishable trajectories (Jacob and Linkow, 2011) or focused on specific adolescent subgroups (Adelman, 2005; Alexander et al., 2008; Hanson, 1994; Trusty, 2000; Trusty and Harris, 1999). Second, we separate patterns of change and stability in bachelor’s degree expectations from degree attainment, allowing an examination of their interrelationship. In past studies, youth who meet their educational goals (assessed at one point in time) are distinguished from those who do not, who may have dropped their plans or held on to no avail. For youth who achieve their plans, there is no attention to the course of their expectations leading up to degree attainment. And for those who maintain their expectations for some period of time after high school, there is no attention to whether those expectations are later met.2 1.2. Analysis strategy We examine persistence and change in expectations to earn a 4-year degree, and the relationship between these expectation trajectories and bachelor’s degree attainment, separately for two phases of young adulthood (see Fig. 1). The early phase begins in the senior year of high school and extends 5–6 years later. Then, for those who have not earned a 4-year degree by this age, we extend our analysis to what happens in the next 5–6 years (ending at approximately age 29–30), the late phase. We first present descriptive findings on the prevalence of four types of bachelor’s degree expectation trajectories: stable high, stable low, unstable, and dropped plans. We then estimate models assessing the relationship between socioeconomic status and membership in the four trajectory categories, followed by models assessing both the relationship between the bachelor’s degree expectation trajectory categories and bachelor’s degree attainment and the extent to which this relationship helps explain the higher rate of bachelor’s degree receipt of higher SES youth. We analyze the two phases of young adulthood separately for several interrelated reasons. The majority of bachelor’s degrees are earned by the end of the early phase, which also captures the socially normative life course timing for earning a 4year degree. Our approach allows us to capture the processes involved for ‘‘late bloomers’’ and ‘‘slow goers,’’ and examine how socioeconomic status, persistent expectations, and attainment are interrelated in the later 20s as well as the early 20s. Recent studies demonstrate the importance of looking into the late 20s to see the full impact of bachelors’ degree expectations (e.g., Andres et al., 2007). It also recognizes that those who have earned bachelor’s degrees during the early phase are no longer at risk of holding expectations to earn a bachelor’s degree afterwards. We follow this examination of SES, bachelor’s degree expectation trajectories, and degree attainment with a closer look at whether competing commitments arising from work and family formation, and institutional sorting (2- vs. 4-year college 2

An exception to this is Bozick et al.’s (2010) study of expectation trajectories assessed up through high school.

822

M.K. Johnson, J.R. Reynolds / Social Science Research 42 (2013) 818–835

attendance) are linked to changes in bachelor’s degree expectations. To do so, we move away from the expectation trajectory categories, so as to be able to more clearly link the timing of holding high or low expectations to earn a bachelor’s degree with enrollment and movement into marriage, parenthood, and full-time employment. In addition to controlling for academic factors, which we have noted are associated with post-high school educational expectations and attainment, our analysis controls for race/ethnicity and gender. Studies show Black and other nonwhite young adults have more persistent bachelor’s degree expectations after high school than do whites (Alexander et al., 2008; Hanson, 1994; Trusty, 2000; Trusty and Harris, 1999). There are also racial/ethnic differences in educational attainment, with Blacks, and especially Hispanics, less likely than whites to earn 4-year degrees (Aud et al., 2010). Gender differences in expectations and attainment have changed dramatically over the past 40 years. Girls have caught up with and surpassed boys in their educational expectations in adolescence (Reynolds and Burge, 2008). Young women’s rates of post-secondary schooling have also risen substantially (DiPrete and Buchmann, 2006; Fitzpatrick and Turner, 2007), surpassing males’ in terms of enrollment around 1978 and in terms of BA recipients around the mid 1980s (Conger and Long, 2010). As such, in the cohorts we study (senior year classes of 1987–1990) there should be little to no gender differences.

2. Methods 2.1. Data The data for this research are from the Monitoring the Future (MTF) studies. Though the MTF lacks some of the subject breadth and depth of other national longitudinal studies of youth and education, it is the only national study that allows examination of educational expectations patterns throughout the 20s. The base MTF component involves repeated crosssectional surveys of US high school seniors carried out annually since 1976 (Johnston et al., 2007). The data were collected by the Institute for Social Research at the University of Michigan, using a multistage cluster sampling technique that included 125–140 public and private schools each year in order to obtain a nationally representative sample of high school seniors. Multiple questionnaire forms are used to expand the range of topics addressed, although the questions about educational plans, sociodemographic characteristics and academic measures we draw on in this study are common to all forms. From the original participants in each senior-year survey, a representative subsample of approximately 2400 was chosen to participate in a panel study. While the senior-year questionnaire was administered in school, questionnaires were mailed in the years after high school for the panel participants. Given the focus of the MTF on alcohol and drug use, the panel component oversampled illicit drug users and was stratified by gender (Bachman et al., 2001, p. 29). The data were weighted to adjust for unequal probabilities of selection resulting from the sampling procedures. In this study, we use data from the panel respondents’ senior year (Wave 1) and from six biennial follow-up surveys after high school (Waves 2 thru 7). For half of each cohort, the first follow up occurred 1 year after the base year; it occurred 2 years after the base year for the other half. For example, for participants from the class of 1990, a random half were surveyed in 1991, 1993, 1995, 1997, 1999, and 2001; the other half were surveyed in 1992, 1994, 1996, 1998, 2000, and 2002. The 11–12-year period from the senior year through the 6th follow-up (approximately age 29–30) covers the primary ages of post-secondary enrollment; around 75% of college students enrolled at degree-granting institutions were under age 30 during the time period covered by this study (Snyder et al., 2008: Table 172). It is a unique strength of the MTF panel study that we may follow youth so far into young adulthood, far exceeding the time span covered by most existing studies of expectations and later attainments. Our analyses include four senior-year cohorts, 1987–1990 (N = 9870), selected to represent the most recent cohorts for which we have data available up through the 6th follow-up. Combining four cohorts provides a sufficient sample size to allow analysis of less common trajectories without introducing too wide a span of history as to involve much social change.

2.2. Attrition and non-response There are two primary sources of missing responses in the MTF data. We addressed item non-response in the base year measures by using multiple imputation techniques. The reported regression estimates are the average coefficients from analyses of five different data sets with imputed values for item non-response on the independent variables. The ICE and MICOMBINE commands in Stata (Royston, 2004) generated the imputed data sets and averaged the coefficients across them that we report in the multivariate analyses below. The larger source of missing data is due to attrition from the panel. As described above, we examine the young adult years in two phases in our analyses. In each, we define attrition as those who we were unable to classify into a bachelor’s degree expectation trajectory based on the three relevant waves, and/or those who did not participate in the follow-up in which we assess bachelor’s degree attainment (5–6 years out for the early phase and 11–12 years out for the late phase). One exception to this is that when a respondent did not participate in these follow-ups but we knew from earlier ones that they had completed a bachelor’s degree, we carried this information forward to retain the respondent. So defined, the rate of retention was around 59% for the early phase sample and 51% for the late phase sample, and it dropped slightly across cohorts (from 62 to 55 and from 53 to 47, respectively). Logistic regression analyses of the odds of attrition indicated that it was higher for men,

