Committed to STEM? Examining Factors that Predict Occupational Commitment among Asian and White Female Students Completing STEM U.S. Postsecondary Programs

Abstract

Although it is well known that women have relatively high rates of attrition from STEM occupations in the United States, there is limited empirical research on the views and experiences of female STEM degree-earners that may underlie their commitment to their chosen fields. Utilizing survey data from 229 women completing STEM degrees at two U.S. universities, the present study examines how perceptions of occupational affordances and interactions with others in the field predict their occupational STEM commitment. Additionally, the study employs an intersectional lens to consider whether the patterns of association are different for Asian women and White women. Multivariate regression analyses reveal that although communal goal affordances do not significantly predict women’s occupational STEM commitment, agentic goal affordances are a strong predictor of such commitment. Regarding experiences with others in the field, results reveal that classmate interactions are not associated with STEM commitment, whereas positive faculty interactions do significantly predict such commitment. However, further analyses reveal racial differences in these patterns because agentic goal affordances are much weaker predictors of occupational STEM commitment for Asian women than for White women, and results indicate that faculty interactions are significant predictors of STEM commitment only for White women. Thus, our results strongly suggest that the theoretical models of motivation and support that underlie much of the discussion around women in STEM do not similarly apply to women from all racial backgrounds and that more research is needed that considers how both gender and race simultaneously shape STEM engagement and persistence.

Gender inequality in high-earning and high-status segments of the U.S. labor force remains a stubborn problem, most notably in science, technology, engineering, and mathematics (STEM) fields. Specifically, women in the United States. compose 47% of the total labor force, but less than 30% of the STEM labor force (National Science Board 2016). Research finds that gender inequality in STEM occupations in the United States is driven by the fact that women are less likely to enter STEM majors and advanced degree programs in the first place and, further, that women who have earned STEM degrees are less likely to enter and persist in STEM occupations after graduation (Beede et al. 2011; Glass et al. 2013; Ohland et al. 2008; Xie and Shauman 2003). Although there is large extant body of research focusing on the threshold of initial entrance into STEM postsecondary fields (Correll 2004; Mann and DiPrete 2013; Morgan et al. 2013), there is much less research examining the factors that shape the occupational persistence of female STEM degree-earners. In other words, although, empirically, women’s lower occupational persistence in STEM relative to men’s is well established, the factors behind it are not. Given that so many female STEM degree-earners choose never to enter STEM occupations, and many who do so leave quickly thereafter, it seems likely that the roots behind their ultimate exit reach back to before they graduate, when many women are likely having serious doubts about their attachment or commitment to working in STEM occupations. Consequently, this is the topic of the present study.

First, we examine whether women’s perceptions of the occupational affordances of STEM careers predict their commitment to working in STEM fields. As articulated in role congruity theory, individuals seek future roles that they perceive as likely to fulfill rather than impede their personal values (Diekman et al. 2015). Although a large body of extant research on female under-representation in STEM fields has focused on agentic goals and motivation, such as pursuing personal interests and feeling confident about the ability to be successful (Correll 2001; Wang et al. 2013), more recent work has considered the simultaneous importance of women’s communal motivation, or the value they place in promoting the interests of others beyond the self (Diekman et al. 2010; Stout et al. 2011). Therefore, we make a new contribution to the small emergent literature on this topic, which has typically considered how such factors may relate to college major aspirations or choices, to consider whether, at the critical time-point when women in STEM majors are finishing their degrees and looking to their future careers, perceptions that STEM careers will fill agentic goals and communal goals may shape their commitment to working in STEM fields.

Additionally, we consider whether and how the interactions and experiences that women have had within their STEM degree programs, both with faculty and classmates, may be influential in forming their attachment or commitment to the field. The research on chilly climate extends back for several decades, and coupled with more recent research on stereotype threat, the extant literature provides a robust theoretical and empirical accounting of women’s experiences of isolation and exclusion within male-dominated college majors (Ecklund et al. 2012; Good et al. 2012; Morris and Daniel 2008; Seron et al. 2016; Shapiro and Williams 2012). Although certainly insightful, this existing research has tended to focus on negative interactions as proximate to decisions to leave STEM majors in college, and not further considered whether interactions with others might be predictive of occupational trajectories after degree completion. We contend that even among those women who persist to earn a degree in STEM, their occupational commitment to STEM fields might be dampened by negative interactions with faculty or with classmates. Conversely, positive experiences within STEM classrooms might bolster a sense of belonging or community and therefore increase occupational commitment. Our study will investigate this possibility.

Finally, our study utilizes an intersectional lens to examine whether and how the factors that predict STEM occupational commitment may vary for women from different racial groups. Gender theorists have long called for attention to intersectionality, arguing persuasively that focusing on gender in the aggregate naively assumes that the same set of patterns, obstacles, and experiences extend to all women (Andersen 2005; Browne and Misra 2003). Still, there are nevertheless too few empirical studies that consider the STEM pathways of women from different racial or ethnic subgroups. In particular, the research literature is virtually silent regarding the experiences of women from Asian backgrounds in the United States. who have rates of representation in STEM fields that are more comparable to their same-race male peers and who have the highest relative representation in STEM fields compared to women from all racial backgrounds (National Science Foundation [NSF], 2017). Indeed, perhaps because of Asian women’s stronger numerical presence among those who earn degrees in STEM in the United States., their motivations and experiences are often ignored. We seek to address this oversight by considering whether perceptions of occupational affordances, as well as experiences with classmates and faculty, might function differently in shaping the occupational STEM commitment of Asian women relative to their White female peers.

To address these issues, we utilize data from two recent cohorts of women completing degrees in chemistry and chemical engineering at two universities in the United States. Although women’s under-representation in computer science has been a popular topic in both the press and academia recently (Sax et al. 2017), chemistry-related disciplines are another important location to examine because they undergird a long-standing, hierarchical, and male-dominated industry. Recent national statistics reveal that although women compose almost 60% of bachelor degree-earners across all fields, they remain under-represented among degree-earners in chemistry and chemical engineering (earning approximately 30% of degrees in the latter and a little more than 40% of degrees in the former), placing them at a disadvantage in terms of access to occupations in fields that are generally characterized by strong salaries and low rates of unemployment (Charette 2013; NSF, 2017). Thus, our dataset provides an opportunity to explore whether and why many young women who are at the final stages of successfully completing degrees in generally high status and in-demand (yet male-dominated) fields may not be strongly committed to working in those fields.

