Introduction

In Australia, women dominate jobs in healthcare and social assistance (79%) and education and training (73.2%) (Australian Bureau of Statistics, 2018), but are underrepresented in mathematics-intensive science, technology, engineering and mathematics (STEM) fields such as engineering (12.4%; Kaspura, 2017), information technology (28%; Australian Computer Society, 2018) and astronomy (21%; Maddison et al., 2015). Among the reasons put forward by researchers to explain gendered participation in mathematics-related careers, ‘communal’ goals to help others in society versus ‘agentic’ occupational goals for self-promotion, wealth and status, have been identified as a contributor (e.g., Diekman et al., 2011).

The literature on occupational goals and mathematics-related careers is predominated by cross-sectional studies conducted in secondary or post-secondary educational settings, the latter mostly among U.S. psychology undergraduates, who have already self-selected interests and pathways which may not generalise to other student groups. Comparatively little is known about how adolescent mathematics-related career aspirations translate to actual career choices through longitudinal studies. There are some notable longitudinal exceptions that recruited diverse samples (e.g., Eccles & Wang, 2016; Ferriman et al., 2009; Wegemer & Eccles, 2019). Our study draws upon long-term longitudinal data to investigate whether occupational goals could moderate the relationship between adolescent mathematical career aspirations and actual mathematics-related careers, and how this may differ between women and men.

Gender differences in agentic and communal goals

Due to traditional gender roles, where women focus on caring for others and men are in positions of leadership and power, gender role stereotypes associate women with communal roles and men with agentic roles (Eagly, 1987; Eagly & Wood, 2012). Initial research proposed that it is optimal for individuals to fulfil a balance of communal and agentic goals (Bakan, 1966), and that both men and women value agentic and communal goals (Helgeson, 1994). Meta-analyses of studies of U.S. secondary school and college students revealed that over the decades (1931–1993), as more women entered the workforce, their endorsement of agentic goals increased to a level similar to men’s (Twenge, 2001). Communal goals are also valued by both men and women (Diekman et al., 2010), which Diekman et al. (2017) attributed to men’s and women’s needs to belong and relate.

However, gender differences in communal values have been widely documented. A meta-analysis of data from U.S. undergraduate students between the 1970s and 1990s found that, although women’s reported agentic goals steadily increased in studies over time, women continued to report higher communal goals than men (Twenge, 1997). In the USA, from as young as age 6, boys endorsed lower communal goals than girls (Block et al., 2018). Cross-cultural meta-analyses revealed that compared to men, women were more likely to endorse basic values associated with community, such as tender-mindedness and warmth (among college students and adult participants across five continents; Costa et al., 2001), and benevolence and universalism (among participants aged 15 and older across 19 European countries; Schwartz & Rubel, 2005). A U.S. study among mathematically precocious youth in particular, tracked two cohorts over a decade and two decades respectively (ages 25–35; and 13–33), and found that males favoured agentic occupational goals more than females, whereas females favoured communal occupational goals more than males (Ferriman et al., 2009).

Within-gender comparisons of occupational goals are often not reported, although a German study found that while men and women both valued communal and agentic goals, 60% of women valued communal goals more than agentic, 62% of men valued agentic goals more than communal, and the type of goal individuals considered more important predicted the type of career goal they attained (Pohlmann, 2001). As individuals can consider more than single goals when making a career choice, it is useful to investigate their combinations of agentic and communal occupational goals endorsement, and potential associations with gender and career choice. Within this study, we investigated potential interactions among agentic and communal occupational goals by comparing four groups based on combinations of high and low endorsement of each, potential differences in gender composition, and trajectories from aspired adolescent to actual adult mathematical careers.

Occupational goals and mathematics-related career choice

The volume of evidence on gender differences in communal occupational goals paved the way for the communal goal congruity perspective (Diekman & Steinberg, 2013), which posits that women’s lack of interest in mathematics-related careers is because these are perceived to not afford opportunities to realise their communal occupational goals (Diekman et al., 2010, 2015). Studies conducted among U.S. psychology undergraduates found that mathematical careers (e.g., mechanical engineering, computer science, physical science) were perceived as affording less communal occupational goals than ‘non-STEM’ careers that were historically male-dominated with an agentic focus (e.g., CEO, dentist, lawyer), female-stereotypic careers (e.g., nurse, teacher) (Diekman et al., 2011)—or other careers including education, business, law and medicine (Morgan et al., 2001). Another study among U.S. psychology undergraduates (Carli et al., 2016) revealed that people working in mathematics-related careers were perceived to possess lower communal characteristics than people in other careers. Participants rated their perceptions about agentic and communal traits that an average adult man, woman, or successful scientist would possess. Men were perceived as highly agentic, women as highly communal, and successful scientists were perceived as less communal than either men or women.

In a U.S. study among a sample of diverse undergraduates, participants were asked to think about role models in mathematical STEM careers, male-stereotypic non-STEM, and female-stereotypic non-STEM careers, and to rate how challenging it was to think of an example who had communal or agentic characteristics. Both male and female participants found it more challenging to think of a role model in mathematical STEM (e.g., mechanical engineering, computer science, aerospace engineering, physics) possessing communal characteristics (Fuesting & Diekman, 2017).