M.K. Johnson, J.R. Reynolds / Social Science Research 42 (2013) 818–835

823

racial/ethnic minorities, more recent cohorts, and—of particular concern for our analyses—lower-SES respondents and those with lower expectations to earn a bachelor’s degree. We did not employ imputation techniques to deal with attrition given that imputed y-values can add ‘‘noise’’ to the slope estimates (von Hippel, 2007).3 Instead, the weighted analyses reported below have been adjusted by the inverse of the sampling weight as determined by the research design and the inverse of the predicted probability of being in the sample for the relevant period, as generated from the logistic regression analysis of attrition (see Clarke et al., 2009 for a similar approach using these data). The sample size for the early phase is 5730; for the late phase, in which we also restrict the analysis to those who had not earned a college degree within 5–6 years of their senior year, the sample size is 2788. The discrete-time event history models we use to examine the relationship between expectations and work, family, and schooling across the transition to adulthood are based on 25,560 person-years. 2.3. Measures 2.3.1. Dependent variables To measure bachelor’s degree expectations, the MTF asks high school seniors ‘‘How likely is it that you will do each of the following things after high school?’’ We analyzed the item on graduating from a 4-year college or university. Valid responses are ‘‘definitely won’t, probably won’t, probably will, and definitely will.’’ We dichotomized the responses such that those who think they probably or definitely won’t are distinguished from those who think they probably or definitely will.4 In later waves, we again distinguished those who think they probably or definitely won’t from those who think they probably or definitely will, but we also added a third category capturing those who had met their goal (earned a 4-year degree). An additional response category, ‘‘I’m doing this now,’’ was included in the response options for the follow up waves that did not appear in the senior year. We considered those reporting they were currently doing so not yet having completed a bachelor’s degree, but continuing to expect they will. We classified respondents into four types of bachelor’s degree expectation trajectories for each phase, first based on the first three waves of data (i.e., the 12th grade base year to 3–4 years later), and then, for those who had not yet earned a 4-year degree, based on waves 4 through 6 (5–6 years after the senior year to 9–10 years after). We required non-missing codes in at least two of the three relevant follow-ups for a case to be successfully classified. For those respondents who earned their 4-year degree during the period, we classified them based on their expectations up to the point of earning their degree. The stable high trajectory includes respondents who only reported expecting to earn a 4-year degree. Likewise, the stable low trajectory includes respondents who only reported not expecting a 4-year degree. The unstable trajectory includes respondents who vacillated in their expectations (with two changes) or who ‘‘warmed up’’ (did not expect to earn a 4-year degree at the beginning of the phase, but did so by the end of the phase). Finally, the dropped plans trajectory includes those who began the phase expecting to earn a 4-year degree, but dropped that plan by the end of the phase). These trajectories are the dependent variable in one set of analyses, in which we estimated multinomial logit models to predict membership in each trajectory category. In our analysis of bachelor’s degree attainment, they became independent variables, coded as a set of dummy variables. Stable low expectations for a bachelor’s degree is the reference category in those analyses, but we tested differences between the other categories and report those in the text as well. We assessed bachelor’s degree receipt at the end of each phase of young adulthood (5–6 years beyond the senior year and 11–12 years beyond the senior year). We classified respondents as having earned a 4-year degree or not based on their reports of the highest degree they earned. 2.3.2. Independent variables We measure socioeconomic status in terms of parents’ educational attainment, family size, and family structure as assessed in the senior year survey. Though the MTF does not gather information on income, wealth, or parents’ occupational attainment, parents’ education is arguably the most important dimension to consider when examining educational expectations and attainments (Roksa and Potter, 2011, pp. 304–305). We tested several alternative measurement approaches, and the best-fitting approach was to combine parents’ education and their presence in the household, and also include an indicator of family size. Specifically, our models designate as the reference group those who were living with two parents

3 Concerns over the amount of attrition across follow-ups made us consider the possibility of including imputed values of the dependent variable in the analyses, as opposed to weighting the data by the inverse probability of remaining in the sample. The standard practice in this body of literature is to impute missing values on just the predictor variables (Andrew and Hauser, 2011; Reynolds and Johnson, 2011; Roksa and Velez, 2012; Sabates et al., 2011). We assessed the sensitivity of our results to these data considerations by re-estimating all of the models with imputed values of the dependent variable that would otherwise be missing due to attrition (results available upon request). In general, imputing values of the dependent variable led to smaller standard errors for the slope estimates that resulted in several non-significant associations becoming just significant (e.g., p < .04). In some cases, including imputed values of the dependent variable also changed the magnitude of the slope, resulting in a difference that appears even larger or smaller when comparing odds ratios. We detail these differences in notes 8 and 11, and stress that these results should be interpreted more cautiously. 4 Because we are attempting to chart patterns across multiple time points, this dichotomization was a necessary simplification strategy for identifying patterns as well as determining whether expectations were met or not. We also dichotomize this measure in recognition of the problems of interpersonal comparability for specific responses like ‘‘probably will’’ (Manski, 2004, p. 1338).

824

M.K. Johnson, J.R. Reynolds / Social Science Research 42 (2013) 818–835

Fig. 2. Prevalence of expectation trajectories by phase, high school senior classes 1987–1990.

in their senior year, and whose parents had both completed a high school degree or less. Against this group we compare seven other categories: (1) two parents and one or both have some college but no college degree5; (2) two parents and at least one completed college but did not get a graduate or professional degree; (3) two parents and at least one did get a graduate or professional degree; (4) living with one parent who had a high school degree or less; (5) one parent with some college; (6) one parent who completed a college or a graduate or professional degree; and (7) living with neither parent. This strategy of looking at combinations of parents’ education and family structure is consistent with Martin’s (2012) recent study indicating a differing impact of parental education on youth’s educational outcomes across single and two-parent household structures. We also measure the number of siblings in the household, which ranges from 0 to 6 or more (topcoded at 6). Cohort is based on the year the participant was a senior in high school and is captured by a set of dummy variables with 1987 as the reference. Sex was selfreported. Self-reported race/ethnicity distinguishes Hispanics, non-Hispanic blacks, non-Hispanic whites, and those of other race/ethnicities. High school achievement was measured by grade point average on a nine-point scale from 1 = D or below to 9 = A. Curriculum track distinguished those who self-reported being in a college preparatory curriculum from those in other tracks. Descriptive statistics on the study sample are in Appendix A. To take into account the role transitions that may lead young people to adjust their expectations, we created measures for family formation, employment, and college enrollment. In terms of family formation, respondents were classified as married or not and parents or not (including step-children) based on their current status in each follow-up. We classified respondents’ employment status as working full-time (35 or more hours per week), part-time (1–34 h per week), or not working based on the work hours they reported for March of each survey year. Finally, based on reports of college enrollment at the time of each survey, we classified respondents as currently attending a 2-year college or not, and attending a 4-year college/university or not. 3. Results 3.1. Bachelor’s degree expectation trajectories The prevalence of the bachelor’s degree expectation trajectories during each phase in young adulthood are shown in Fig. 2. We begin the presentation of our findings focused on the early phase, the ages most traditionally associated the pursuit of higher education and most often considered in past research (bars on the left in each pair). Stable high expectations were held by 61.6% of the sample during the early phase. Less than 15% held stable low expectations for earning a 4-year degree. Approximately 12% of the sample fell into the unstable plans trajectory, and a similar percentage dropped their plans over time. Our findings complement past work documenting high expectations among high school students by showing that a large majority of young people have and hold onto bachelor’s degree expectations, at least through the ages traditionally associated with college-going. For those who did not earn a 4-year degree within 5–6 years after their senior year of high school, the prevalence of stable high expectations to do so is lower, yet still substantial at approximately 41%.6 Around 35% held stable low expectations. Put differently, this means that almost two-thirds of young adults without bachelor’s degrees by age 23–24 held expectations to earn one at some time in their mid to late 20s. As before, the group with unstable plans and those who dropped their plans each represented about 12%. Clearly, holding on to bachelor’s degree expectations is not limited to the years immediately after high school. For those without a 4-year degree by age 23–24, a substantial proportion held onto expectations to earn one into their late 20s. To examine the extent to which socioeconomic status influenced young people’s expectation trajectories, we estimated a multinomial logit model. As shown in Tables 1 (for the early phase) and 2 (for the late phase), socioeconomic status was strongly related to stability vs. change in young people’s bachelor’s degree expectations, even controlling high school curriculum track and grades, among other factors. Moreover, it matters even in the later 20s, after the normative years of bachelor’s degree seeking have passed. With eight SES categories and four trajectory categories there are many contrasts that can be made. We give primary attention to the stable high trajectory, which is particularly important for college attendance (Bozick et al., 2010) and, as we later show, earning a bachelor’s degree. We also omit the results for ‘‘dropped plans’’ vs. 5 6

The questionnaire items on parents’ education did not distinguish between 2- and 4-year ‘‘college’’ degrees. Of those in the stable high group during the late phase, 78% also held stable high expectations during the earlier phase as well.

825

M.K. Johnson, J.R. Reynolds / Social Science Research 42 (2013) 818–835

Table 1 Multinomial logistic regression models predicting expectation trajectories in early phase of young adulthood. Odds ratios (z-scores). High school senior classes of 1987–1990 (N = 5730).

Socioeconomic status Two parents, 1 + some collegea Two parents, 1 + collegea Two parents, 1 + grad/prof degreea One parent, high school or lessa One parent, some collegea One parent, college+a Neither parent in homea

[1]

[2]

[3]

[4]

[5]

Stable high vs. stable low

Stable high vs. unstable

Stable high vs. dropped plans

Unstable vs. stable low

Dropped plans vs. stable low

2.56*** (6.10)

1.68** (3.36)

1.33 (1.84)

1.53* (2.41)

1.92** (3.43)

4.62*** (9.83) 9.08*** (9.42)

2.77*** (6.69) 5.30*** (8.26)

1.99*** (4.37) 2.75*** (5.65)

1.67** (2.82) 1.71 (1.86)

2.33*** (4.68) 3.30*** (4.41)

1.82** (3.27)

1.50* (2.00)

1.12 (.58)

1.21 (.93)

1.63* (2.36)

3.59*** (3.60) 5.53*** (6.00) 1.21 (.78)

1.20 (.65) 3.03*** (4.35) .87 ( .54)

1.11 (.38) 2.58** (3.47) .91 ( .36)

2.98** (2.89) 1.82 (1.80) 1.38 (1.35)

3.24** (3.06) 2.14* (2.14) 1.33 (1.04)

.90** ( 3.05) 1.06 (.59) 1.18 (.85) 1.07 (.36) 1.38 (1.38) 4.91*** (13.70)

