Abstract
Women were the driving force behind the expansion of higher education in the United States. Their rise in enrollment developed alongside a rapid increase in educational expectations. This study explores the question of how these landmark shifts in women’s expanded expectations, attainment, and the (mis)alignment of the two have influenced mental health now that these women have entered midlife. Using data from the sophomore cohort of the 1980 High School and Beyond (HS&B:So) study through its 2014 midlife follow-up, results show that for women in the sophomore cohort (n = 2,010), exceeding educational expectations is significantly associated with an increase in experiencing serious mental illness in midlife compared to meeting or missing educational expectations. These results suggest that overachieving educational expectations carry distinct psychological consequences.
Women were the driving force behind the expansion of higher education in the United States (Fischer and Hout 2006; Goldin 1999; Goldin, Katz, and Kuziemko 2006). From 1970 to 1985, women’s enrollment accounted for 80 percent of total enrollment growth in all institutions of higher education (National Center for Education Statistics [NCES] 1987). This rise in enrollment developed alongside a rapid increase in educational expectations among young girls, particularly those living in disadvantaged households (Jacob and Linklow 2011; Kets de Vries 2005; Lehmann 2014). Combined rapid changes in women’s educational expectations and postsecondary enrollment fostered an environment for misaligned expectations—expectations that do not fit with other life goals or levels of ability (Schneider and Stevenson 1999).
This study asks how these landmark shifts in women’s educational expectations, attainment, and the (mis)alignment of the two have influenced mental health now that this cohort of women has entered midlife. There was a surge of research in the 1980s on misaligned educational expectations and mental health for U.S. young adults, not specific to women, that largely focused on “missed expectations.” It was born out of concerns of relative deprivation and self-discrepancy, which predicted that unattained educational goals would result in adverse mental health outcomes such as depression, anxiety, and self-doubt (Dannefer 1984; Higgins 1987, 1989; Morgan 2004; Walker and Pettigrew 1984). However, empirical research on young adults did not support these theoretical predictions or concerns (Reynold and Baird 2010).
Yet, there are reasons to believe that misaligned educational expectations may have mental health implications for midlife women—especially if they exceeded their educational expectations. On the one hand, exceeding educational expectations may be beneficial simply because educational attainment is an important determinant of mental health. Higher levels of attainment are associated with reduced odds of depressive symptoms (J. O. Lee et al. 2013; Schieman and Koltai 2017), suicide (Phillips and Hempstead 2017), and psychological distress (Mandemakers and Monden 2013). On the other hand, exceeding educational expectations—especially in the 1980s—exposed young women to a changing, and at times, unwelcoming, educational, and occupational landscape (Chamberlain 1988; Goldin 2006). The chronic stress of unexpectedly engaging with gendered institutions of education and subsequent work environments could adversely impact mental health outcomes at midlife (Crocker 1999; Link et al. 1997; Meyer 2003; Steele 1997). This is particularly plausible given that cumulative exposure to gender bias, in combination with midlife hormonal changes, contributes to women’s sustained higher prevalence of serious mental illness (SMI) (Almeida and Fletcher 2022).
The present study explores these two competing notions by investigating the relationship between misaligned educational expectations and SMI among midlife women in the sophomore cohort of the 1980 High School and Beyond (HS&B) study who participated in its 2014 midlife follow-up. The study focuses on women because the prevalence of SMI, although relatively rare, is nearly twice as high among women (7 percent) than men (4 percent) (Substance Abuse and Mental Health Services Administration [SAMHSA] 2021). In addition, the HS&B female cohort grew up and considered educational attainment trajectories during the expansion of postsecondary educational opportunities and, in turn, the dramatic shifts in women’s expectations and attainment. Moen (2022) has argued that a period of social change, such as that experienced through educational institutions by this cohort of women, presents a “strategic research site” to investigate the mental health of individuals experiencing structural bias and constraints. As educational expectations continue to shift despite structural constraints on achievement (Goyette 2008; Reynolds and Johnson 2011), understanding the relationship between misaligned expectations and mental health for women at this juncture of social change can inform future research on misaligned expectations and mental health for structurally constrained groups, such as racially and ethnically minoritized students and first-generation postsecondary students.
Background
Misaligned educational expectations signal a disconnect between student preparation, ability, and attainment. When a student does not understand feasible opportunities or is unprepared to meet their educational expectations, their expectations are deemed misaligned (Schneider and Stevenson 1999). Across the literature on misaligned expectations, unattained or missed expectations are generally the focus (Cundiff 2017; Reynold and Baird 2010; Schneider and Stevenson 1999).
Self-discrepancy theory and relative deprivation theory are the two primary perspectives used to argue that misaligned expectations can contribute to poor mental health outcomes. Self-discrepancy theory predicts distress due to misalignment between an individual’s actual self (or actual educational attainment) and their ideal self (educational expectations) (Higgins 1987, 1989). Relative deprivation theory predicts distress due to individuals feeling as if they have been denied a reward (expected educational attainment) to which they were entitled (Walker and Pettigrew 1984). These theories can be used to motivate a research expectation that missing educational expectations could be associated with higher odds of SMI among midlife women.