Our focal outcome is occupational commitment, a measure of affective attachment to STEM that captures the future desire to be in the field and the personal meaning and sense of belonging attached to it (Singh et al. 2013). Importantly, past research has found that commitment is highly predictive of subsequent persistence and satisfaction in an occupation (Fouad et al. 2016; Meyer et al. 2002). Our analyses will address three general questions. First, are young women’s perceptions of STEM fields as affording agentic as well as communal goals predictive of their occupational commitment to STEM? Second, are young women’s reports of positive experiences with faculty and classmates predictive of their occupational commitment to STEM? And finally, is the association between these factors and women’s occupational commitment different for Asian women than for White women? In addressing these questions, our study will provide new insights into the factors that ultimately underlie the attrition of female STEM degree-earners from STEM occupations, and subsequently it contributes to patterns of gender inequality in the labor force.

Gender as a Social System

Our study is situated in the recognition that gender is a multilevel social system that is created (and recreated) at the societal or macro level of institutions, at the interactional level within the local contexts of everyday life, and at and within the level of the individual (Ridgeway and Correll 2004; Risman 2004). Vital to the creation and maintenance of this system are cultural beliefs about gender, including beliefs about the personal characteristics and related behaviors that distinguish women from men. Such broadly acknowledged and endorsed beliefs function to shape the expectations people bring with them into social relations and interactions with others, and they are often accepted and internalized by individuals (Ridgeway 2001). Therefore, building on the insights of gender theorists, in the present paper we examine how factors that are internalized by young women (e.g., their agentic and their communal motivation), as well as factors present in their interactions with others (e.g., positive or negative experiences with faculty and with classmates), function to either encourage or discourage women’s commitment to male-dominated STEM fields.

Agentic and Communal Motivation

Psychological theories of decision-making, such as role congruity theory, posit that individuals will seek out and subsequently persist more in roles, including adult occupations, that are perceived as consistent with their goals and motivation (Diekman et al. 2017; Eagly and Karau 2002). Yet as individuals’ values and beliefs are created within a gender system, gender plays a strong role in shaping what individuals value and what they perceive as consistent with those values. Related to the focus of our study on women’s representation in traditionally male-dominated STEM fields, recent research has argued that the social construction of communal goals as most appropriate for women (such that female youth are raised to value and prioritize the concerns and well-being of others more so than male youth) has implications for their college major aspirations and choices (Diekman et al. 2010; Eagly 1997; Ridgeway 2001). Specifically, we argue that women’s under-representation in STEM fields is at least partly due to their higher prioritization of communal goals and, relatedly, the prevalent belief that STEM occupations are incongruent or do not afford the opportunity to realize such goals (Abele and Wojciszke 2007; Brown et al. 2015; Diekman et al. 2010). Although the limited research on young women’s communal goals focuses primarily on how they function as deterrents to either aspirations or actual entrance into a STEM college major, we posit that the importance of such goals is surely not limited to views and decisions around the transition into college. Rather, at the juncture when women are preparing to enter the workforce, perceptions of STEM fields as affording communal, other-oriented goals could bea strong motivator of commitment to working in STEM fields whereas, conversely, perceptions of STEM as incongruent with such goals likely predicts reduced commitment.

In contrast to the relatively small emergent body of empirical work investigating the role of communal motivation in STEM-related decisions, there is a very large extant body of research that has focused on the role of agentic motivation, or values and goals that promote the interests of the self or individual. The broad umbrella of agentic goals includes male-stereotyped attributes such as a preference for power and competition, as well as characteristics considered more gender-neutral (at least among recent generations), such as a desire for self-satisfaction and the need to feel competent and able (Abele and Wojciszke 2007; Ridgeway 2001). Arguably, the bulk of the social-psychological literature on women’s under-representation in STEM fields has concentrated on these later traits. For example, the expectancy-value model developed by Eccles and colleagues articulates how individuals’ expectations for success and interest in STEM domains drives their decision-making, such that those who feel that they are good at and like STEM are more likely to choose to take advanced math and science courses in high school as well as aspire toward and enter STEM fields in college (Correll 2001; Eccles 1994, 2007; Wang et al. 2013). Importantly, from this perspective, young men and women are not necessarily motivated by different types of goals; rather gender shapes STEM-related outcomes because social stereotypes and biased expectations lead female students to under-estimate their ability in STEM-related fields and to have less interest and affect toward such fields. Put differently, this body of research recognizes that both men and women are motivated by agentic goals, such as personal interest and self-efficacy, but due to cultural beliefs about gender, individuals belonging to different genders tend to generally develop interests and self-efficacy in different domains (e.g., men in STEM fields and women in the humanities).

Indeed, empirical evidence confirms that compared to their male peers, female students typically have relatively low levels of math and science interest, as well as low levels of math and science self-efficacy (despite comparable levels of achievement). Subsequently, these differences contribute to the creation of the gender gap in choice of STEM college majors (Correll 2001; Eccles 2015; Lent et al. 2005). We build on the extant research in this area, which typically focuses on understanding why young women do not aspire to or subsequently choose to study STEM in college despite being well-qualified to do so, to consider how agentic motivation might contribute to women’s occupational commitment to STEM among degree-earners. Specifically, to the extent that women perceive that STEM occupations are consistent with some of their agentic goals (such that they view STEM occupations as places where they can utilize their skills and abilities and do work that is exciting), then they will likely feel committed to such fields. Conversely, if they view STEM fields as incompatible with their talents and interest, then their sense of commitment could suffer.

In sum, we concur with scholars who posit that women’s decision-making is complex and shaped by multiple and simultaneous factors (Ridgeway 2011; Risman 2004). Specifically, recent cohorts of young women have been found to similarly value both agentic goals and more traditionally feminine communal goals because these are not necessarily contradictory sources of motivation (Diekman et al. 2017). Therefore, our study will focus on women who have persisted to finish degrees in male-dominated STEM fields to understand the extent to which their commitment to STEM occupations is related to their perceptions of the agentic, as well as the communal, affordances of STEM fields.