.98 ( .78) .85 ( 1.57) 1.50* (1.96) 1.28 (1.20) 1.62 (1.88) 3.05*** (10.57)

1.01 (.31) .78* ( 2.08) 2.53*** (3.85) 1.96** (2.93) 1.18 (.62) 2.46*** (5.98)

.93 ( 1.92) .97 ( .23) 2.00** (2.64) 1.64* (1.97) 1.01 (.02) 3.96*** (9.47)

1.31*** (9.01)

1.24*** (7.47)

1.04 (1.11)

1.10** (2.65)

Demographic and academic controls Number of siblings .91** ( 3.23) Femaleb .83 ( 1.78) African Americanc 2.99*** (4.75) Hispanicc 2.10*** (3.50) Other race/ethnicityc 1.63* (2.15) College prep high school 12.06*** (20.23) programd High school grades 1.36*** (10.19)

Notes: The model also included a dichotomous indicator of whether the first follow-up occurred 1 or 2 years after the senior year and dummy variables representing senior year cohort. Reference categories for dummy variables include (a) two parents, both with high school or less education; (b) male; (c) non-Hispanic white; and (d) general or vocational/tech program. Wald Chi-Square (54) = 1236.58; Pseudo-R2 = .18. * p < .05. ** p < .01. *** p < .001.

‘‘unstable plans’’ as there were very few factors that significantly differentiated between these two trajectories (results available upon request). Generally, in the early phase, those who lived in a two-parent family where at least one parent had a graduate/professional degree were the most likely to hold high stable expectations to earn a bachelor’s degree over any other trajectory (columns 1–3, Table 1), followed by living with a single parent who had a college degree (or more) or with two parents where at least one had a college degree. Young people from these three SES categories, in which at least one parent had a college degree or more, were the most likely to hold stable high expectations in the late phase as well (columns 1–3, Table 2). Education levels of parents contributed to young people’s stable high expectations within single parent and two-parent families, and at comparable parent education levels, young people from single-parent families were as likely or more likely than those from two-parent families to hold stable high expectations over other trajectories. Persistent plans for a bachelor’s degree in the transition to adulthood, even in the later 20s (for those who did not earn bachelor’s degrees in the first 5–6 years after leaving high school), thus could have served as a mechanism linking family socioeconomic status and bachelor’s degree attainment. Net of other controls, family size also predicted bachelor’s degree expectation trajectories. In the early phase alone, young people from larger families were less likely to hold stable high expectations compared to holding stable low or unstable expectations, and were less likely to drop their plans than hold unstable ones. Differences across parent SES categories were strongest in contrasting stable high trajectories with other trajectories, though they also distinguished whether young people held unstable plans or dropped their plans over holding stable low expectations to earn a bachelor’s degree (columns 4 and 5). Higher parental education, whether in a two-parent or single-parent family, generally predicted dropping plans or having unstable plans over holding stable low expectations to earn a bachelor’s degree in the early phase. (While the magnitude of the coefficient for a single parent with some college is larger than for a single parent with a college degree or more in both of these comparisons, the difference is not statistically significant). Thus, young people from families with more educated parents were unlikely to consistently not plan on a 4-year degree for themselves. The control variables were also related to the expectation trajectories of young people in important ways during both phases of young adulthood. Minorities, especially Blacks and Hispanics, were disproportionately overrepresented in stable high vs. stable low and dropped plans trajectories, and in general were less likely to be in the stable low trajectory vs. other trajectories compared to whites. This is consistent with the findings of previous studies showing greater persistence in bachelor’s degree expectations among nonwhites (Alexander et al., 2008; Hanson, 1994; Trusty, 2000; Trusty and Harris, 1999), persistence which may be partly due to black young adults’ greater tendency to attend college, net of family SES, academic achievement, and expectations to go to college (Bennett and Xie, 2003; cf. Roksa et al., 2007). Previous studies have also

826

M.K. Johnson, J.R. Reynolds / Social Science Research 42 (2013) 818–835

Table 2 Multinomial logistic regression models predicting expectation trajectories, late phase of young adulthood. High school senior classes of 1987–1990. Odds ratios (N = 2788; limited to those who had not completed a degree within 5–6 years of high school).

Socioeconomic status Two parents, 1 + some collegea Two parents, 1 + collegea Two parents, 1 + grad/prof degreea One parent, high school or lessa One parent, some collegea One parent, college+a Neither parent in homea Number of siblings

[1]

[2]

[3]

[4]

[5]

Stable high vs. stable low

Stable high vs. unstable

Stable high vs. dropped plans

Unstable vs. stable low

Dropped plans vs. stable low

1.56** (2.73)

1.05 (.19)

1.56 (1.90)

1.49 (1.75)

1.00 ( .02)

**

**

2.01 (3.03) 1.65 (1.60)

1.45 (1.64) 1.11 (.30)

1.21 (.70)

1.52 (1.78)

.84 ( .44) 2.28* (2.17) .81 ( .62) 1.00 ( .02)

1.40 (.63) 3.17** (2.61) 1.23 (.60) 1.01 (.12)

2.67* (2.24) 2.68* (2.29) 1.31 (.85) .97 ( .56)

.87 ( .84) .99 ( .05) 1.12 (.41) .98 ( .08) 2.70*** (6.05)

.80 ( 1.49) 1.19 (.65) 1.76 (1.91) 1.69 (1.53) 1.93*** (4.06)

1.16 (.97) 3.43***(4.00) 2.94*** (3.77) 1.49 (1.19) 1.41* (2.05)

1.27 (1.61) 2.86*** (3.48) 1.86* (2.03) .86 ( .43) 1.98*** (4.22)

1.10* (2.12)

1.13** (2.64)

1.02 (.40)

***

2.73 (5.81) 3.75*** (5.97)

1.36 (1.36) 2.28** (2.94)

1.36 (1.64)

1.12 (.42)

*

2.23 (2.20) 6.11*** (5.84) 1.06 (.21) .97 ( .91)

Demographic and academic controls 1.02 (.16) Femaleb African Americanc 3.38*** (4.90) Hispanicc 3.28*** (5.28) Other race/ethnicityc 1.45 (1.43) College prep high school 3.82*** (11.69) d program High school grades 1.12** (3.42)

1.59 1.92 .86 .96

(.91) (1.70) ( .39) ( .79)

1.88 (2.74) 3.37*** (3.88) .90 ( .45)

.99 ( .19)

Notes: The model also included a dichotomous indicator of whether the first follow-up occurred 1 or 2 years after the senior year and dummy variables representing senior year cohort. Reference categories for dummy variables include (a) two parents, both with high school or less education; (b) male; (c) non-Hispanic white; and (d) general or vocational/tech program. Wald Chi-Square (54) = 413.46; Pseudo-R2 = .09. * p < .05. ** p < .01. *** p < .001.

found minorities’ expectations to be more strongly shaped by family members’ encouragement and less by academic factors compared to non-Hispanic whites’ expectations (Cheng and Starks, 2002; Alexander et al., 1994; Qian and Blair, 1999). For the most part there were no gender differences in expectation trajectories, though males were more likely than females to hold unstable expectations compared to stable low expectations in the early phase. The academic factors, like SES, distinguished membership across all trajectories during both phases of young adulthood. Being in a college preparatory track was especially important for sustaining bachelor’s degree expectations over any other trajectory. The considerable influence of curricular program may reflect their role in signaling to students in AP and honors classes the certainty of their getting a bachelor’s degree, as well as conveying to students in the general classes their poorer prospects. 3.2. Educational attainment These expectation trajectories, especially having stable plans to earn a bachelor’s degree, were highly predictive of earning a bachelor’s degree. As shown in Table 3 (section on left), 35% of the sample as a whole had earned a 4-year degree within 5–6 years of their senior year, and among those who did not, just over 24% eventually did within 11–12 years of their senior year (section on right). Over half of those who held stable high expectations for 4-year degrees during the early phase had Table 3 Educational attainment overall and by expectation trajectories, high school senior classes of 1987–1990. % with BA at the end of early phase of young adulthood Overall Expectation trajectorya Stable low Stable high Unstable Cool down N a

% with BA at the end of late phase of young adulthoodb

35.02

24.27

1.75 54.33 5.70 5.16

2.33 56.23 3.73 1.85

5730

2788

Expectation trajectory is based on the first three data collections within each phase of young adulthood, and attainment is assessed as of the fourth. b Analyses in the late phase are limited to those without a 4-year degree within the first 5–6 years after the senior year of high school.