Another U.S. study of psychology undergraduates asked them to rate each of their agentic and communal occupational goals, interest in mathematics-intensive STEM (e.g., mechanical engineer, computer scientist), male-stereotypic (e.g., lawyer, dentist) and female-stereotypic careers (e.g., social worker, nurse), and how much they considered each to provide for agentic and communal occupational goals. More women endorsed communal occupational goals than men, and communal goal endorsement negatively predicted interest in mathematical STEM and male-stereotypic non-STEM careers. There was no gender difference in agentic occupational goals, which positively predicted interest in mathematical STEM careers, even controlling for self-efficacy and experience in mathematics and science (Diekman et al., 2010). Collectively, these studies support the communal goal congruity perspective that women’s higher communal occupational goals may steer them away from mathematics-related careers. This appears to be the case even controlling for achievement-related constructs such as mathematics self-concept (Wegemer & Eccles, 2019) or self-efficacy (Diekman et al., 2010).

To understand the relations among occupational goals and career choice over time among diverse youth who had not already self-selected into professional career training pathways, six waves of longitudinal data were examined from the Maryland Adolescent Development In Context Study (‘MADICS’; participant ages 13, 14, 17, 19, 21 and 26; see Cook et al., 2002). Wegemer and Eccles (2019) explored how altruistic occupational values, mathematics ability self-concept, gender and femininity (in terms of gender self-schema) related to STEM career choices for ‘MPECS’ (Mathematics, Physical Sciences, Engineering and Computer Sciences) vs. ‘HBMS’ (Health, Biological and Medical Sciences) within each timepoint. Cross-sectional analyses were chosen as participants’ STEM career choices fluctuated across secondary school, college major and actual career.

Occupational altruism, gender and femininity predicted concurrent career choices for life sciences (HBMS) and away from mathematically intensive (MPECS) fields. Femininity moderated the relationship between gender and STEM choice, and regardless of gender, participants who endorsed a feminine self-schema were more likely to choose HBMS over MPECS. The strength of relationship between occupational altruistic values and choice of HBMS over MPECS increased across waves, especially 3 years after high school (wave 6). Occupational altruism mediated the relation between gender and HBMS, and femininity and HBMS, more strongly than mathematics ability self-concept—highlighting the importance of occupational goals in shaping STEM career choices.

The fact that STEM career choices fluctuated across waves suggests that adolescent aspirations may not translate to actual career outcomes, due to a range of influences between secondary school and career entry. This underlines the importance of understanding how key factors may sustain or disrupt mathematical career trajectories, beyond the large literature focused on adolescent aspirations in secondary school settings (e.g., Lazarides et al., 2020; Mau & Li, 2018; Watt et al., 2012, 2017).

The present study

The reviewed studies found support for the communal goal congruity perspective: that individuals perceive mathematics-related careers as lacking in affordances to fulfil communal goals, and that occupational communal goals play a key role in steering, especially women’s, career choices away from mathematics. As outlined earlier, these effects were found to hold above the influences of mathematical self-concepts or self-efficacy. However, much of the evidence comes from U.S. cross-sectional analyses with psychology undergraduate students who may have already self-selected career interests and pathways. An exception is Wegemer and Eccles (2019), who utilised a longitudinal dataset and a broad sample commencing in secondary school, but conducted cross-sectional analyses. While the reviewed studies demonstrate that occupational goals mediate the relationship between gender and concurrent career choice, how the combination of agentic and communal occupational goals may affect mathematical career choice trajectories for women and men has yet to be explored.

We extend existing research by drawing upon long-term longitudinal data in the Australian context to investigate gender differences in agentic and communal occupational goals and their combination, and to explore the relation between adolescent aspired and adult actual mathematical career choices, and potential dependency on occupational goal combinations, over and above secondary school mathematical achievements. The four occupational goal groups that were examined are Low-Agentic/Low-Communal (LA/LC), Low-Agentic/High-Communal (LA/HC), High-Agentic/Low-Communal (HA/LC), High-Agentic/High-Communal (HA/HC). Based on the reviewed literature, we proposed the following hypotheses:

H1

There will be a gender difference in the valuing of communal career goals, being higher for women. However, men and women will value agentic goals similarly.

H2

Relatively more women than men will be in the occupational goal group holding higher communal than agentic goals. While there were no directional hypotheses, exploratory analyses also examined how the four occupational goal groups related to gender and background mathematical achievement.

H3

The trajectory from mathematics career aspirations to actual careers will be moderated by combinations of agentic and communal goals as follows:

  1. (a)

    sustained or increased, when individuals hold High-Agentic/Low-Communal (HA/LC) goals;

  2. (b)

    declining, when individuals hold Low-Agentic/High-Communal (LA/HC) goals.

No hypothesis was advanced for trajectories of High-Agentic/High-Communal (HA/HC) or Low-Agentic/Low-Communal (LA/LC) groups.