Interactions with Classmates and Faculty

As we mentioned earlier, prevailing cultural beliefs about gender shape not only the personal goals that individuals develop and internalize, such as the occupational goals we described, but also the expectations that individuals have of others when they engage in personal interactions. Due to men’s higher social status, expectations generally favor them in social situations, particularly in those contexts where individuals are expected to demonstrate skills or proficiencies (Ridgeway 2011; Ridgeway and Correll 2004). But, within a local context that is a stereotypically male domain (such as STEM college classrooms), men’s presence is viewed as an obvious extension of essentially masculine characteristics; therefore gendered expectations are particularly pronounced and, consequently, men’s performance and behavior is viewed very favorably. Conversely, women’s presence in these same classrooms transgresses traditional gender norms and activates stereotypes and beliefs about their presumed lower ability (Ehrlinger et al. 2018; Grunspan et al. 2016). Thus, it is not surprising that empirical research has found that interactions with peers and with faculty in STEM postsecondary classrooms can be particularly negative and exclusionary for female students. Indeed, prior research has pointed toward the “chilly climate” that women may encounter due to the gendered expectations of both faculty and peers (Sandler and Hall 1982).

Regarding interactions with faculty, research has found that professors and college instructors often favor male students in STEM classrooms, acknowledging their ideas and offering positive feedback. In contrast, female students tend to be the subject of more disparaging comments or are simply ignored (Moss-Racusin et al. 2012; Seymour and Hewitt 1997). Even when men and women in the classroom are equally subjected to faculty’s unrealistic expectations of performance and asked to prove their worth, men are doing so within a context that is viewed as naturally belonging to them. Women, on the other hand, must additionally contend with negative stereotypes and subsequent concerns of confirming such stereotypes (Cheryan 2012; Lane et al. 2012). As such, research on stereotype threat has documented the negative impact of STEM classroom environments cultivated by faculty on female students’ academic performance, as well as their sense of belonging (Good et al. 2012; Stout et al. 2011). At the same time, some studies have found evidence that positive experiences with STEM faculty, such as interactions that are supportive and encouraging, can be very beneficial for female students’ sense of belonging and future aspirations in STEM (Amelink and Creamer 2010; Lawson et al. 2018; Seron et al. 2016). Thus, the messages sent by STEM faculty are clearly important in shaping female students’ perceptions of whether or not the field is an appropriate place for them.

Although faculty signal who belongs in STEM fields from a position of authority and expertise, the signals sent by peers in the classroom may be no less important. Young women and men look to their peers for evidence of what is socially acceptable and appropriate, particularly with regard to gender roles (Eccles 2009). Yet again, women pursuing STEM fields are transgressing traditional gender norms, whereas their male peers are not. Therefore, positive interactions with peers in the field could provide needed validation and support to counteract the larger societal gender narratives about women’s presumed natural inferiority in math-related domains (Stout et al. 2011). Indeed, prior research offers strong evidence that experiences and interactions with peers in the classroom play a critical role in contributing to a local environment that is either supportive or discouraging of female students’ intentions and actual pursuit of STEM fields in college (Leaper 2015; Leslie et al. 1998).

Stepping back, the extant theoretical as well as empirical literature has highlighted the important role of faculty and peers in contributing to the maintenance of gender inequality in STEM fields, or alternatively, contributing to the deconstruction of inequality by encouraging and supporting young women to pursue and remain in STEM fields. Yet at the same time, most empirical studies on this topic have focused on early intentions and decisions within STEM pathways, such as decisions to take advanced courses in high school, to aspire toward or to declare a STEM major at the beginning of college, or to persist in a STEM major until degree completion. There is little research that extends this logic to consider how the influence of experiences with classmates and faculty within STEM fields of study might reach past the actual completion of the degree to impact women’s thoughts and feelings about their occupational trajectories.

Specifically, we suggest that as young women near completion of their STEM degrees and contemplate their future careers, the experiences they had with faculty and peers throughout their program (whether positive or negative) could have a powerful influence on their level of attachment or commitment to the STEM labor force. Women in STEM degree programs may endure years of negative interactions for a variety of reasons, including logistical difficulties associated with switching programs or a sense of personal perseverance or grit. However, this does not necessarily mean that such interactions do not influence their future STEM trajectories. Rather, it seems likely that discouraging experiences with others in the field would dampen women’s enthusiasm for committing to a lifetime career in STEM. The opposite pattern is also likely to be true, such that at the time point where they are finishing their degree and contemplating their future, women who have had more positive experiences with peers and faculty in their field would feel more committed to a STEM occupation. Our study will examine this issue.

Intersection of Gender and Race: Experiences of Asian Women

Finally, we contribute to the limited but expanding research literature that considers racial or ethnic differences in gendered STEM experiences and outcomes. Although gender theorists have long acknowledged the need to consider gender as an identity that is simultaneous and intersecting with other social identities, including race and ethnicity (Andersen 2005; Browne and Misra 2003), the empirical literature on gender inequality in STEM fields has been relatively slow to incorporate this lens. Indeed, there is a somewhat surprising lack of attention to the experiences of women from Asian backgrounds, who have the highest proportionate representation among women in male-dominated STEM fields in the United States. For example, Asian women in the United States comprise only 6% of female bachelor degree-earners overall but about 15% of female bachelor degree-earners in both chemistry and chemical engineering. Indeed, Asian women in the United States choose STEM fields at rates that are more similar to their male peers when compared to other racial/ethnic groups (Beasley and Fischer 2012; NSF, 2017). It is perhaps the very fact that female Asian students have a high representation among women in postsecondary STEM fields that translates into little attention from researchers and policymakers on understanding their experiences. We believe this is problematic because theoretical models of women’s choices and trajectories in STEM fields have been tested on predominantly White samples. They have done very little to consider whether or not such models work to explain the choices and trajectories of women from the racial group with the largest relative presence in STEM postsecondary fields, who are nevertheless also at risk of early attrition from STEM occupations. In our study, we seek to address this oversight, and thus we consider whether and how perceptions of the occupational affordances of STEM careers, as well as experiences with classmates and faculty, might function differently in shaping the STEM commitment of Asian women compared to White women in the United States.