Table 4 Logistic regression models of 4-year degree attainment, high school senior classes of 1987–1990. Odds ratios (z-scores). Early phase of young adulthood [1]

[2]

Late phase of young adulthood [3]

[1]

[2]

[3]

[4]

1.44** (2.95) 2.34*** (7.81) 3.28*** (10.02) .98 ( .12) 1.19 (.75) 2.00*** (4.08) .81 ( .84) .87*** ( 5.97)

1.16 (1.10) 1.65*** (4.15) 2.12*** (6.00) .83 ( .97) .94 ( .25) 1.29 (1.38) .81 ( .73) .88*** ( 4.74)

1.15 (1.05) 1.64*** (4.08) 2.10*** (5.95) .82 ( 1.01) .93 ( .31) 1.27 (1.32) .80 ( .75) .88*** ( 4.75)

1.60** (2.75) 2.58*** (5.73) 4.53*** (7.65) 1.19 (.81) 1.95 (1.88) 2.53*** (3.74) 1.14 (.42) .95 ( 1.46)

1.40 (1.57) 1.86** (3.05) 3.02*** (4.30) .94 ( .23) 1.61 (1.20) 1.28 (.86) 1.28 (.60) .94 ( 1.05)

1.23 (.95) 1.66* (2.44) 2.46*** (3.64) .80 ( .83) 1.40 (.79) 1.11 (.36) 1.08 (.18) .93 ( 1.23)

1.27 (1.16) 1.72** (2.62) 2.74*** (4.12) .85 ( .68) 1.56 (1.08) 1.22 (.72) 1.11 (.25) .94 ( 1.15)

Demographic and academic controls Femaleb African Americanc Hispanicc Other race/ethnicityc College prep HS programd High school grades

.96 ( .90 ( .47*** .75 ( 5.23*** 1.52***

1.00 (.04) .71 ( 1.79) .37*** ( 4.81) .62** ( 2.81) 2.95*** (10.95) 1.45*** (14.85)

1.00 (.03) .70 ( 1.83) .37*** ( 4.78) .62** ( 2.88) 2.91*** (10.73) 1.45*** (14.83)

.83 ( 1.66) 1.03 (.12) 1.08 (.36) 1.86** (2.60) 2.94*** (9.36) 1.24*** (6.49)

.82 ( 1.39) .57* ( 2.05) .59* ( 2.18) 2.01* (2.25) 1.71*** (3.68) 1.23*** (5.15)

.86 ( 1.07) .58* ( 2.01) .54* ( 2.37) 2.09* (2.34) 1.48* (2.55) 1.21*** (4.55)

.85 ( 1.23) .66 ( 1.50) .61 ( 1.87) 1.59 (1.62) 1.58** (3.14) 1.19*** (4.25)

45.91*** (16.01) 1.69 (1.20) .77 ( .56)

35.20*** (14.49) 1.09 (.21) .60 ( 1.04) 2.87*** (4.75)

.51) .56) ( 4.00) 1.77) (18.29) (18.07)

Expectation trajectory Stable highe Unstablee Dropped planse Senior yr. plans Stable high (W1-6)f Unstable (W1-6)f Dropped plans (W1-6)f Wald Chi-Square (df) Pseudo R2

1082.34*** (18) .28

N

5730

24.79*** (9.33) 2.79** (2.66) 1.48 (1.02)

998.49*** (21) .38 5730

9.85*** (4.49) 1.88 (1.49) .59 ( .98) 2.54* (2.46)

985.13*** (22) .38 5730

315.17*** (18) .16 2788

582.04*** (21) .44 2788

608.58*** (22) .45 2788

65.75*** (6.91) 6.58** (3.05) 1.97 (1.04) 609.02*** (21) .38 2.788

Notes: Cases for the late phase limited to those who had not earned a 4-year degree within 5–6 years after the senior year. Each model also included a dichotomous indicator of whether the first follow-up occurred 1 or 2 years after the senior year and dummy variables representing senior year cohort. Reference categories for dummy variables include (a) two parents, both with high school or less education; (b) male; (c) nonHispanic white; (d) general or vocational/tech program; (e) stable low and (f) stable low (W1-6). * p < .05. ** p < .01. *** p < .001.

M.K. Johnson, J.R. Reynolds / Social Science Research 42 (2013) 818–835

Socioeconomic status Two parents, 1 + some collegea Two parents, 1 + collegea Two parents, 1 + grad/profa One parent, high school or lessa One parent, some collegea One parent, college+a Neither parent in homea Number of siblings

827

828

M.K. Johnson, J.R. Reynolds / Social Science Research 42 (2013) 818–835

earned one within 5–6 years after the senior year, far out-pacing those from any other trajectory. Though the number of young people who held stable high expectations during the late phase of young adulthood was much smaller than in the early phase, it is remarkable how similar the rate of success was for those who did (54.33 in the early phase; 56.23 in the late phase). Even more remarkable are the extremely low completion rates for any other trajectory – faltering or late blooming expectations rarely resulted in a bachelor’s degree. In part, this suggests that young adults who initially planned to get a degree only let go of those plans once the prospective chances of degree completion approach zero. To examine these patterns in a multivariate context, we estimated logistic regression models for 4-year degree attainment. We estimated a series of models for each phase of young adulthood (see Table 4, Models 1–3). The first included socioeconomic status along with controls for baseline characteristics. The second added dummy variables representing the bachelor’s degree expectation trajectories to assess the relationship between the trajectories and attainment as well as to see whether the effects of SES are reduced. The third model added senior year expectations to earn a 4-year degree to evaluate whether the expectation trajectories continued to predict educational attainment above and beyond expectations from the senior year—a critical point in decision making and the measure of expectations used in most prior research. This model enables an assessment of whether the impact of the trajectories represents more than just leaving high school with high expectations. Consistent with a large body of previous evidence, Model 1 demonstrates that SES had a strong effect on young people’s own educational attainment, even controlling for academic factors.7 An important and new finding of this analysis, however, is that the benefits of parental SES also are substantially mediated by the over-time trajectories of educational expectations. That is, the estimated effects of SES decreased substantially once expectation trajectories were included in Model 2. Past research has focused on the mediating role of adolescent expectations only, which do not necessarily signify persistent expectations through young adulthood. The logits for having lived with two parents with some college, a college degree, or a graduate degree (compared to having lived with two parents with a high school degree or less) were reduced by 37–59% between Models 1 and 2 in the early phase, and 27–35% in the late phase. An ever greater reduction was observed for having lived with one parent who had earned a college degree or more, where there was a 63% change in the logit in the early phase and a 73% change in the logit in the late phase. The considerable mediation of the influences of SES is consistent with the sizable effects of SES on the odds of being in stable high trajectories and the dramatic influence on attainment of maintaining stable bachelor’s degree expectations over any other trajectory, as seen in Tables 1–3. Those with stable high expectations were by far the most likely to earn a 4-year degree. For example, compared to consistently reporting no bachelor’s degree expectations, holding high stable plans to do so increased the odds of earning a 4year degree by a factor of over 25 in the early phase and over 45 in the late phase.8 Unstable plans also increased attainment somewhat in the early phase relative to having stable low expectations, but it is clear from Table 3 that only those with stable high plans had appreciable rates of degree attainment. Do post-secondary trajectories of expectations tell us significantly more than just asking high school seniors what they expect to achieve? Model 3 demonstrates that holding stable high expectations to earn a bachelor’s degree greatly mattered, even with senior year expectations controlled. For example, controlling senior year plans, those with stable high expectations over the next 3–4 years had higher odds of earning a 4-year degree within 5–6 years of their senior year by a factor of almost 10 compared to those who never expected to earn one. In the late phase, those with stable high expectations still had higher odds of earning a 4-year degree by the end of their 20s by a factor of over 35 compared to those with stable low expectations across the previous 4 years. For the late phase only, we also estimated a fourth model with expectation categories defined over the full period since high school instead of only during the late phase itself. Here, holding stable high plans, which meant consistently planning to earn a bachelor’s degree across six measurements (from the senior year until 9–10 years later) increased the odds of completing a bachelor’s degree by a factor of over 65 compared to those with stable low plans. Holding unstable plans also increased the odds of earning a bachelor’s degree by over 6 compared to never planning one. This longer-term trajectory explains considerably less variance in bachelor’s degree attainment, however, than the expectation trajectories considered for the late phase years alone.9 While the control variables are not of key interest in these models, the pattern of racial/ethnic differences across models is worthy of note. In both the early and late phases, the Black–white and Hispanic–white differences in the odds of degree completion grow after accounting for racial differences in the distribution of expectation trajectories, that is, the odds ratio

7 The effects of socioeconomic background for the early and late phases are not directly comparable, as they are estimated for different samples (i.e., the late phase analysis is performed only on those who had not already earned a bachelor’s degree by wave 4 of the study). 8 Model estimates based on multiple imputation of the outcome measure as well as the predictors were quite different than those in Table 4, though still pointing to the large influence of stable high plans on degree attainment. Because the rates of degree attainment were estimated to be a little higher among those in the other trajectory categories (11% for stable low, 14% for unstable, and 17% for cooled down) with imputed y values, the logits for stable high in these models were smaller (1.77 vs. 3.21 for the early phase; 1.78 vs. 3.83 in the late phase). 9 In analyses not shown, we pursued several other strategies for assessing the effects of expectation trajectories across the entire study (senior year through the following 9–10 years) on educational attainment as of the final follow-up. These included adding both sets of trajectories to the model (early-stable, laterstable, early-unstable, etc.), and using a single measure equal to the proportion of waves in which the respondent reported expecting to earn a 4-year degree. This last measure was strongly related to attainment and did surprisingly well in terms of explaining variance in educational attainment. Yet, the results in Table 3 suggest that persistence is the key, and that lowering expectations for just a single year (as unstable or dropped plans) is associated with dramatically lower odds of degree completion.