Considering how exceeded expectations and SMI may be related among midlife women is more complicated. The extraordinary growth in women’s postsecondary enrollment and subsequent entry to the workforce followed an evolution in women’s decision-making and long-term expectations (Goldin et al. 2006). As available workplace expectations for women began to shift, high-school-aged girls started to invest more heavily in their academic achievement, college enrollment, and workforce preparation (Goldin 2006). These changes throughout high school led some women to exceed their expectations and attain higher levels of education than anticipated, and higher levels of educational attainment are associated with better midlife health outcomes (J. O. Lee et al. 2013; Lynch 2003; Ross and Wu 1996).
Nonetheless, recent evidence suggests rising SMI in midlife adults is concentrated among women (Almeida and Fletcher 2022; Seeman 2010) and adults with at least some postsecondary education (Infurna et al. 2021; Weinberger et al. 2018). Homan’s (2019) structural sexism perspective and theory about the stress process can be combined to provide a framework to consider how the unique educational and social experiences of women who exceeded their educational expectations produce an increased risk of SMI.
Structural sexism is defined as a “systematic gender inequality in power and resources” that exists at the micro, meso, and macro levels of the United States’ social structure (Homan 2019). Importantly, the processes that create this systematic gender inequality are not always directly perceived as discriminatory by individuals (Ridgeway 2011; C. West and Zimmerman 1987). However, even when not perceived individually as discriminatory, gender-stratified opportunities and constraints rooted in systemic inequality can influence individual stress and associated health outcomes (Homan 2019; Krieger 2014).
Increased exposure to social stressors is also associated with psychological distress (Erving et al. 2022; Pearlin and Bierman 2013; Pearlin et al. 1981; Turner 2013). In particular, sociological stress scholars have emphasized that personal distress can often be traced back to social structures and stress produced by an individual’s position within the structure (Moen 2022; Pearlin 1989; Phelan 2024). A specific application of stress theory describes how stigma, prejudice, and discrimination create a stressful social environment that contributes to the prevalence of mental illness in a specified minority population (Meyer 2003). This theory was originally and is most commonly applied to the disproportionately high mental illness prevalence among sexual minorities (Meyer 2003). Yet, it may also be salient to women who exceeded educational expectations during a time of educational expansion characterized by transitioning gender ideologies, underrepresentation of women in faculty and administrative positions, and a maintained culture of gender discrimination throughout educational institutions.
At the macro-level, structural sexism is characterized by gendered cultural norms and a gender-stratified distribution of resources within society’s major institutions, including educational institutions (Homan 2019; Ridgeway and Correll 2004; Risman 2004). As women’s academic expectations and enrollment in higher education increased, they did so within gendered organizations. All levels of educational institutions pattern advantage, control, action, and identity through a male/female lens that tends to favor a “masculine ethic” (Acker 1990) that rewards characteristics stereotypically assigned to educationally advantaged men (Kanter 1977) and subsequently undervalues those belonging to women (J. J. Lee and Mccabe 2021; Musto 2019; Sadker and Sadker 1994). Although enrollment in postsecondary institutions reached gender parity by 1980 (NCES 1987), gender parity was not achieved among the faculty, staff, and administration (Chamberlain 1988; Goldin et al. 2006; M. S. West 1994). By the mid-1980s, overt discriminatory policies, like admission quotas, had been removed (Chamberlain 1988). However, legal challenges that narrowed the scope and enforcement of Title IX contributed to the maintenance of gender discrimination throughout educational institutions moving forward (Chamberlain 1988).
At the meso-level, structural sexism is reflected in interpersonal interactions and behavior within institutions, and is reflected in organizational practices that inequitably distribute power and resources (Homan 2019; Ridgeway and Correll 2004; Risman 2004). In 1982, surveys of classroom behaviors revealed that women were less likely to be called upon or coached by faculty, more likely to be interrupted by faculty and male students, and experienced common unwelcome sexual attention (Chamberlain 1988; Hall and Sandler 1982). Unfortunately, this “chilly” climate toward women persists (Blithe and Elliott 2020; J. J. Lee and Mccabe 2021), and gender inequity in student treatment has been documented as early as primary and secondary education (Buchmann, DiPrete, and McDaniel 2008; Leaper and Brown 2014; Musto 2019; Sadker and Sadker 1994).
Beyond the classroom, higher levels of education led to entry into nontraditional careers or overeducation for traditional gendered pathways (Goldin 2006). Within the workplace, women with higher levels of education generally report worse job satisfaction (Miller 1980). When employment contributes to role overload through interaction with other gendered social roles (mother, wife, caretaker), job satisfaction decreases and carries a psychological toll for women (Hodson 1989; Pearson 2008). If stepping out of the workforce or into more traditional gendered work pathways, exceeding educational expectations may lead to overeducation—holding more education than required for your occupation—which diminishes the protective effect of education on psychological well-being (Bracke, van de Straat, and Missinne 2014).