First, there is reason to expect that racial differences in emphases on communal and agentic goals could lead to different patterns between goal affordances and commitment to STEM occupations. Some scholars have argued that broadly speaking, Asian cultures promote a social orientation that places a stronger priority on interdependence and connections between others and, at the same time, places a weaker emphasis on independence, self-direction, and self-expression (Lee and Zhou 2015; Varnum et al. 2010). Thus, Asian cultures may more strongly cultivate communal goals compared to the agentic goals that are more characteristic of individualistic cultures. To the extent that this is the case, then theoretical models that rely on the predictive power of agentic characteristics such as personal interest and confidence may not be well suited to explain the academic and related occupational choices of Asian youth. A few studies provide some empirical support for this supposition. For example, a study by Chen and Stevenson (1995) of U.S. high school students found that interest in math was a weaker predictor of math performance for students from Asian backgrounds compared to their White peers. A similar study by Tang et al. (1999) found that math interest was not a significant predictor of career choice for youth from Asian backgrounds. Specific to the focus of our study, such findings suggest that agentic goal affordances might be weaker predictors of occupational commitment to STEM for Asian women than for White women. Relatedly, communal goal affordances may be stronger predictors of occupational commitment to STEM for the former than the latter group. Our analyses will investigate this possibility.

Furthermore, some research suggests that the relatively high representation of Asians in STEM fields in the United States is at least partially due to a higher cultural emphasis on the value of STEM fields and that generally speaking, these fields are not as strongly associated with masculinity in Asian cultures (Hanson and Meng 2008; Lee and Zhou 2015; Tang et al. 1999). Relatedly, as STEM postsecondary fields tend to be disproportionately composed of Asian youth, and stereotypes of math ability favor Asians over Whites (Hanson and Meng 2008; Shen et al. 2011; Shih et al. 1999), Asian female students may not confront the same feelings of exclusion and lack of belonging in STEM classrooms compared to women from other racial/ethnic backgrounds. Indeed, there are a handful of experimental studies that have examined whether activating the Asian-math stereotype counters the salience of gender-math stereotypes, leading to higher performance on math tasks for Asian women (Cheryan and Bodenhausen 2000; Gibson et al. 2014; Moon and Roeder 2014). If Asian female students perceive that stereotypes and expectations of female inferiority in math fields do not particularly resonate with or apply to them, then they may be less reliant on support and encouragement from faculty and/or peers to bolster their sense of belonging (Williams and Phillips 2016). As such, positive experiences with faculty and classmates may be less important for shaping their occupational commitment to STEM. Building on the very limited research in this area, our study will therefore examine whether classmate and faculty interactions are potentially weaker predictors of STEM commitment among Asian women than among White women.

Current Study

The present study focuses on the critical time-point when women in male-dominated STEM fields are finishing their degrees and looking to their future careers to examine whether their perceptions of the occupational affordances of STEM careers, as well as their experiences with others in the fields, predict their commitment to working in STEM fields. Further, we examine potential racial variation in whether and how these factors predict commitment, with a focus on examining differences between Asian women and White women attending U.S. universities. Because prior research has found that feelings of occupational commitment are highly predictive of subsequent occupational decision-making and satisfaction, such that individuals who express low levels of commitment are very unlikely to persist (Fouad et al. 2016; Singh et al. 2018), the results from our study will contribute to our understanding of the factors behind women’s departure from and ultimately low levels of representation in STEM occupations.

Specifically, we will test four hypotheses. First, we expect that perceptions that STEM fields are congruent with agentic goals will predict greater occupational commitment to STEM (Hypothesis 1a) and that this association will be stronger for White women compared to Asian women (Hypothesis 1b). Second, we expect that perceptions that STEM fields are congruent with communal goals will also predict greater occupational commitment to STEM (Hypothesis 2a) but that this association will be stronger for Asian women compared to White women (Hypothesis 2b). Third, we hypothesize that positive interactions with classmates will predict greater occupational commitment to STEM (Hypothesis 3a) and that this association will be stronger for White women compared to Asian women (Hypothesis 3b). Fourth, similar to interactions with classmates, we propose that positive interactions with STEM faculty will predict greater occupational commitment to STEM (Hypothesis 4a), and that this association will be stronger for White women compared to Asian women (Hypothesis 4b).

Data and Methods

Participants

Our study utilizes survey data collected from female students in two graduating cohorts (2016 and 2017) in chemistry and chemical engineering programs at two universities in the United States: one a large, selective public university and the other, a private selective university. Before our study began, Institutional Review Board (IRB) approval was obtained from both institutions. Both undergraduate and graduate students were recruited via email to complete a survey about their experiences in their program. Students were surveyed online during the end of the Spring semester of the last year of their degree program, just before their graduation (which was later confirmed as occurring for all participants). The response rate was about 60% at the public university and about 70% at the private university.

We collected complete surveys from 252 women (M_age = 23.01, SD = 2.51, range = 21–35). Students self-identified the category that best described their race/ethnicity, such that 144 (57%) identified as non-Hispanic White; 85 (34%), as Asian; 19 (7%), as Hispanic; and 4 (1%), as Black. The racial-ethnic representation of the sample is consistent with that of the chemistry and chemical engineering programs in both universities and, generally speaking, consistent with national patterns (NSF, 2017). Given their very low numbers, we do not include Hispanic and Black women in the analyses presented here. Our final analytic sample includes a total of 229 women who racially identify as Asian or White.