829

M.K. Johnson, J.R. Reynolds / Social Science Research 42 (2013) 818–835 Table 5 Family, work, and school roles by years after high school; overall and by parents’ education and living arrangements. Years beyond senior year of high school: 1–2

3–4

5–6

7–8

9–10

11–12

Percent married Overall (%) Two parents, both high school or less Two parents, 1 + some college Two parents, 1 + college Two parents, 1 + grad/prof One parent, high school or less One parent, some college One parent, college+ Neither parent in home

5.8 7.8 5.3 3.6 2.1 9.2 7.5 3.7 16.3

14.0 20.1 13.9 9.9 6.3 18.2 10.2 5.6 33.3

25.9 32.5 29.6 22.5 15.7 27.6 22.3 14.7 43.7

39.5 46.8 41.5 37.6 30.9 40.4 32.3 25.2 52.1

50.5 54.9 55.0 50.5 44.8 50.5 45.1 36.8 56.7

59.3 65.4 61.8 58.5 55.4 56.3 53.4 49.8 64.1

Percent have child Overall (%) Two parents, both high school or less Two parents, 1 + some college Two parents, 1 + college Two parents, 1 + grad/prof One parent, high school or less One parent, some college One parent, college+ Neither parent in home

5.0 5.0 3.7 2.7 1.6 11.6 5.8 5.0 18.4

12.0 15.7 10.7 7.3 4.7 19.2 13.4 7.9 34.9

19.2 26.4 17.9 12.4 7.0 30.2 22.3 13.1 43.8

29.1 37.6 31.4 21.4 12.4 39.1 29.9 18.9 59.5

37.6 47.0 40.9 30.0 22.3 47.4 36.1 24.8 64.0

48.0 59.6 52.0 41.7 31.4 56.3 43.7 34.5 70.1

Percent working full-time Overall (%) Two parents, both high school or less Two parents, 1 + some college Two parents, 1 + college Two parents, 1 + grad/prof One parent, high school or less One parent, some college One parent, college+ Neither parent in home

30.2 40.6 31.3 21.9 14.3 39.5 39.0 25.1 48.0

40.1 50.7 42.2 32.7 22.1 49.9 53.6 35.3 56.1

62.2 65.2 68.0 61.6 54.7 62.4 63.3 60.3 57.7

74.1 74.0 77.1 78.9 68.0 75.0 74.4 71.9 66.5

78.2 77.2 80.1 81.7 77.1 80.3 71.9 77.5 70.7

77.1 75.7 78.3 80.2 76.5 76.9 74.3 79.7 69.2

Percent working part-time Overall (%) Two parents, both high school or less Two parents, 1 + some college Two parents, 1 + college Two parents, 1 + grad/prof One parent, high school or less One parent, some college One parent, college+ Neither parent in home

37.2 35.3 38.9 43.0 35.4 36.2 37.5 35.5 25.2

34.2 29.8 35.9 38.6 39.3 32.0 26.6 36.7 20.3

21.7 18.4 19.8 22.1 26.9 23.1 25.0 25.0 16.3

13.5 12.0 11.6 11.6 18.0 12.1 14.5 16.2 16.3

11.2 11.2 10.5 9.5 13.0 10.2 11.0 11.7 15.6

10.8 10.9 11.0 9.4 11.1 9.6 11.1 9.2 17.3

Percent enrolled 2-year Overall (%) Two parents, both high school or less Two parents, 1 + some college Two parents, 1 + college Two parents, 1 + grad/prof One parent, high school or less One parent, some college One parent, college+ Neither parent in home

18.7 19.3 21.7 20.8 11.7 20.5 24.8 14.2 18.0

10.3 10.4 11.8 11.0 6.1 10.2 12.9 12.3 11.5

6.9 7.8 6.9 7.7 3.7 6.8 6.3 6.9 7.1

5.0 5.1 5.2 4.3 4.2 6.0 8.5 3.0 6.5

3.3 3.9 3.5 2.4 1.8 4.9 6.6 3.4 4.0

2.9 2.7 2.3 2.7 1.8 4.8 3.9 2.5 5.8

Percent enrolled 4-year Overall (%) Two parents, both high school or less Two parents, 1 + some college Two parents, 1 + college Two parents, 1 + grad/prof One parent, high school or less One parent, some college One parent, college+ Neither parent in home

43.0 28.4 40.8 52.0 70.0 28.5 35.3 52.5 18.8

40.6 26.8 40.2 50.1 62.9 25.9 34.1 49.6 17.9

14.7 11.4 14.4 16.7 18.7 10.9 16.9 20.1 10.2

7.0 6.3 7.2 8.4 6.6 4.9 8.9 8.5 6.4

4.2 4.3 4.6 5.3 2.2 3.8 5.8 2.6 4.0

3.2 3.4 2.7 2.5 2.5 4.8 4.9 3.5 3.3

moves further away from 1.0 when comparing Model 1 to Models 2 and 3. This pattern suggests that holding stable high expectations is one way minorities did better than what would be expected based on their socioeconomic status and their

830

M.K. Johnson, J.R. Reynolds / Social Science Research 42 (2013) 818–835

Table 6 Discrete-time event history models for bachelor’s degree expectations. High school senior classes of 1987–1990. Odds ratios (z-scores). N = 25,560 personyears. [1] Socioeconomic status Two parents, 1 + some collegea Two parents, 1 + collegea Two parents, 1 + grad/prof degreea One parent, high school or lessa One parent, some collegea One parent, college+a Neither parent in homea Number of siblings Time and roles Wave Marriedb Parentc Employed part-timed Employed full-timed Attending 2-year college/universitye Attending 4-year college/universitye Wave  Married Wave  Attending 2-year college/university Wave  Attending 4-year college/university Demographic and academic controls Femalef African Americang Hispanicg Other race/ethnicityg College prep high school programh High school grades Constant Wald Chi-Square (df) Pseudo R2

[2] 1.51*** (7.05) 2.02*** (12.55) 2.83*** (14.35) 1.23** (3.40) 1.63*** (4.60) 2.37*** (9.50) 1.08 (.74) .97** ( 2.62) .72*** ( 32.51)

.96 ( 1.17) 2.05*** (11.00) 1.77*** (8.57) 1.33*** (3.84) 3.25*** (32.79) 1.13*** (12.37) 1.06 (.66) 3790.75*** (19) .17

[3] 1.45*** (5.48) 1.71*** (8.04) 2.08*** (8.79) 1.31*** (4.14) 1.55*** (3.71) 2.07*** (7.28) 1.26* (2.28) 1.00 ( .17)

.88*** ( 10.45) .82*** ( 4.43) .72*** ( 6.76) 1.00 (.07) .88* ( 2.33) 4.35*** (24.32) 22.38*** (34.16)

1.00 ( .12) 2.16*** (11.33) 2.00*** (10.28) 1.30** (3.16) 2.29*** (20.94) 1.06*** (5.73) .48*** ( 6.79) 3712.73*** (25) .30

1.45*** (5.46) 1.72*** (8.04) 2.08*** (8.76) 1.31*** (4.13) 1.54*** (3.66) 2.09*** (7.34) 1.27* (2.31) 1.00 ( .24) .84*** ( 10.95) .52*** ( 5.16) .72*** ( 6.70) 1.01 (.09) .88* ( 2.36) 2.78*** (6.77) 14.04*** (10.55) 1.10*** (3.76) 1.13** (3.05) 1.14 (1.85) 1.00 ( .04) 2.17*** (11.28) 2.01*** (10.28) 1.30** (3.19) 2.30*** (20.95) 1.07*** (5.81) .56*** ( 5.02) 3831.99*** (28) .30

Notes: Each model also included a dichotomous indicator of whether the first follow-up occurred 1 or 2 years after the senior year and dummy variables representing senior year cohort. Reference categories for dummy variables include (a) two parents, both with high school or less education; (b) not married; (c) has no children; (d) not employed; (e) not attending college; (f) male; (g) non-Hispanic white; and (h) general or vocational/tech program. * p < .05. ** p < .01. *** p < .001.

academic histories. Stable trajectories may thus underlie some of the observed racial and ethnic gaps in college completion rates among those who matriculate.