At the micro-level, structural sexism is characterized by internalized gender ideologies that reinforce gendered inequalities (Homan 2019; Ridgeway and Correll 2004; Risman 2004). Minority stress theory highlights how stressors of these gendered education institutions can lead to internalized sexism and disidentification. Ongoing stressors from structural forces can lead to self-stigmatization through the ability to imagine the perspective of others and anticipate their response to your actions (Meyer 2003). In the context of educational expectations, young women in the 1980 sophomore high school cohort were exposed to a changing educational landscape accompanied by a shift in gender values (Goldin 2006). An internalized sexism from structural forces within and beyond educational institutions could lead to the adoption of low educational expectations. However, as the educational landscape continued to expand, women participating in these opportunities, and completing more education than they had expected, were engaged in a form of stereotype threat against negative perceptions of women in educational institutions. When surpassing their own internalized educational expectations shaped by structural forces within and beyond educational institutions, women at any level of education could experience the micro-level stressors of self-stigmatization and stereotype threat. The chronic stressor of stereotype threat and air of stigmatization does not require a concrete event for the mental health consequences to be actualized (Crocker 1999; Link et al. 1997; Meyer 2003; Steele 1997). As a result, exceeding educational expectations may be associated with an increased prevalence of SMI among women in this cohort through micro-level structural sexism processes.
The present study draws on theoretical perspectives of perceived consequences of unattained educational expectations and the application of a structural sexism perspective to exceeding expectations to explore the relationship between misaligned educational expectations and SMI among midlife women. While the structural sexism framework provides a theoretical context to investigate the impact of exceeding expectations in gendered institutions, this study does not aim to test the theory empirically. Rather, the theoretical context provides a framework for this initial step in understanding the relationship between misaligned expectations and mental health for a structurally constrained group. This study advances prior work on misaligned expectations and mental health by recognizing exceeded expectations as a potentially important category of misalignment for midlife mental health, focusing attention on a cohort of women who had increased opportunities due to higher education expansion, and investigating mental health at a time in the life course where the second wave of mental illness onset is observed among women. Results offer insight into gender-specific relationships between education and SMI that contribute to a growing literature on the disproportionate burden of SMI among women in midlife.
Data and Methods
Data and Sample
Data for this study come from the HS&B study, a nationally representative sample of approximately 30,000 high school sophomores and 30,000 seniors in the United States in 1980. Follow-ups for HS&B were conducted in 1982 (n = 14,830), 1984 (n = 13,680), 1986 (n = 13,430), and 1992 (n = 12,640). In 2014, 60 percent of the eligible sophomore sample members completed a midlife follow-up survey (n = 8,790), and a subsample completed an extended version of the survey that included questions related to SMI (n = 3,710). All unweighted sample sizes have been rounded to the nearest 10, as required by the NCES-restricted data license. Midlife data have also been collected for the senior cohort, but these data are not yet available from the NCES.
This study utilizes the data from the female sample members of the sophomore cohort who participated in the midlife follow-up and answered questions about SMI. Just over half of the 2014 extended midlife survey respondents were women (n = 2,010). To retain women with missing data on at least one study covariate (N = 420), I multiply impute missing data using the Stata 17.0 mi function. It replaces missing values that were randomly drawn from a posterior predictive distribution conditioned on the observed values (von Hippel 2020). For this analysis, imputations were performed using “chained” equations over 10 iterations. The multiple imputation process averages the values of the parameter estimates across M samples to produce a single-point estimate and produces the standard errors through averaging the squared standard errors of the M estimates and calculating the variance of the M parameter estimates (Allison 2000). Results from the imputed analyses were substantively similar to analyses using listwise deletion. The 2014 HS&B midlife follow-up panel weights are applied to adjust for unit nonresponse, initial selection, and panel attrition (Muller et al. 2019).
Measures
I measure SMI, the study’s dependent variable, using the Kessler Psychological Distress Scale (K6), a 6-item self-report measure of psychological distress that has been validated as an assessment for SMI in the general population (Kessler et al. 2003, 2010). The K6 items measure self-reported feelings of nervousness, hopelessness, restlessness, worthlessness, effort, and depression within the past month. Each item includes responses coded from 0 to 4(0 = None of the time, 1 = A little of the time, 2 = Some of the time, 3 = Most of the time, and 4 = All of the time). The scale can be used as a count of adverse mental health symptoms or condensed into a dichotomous variable indicating the presence or absence of SMI. For this study, I dichotomize it using a score of 13 or more to indicate SMI (1 = yes) in line with previous research that has validated this cut-off in clinical settings (Kessler et al. 2003, 2010). This operationalization reflects the original design of the K6, which was developed as a brief screening tool for nationally representative surveys to assess whether individuals likely had SMI, or a mental health disorder diagnosable by the Diagnostic and Statistical Manual of Mental Disorders (DSM; Kessler et al. 2002, 2003; Pratt 2009). Dichotomizing the K6 produces a screening measure for the clinical presence of disorders such as major depressive disorder, panic disorder, generalized anxiety disorder, bipolar disorder, and schizophrenia (Cairney et al. 2007; Umucu et al. 2022).
Sensitivity analyses using Kessler scores as a continuous variable produced substantively similar results to the main study results shown in Table 3 (see Supplemental Appendix). The continuous measure taps a more general indicator of psychological distress severity, but the substantive meaning of moving one point higher or lower on the scale is not as clear as the meaning of scoring high enough to approximate SMI. For example, a one-point increase on the K6 scale could differentiate between individuals with minimal distress (i.e., the difference between a score of 1 and 2) or extreme severity of SMI (e.g., the difference between 19 and 20).