Variables

Our dependent variable captures women’s future commitment to working in STEM. We use a scale originally designed by Meyer and Allen (1991) that has been used in a large body of research on commitment to occupations and found to be highly predictive of subsequent satisfaction and retention or attrition (for example, see Aven et al. 1993; Singh et al. 2018). Because it captures the desire to be in the field in the future as well as feelings of attachment, it is also complementary with measures of inclinations toward STEM often used in research on gender. It is comprised of the following five items: (a) “I would be very happy to spend the rest of my career working in chemistry/chemical engineering,” (b) “I enjoy discussing chemistry/chemical engineering with people outside it,” (c) “I think that I could easily become as attached to another profession as I am to chemistry/chemical engineering,” (d) “Chemistry/chemical engineering has a great deal of personal meaning for me,” and (e) “I do not feel a strong sense of belonging to chemistry/chemical engineering” (α = .78). Responses ranged from 1 (strongly disagree) to 7 (strongly agree), with the third and fifth items reverse coded so that a higher averaged overall score corresponds to a higher level of commitment.

To capture students’ perceptions of both the agentic and communal occupational affordances of STEM fields, we created two scales derived from other research on this topic (Diekman et al. 2010, 2015). Beginning with agentic affordances, we included two items: (a) students’ report of whether their field (chemistry/chemical engineering) “will allow me to use my talents and abilities” and (b) their report of whether their field “will allow me to do exciting work” (α = .83). To measure perceptions of communal occupational affordances, we also create a scale from two variables: (a) students’ report of whether their field “will allow me to do work that can make a difference in people’s lives” and (b) whether their chosen field “will allow me to have a career that is valued by my family” (α = .78). Response categories for both measures range from 1 (strongly disagree) to 7 (strongly agree).

Additionally, drawing on previous measures of peer and faculty interactions in college (Kuh and Hu 2001; Sax et al. 2005), we include variables to capture students’ experiences within the major. Students were asked: “How satisfied are you with the following aspects of your major?” The first variable captures positive classmate interactions, and it is created by taking the average of students’ reports of satisfaction with (a) “social interactions with classmates” and (b) “academic interactions with classmates” (α = .85). Following the same prompt, the second variable is students’ reports of satisfaction of “experiences with faculty”. Categories of response for all items ranged on a scale from 1 (very dissatisfied) to 7 (very satisfied). Thus, for each variable, a higher score indicates more satisfying interactions.

Because occupational commitment might vary based on other characteristics of students, our multivariate models also include controls for type of major (coded 1 for chemical engineering vs. 0 for chemistry), degree level (coded 1 for undergraduate vs. 0 for graduate), and institution (coded 1 for public selective vs. 0 for private selective university). Additionally, we include students’ self-reported grade point average (GPA) as a control because those students who have been more successful in their degree programs might logically feel more committed to the field. Race is included in all models, distinguishing between Asian students (coded 1) and White students (coded 0 as the reference).

Analysis Plan

After presenting descriptive statistics, we turn to multivariate models. Specifically, we use linear regression analyses to predict female students’ STEM commitment. In the first model we include control variables and measures of agentic and communal occupational affordances (to test Hypotheses 1a and 2a, respectively) as well as classmate and faculty interactions (to test Hypothesis 3a and 4a, respectively). Due to the relatively small sample size and related collinearity issues with including multiple interactions in the same model (Echambadi and Hess 2007), we then test the interaction between students’ race and each of the four focal independent variables separately in Models 2–5 to see whether the effects are stronger or weaker for Asian women compared to their White peers (Hypotheses 1b, 2b, 3b, and 4b). All analyses were conducted using the statistical software package Stata.

Results

Preliminary Results

As seen in Table 1, the overall mean for STEM commitment is 4.17, which corresponds to an average response of neutral, or Bneither agree nor disagree.” The means between White and Asian women are not statistically different, t(227) = 1.13, p = .260, d = .16,. Therefore, it appears that on average, in the last semester of their STEM degree program, young women from both racial groups have quite an ambivalent sense of occupational commitment to their STEM field. Although Asian women have higher relative rates of representation among degree-earners nationwide in both fields as compared to their White female peers (NSF, 2017), the occupational commitment of Asian women and White women in our sample is comparable.

Table 1.

Descriptive statistics overall and separately for Asian and White women

Variables	Overall	Asian women	White women
	M (SD)	M (SD)	M (SD)
Dependent Variable
Occupational Commitment to STEM Field	4.17 (1.27)	4.04 (1.13)	4.24 (1.34)
Independent Variables
Goal Affordances
Agentic Goal Affordances	5.34 (1.34) a	5.19 (1.27)	5.43 (1.37)
Communal Goal Affordances	5.61 (1.22)	5.42 (1.14)	5.73 (1.26)
Interactions with Others
Positive Classmate Interactions	5.64 (1.35) b	5.59 (1.39)	5.66 (1.33)
Positive Faculty Interactions	5.06 (1.46)	4.87 (1.39)	5.17 (1.50)
Control Variables
STEM Field (Chemical Engineering vs. Chemistry)	.56 (.50)	.71 (.46) c	.48 (.50)
Degree Level (Undergraduate vs. Graduate)	.74 (.44)	.79 (.41)	.72 (.45)
Institution (Public vs. Private)	.39 (.49)	.34 (.48)	.42 (.50)
Grade Point Average	3.50 (.34)	3.54 (.32)	3.48 (.35)

n = 85 for Asian women, n = 144 for White women, n = 229 for overall analytic sample

^athe mean for all female participants on agentic goal affordances is significantly different from the mean on communal goal affordances (results of paired t-test, p < .001, d = .22).

^bthe mean for all female participants on positive classmate interactions is significantly different from the mean on positive faculty interactions (results of paired t-test, p < .001, d = .41).

^cthe mean for Asian women on STEM Field is significantly different from the mean for White women (results of independent t-test, p < .001, d = .45)

Turning to our independent variables capturing occupational affordances, respondents have an average of 5.34 (leaning toward “slightly agree”) on the scale for agentic occupational affordances, whereas the mean for communal occupational affordance is 5.61 (leaning toward “agree”). Paired samples t-tests confirm that there is a significant difference between the two means, such that across the full sample, women report significantly higher levels of communal occupational affordances for STEM fields compared to agentic occupational affordances, t(227) = 4.59, p < .001, d = .22, although this difference is small in magnitude. Additionally, results of independent samples t-tests reveal that the means for both agentic, t(228) = 1.31, p = .193, d = .18 and communal, t(227) = 1.87, p = .063, d = .25, occupational affordances are statistically equivalent between White students and Asian students.