3.3. Young adult roles and changing expectations During the decade or so following high school, young people may revise their plans to obtain a bachelor’s degree as they encounter competing demands related to family or paid work, fail to get into or make the grade in bachelor’s degree-granting institutions. To the extent that parental SES shapes the likelihood and/or the timing of these experiences, then they may partly explain SES variations in expectation trajectories and, ultimately, degree attainment. We next consider marriage, parenthood, full-time and part-time employment, and enrollment at 2-year and 4-year institutions in this regard. We focus on the years beyond high school, since few high school seniors were employed full-time or enrolled in colleges/universities and even fewer (1% or less) were married or had children. Family socioeconomic influences are apparent in the family, work and school roles of respondents in the transition to adulthood (see Table 5). Young adults from two parent families where at least one parent had a graduate/professional degree postponed marriage, parenthood, and full-time work the most, and went to 4 year schools at the highest rates in the first 3– 4 years after high school. Those with a college-educated parent, whether from a one- or two-parent home, were next in terms of postponing family formation and attending 4-year schools. Young people from two parent families in which at least one parent had a graduate degree were also particularly less likely to attend 2-year schools in the first couple of years after high school. In contrast, young people with less educated parents more quickly entered marriage, parenthood, and full-time work. Those not living with either parent in their senior year of high school were especially likely to be married and parents, particularly in the early phase; this earlier family formation may actually predate their senior year living arrangements for some.

M.K. Johnson, J.R. Reynolds / Social Science Research 42 (2013) 818–835

831

Do these role transitions explain the differences across family socioeconomic status in persistence and change in bachelor’s degree expectations? We address this question next by using discrete time event history models that relate bachelor’s degree expectations at a given follow-up to young adults’ family, work and school roles at the same time. Included in the model is a linear term for time that we use to capture the general decline in bachelor’s degree expectations across waves and young adulthood, and to test the possibility that the influences of these competing roles vary with age. Two characteristics of these models are important to note. First, given our interest in the timing of role adoptions and expectations, the dependent variable is now year-specific expectations and not the overall expectation trajectory. Second, the model treats bachelor’s degree expectations as a non-absorbing event, attitudes that may change in the next follow-up depending on whether or not they occur within a persistent expectation trajectory. Table 6 presents the results from three models. The first excludes family, work, and school roles and its estimates of the influence of socioeconomic status serve as the baseline; the second adds family, work, and school roles; and the third presents evidence of the time-contingent effects of marriage and school enrollment. Family, work and school roles significantly shaped young people’s expectations to complete a bachelor’s degree, and did account for a considerable proportion of the effects of socioeconomic status. For example, Model 1 shows that young adults with two parents, at least one of whom earned a graduate or professional degree, had nearly three times the odds of expecting to get a college degree as compared to young adults whose two parents had high school educations or less. The results in Model 2 confirm that part of the influence of socioeconomic status was due to associated differences in the prevalence and timing of family, work and school roles.10 Being married and having children reduced the odds of expecting to get a college degree, by around 18% and 28% respectively, according to Model 2. Working full-time vs. not working reduced the odds by 12%, while working part-time was not significant. These influences pale in comparison to those for attending 2- or 4-year colleges or universities,11 and college attendance is also the primary mediator of the influence of socioeconomic status on college expectations.12 All together, adjusting for family, work, and school roles reduced the logits of the SES categories by 9–27%. There are two notable exceptions, however. Adjusting for these roles actually increased the odds ratio from 1.23 to 1.31 for those with a single parent with a high school education or less and from 1.08 to 1.26 for those who lived in households with neither parent (both compared to those from a two parent family with both parents having high school educations or less). In both cases, it was the inclusion of both parenthood and college enrollment that produced the change across models. Thus, young people from these SES groups were more likely to hold plans for earning a bachelor’s degree than would be expected given their childbearing and college attendance patterns. We examined whether the influence of work, family and school roles varied by their timing by including interaction terms between the roles and the wave of assessment. We found significant interactions for marriage and college enrollment, but not parenthood and employment. As shown in Model 3, the negative consequences of marriage for college expectations weakened over time (or with age). In contrast, college attendance had stronger effects in later years, though the interaction term for attendance at 4-year institutions did not quite reach conventional levels of significance (p = .09). The results for college attendance suggest that expectations among those who are not currently enrolled are harder to hold on to outside of the normative degree-seeking ages in the early 20s. While the results are consistent with our expectation that family, work, and school roles in young adulthood would partially explain the effects of SES on expectations to earn a bachelor’s degree, it is also true that there remain persistent effects of SES after considering these roles. Parental education levels, in both single and two-parent families, still exert an important stratifying effect on expectations, regardless of whether young adults are enrolled in school, married, parenting, or working full-time. 4. Discussion Recent theoretical work on the status attainment model has directed attention to the unique importance of stable high ambitions. Bozick et al. (2010) argue that familial socioeconomic status facilitates having stable high expectations for educational attainment and that the overall stability of expectations matters more for eventual attainment than does a snapshot of plans at any one time. Our findings from analyses of the high school classes of 1987–1990 affirm this pattern and extend it into the post-high school years and to degree attainment (vs. enrollment). Young people from more socioeconomically privileged families were more likely to hold stable expectations to complete 4-year degrees, and these stable expectations helped explain their greater likelihood of earning one. Senior year expectations helped, but the real key was whether they were sustained over time. 10 In preliminary analyses we found that some of these influences, such as parenthood and full-time employment, varied by gender. A full consideration of the gendered nature of post-secondary trajectories was beyond the scope of this study, but see Settersten et al. (2004) for an overview. 11 Model estimates based on multiple imputation of the outcome measure as well as the predictors were quite different, though still pointing to the large influence of college enrollment. With imputed x and y values, the logits for 2-year and 4-year enrollment in Model 2 were .75 and 1.63 (compared to 1.47 and 3.11 in Table 6). 12 For example, in analyses not shown, adding the two attendance measures reduced the logit for having lived with two parents where at least one had a graduate degree by 27% and diminished the logit for having lived with two parents where at least one had a college degree by 22%. Adding the two attendance measures reduced the logits for those from single parent families where the parent had a college degree or more, and those where the parent had some college, by 13% and 10% respectively. Once attendance is in the model, adding parental and marital roles further reduced these logits by 0–4%, and adding employment reduced them by 1% or less. Adding family and work roles without enrollment in the model resulted in small changes to these slope estimates.