My primary independent variable is misaligned educational expectations. It is constructed by comparing 1980 reports of educational expectations during women’s sophomore year to the highest degree women earned by 2014. During their sophomore year, HS&B respondents reported whether they expected to earn less than a high school degree, a high school degree, an associate’s or a vocational degree, a four-year college degree, or a graduate or professional degree. In 2014, respondents reported whether they attained these levels of education. Individuals with aligned expectations (reference category) reported the same educational expectations and attainment in 1980 and 2014, respectively. These individuals are compared to respondents who had higher educational expectations than attainment levels (missed expectations) and respondents with higher attainment levels than educational expectations (exceeded expectations).
Because misaligned expectations can occur at varying levels of attainment (i.e., expected a high school degree but earned a college degree; expected a professional degree but earned a high school degree), I not only use the 2014 educational attainment measure to construct misaligned expectations; I also control for it in multivariate models (reference = less than a high school degree). Note that HS&B participants also reported educational attainment in 1992. In supplementary models, I estimated educational attainment and constructed my measure of misaligned educational expectations using the 1992 measure. Results (available upon request) from models utilizing the 1992 attainment variables were substantively and statistically similar to results shown in this article.
I also control for a measure of 1980 depression and unhappiness to account for its long-term consequences for later life mental health (Copeland et al. 2021; Johnson et al. 2018) and educational attainment (Breslau et al. 2008; Simson, Brekke, and Hardoy 2021; Wickersham, Dickson, et al. 2021; Wickersham, Sugg, et al. 2021). The measure is a self-reported rank of the frequency of feeling depressed or unhappy in the last month (0 = Never, 1 = Once, 2 = Several times, 3 = A lot). I dichotomize responses with never (= 1) serving as the reference category. Note that this measure is quite blunt given that it is a single measure including feelings of unhappiness and not simply depression, but it is the only available high school measure of psychological distress.
Other control variables account for sociodemographic background, cognitive ability, and noncognitive skills, all measured in adolescence. Respondent’s self-identified race is dichotomized as White (=1) and non-White due to sample size constraints. Family socioeconomic status in 1980 is a continuous composite measure that sums standardized values of father’s occupation and education, mother’s education, family income, and material possessions that is centered at 0 and ranges from −3 to 3. I use the 1980 sophomore math standardized test scores as an indicator of academic achievement because of its strong correlation with educational attainment (S. M. Lee et al. 2008; Thompson et al. 2006) and skills (Alexander and Pallas 1984; Grogger and Eide 1995; Murnane, Willett, and Levy 1995). The 1980 math standardized test included 38 items to measure skills in basic numeracy, algebra, and geometry (Rock et al. 1985). Test scores range from 25 to 69 and are condensed into quartiles. Locus of control serves as a measure of noncognitive skill in adolescence, reflecting the extent to which students feel they control their destinies. The measure is a composite scale of four questions regarding perceived internal control (Rotter 1966). The scale results in a continuous standardized score centered at 0 and ranges from −3 to 3. As indicated by a higher locus of control score, a higher level of perceived control is associated with fewer mental health symptoms (Awaworyi Churchill et al. 2020, Yu and Fan 2016).
I also control for midlife employment status and health status, and include a theoretical midlife confounder, job satisfaction. Employment status (1 = yes) is a dichotomous measure of employment (currently working for pay and not reporting retirement or disability) as of the interview in 2014. Employment has been found to have a robust positive relationship with mental health (Murphy and Athanasou 1999). Self-assessed health is a self-reported measure from the 2014 questionnaire that indicates whether respondents perceive their health to be poor, fair, good, very good, or excellent. I dichotomize it to compare those with excellent or good health (= 1) to those with fair or poor health. Last, job satisfaction is based on responses to a 5-item Likert scale that ranges from being very dissatisfied to very satisfied with a respondent’s current or most recent job in 2014. The measure is then dichotomized into an indicator of dissatisfaction (1 = yes). Among women, job satisfaction is theorized to reflect dissatisfaction with occupational experiences (Miller 1980), overeducation (Bracke et al. 2014; Bracke, Pattyn, and von dem Knesebeck 2013), and role overload (Hodson 1989; Pearson 2008). Ultimately, job dissatisfaction has been associated with adverse mental health and psychological concerns (Faragher, Cass, and Cooper 2005).
The final set of confounders includes a more comprehensive set of workplace and family characteristics that reflect experiences with macro-level (e.g., male-dominated occupations and societal constraints on women’s workforce participation) and meso-level (e.g., coworker relationships and role overload) structural sexism. Workplace characteristics include 1992 relationships with coworkers (1 = good relationships with coworkers), 1992 to 2014 movement between male- and female-dominated occupations, and 2014 occupation. The 1992 to 2014 movement between male- and female-dominated occupations captures the gender composition of women’s reported careers in 1992 and 2014 and establishes whether they remained in female- or male-dominated occupations or moved between male- and female-dominated occupations. Their 2014 occupation is a four-category variable of occupational codes. The family characteristics include 2014 marital status (1= married), number of children in 2014, and 2014 parenting and caregiving stress. Parenting and caregiving stress are categorical variables that include not experiencing parenting or caregiving, experiencing parenting or caregiving as never or rarely stressful, and experiencing parenting or caregiving as sometimes or always stressful.
Analytic Strategy
I begin by estimating descriptive statistics about the study sample, which are shown in Table 1. I then present results from a bivariate analysis that compares women’s high school expectations and educational attainment in Table 2. Estimates in this table demonstrate the degree to which women’s expectations and attainment are aligned.
Weighted Descriptive Statistics for the Study Sample.
