High Hopes: Gender Trends in Educational Expectations for Graduate and Professional School, 1976-2019

ABSTRACT

Educational expectations have increased over time, with greater increases among young women than men, yet research focused on expectations for post-baccalaureate degrees is limited. We investigate young men’s and women’s plans to attend graduate or professional school using Monitoring the Future data from 12th graders for 1976 to 2019, focusing on how academic performance and work and family values may be associated with post-baccalaureate expectations. We find that young women’s expectations for graduate or professional school began to exceed young men’s in the early 1990s and continued to do so afterward, although expectations for post-baccalaureate schooling declined some in recent years, especially among young men. Results also indicate that the gender gap over time is driven partially by more young women than men with B or lower average grades holding post-baccalaureate expectations. Work values may foster these high expectations, especially for lower-achieving young women. Finally, we examine whether post-baccalaureate expectations translate into higher attainments, and results suggest that higher-achieving students are better positioned to meet their post-baccalaureate expectations. Collectively, our findings suggest that sociocultural factors promoting women’s participation in the public sphere may encourage some young women to form high-level expectations that they are not academically equipped to meet.

KEYWORDS:

• Gender
• post-baccalaureate educational expectations
• academic performance

Disclosure Statement

No potential conflict of interest was reported by the authors.

Supplementary Material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/00380253.2023.2177207

Notes

1. Educational expectations, as measured by reporting one probably or definitely will complete a given level of schooling, also have been referred to as educational ambitions (Reynolds and Johnson 2011). The terms “educational expectations” and “educational ambition” contrast with “educational aspirations,” which has been treated as an indicator of how much education one ideally would like to or desires to achieve and therefore is considered a poorer indicator of attainment (e.g., S. Hanson 1994; Marini and Greenberger 1978).

2. Women first earned a larger share of bachelor’s degrees in 1981–1982, master’s degrees in 1986–1987, and doctoral degrees in 2005–2006 (Aud et al. 2012). Women first outnumbered men in law school in 2016 (Pisarcik 2019) and in medical school in 2019 (American Association of Medical Colleges 2019).

3. Educational expectations have been conceived of differently in other work, see, for example: Fishman (2019); Frye (2012); Morgan (2004).

4. As a robustness check, we ran two multinomial logistic regression models. The first model has three outcomes and compares those who are certain of their plans to attend graduate school (“Definitely Will”) to those who are not planning to attend graduate school (“Definitely Won’t” responses) and to those who are not sure (“Probably Will” and “Probably Won’t” responses). The “Probably Will” and “Probably Won’t” responses are collapsed into a single category and serve as the reference category in this model. The second model includes all four response categories as separate outcomes with “Probably Won’t” as the reference category. The results we present here are qualitatively unchanged under these conditions; therefore, we only present the results of the binary logistic regression models. Multinomial logistic regression results are available upon request.

5. Given how the MTF data were collected, we are forced to operationalize gender as strictly binary. This is a significant limitation of MTF and most other large-scale, repeated cross-sectional datasets (Garvey et al. 2019; Westbrook and Saperstein 2015).

6. Because less than two percent of the sample reported average grades of D or below, we collapsed the categories for D average or below and C average into a single category to avoid skewed coefficients.

7. Results with sample weights are substantively unchanged from those without sample weights. We present models with sample weighting.

8. Our regression takes the following form:

$\begin{array}{rl}ln\left(\frac{\hat{p}}{1-\hat{p}}\right)& =\alpha +{\beta }_1\left(Gender\right)+\hspace{0.17em}{\beta }_2\left(Year\right)+\hspace{0.17em}{\beta }_3\left(Yea{r}^2\right)+\hspace{0.17em}{\beta }_4\hspace{0.17em}\left(Grades\right)+\hspace{0.17em}{\beta }_5\left(Prep\right)+{\beta }_6\left(Cont.\right)\\ & +\hspace{0.17em}{\beta }_7\left(Values\right)+\hspace{0.17em}{\beta }_8\left(\int .\right)\end{array}$

where $\hat{p}$ is the probability of planning to attend graduate school in the future. Gender is an indicator variable for women with the corresponding coefficient captured in β_1. Year and Year² variables are captured with corresponding coefficients in β₂ and β₃, respectively. Grades is a variable with indicators for A and B average students with C average or below students as the reference category and the corresponding coefficients captured in β₄. Prep is an indicator for students who participated in a college preparatory high school program with the corresponding coefficient captured in β_5. Cont. is a vector of all of our control variables with the accompanying β₆ vector of coefficients. Values is a variable with indicators for the importance of family and importance of paid work indicators that are included in some of our models with the accompanying β₇ vector of coefficients. We also include several interactions in the Int. vector with the accompanying β₈ vector of coefficients.

9. The proportion of MTF seniors with averages of C or lower declined over time, which is consistent with other research (Pattison, Grodsky, and Muller 2013). Supplemental figures showing grade changes over time for the women and men in our study are found in the online appendix (Figures A-3 and A-4, respectively).

10. We use “pattern” here to signal a statistical association and not a causal relationship. While the theoretical tradition we are building on here is grounded in causal assumptions, the statistical tests that we run cannot fully establish causality. We view the term “pattern” as a relatively neutral term, and we do not imply any form of causal relationship in its use.

11. We use the panel data weight, V106, that is provided with the data files and is recommended for use with all analyses. This follow-up weight variable accounts for the oversampling of drug users into the longitudinal sample by a ratio of 3:1 (and has values of either 1 or .3333).

12. To calculate the attrition weight, we first use logistic regression to calculate the predicted probability of participating in the sixth follow-up (when respondents were ages 29–30) based on the following indicators: gender (male = 1, female = 0), indicators for Black, Hispanic, and respondents from other racial groups with White respondents as the reference category, whether at least one parent has a college degree, the number of parents in the household, whether the respondent had at least a C plus (77–79%) average in high school, whether the respondent planned to attend a four-year college, the number of days cut from school in the past four weeks, frequency of past 30-day cigarette use, frequency of binge drinking (five or more drinks) in the past two weeks, frequency of marijuana use in the past 12 months, and dummy variables for each cohort. To generate the attrition weight, we multiply the inverse of the probability of participating in the fourth, fifth, or sixth follow-up waves by the sampling weight variable (V106).

Additional information

Notes on contributors

S. Abby Young

S. Abby Young is a doctoral degree student at the University of Oklahoma. Her research currently focuses on the application of diverse statistical modeling techniques to examine change in gender attitudes and expectations over the life course.

Ann M. Beutel

Ann M. Beutel is an associate professor of Sociology at The University of Oklahoma. Her research focuses on the influence of social location on the values, attitudes, and expectations of adolescents and adults and on the relationship between gender and experiences in education and the workplace. Her work has appeared in such journals as the American Sociological Review, Social Science Research, Sociological Perspectives, and Demographic Research.

Stephanie W. Burge

Stephanie W. Burge is an associate professor in the Department of Sociology at the University of Oklahoma. Her research investigates the transition to adulthood, with an emphasis on gender differences in adolescents’ educational and career plans and their link to eventual attainments in young adulthood.