832

M.K. Johnson, J.R. Reynolds / Social Science Research 42 (2013) 818–835

This is not only true during the late teens and early 20s, the normative period for pursuing a bachelor’s degree, but also during the later 20s. Expectation trajectories continued to be strongly stratified by SES at these older ages. This pattern of results demonstrates that social background does not exert its full influence on ambitions by the senior year of high school (Johnson, 2002; McClelland, 1990). Completion of a 4-year degree continued to be stratified by SES as well, and stable high expectations were an important mediator of the effect of SES on attainment in the late phase as it was in the early phase. While persistent high expectations are somewhat less common in the later 20s than the early 20s, similar rates of those with stable high expectations (just over half) completed their degree during each observation period. Consistent and high expectations matter even in the later 20s. These comparable ‘‘success’’ rates do not equate to similar ongoing outcomes, however, since the timing of post-secondary education modifies its ultimate value (Elman and O’Rand, 2004). It is notable how common expectations to earn a 4-year degree were throughout the transition to adulthood. Considerable attention is given the rise in adolescent educational ambitions, becoming even more common since the cohort we study (Bachman et al., 2008). Our findings document just how common they are even as years have passed since leaving high school, with remarkable levels of stability in expectations. Over 60% of our sample of high school seniors between 1987 and 1990 held stable high expectations from their senior year to 3–4 years later. Only 11% can be said to have truly dropped their plans, i.e., having started with expectations to earn a bachelor’s degree but no longer holding those expectations 3– 4 years later. Among those who had not yet earned a degree by this time, age 23–24, around 40% held stable high expectations from age 23–24 to age 27–28. Another 23% held expectations to earn a 4-year degree at least once during these older ages. These findings underscore just how salient the bachelor’s degree is in the US, even at ages in which rates of BA receipt are reduced. While there is considerable stability in expecting to earn a bachelor’s degree, and the prevalence of these degree expectations at various ages is indeed notable, the change in expectations we document is also of consequence to current debates. About a quarter of the sample changed their expectations about whether they would earn a bachelor’s degree in the first 3– 4 years after their senior year of high school, sometimes more than once, supporting Andrew and Hauser’s (2011) conclusion that educational expectations are not a static mental construct. Indeed the findings underscore that expectations are really put to the test after the high school years. Previous research addresses the role of academic and significant others’ feedback in shaping class-stratified trajectories of educational expectations over time (Andrew and Hauser, 2011; Bozick et al., 2010; Jacob and Linkow, 2011), but has given less attention to other factors that may sustain or prompt goal revision, especially in the years after secondary schooling. Our results demonstrate that marriage, parenthood, and full-time work were significantly and negatively associated with expecting to earn a 4-year degree in the years after high school. Marriage mattered more the earlier it occurred; school attendance supported bachelor’s degree expectations more at older ages. Although attendance at 2-year, and especially 4-year, institutions was the primary mediator of the influence of SES background on expectations, family and work roles were nontrivial mediators as well. Thus our findings support the argument that differentiated pathways to adulthood be more fully incorporated into models of post-secondary educational processes (Pallas, 2003; Roksa and Velez, 2012). Importantly, however, a strong impact of SES background on bachelor’s degree expectations remained after accounting for these roles, which may stem from a variety of unaccounted for factors including academic performance in post-secondary courses, the continuity of enrollment, and significant others’ support and expectations (Bozick et al., 2010; Goldrick-Rab, 2006). Financial resources and competing commitments to the natal family (e.g., role obligations in the face of parental financial problems, illness, or death) may also be important. A recent report (Nichols, 2011) argues familial income inequality (strongly associated with family structure and parental educational attainment) is a major obstacle to President Obama’s national education goals that include having the highest proportion of college graduates in the world by 2020. A number of limitations of this study should also be acknowledged. First, our analysis does not include the most recent cohorts, and educational expectations have continued to rise since the time our sample left high school (Bachman et al., 2008). College affordability has also eroded (Kane, 2007), and the earnings advantage to a BA increased (Baum et al., 2010). Thus there is a continued need to examine educational expectations, attainments, and their relationship to socioeconomic background as more recent cohorts make their way through the transition to adulthood. The rise in adolescent expectations will have reduced the proportion of young people holding stable low expectations by definition. The group who persists, holding stable high expectations to earn a bachelor’s degree, may have grown, as a bachelor’s degree have come to be seen as even more critical to young people’s future and perceptions of ongoing opportunity to return to school are maintained or increased. Reduced affordability and high college drop-out threaten the ability of young people to achieve their expectations, however. It may be especially important to examine gendered processes in later cohorts as well. Our data stem from a time in which young women had caught up with young men in terms of their educational plans and expectations, but had not yet really exceeded them. Our findings are inherently context-dependent, and as newer cohorts advance through their young adult years it will be important to evaluate whether the patterns we observe persist for them. Second, panel attrition may have biased our results. While we made efforts to minimize this risk, those retained in the study were more advantaged than those lost to attrition. This means, for example, that the true population rates at which high school graduates held on to their bachelor’s degree plans or obtained those degrees are likely to be lower than the rates we report based on the MTF panel data. Third, our focus with degree completion was specifically on 4-year degrees, in line with our measure of educational expectations. While many young people failed to achieve 4-year degrees within the decade following their senior year of high school, some likely completed 2-year degrees and other educational credentials. It would be desirable in future work to examine the implications of maintaining expectations for a 4-year degree for alternative edu-

833

M.K. Johnson, J.R. Reynolds / Social Science Research 42 (2013) 818–835

cational outcomes as well. Expectations to earn a bachelor’s degree, whether held sporadically or consistently, may facilitate completion of a lesser degree in lieu of the bachelor’s. Finally, our focus on change and stability in expectations to earn a bachelor’s degree masks heterogeneity in what other levels of education young people expect to achieve over time. While a full accounting of plans to earn graduate or 2-year degrees is beyond the scope of this article, we note here some of the larger patterns to stimulate further consideration of this issue. Among those who held stable high bachelor’s degree expectations in the early phase, half also held stable plans to attend graduate/professional school and 26% warmed up to such plans during the period. Among those who dropped their bachelor’s degree expectations in the early phase, half held on to, or newly formed, expectations to earn a 2-year degree instead; 26% started with expectations to earn a 2-year degree (on the way to their expected bachelor’s degree), but dropped those 2-year degree plans along with their 4-year degree plans. So how are socioeconomic status and expectations playing out in the post-high school years? The findings of this study emphasize the ongoing large benefits to young adults from higher SES backgrounds, even among those who do not complete their bachelor’s degrees in the more normative four to 6 years after twelfth grade. They further indicate that a substantial reason SES background continues to matter is through promoting stable expectations to complete a 4-year degree; senior year expectations matter little if they do not persist. Later formed expectations also do not matter much if not embedded in a trajectory of persistence. Role pathways in the transition to adulthood differ by socioeconomic background and provide one mechanism through which educational goals are sustained or revised. Pursuing one’s goals, though enrollment in 2-year, and especially 4-year, institutions of higher education facilitates holding bachelor’s degree expectations in the years after high school, though delayed family formation and full-time work contribute as well. Acknowledgments This material is based upon work supported by the National Science Foundation under Grant No. 0647333. The Monitoring the Future data were made available by John Schulenberg, Jerald G. Bachman, Lloyd D. Johnston, and Patrick M. O’Malley of the Survey Research Center, Institute for Social Research, University of Michigan. The collectors of the data do not bear any responsibility for our analyses or interpretations. The Monitoring the Future Study is funded by a grant from the National Institute on Drug Abuse (DA01411). We thank Beth Fussell, Christine Horne, Julie Kmec, and Alair MacLean for comments on an earlier draft.

Appendix A Baseline characteristics of study sample. Weighted percentage Two parents, both high school or less Two parents, 1 + some college Two parents, 1 + college Two parents, 1 + grad/prof degree One parent, high school or less One parent, some college One parent, college+ Neither parent in home Missing

12.57 14.35 18.24 14.24 9.56 2.80 5.39 5.52 6.33

0 siblings 1 sibling 2 siblings 3 siblings 4 siblings 5 siblings 6 + siblings Missing

4.86 20.51 20.73 14.16 7.98 5.19 8.76 17.81

Male Female Missing

48.98 51.02 .00

White

74.48 (continued on next page)

834

M.K. Johnson, J.R. Reynolds / Social Science Research 42 (2013) 818–835

Appendix A (continued) Weighted percentage African American Hispanic Other race/ethnicity Missing

11.07 7.34 6.19 0.91

Non-college prep track College prep track Missing

46.37 51.43 2.20

A (93–100) A (90–92) B+ (87–89) B (83–86) B (80–82) C+ (77–79) C (73–76) C (70–72) D (69 or below) Missing

8.82 10.96 16.36 20.20 15.27 12.38 8.98 2.91 1.20 2.92

N (unweighted)

9870

References Adelman, Clifford, 2005. Moving into Town – and Moving On: The Community College in the Lives of Traditional-age Students. U.S. Department of Education, Washington, DC. Alexander, Karl, Entwisle, Doris, Bedinger, Samuel, 1994. When expectations work: race and socioeconomic differences in school performance. Social Psychology Quarterly 57, 283–299. Alexander, Karl, Bozick, Robert, Entwisle, Doris, 2008. Warming up, cooling out, or holding steady? Persistence and change in educational expectations after high school. Sociology of Education 81, 371–396. Andres, Lesley, Adamuti-Trache, Maria, Yoon, Ee-Seul, Pidgeon, Michelle, Thomsen, Jens Peter, 2007. Educational expectations, parental social class, gender, and postsecondary attainment: a 10-year perspective. Youth & Society 39, 135–163. Andrew, Megan, Hauser, Robert M., 2011. Adoption? Adaptation? Evaluating the formation of educational expectations. Social Forces 90, 497–520. Astone, Nan Marie, Schoen, Robert, Ensminger, Margaret, Rothert, Kendra, 2000. School reentry in early adulthood: the case of inner-city African Americans. Sociology of Education 73, 133–154. Aud, Susan, Ann Fox, Mary, KewalRamani, Angelina, 2010. Status and Trends in the Education of Racial and Ethnic Groups (NCES 2010-015). U.S. Department of Education, National Center for Education Statistics. U.S. Government Printing Office, Washington, DC. Bachman, Jerald G., Johnston, Lloyd D., O’Malley, Patrick M., 2001. The Monitoring the Future Project After Twenty-seven years: Design and Procedures. Monitoring the Future Occasional Paper 54. Institute for Social Research, Ann Arbor, MI. Bachman, Jerald G., Johnston, Lloyd D., O’Malley, Patrick M., 2008. Monitoring the Future: Questionnaire responses from the Nation’s High School Seniors. Institute for Social Research, Ann Arbor, MI. Baum, Sandy, Ma, Jennifer, Payea, Kathleen, 2010. Education Pays, 2010: The Benefits of Higher Education for Individuals and Society. The College Board, Washington, DC. Bennett, Pamela R., Xie, Yu, 2003. Revisiting racial differences in college attendance: the role of historically Black colleges and Universities. American Sociological Review 68, 567–580. Bozick, Robert, Alexander, Karl, Entwisle, Doris, Dauberand, Susan, Kerr, Kerri, 2010. Framing the future: revisiting the place of educational expectations in status attainment. Social Forces 88, 2027–2052. Cheng, Simon, Starks, Brian, 2002. Racial differences in the effects of significant others on students’ educational expectations. Sociology of Education 75, 306–327. Clark, Burton R., 1960. The ‘cooling-out’ function in higher education. American Journal of Sociology 65, 569–576. Clarke, Philippa, O’Malley, Patrick M., Johnston, Lloyd D., Schulenberg, John E., 2009. Social disparities in BMI trajectories across adulthood by gender, race/ ethnicity and lifetime socio-economic position. International Journal of Epidemiology 38, 499–509. Conger, Dylan, Long, Mark C., 2010. Why are men falling behind? Gender gaps in college performance and persistence. The Annals of the American Academy of Political and Social Science 627, 184–214. Dariotis, Jacinda K., Pleck, Joseph H., Astone, Nan M., Sonenstein, Freya L., 2011. Pathways of early fatherhood, marriage, and employment: a latent class growth analysis. Demography 48, 593–623. DiPrete, Thomas A., Buchmann, Claudia, 2006. Gender-specific trends in the value of education and the emerging gender gap in college completion. Demography 43, 1–24. Domina, Thurston, Conley, Anne Marie, Farkas, George, 2011a. The case for dreaming big. Sociology of Education 84, 118–121. Domina, Thurston, Conley, Anne Marie, Farkas, George, 2011b. The link between educational expectations and effort in the college-for-all era. Sociology of Education 84, 93–112. Elder, Glen H., 1994. Time, human agency, and social change: perspectives on the life course. Social Psychology Quarterly 57, 4–15. Elder Jr., Glen H., Johnson, Monica Kirkpatrick, Crosnoe, Robert, 2003. The emergence and development of life course theory. In: Mortimer, J., Shanahan, M. (Eds.), Handbook of the Life Course. Plenum, New York, pp. 3–22. Elman, Cheryl, O’Rand, Angela M., 2004. The race is to the swift: socioeconomic origins, adult education and mid-life economic status. American Journal of Sociology 110, 123–160.