Source. U.S. Department Of Education, National Center For Education Statistics, High School & Beyond, 1980 1992, 2014.
Note. All percentages do not sum to 100% due to rounding error.
Women’s Educational Expectations and Attainment (Weighted Percentages).
Note. Shaded cells indicate alignment between 1980 educational expectations and 2014 degree.
Source. U.S. Department of Education, National Center for Education Statistics (NCES), High School & Beyond, 1980, 1992, 2014.
Signifies suppressed estimates due to NCES guidelines of reporting estimates based on small cell sizes when analyzing restricted-use versions of the HS&B data.
Next, I estimate a series of logistic regression models that show the relationship between misaligned expectations and SMI. Model 1 includes misaligned expectations, educational attainment, and depression in adolescence. Model 2 adds high school race, family socioeconomic status, standardized mathematics test scores, and locus of control to Model 1. Model 3 then incorporates midlife employment status, job satisfaction, and self-assessed health into Model 2. The fourth model adds more detailed workplace and family characteristics to Model 3. Results from all four models are shown in Table 3.
The Estimated Odds of Experiencing Serious Mental Illness in Midlife among Women.
Source. U.S. Department of Education, National Center for Education Statistics, High School & Beyond, 1980, 1992, 2014.
Signifies that the estimated effect of exceeded educational expectations is significantly larger than missing educational expectations.
p < .10. *p < .05. **p < .01 ***p < .001 (two-tailed test).
To aid in the interpretation of odds ratios and comparison across models shown in Table 3, I estimate average marginal effects (AMEs) of misaligned educational expectations on SMI, with other covariates held at their means. These AMEs are shown in Table 4. AMEs present the association in the natural metric of the dependent variable (Mize, Doan, and Long 2019; Williams and Jorgensen 2023) and facilitate cross-model comparisons by avoiding the concern of rescaling across models (Breen, Karlson, and Holm 2018; Mize et al. 2019). The AMEs in Table 4 help depict how missed and exceeded expectations are related to SMI across models generally and at different levels of educational attainment.
Average Marginal Effects of Educational Expectations on Serious Mental Illness in Midlife by Attainment Level.
Source. U.S. Department of Education, National Center for Education Statistics, High School & Beyond, 1980, 1992, 2014.
Note. Standard errors in parentheses. All values that have statistical significance have stars. The bold values in the “Missed” line do not have statistical significance and should not have stars.
p < .10. *p < .05. **p < .01 (two-tailed test).
Note that the HS&B is a school-based sample. Thus, I use appropriate weights and corrections for design effects to account for the HS&B study design and to produce population-representative estimates.
Results
Table 1 presents weighted descriptive statistics for the sample. Among women in the sample, 3.4 percent have K6 scores that approximate SMI diagnoses. This prevalence is lower than that of women in the general U.S. population. This reflects the HS&B sample design. Supplementary analysis available upon request indicates that SMI prevalence is highest among study sample members who dropped out of high school (8.0 percent). The HS&B study began in respondents’ sophomore year, thus excluding individuals who dropped out prior to Grade 10. In 1982, the normative year for high school graduation among HS&B sophomores, 13.3 percent of all U.S. women dropped out of high school (Chapman et al. 2011). In contrast, just 3 percent of the study sample dropped out of high school, meaning that it underrepresents midlife women who dropped out of high school and had the highest odds of SMI.
Row 2 of Table 1 indicates that 80.6 percent of women reported feelings of general unhappiness or depression at least once in the last month during their sophomore year. Given how frequently high school unhappiness and depression is reported on this very general survey item, it is unsurprising that it is not strongly correlated (.05) with midlife SMI.
The next rows in Table 1 indicate that more than half of women in the sample experienced misaligned educational expectations, with 37.1 percent of women missing expectations and nearly one quarter of women exceeding expectations. With respect to educational attainment, 26.1 percent earned a four-year degree and 61.4 percent attained some level of postsecondary education (Associate’s, Vocational, College, Graduate, or Professional degrees) by midlife. The majority (77.1 percent) of women in the sample were employed at midlife, job dissatisfaction is relatively low with only 11.6 percent of women reporting it, and 88.1 percent of women in the study reported excellent or good health.
As shown in the bottom rows of Table 1, the majority of women experienced good relationships with their coworkers in 1992 (95.8 percent) and stayed in female-dominated occupations from 1992 to 2014 (62.7 percent). In addition, 67 percent of women were married in 2014 and the average number of children per woman was .9. Last, in 2014, most women were not actively parenting (48 percent) or experiencing low levels of parenting stress (32.7 percent), and most women were not engaged in caregiving (76.6 percent).
Table 2 presents estimates showing how educational expectations are aligned or misaligned at different levels of academic attainment. Estimates on the diagonal that are shaded gray represent women with aligned expectations. Less than 1 percent of women expected and attained less than a high school degree, and women who expected and attained a high school degree are the largest share of the sample (20.1 percent), followed by those who expected and attained a college degree (11.3 percent).
Estimates above the diagonal represent women whose attainment exceeded their educational expectations. The largest share of women who exceeded their educational expectations are those who expected a high school degree but earned an associate or a vocational degree (9.1 percent). A similar proportion of women earned a graduate degree after expecting a college degree (6.5 percent) and a college degree after expecting an associate’s or a vocational degree (6.1 percent).