Regarding independent variables capturing experiences with others in the field, the mean for classmate interactions is 5.64, such that on average female students lean toward agreeing that interactions are positive. The mean for faculty interactions is lower at 5.06 (or “slightly agree”). Paired samples t-tests confirm that this difference is statistically significant and moderate in size, t(228) = 5.67, p < .001, d = .41, so that on average female students report more satisfaction with their interactions with peers in their field compared to faculty. As with the measures of occupational affordances, although the means for White women on both variables appear somewhat larger than those for their Asian peers, the differences between racial groups on both variables are not statistically significant (peers: p = .686, d = .05; faculty: p = .138, d = .20).

Table 1 also displays means and standard deviations for control variables in the analyses. The only significant racial difference is for STEM field such that Asian women are more likely to be in chemical engineering than White women (p < .001). There are no racial differences in institution attended, degree level, or grade point average. Further, in supplemental analyses, we found that there are no racial differences in either age of respondent or level of paternal education.

Multivariate Results

Table 2 displays the results of multivariate regression models predicting students’ occupational commitment to STEM. Model 1 includes control variables and measures of occupational affordances and experiences with others in the field. Beginning with affordances, we observe that agentic occupational affordances are positive and statistically significant (B = .519, p < .001), indicating that the more that women perceive STEM fields to be consistent with their agentic motivation, the more they are committed to working in STEM. This supports Hypothesis 1a. In contrast, we do not find support for Hypothesis 2a because the coefficient for communal occupational affordances is close to zero and not statistically significant (B = −.097, p = .314). Moving to experiences with others in the major, we do not find support for Hypothesis 3a because positive interactions with classmates are not a significant predictor of women’s occupational commitment (B = −.052, p = .406). However, interactions with faculty are significant (B = .225, p < .001) in support of Hypothesis 4a. Specifically, higher levels of positive faculty interactions predict higher levels of occupational STEM commitment.

Table 2.

Predicting female students’ STEM occupational commitment: Results from linear regression models

	Model 1	Model 2	Model 3	Model 4	Model 5
	M (SE)	M (SE)	M (SE)	M (SE)	M (SE)
Occupational Affordances
Agentic Goal Affordances	.519*** (.072)	.529*** (.072)	.608*** (.082)	.508*** (.073)	.507*** (.071)
Communal Goal Affordances	−.097 (.096)	−.040 (.116)	−.090 (.094)	−.097 (.096)	−.099 (.095)
Interactions with Others
Positive Classmate Interactions	−.052 (.063)	−.053 (.062)	−.058 (.061)	.005 (.083)	−.035 (.062)
Positive Faculty Interactions	.225*** (.062)	.225*** (.062)	.224*** (.063)	.233*** (.061)	.296*** (.070)
Race Interactions
Asian * Agentic Goal Affordances		−.270* (.116)
Asian * Communal Goal Affordances			−.201 (.147)
Asian * Positive Classmate Interactions				−.140 (.111)
Asian * Positive Faculty Interactions					−.216* (.099)
Control Variables
Asian (vs. White)	.019 (.141)	1.440* (.652)	1.125 (.843)	.805 (.658)	1.090* (.510)
Chemical Engineering (vs. Chemistry)	−.242 (.126)	−.225 (.123)	−.215 (.125)	−.23 (.126)	−.206 (.127)
Grade Point Average (GPA)	.215 (.234)	.171 (.225)	.168 (.225)	.191 (.229)	.192 (.229)
Undergraduate program (vs. Graduate)	−.124 (.149)	−.150 (.147)	−.153 (.148)	−.123 (.148)	−.177 (.154)
Constant	.544 (.975)	.225 (1.035)	.344 (1.039)	.312 (1.028)	.239 (.962)
R2	.394	.412	.402	.400	.407

n = 229. Robust standard errors are in parentheses

^*p < .05.

^**p < .01.

^***p < .001

Additionally, we note here that as seen in Model 1, none of the control variables significantly predict STEM commitment. In other words, students at both institutions have comparable levels of STEM commitment as do students in both fields, students at different degree levels, and students with different grade point averages. Further, consistent with descriptive statistics in Table 1, Asian and White female students report comparable levels of STEM commitment.

We now turn to results of regression models that include race interactions. Beginning with the results in Model 2, there is a significant negative interaction between race and agentic goals (B = −.270, p = .021). This indicates that compared to their White peers, agentic occupational affordances are weaker predictors of occupational commitment for Asian female students, consistent with Hypothesis 1b. The interaction term between race and communal goals is not significant (p = .172), nor is the interaction with classmate experiences (p = .207); thus, we do not find support for Hypotheses 2b or 3b. However, consistent with Hypothesis 4b, there is a negative and statistically significant interaction between race and faculty experiences in Model 5 (B = −.216, p = .031), revealing that compared to White women, Asian women’s occupational STEM commitment is not as strongly associated with positive interactions with faculty in the field.

To better illustrate the significant race interactions with both agentic goal affordances and faculty interactions, we calculated predicted outcomes for each of these focal variables by race, holding all other variables in the model at the mean (utilizing the margins post-estimation command in Stata). First, Fig. 1 displays racial differences in the association between agentic occupational affordances and predicted levels of STEM commitment. For White women, as their perceptions that STEM fields afford the opportunity to fulfill agentic goals increase from one standard deviation below the mean to one standard deviation above, their commitment increases by more than 1.5 points. In contrast, for Asian women, the increase is almost half that observed for White women. Follow-up tests calculating simple slopes revealed that the slope is significant for both White women (p < .001), and Asian women (p = .026); thus, although agentic occupational affordances are associated with higher STEM commitment for both groups, the link is clearly stronger for White women compared to their Asian peers.

Fig. 1.