M.K. Johnson, J.R. Reynolds / Social Science Research 42 (2013) 818–835

835

Elman, Cheryl, O’Rand, Angela M., 2007. The effects of social origins, life events, and institutional sorting on adults’ school transitions. Social Science Research 36, 1276–1299. Fitzpatrick, Maria D., Turner, Sarah E., 2007. Blurring the boundary: changes in the transition from college participation to adulthood. In: The Price of Independence. The Economics of Early Adulthood, edited by Sheldon Danzinger and Cecilia Rouse. Russell Sage, New York, pp. 107–137. Goldrick-Rab, Sara, 2006. Following their every move: an investigation of social-class differences in college pathways. Sociology of Education 79, 61–79. Goyette, Kimberly A., 2008. College for some to college for all: social background, occupational expectations, and educational expectations over time. Social Science Research 37, 461–484. Haller, Archibald O., Portes, Alejandro, 1973. Status attainment processes. Sociology of Education 46, 51–91. Hanson, Sandra L., 1994. Lost talent: unrealized educational aspirations and expectations among U.S. youths. Sociology of Education 67, 159–183. Ingels, Steven J., Dalton, Ben W., 2008. Trends Among High School Seniors, 1972–2004 (NCES 2008-320). National Center for Education Statistics, Institute for Education Sciences, U.S. Department of Education. Washington, DC. Jacob, Brian A., Linkow, Tamara Wilder, 2011. Educational expectations and attainment. In: Duncan, G., Murnane, R. (Eds.), Whither Opportunity: Rising Inequality, Schools, and Children’s Life Chances. Russell Sage Foundation, NY, pp. 133–162. Johnson, Monica Kirkpatrick, 2002. Social origins, adolescent experiences, and work value trajectories during the transition to adulthood. Social Forces 80, 1307–1341. Johnston, Lloyd D., O’Malley, Patrick M., Bachman, Jerald G., Schulenberg, John E., 2007. Monitoring the Future National Survey Results on Drug Use, 1975– 2006. Vol. I: Secondary School Students. National Institute on Drug Abuse, Bethesda, MD (NIH Publication #07-6205). Kane, Thomas J., 2007. Public intervention in postsecondary education. In: Hanushek, E., Welch, F. (Eds.), Handbook of the Economics of Education. Elsevier Science, Amsterdam. Kao, Grace, Tienda, Marta, 1998. Educational aspirations of minority youth. American Journal of Education 106, 349–384. Manski, Charles, 2004. Measuring expectations. Econometrica 72, 1329–1376. Martin, Molly A., 2012. Family structure and the intergenerational transmission of educational advantage. Social Science Research 41, 33–47. McClelland, Katherine.E., 1990. The social management of ambition. The Sociological Quarterly 31, 225–251. Morgan, Stephen L., 2004. Methodologist as arbitrator: five models for black–white differences in the causal effect of expectations on attainment. Sociological Methods & Research 33, 3–53. Nichols, Andrew Howard, 2011. Developing 20/20 Vision on the 2020 Degree Attainment Goal: The Threat of Income-Based Inequality in Education. The Pell Institute for the Study of Opportunity in Higher Education. Pallas, Aaron M., 2003. Educational transitions, trajectories, and pathways. In: Mortimer, Jeylan T., Shanahan, Michael J. (Eds.), Handbook of the Life Course. Kluwer Academic/Plenum Publishers, New York, pp. 165–184. Plank, S.B., Jordan, W.J., 2001. Effects of information, guidance, and actions on postsecondary destinations: a study of talent loss. American Educational Research Journal 38, 947–979. Qian, Zhenchao, Blair, Sampson Lee, 1999. Racial/ethnic differences in educational aspirations of high school seniors. Sociological Perspectives 42, 605–625. Reynolds, John, Burge, Stephanie, 2008. Educational expectations and the rise in women’s post-secondary attainments. Social Science Research 37, 485–499. Reynolds, John, Johnson, Monica Kirkpatrick, 2011. Change in the stratification of educational expectations and their realization. Social Forces 90 (1), 85– 109. Reynolds, John, Stewart, Mike, Sischo, Lacey, McDonald, Ryan, 2006. Have adolescents become too ambitious? High school seniors’ educational and occupational plans, 1976–2000. Social Problems 53, 186–206. Roderick, Melissa, Nagaoka, Jenny, Coca, Vanessa, Moeller, Eliza, 2008. From High School to the Future: Potholes on the Road to College. Consortium on Chicago School Research, Chicago, IL. Roksa, Josipa, Potter, Daniel, 2011. Parenting and academic achievement: intergenerational transmission of educational advantage. Sociology of Education 84, 299–321. Roksa, Josipa, Velez, Melissa, 2012. A late start: delayed entry, life course transitions and bachelor’s degree completion. Social Forces. http://dx.doi.org/ 10.1093/sf/sor018. Rosenbaum, James, 2011. The complexities of college for all: beyond fairy-tale dreams. Sociology of Education 84, 113–117. Royston, Patrick, 2004. Multiple imputation of missing values. Stata Journal 4 (3), 227–241. Ruggles, Steven, J., Alexander, Trent, Genadek, Katie, Goeken, Ronald, Schroeder, Matthew B., Sobek, Matthew, 2010. Integrated Public Use Microdata Series: Version 5.0 [Machine-readable database]. University of Minnesota, Minneapolis. Sabates, Ricardo, Harris, Angel L., Staff, Jeremy, 2011. Ambition gone awry: the long-term socioeconomic consequences of misaligned and uncertain ambitions in adolescence. Social Science Quarterly 92, 959–977. Sandefur, Gary D., Eggerling-Beck, Jennifer, Park, Hyunjoon, 2005. Off to a good start? Postsecondary education and early adult life. In: Settersten, R., Furstenberg, F., Rumbaut, R. (Eds.), On the Frontier of Adulthood: Theory, Research, and Public Policy. University of Chicago Press, Chicago, pp. 292–319. Schneider, Barbara, Stevenson, David, 1999. The Ambitious Generation: America’s Teenagers, Motivated But Directionless. Yale University Press, New Haven. Snyder, Thomas D., Dillow, Sally A., Hoffman, Charlene M., 2008. Digest of Education Statistics 2007 (NCES 2008-022). National Center for Education Statistics, Institute of Education Sciences, U.S. Department of Education, Washington, DC. Trusty, Jerry, 2000. High educational expectations and low achievement: stability of educational goals across adolescence. Journal of Educational Research 93, 356–365. Trusty, Jerry, Harris, Morag B. Colvin, 1999. Lost talent: predictors of the stability of educational expectations across adolescence. Journal of Adolescent Research 14, 359–382. Turner, Ralph H., 1960. Sponsored and contest mobility and the school system. American Sociological Review 25, 855–867. Uno, Mayumi, Mortimer, Jeylan T., Kim, Minzee, Vuolo, Michael, 2010. ‘‘Holding on’’ or ‘‘Coming to Terms’’ with educational underachievement: a longitudinal study of ambition and attainment. New Directions for Child and Adolescent Development 130, 41–56. von Hippel, Paul T., 2007. Regression with missing Ys: an improved strategy for analyzing multiply-imputed data. Sociological Methodology 37, 83–117.