Estimates below the diagonal represent women whose attainment missed their educational expectations. The largest share of women in this category are women who expected an associate’s or a vocational degree but earned a high school degree (7.9 percent). Nearly the same proportion of women expected to earn a graduate degree but earned a college degree (4.9 percent) and expected to earn a college degree but earned an associate or a vocational degree (4.8 percent).
Now that I have contextualized the nature of aligned and misaligned expectations, I present odds ratios from four logistic regression models in Table 3 that estimate the relationship between misaligned educational expectations and midlife SMI. Estimates from Model 1 suggest women are significantly more likely to experience SMI in midlife if they have exceeded their educational expectations compared to having met their expectations. Net of educational attainment and adolescent unhappiness or depression, the odds of experiencing SMI at midlife are more than six times higher for women who have exceeded their expectations versus met them. Note that supplementary analyses also revealed that women who exceeded their educational expectations were significantly more likely to experience SMI than women who missed their educational expectations.
Model 1 also indicates that earning any educational credential also reduces the odds of SMI in midlife relative to not earning a high school degree. In addition, depression in adolescence is not significantly related to SMI in midlife. The lack of a relationship between high school depression and SMI in midlife likely reflects the general nature of the item tapping high school unhappiness and depression.
Adding variables indicating high school sociodemographic background, math test scores, and locus of control to Model 2 slightly reduces the odds of experiencing SMI in midlife among those who exceeded their expectations relative to those who met or missed expectations. None of the new Model 2 covariates are significantly related to SMI at the P≤ .05 level net of other covariates.
When I add midlife employment status, job (dis)satisfaction, and self-assessed health to Model 3, the estimated odds of experiencing SMI in midlife remain high and statistically significant for women who exceeded educational expectations compared to those who met and missed their educational expectations. Each of these measures also has estimated effects on SMI. Being employed for pay is significantly associated with a 76.5 percent decrease in the odds of experiencing SMI in midlife. In contrast, reporting dissatisfaction with a current or most recent job increases the odds of experiencing SMI by 500 percent compared to not expressing dissatisfaction. Last, for those who report good health, the odds of experiencing SMI decrease by 95 percent compared to those with fair or poor health. Adding these midlife measures also reduces the estimated effect of earning a college degree on reduced odds of SMI to the point where it is at the bounds of statistical significance at the P < .10 level. This is unsurprising given that these midlife characteristics are all strongly correlated with receiving at least a college education (supplementary analyses available upon request). Model 4 adds a more comprehensive set of family and work characteristics. This addition largely did not impact the significance and magnitude of the relationship between exceeding educational expectations and experiencing SMI. Working in a sales or office-related occupation did have a significant association with the SMI odds net of other covariates, but employment for pay and job (dis)satisfaction continue to have large, statistically significant relationships with the likelihood of SMI.
Table 4 shows the AMEs derived from the models shown in Table 3. The top panel of results indicates that missed educational expectations are not related to SMI across models, consistent with results shown in Table 3. This panel of results also shows that at every level of educational attainment, missed educational expectations and SMI are unrelated.
Conversely, the bottom panel of results in Table 4 indicates that exceeded expectations are related to SMI generally and at different levels of attainment. In Model 1, the AME for exceeding expectations is 0.061, meaning that women who exceeded their educational expectations were six percentage points more likely to experience SMI in midlife compared to women who met their educational expectations. The addition of high school characteristics in Model 2 reduces the AME of exceeding expectations by .005 (or 8 percent). In Model 3, which includes high school characteristics from Model 2 and midlife employment, job (dis)satisfaction, and self-assessed health, there is a reduction in the AME of exceeding expectations by .021 (or 38 percent) to 0.035. In Model 4, the addition of workplace and family characteristics reduces the AME of exceeding educational expectations by .004 (or 11 percent) to .031. That is, women who exceeded their educational expectations were 3.1 percentage points more likely to experience SMI than women who met their expectations. Of note, the AME of midlife job (dis)satisfaction is similar in magnitude to exceeding expectations in Model 3 and Model 4. Women who reported dissatisfaction in their current or most recent job were 3.4 percentage points more likely to experience SMI compared to women who reported being satisfied. In supplementary analyses, I tested the relationship between job satisfaction and exceeding expectations and found no significant relationship. This suggests that exceeding expectations and job dissatisfaction have separate independent effects on SMI. Notably, job dissatisfaction is measured at the same time as SMI which may fail to capture the effect of job dissatisfaction across the life course.
Table 4 shows that relationships between misaligned educational expectations and SMI are observed across all educational attainment levels at the .05 level or .10 level except for one, graduate degree attainment. Among women with a graduate or professional degree, the relationship is in the expected association but not statistically significant, likely due to the small number of cases in this category. Only 240 women earned a graduate or professional degree, and among them, few experienced SMI (suppressed according to NCES guidelines).
Before turning away from results, I want to note that to thoroughly understand these associations among women, I also considered whether similar results were found among men. They were not (shown in Table 5).
The Estimated Odds of Experiencing Serious Mental Illness in Midlife among Men.
Source. U.S. Department Of Education, National Center For Education Statistics, High School & Beyond, 1980 1992, 2014.
Note. 2.7 percent of men experienced SMI.
p < .10. *p < .05. **p < .01 (two-tailed test).