Association between agentic goal affordances and STEM occupational commitment. The slopes for both groups are statistically significant (p < .001 for White women and p < .05 for Asian women). The original response scale ranged from 1 to 7; the y-axis was truncated to more clearly display the association

Figure 2 displays parallel results for interactions with faculty. Once again we see that the positive associations between the focal variable and STEM commitment diverges for Asian women and for White women. Specifically, as satisfaction with faculty interactions increases from one standard deviation below the mean to one standard deviation above the mean, White women’s STEM commitment increases about .9 of a point on the commitment scale. For Asian women the increase in commitment is much smaller, about .2 of a point on the commitment scale. Further, follow-up tests calculating simple slopes revealed that this slope for Asian women was not significant, (p = .503), whereas the slope for White women was significant (p < .001).

Fig. 2.

Association between positive faculty interactions and STEM occupational commitment. The slope for White women is statistically significant (p < .001); the slope for Asian women is not significant (p > .05). The original response scale ranged from 1 to 7; the y-axis was truncated to more clearly display the association

Supplemental Analyses

To ensure the robustness of our results, we ran models with additional control variables. First, we conducted models with parental education level as a proxy for social class; it was not a significant predictor nor did its inclusion change our results. Furthermore, we added a control for whether or not respondents were born in the United States because the Asian women in our sample were more likely to be born abroad than White women were. The measure was not a significant predictor of occupational commitment and its inclusion did not change our results. Additionally, we ran models that included the responses of the very small number of Black and Hispanic women, which revealed that there was not a main effect for either Black or Hispanic on the outcome, nor were there significant interactions between Black and Hispanic and communal or agentic affordances, or classmate or faculty interactions. These null effects are not surprising given their very low numbers in the sample. Finally, we conducted analyses that were limited to undergraduate students only, and additional analyses separately for those in chemical engineering and those in chemistry. In all cases, the results were parallel to those shown here.

Discussion

In this study, we suggest that the roots behind women’s comparatively high exit rates from STEM occupations in the United States likely lead back to their thoughts and experiences while they were pursuing their degree. As women are completing their degrees and looking ahead to life after graduation, many are likely having serious doubts about whether they want to work in STEM, yet there is very limited literature that considers this topic. We posit that while young women are immersed in their degree programs, they are exposed to daily information about how this field fits, or does not fit, with their values and priorities. Additionally, the signals and messages they receive from those around them as they pursue this gender non-normative field likely also shape their sense of commitment to the field. Therefore, we examined whether perceptions of agentic and communal occupational affordances, as well as experiences with classmates and faculty, are associated with women’s commitment to STEM fields. In doing so, we employ an intersectional lens and further considered whether among female students attending two U.S. universities, the patterns vary between White women and their Asian female peers, a group that is often ignored in the literature despite (or perhaps because of) their relatively high levels of representation among women in STEM fields of study.

Beginning with occupational affordances, we hypothesized that when young women perceived that STEM fields were consistent with agentic motivation (e.g., being able to use their skills and talents and do work that was interesting and exciting), they would have higher levels of occupational commitment (Hypothesis 1a); and drawing on recent literature that highlights the importance of communal goals for women, we hypothesized that perceptions of communal goal affordances would also predict their commitment (Hypothesis 2a). Further, we hypothesized that agentic goal affordances would matter less for Asian women compared to White women (Hypothesis 1b) and that communal goal affordances would matter more for Asian women than White women (Hypothesis 2b). Our quantitative analyses provide empirical support for both of our hypotheses regarding agentic goals, and neither of our hypotheses for communal goals. Specifically, as seen in Table 2, we found that agentic occupational affordances were a strong predictor of occupational commitment; however, interaction effects revealed that this effect was statistically and significantly weaker for Asian women than for White women. In contrast, we found evidence of neither a main effect nor an interaction effect with race for communal goal affordances, indicating that such goals do not significantly predict occupational commitment for either White women or Asian women in our sample.

In terms of making sense of these results, we note that our findings regarding agentic occupational affordances are consistent with a large body of literature on the importance that individuals, including women, place on feeling that a particular field suits their abilities and interests (Correll 2001; Eccles 2007; Wang et al. 2013). Yet as noted before, the literature on gender and STEM has typically focused on how this plays out in terms of early aspirations or decisions to declare STEM majors in college. Our study extends this literature by demonstrating that such factors continue to matter upon degree completion, as women contemplate their level of commitment to STEM occupations, which is itself known to be a strong predictor of actual occupational persistence (Fouad et al. 2016). Even so, our study offers a critical caveat to this finding, as our results indicated that agentic occupational affordances are relatively weak predictors of commitment for Asian women in our sample. While there is very limited prior research on Asian female students in the United States within the gender/STEM literature, this result is consistent with some prior research comparing motivation for achievement and career choice between students from different racial backgrounds, which found evidence that typical agentic motivators like interest and self-efficacy were weaker predictors in general for Asian students compared to White students (Chen and Stevenson 1995; Tang et al. 1999). As such, this speaks to the need for theoretical frameworks commonly invoked in gender research, such as role congruity and expectancy-value theory, to more explicitly address how both gender and race, as simultaneous social identities, likely shape not only how individuals perceive certain occupations, but also the importance of such perceptions in shaping future experiences and decisions.

Unlike our hypotheses for agentic goal affordances, our hypotheses that communal goal affordances would predict STEM occupational commitment for women, and that this effect would be stronger for Asian women, were not supported. Looking back at the descriptive statistics displayed in Table 1, it is worth noting that women’s perceptions that STEM fields were consistent with communal motivation was on average quite high. Perhaps this indicates that women who felt that STEM fields were very inconsistent with such goals never entered a STEM major to begin with, or if they did enter, left very quickly (Diekman et al. 2010). In contrast, women who chose and persisted through a STEM degree (both White women and Asian women) may stand apart from the majority of other women by perceiving that STEM fields are consistent with communal goals; at the same time they may also stand apart from their gender in terms of being more motivated by agentic occupational goals than communal ones. Given that research on women who complete male-dominated STEM degrees in the United States is quite limited, our discussion is speculative in nature and warrants further consideration. For example, future studies could compare both the agentic and communal occupational affordances of women from diverse racial backgrounds in different types of fields (including STEM and non-STEM) and at different points in their educational and occupational trajectories.

Beyond occupational affordances, we also hypothesized that occupational commitment to STEM fields would be strengthened as a result of positive experiences with classmates (Hypothesis 3a) and faculty (Hypothesis 4a), and that such effects would be stronger for White female students compared to Asian female students (hypotheses 3b and 4b). Here again we find mixed support for our hypotheses, as we found empirical support for those regarding faculty interactions but not for those regarding classmate interactions. Beginning with the former, we found that women who reported higher levels of satisfaction with their interactions with faculty had significantly higher levels of occupational commitment to STEM; yet subsequent analyses revealed that this increase was significant only for White women, and not for their Asian peers. The positive results for White women are consistent with research that suggests that women in male-dominated STEM fields look to others for validation and encouragement that they belong in these fields (Seron et al. 2016; Stout et al. 2016). At the same time, given that Asian women are relatively highly represented in such fields, they may be less likely to look for such support. Further, validation from faculty may be less important due to a cultural emphasis on the value and importance of STEM fields more generally (Lee and Zhou 2015). If so, this suggests that the further development of theoretical frameworks that consider both gender and race as socially constructed and inter-related systems of inequality could help us to better understand how prevailing cultural beliefs shape the form and the meaning of interactions both inside and outside of STEM classrooms.

Finally, our results regarding interactions with classmates are somewhat surprising in the context of the research cited above, as we found no evidence that satisfaction with interactions with classmates in in their field predicted occupational commitment in our multivariate models. This null finding could be because feedback from current peers (either positive or negative in nature) might be somewhat discounted or carry little weight as women plan their long-term occupational future. We further address this null finding in the section below.

Limitations

As with any study, ours has certain limitations. First, we note that our measures of interactions with faculty and classmates were gender non-specific. Given that some research on same-gender peers and role models finds that they can serve as a source of ‘innoculation’ against gender-STEM stereotypes (Stout et al. 2011; Young et al. 2013), it is possible that we might have observed significant classmate effects or found even stronger faculty effects if respondents reported about female classmates and faculty, respectively. Yet at the same time, we note that research on this topic is not conclusive, as some studies have found that the gender composition of peers and faculty does not matter for predicting women’s STEM-related outcomes (e.g., Bettinger and Long 2005; Pennington et al. 2018), while other studies have found evidence suggesting that same-gender peers may sometimes discourage female students’ pursuit of STEM fields, and male students may sometimes act as critical allies (Pietri et al. 2018; Robnett and Leaper 2013). Given that by definition, men comprise the majority of both classmates and faculty within these male-dominated fields, we suggest that future research should not only consider how the gender of faculty and classmates predicts women’s occupational commitment in STEM, but also consider more directly what male classmates and faculty can do (or refrain from doing) that would be most impactful for increasing the representation of racially diverse women in STEM occupations.

Our focus on intersectionality is also necessarily limited. In addition to the absence of a sufficient number of Black and Hispanic women in our data to study, we are also limited by the inability to disaggregate within the broad racial category of Asian. This likely obscures substantial heterogeneity of background, as we are not able to distinguish those from East Asian or South Asian backgrounds, for example. Further, while our study strongly suggests that the theoretical models of motivation and support that underlie much of the discussion around women in STEM do not necessarily apply to the experiences and obstacles faced by Asian women in the United States, we are not able to provide an alternative account of what does matter strongly for Asian women’s decision-making. And importantly, while Asian women have relatively high rates of representation among female STEM degree-earners in the United States, they nevertheless share with their White female peers a high risk of departing STEM occupations (NSF, 2017), a fact echoed by the seemingly low levels of occupational commitment of both groups in our sample. Thus, while our study contributes new information to the very limited literature on Asian women in STEM in the United States, clearly much more theoretical and empirical work is needed.

Finally, we note that our conclusions are limited by the cross-sectional nature of our data. We cannot ascertain, for example, whether women’s occupational commitment changes over the course of their education (either increases or decreases) in relation to the factors we consider. For example, it is possible that those who begin their program with relatively high levels of commitment to the field may seek out interactions with faculty, thus contributing to the pattern we observe here. Yet of course our results also point to variation by race, which should be simultaneously considered in longitudinal research on this topic. We look to future research to further unpack the interplay of gender and racial identities over time, both throughout and across the transition between women’s STEM educational and occupational paths. We think that qualitative research in particular might lead to more critical insights and understanding on this important topic.

Practice Implications

The results of our study have several implications for programs and interventions aimed at promoting women’s representation in STEM fields. First, currently there are many efforts at highlighting the social benefits of STEM fields, which is consistent with the idea that communal occupational affordances are an important driver at increasing women’s presence. For example, national ad campaigns by Exxon and other large corporations, as well as smaller scale efforts within departments to include service projects in STEM curriculum, all fall under this umbrella, may be worthwhile for many reasons. However, it is important to note that such efforts are likely to be insufficient at increasing women’s occupational persistence in STEM if corresponding efforts are not also made to facilitate women’s beliefs that STEM occupations are also consistent with their agentic beliefs, such that they feel that such occupations are congruent with their skills and abilities (Cech et al. 2011).

Finally, our findings regarding the much weaker (or null) effects for Asian women compared to White women points to the need for further attention to intersectionality. For example, we suggest that organizations dedicated to supporting women in STEM fields in the United States (such as national organizations like the Society for Women Engineers (SWE), as well as local organizations and clubs at the university level) need to pay special attention to ensure that they are aware of and responsive to the needs and priorities of women from different racial and ethnic backgrounds; this includes focusing on those from groups with relatively high levels of representation (e.g., Asian women), as well as those from racial and ethnic backgrounds who have much lower levels of representation.

Conclusion

The relatively low presence of women in STEM occupations is a critical manifestation of gender inequality in the contemporary United States. Alleviating this disparity will require not only continued attention to the factors that deter young women from pursuing STEM fields of study in the first place, but also a greatly increased focus on factors that shape the decision-making of women who are earning degrees in these fields. While this study makes a contribution to the latter, there is clearly much more work to be done, particularly that which increases our understanding of how the social systems of gender and race intersect to shape women’s attitudes, experiences, and outcomes in STEM fields.