Consistent with population trends that show women experience SMI at higher rates than men, 2.7 percent of men in the sample experienced SMI compared to 3.4 percent of women in the study sample. In addition, while there are no significant effects of misaligned expectations on SMI among men, the effect of exceeding expectations is negative. In contrast to the findings for women, this would suggest that men who exceeded expectations would be less likely to experience SMI, compared to those who met their expectations. These findings align with theoretical predictions that overachieving may be distinctly distressing for women through experiences of structural sexism, and we would not expect to find the same among men, the socially dominant group.
Discussion
Women who would have expected to graduate on time as members of the high school class of 1982 experienced a rapidly changing landscape of educational expectations, educational attainment, and occupational opportunities. The availability of college expanded, and more options were available for women to seek postsecondary education. Unfortunately, the embedded gender bias within the educational institutions with which they were engaged did not change as rapidly as the opportunity to enter these spaces. I posited that this could have unforeseen consequences for mental health at midlife by considering the psychological toll of micro-level structural sexism processes and recognizing that psychological distress can result from the chronic stress of engaging in stereotype threat and self-stigmatization.
For women in the study, my results were in line with theoretical expectations. Exceeding expectations is significantly associated with increased odds of experiencing midlife SMI compared to meeting or missing educational expectations. Higher levels of education continue to be associated with a lower likelihood of SMI compared to those with less than a high school education, which aligns with existing literature on the physical and mental health benefits of education (J. O. Lee et al. 2013; Mandemakers and Monden 2013; Phillips and Hempstead 2017; Schieman and Koltai 2017). However, regardless of degree level, exceeding educational expectations is associated with an increased likelihood of SMI. These results suggest that overachieving educational expectations carries distinct psychological consequences. It is possible that processes of self-stigmatization and participation in stereotype threat explain why exceeding expectations but not missing expectations has adverse psychological consequences, especially when experienced by a socially oppressed group (e.g., women experiencing structural sexism). The lack of a statistically significant effect for exceeding expectations among men, the socially dominant group, lends additional support to this argument. Data limitations do not allow a formal test of this theoretical idea, but future research should examine this possibility.
I have largely framed the psychological toll of exceeding educational expectations for women within the cohort-specific context of a rapidly changing educational landscape. This approach may limit the applicability of the results to the cohort of women currently in midlife. However, women continue to be engaged in gendered institutions at every level of education (American Association of University Professors [AAUP] 2023; Blithe and Elliott 2020; J. J. Lee and Mccabe 2021; Musto 2019). In exploratory analyses, I have replicated the findings from this study using the Wisconsin Longitudinal Study (WLS). Women in the WLS graduated high school in 1957—25 years earlier than the women in HS&B (Herd, Carr, and Roan 2014). Arguably, women in this earlier cohort faced steeper educational barriers and psychological tolls to unexpected overachievement with only 5 percent of women holding a bachelor’s degree in the United States by 1960 (NCES 2018) and without the legal protections of Title IX of the Education Amendments of 1972 (U.S. Department of Justice 2023).
In contrast, similar preliminary analyses using the public-use version of the National Longitudinal Study of Adolescent to Adult Health (Add Health) study did not replicate findings. Women in the Add Health study graduated high school in the late 1990s (Harris et al. 2009) and over 80 percent expected to attend college which leaves little room to exceed educational expectations. The lack of replicability in the Add Health study may be attributable to measurement differences and or sample differences. Yet, it may also reflect a diminishing psychological toll of unexcepted overachievement among women as educational opportunities became more accessible. To this end, future research should thoroughly examine this relationship among other cohorts of U.S. women to consider the persistence of these observed effects.
To more critically examine the role of structural sexism as a contributor to psychological distress, this study has included measures of adulthood workplace and family characteristics as confounding variables. These measures (e.g., job satisfaction, movement between male- and female-dominated occupations, and parenting/caregiver stress) are theoretically and empirically supported confounders that have the potential to mediate the relationship between exceeding educational expectations and women’s midlife SMI. However, this study found little support for these measures as statistically significant contributors to women’s midlife likelihood of experiencing SMI, and among those with a statistically significant association with SMI, there is little evidence to suggest mediation. For example, the estimated effect of job satisfaction on SMI may provide additional support for structural sexism as a contributing stressor to psychological distress. In the fully specified model, women who express job dissatisfaction are 3.4 percentage points more likely to experience SMI versus those that do not experience job dissatisfaction. Supplementary analyses (available upon request) revealed that there is not a significant difference in job dissatisfaction between those who exceeded their expectations and those who did not. So, while this result is large and significant, it appears to be independent from the effect of exceeding on SMI. This finding is unexpected given that higher levels of education and perceptions of overeducation have been identified as contributors to job dissatisfaction (Bracke et al. 2014; Miller 1980). In addition, prior research has supported the notion that at the meso-level of structural sexism, the accumulation of stress from competing gendered roles (e.g., employee, mother, wife, caretaker) and cumulative exposure to more explicit gender bias may contribute to job dissatisfaction and psychological distress (Almeida and Fletcher 2022; Hodson 1989; Pearson 2008). Yet, the study’s final model does not provide statistically significant evidence for an adverse association between coworker relationships, occupation in a male-dominated field, nor caregiving stress and SMI. Due to varying measurement timing and sample size restrictions, the present study does not formally test mediation for these confounding variables in the relationship between exceeding educational expectations and SMI. Future research is strongly encouraged to explore the mediating role of workplace and family characteristics in the relationship of women’s overachievement and SMI.
This study provides new insight into the relationship between misaligned educational expectations and SMI through the extension of analyses into midlife, focus on exceeded expectations as a category for misalignment, and focus on women. To date, much of the research on the mental health consequences of misaligned expectations have been constrained to missed expectations and outcomes measured during early adulthood (Cundiff 2017; Kao and Tienda 1998; Mello 2009; Reynold and Baird 2010). The focus on missed expectations is understandable and based on relevant theory that strongly suggests that it should be related to psychological well-being. Yet, my study results show that overlooking the potential psychological consequences of overachievement underplays the importance of misaligned expectations—especially for structurally oppressed groups that are not expected to achieve at high levels. Study results show that prior assumptions of exceeding expectations operating similar to meeting expectations may suppress the unique effects of different directions of misalignment. It is also understandable that young adults were largely the focus of previous research, but the rising prevalence of SMI beyond age 50 combined with the changing landscape of educational experiences into midlife make it important to think about longer-term consequences of misaligned expectations.
While this study provides new evidence for an adverse relationship between misaligned expectations and SMI, there are limitations. First, I dichotomized race to signify whether women were White or non-White due to cell size issues, but this limits my ability to investigate whether findings are consistent for women in different racial/ethnic groups. Future research should take on this line of investigation given that educational expectations, attainment, and SMI all vary by gender and race in the United States.
From the early 1970s to the early 1990s, Black, White, and Hispanic girls’ educational expectations increased at a faster rate than boys’ of the same race/ethnicity (Reynolds and Burge 2008). However, the magnitude of gender-related change in expectations varied by race and ethnicity. White female students were particularly advantaged with regard to family and school contributors to expectation growth, while Black and Hispanic girls experienced less gender-related change in parental encouragement and college preparatory coursework than their White peers (Reynolds and Burge 2008). In addition, it is important to recognize the intersectional perspective of racially and ethnically minoritized women within educational institutions. Gendered racism and gendered racial microaggressions, which capture unique stressors at the intersection of race and gender, are negatively associated with Black women’s mental health across a variety of mental health indicators (Erving et al. 2022; Jones et al. 2021; Martins, Souza de Lima, and Santos 2020; Moody and Lewis 2019; Wright and Lewis 2020). Future research should incorporate an intersectional perspective to consider the role of gendered and racialized structures in shaping women’s educational expectations, attainments, and relationships with mental health.
Another study limitation is that women who dropped out of high school are underrepresented in the study due to HS&B data collection not beginning until students’ sophomore year. Improved representation of women who did not complete high school may provide insight into unexpected high school incompletion as an important case of misalignment for midlife mental illness.
Despite study limitations, highlighting the relationship between exceeded educational expectations and adverse mental health outcomes among women in midlife has implications for future research on misaligned expectations in changing social environments. The women in this study were in a position of social change that could illuminate the mental health consequences for individuals experiencing structural bias and constraints (Moen 2022). At present, the negative relationship between unexpected achievement and SMI provides new insight into the rising prevalence of SMI among midlife adults, particularly women, in the United States. Renewed attention to the relationship between misaligned expectations (including exceeding) and mental health should be considered in future research. Ongoing changes in educational expectations (Goyette 2008; Reynolds and Burge 2008; Reynolds and Johnson 2011) and the stratification of attainment, even among those who return to education (Grodsky et al. 2021), present a further changing educational landscape, especially among groups structurally constrained in their attainment including racially and ethnically minoritized students and first-generation postsecondary students. Beyond investigating the associations between misalignment and mental health, future research should also consider the underlying mechanisms of this relationship, such as the possibility of upward mobility strain or changes in subjective social status and the role of workplace and family mediators.
Supplemental Material
sj-docx-1-smh-10.1177_21568693251315167 – Supplemental material for Midlife Women’s Misaligned Educational Expectations and Serious Mental Illness
Supplemental material, sj-docx-1-smh-10.1177_21568693251315167 for Midlife Women’s Misaligned Educational Expectations and Serious Mental Illness by Kelsey Shaulis in Society and Mental Health
Footnotes
Acknowledgements
I would like to sincerely thank Dr. Michelle Frisco for her unwavering guidance and assistance. I would also like to thank Drs. Sarah Damaske and Jeremy Staff, as well as the EdSHARe Team for their feedback. Note that this manuscript has been subject to disclosure review and has been approved by the U.S. Department of Education’s Institute for Education Sciences in line with the terms of the HS&B restricted use data agreement.
Funding
The author(s) disclosed receipt of the following financial support for research, authorship, and/or publication of this article: I acknowledge support provided by the Population Research Institute at Penn State University and the Social Environments and Population Health Training Grant (NICHD, T32HD007514). PRI is funded by the NICHD (P2CHD041025). I also acknowledge the generous support of the High School and Beyond Midlife Follow-up project. The 2013–2015 wave of HSB data collection was supported by the Alfred P. Sloan Foundation (2012-10-27); the National Science Foundation (HRD1348527, HRD1348557, DRL1420691, DRL1420330, and DRL1420572); the Institute on Education Sciences (R305U140001 and R305U180002); and the Spencer Foundation (201500075 and 20160116). Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author and do not necessarily reflect the views of NICHD, NSF, IES, and other funders.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