Method

Participants and procedure

The study sample consisted of 279 participants (n = 153/54.84% women) from the Study of Transitions and Education Pathways Study (STEPS; see www.stepsstudy.org). At T1, participants were recruited from three coeducational government secondary schools in metropolitan Sydney, Australia, who were matched on middle to upper-middle socioeconomic backgrounds based on the Index of Education and Occupation from census data (Australian Bureau of Statistics, 1991) and who agreed to participate in the study. Self-report surveys at T1 were administered during class hours by the second author and two trained assistants, and the T2 follow-up survey was conducted online using the Qualtrics platform. Data were collected under ethical approvals from The University of Sydney (T1; Ref. 95/6/10) and Monash University (T2; Ref. CF11/2681-2011001571). Informed parent/guardian consent was obtained during participants’ secondary schooling and informed participant consent at the adulthood follow-up.

The T2 adulthood survey was conducted between 2015 and 2019, approximately 20 years after participants’ latest timepoint during secondary school in 1998, at grade 9, 10, or 11 for sequential cohorts (typically referred to as Years 9, 10 and 11 in the Australian context). At T1, present study participants’ ages ranged from 13 to 17 years (M = 13.29, SD = 1.03) and consisted of 43.8% girls. At T2, ages ranged from 31 to 37 years (M = 33.36, SD = 1.04). Of the initial 1,367 participants at T1, there was an overall retention rate of 27.5% at T2, similar to other long-term longitudinal studies (e.g., 25.7% after 30 years in the Michigan Study of Adolescent and Adult Life Transitions [MSALT]; see Eccles et al., 1989; Dicke et al., 2019). Compared with the initial attrited sample (n = 991) and other retained participants (n = 97, who did not respond to both questions about aspired and actual careers) (see Appendix Table 6), participants in the present study contained a higher proportion of women and individuals who had more mathematics-related actual careers at T2 than other retained participants; they did not differ on mathematics-related career aspirations or prior mathematics achievement at T1.

Participants who were retained in the current study (i.e., 320 of the retained 376 study participants) answered questions about career aspirations (T1) and actual careers (T2); low proportions of data were missing completely at random (MCAR; χ2 [N = 320, df = 31] = 42.17, p = .087) for agentic and communal goals (11.25%) and grade 9 mathematics achievement score (6.56%). Listwise deletion was used to treat these low proportions of missing data resulting in the final study sample of 279 participants.

Participants answered about their gender and birth date at both timepoints. At Time 1, additional demographic information included participants’ gender, date of birth, language spoken at home, country of birth, and parents’ countries of birth and highest levels of education (see Table 1). 75.6% of participants were born in Australia or culturally similar countries (i.e., New Zealand, United Kingdom, USA); the second largest group came from Asian countries at 11.8%. The same pattern emerged for mothers’ and fathers’ birth countries. For most participants, language spoken at home was English (80.3%), followed by Asian languages (8.6%). Approximately half of mothers and fathers had completed university level education, followed by high school (22.9% of mothers, 18.3% of fathers).

Table 1 Participants’ demographic background information at T1
Full size table

Measures

Time 1—adolescence

Mathematics achievement background

Mathematics achievement was measured using a standardised Progressive Achievement Test (Australian Council for Educational Research, 1984) when participants were in grade 9. The test consisted of 28 multiple-choice items with 5 response options each. Correct answers were scored 1, 0 otherwise, and summed to create a total score.

Secondary school career aspirations

Participants answered an open-ended question about their secondary school career aspirations, “What career are you mainly considering for your future?”. These responses were quantified in terms of mathematics-relatedness using the U.S. occupational coding system O*NET™ 98: The Occupational Information Network (U.S. Department of Labor, Employment & Training Administration, 1998), since a similarly comprehensive database was not available in Australia. Participants’ verbatim responses were matched with titles listed on the database, which rated the mathematics-relatedness of these occupations from 0 (none), 1 (low), 2 (average), to 3 (high). Examples of jobs in the category for 0 (none) included author, music and sport; 1 (low) included electrician, counselling/social work and hospitality; 2 (average) included architecture, politics and medical doctor; and 3 (high) included accountancy, astronomy and engineering.

Time 2—adulthood

Occupational goals

Five items used to tap occupational goals were sourced from the Motivations for Career Choice scale (MCC; Watt & Richardson, 2017). As part of this unique longitudinal dataset, items were included which resembled those recently developed in the literature to measure occupational agentic and communal goals (Diekman et al., 2010), accordingly selected to address our central research question. Participants rated items from 1 (not at all) to 7 (extremely) in response to the stem, “It is important for me to have a career that …”. Two items resembling agentic goals (e.g., “… is a high-status career”; ω = 0.65) and three items resembling communal goals (e.g., “… allows me to make a worthwhile social contribution”; ω = 0.72) were averaged to form composite measures. Construct reliabilities were acceptable for communal and marginal for agentic goals, for which reliability was just below the generally accepted 0.70 cut-off (Nunnally, 1978). The full list of items and reliability coefficients is presented in Table 2.

Table 2 Occupational goal construct items and McDonalds’ omega reliability coefficients
Full size table
Occupation

Participants reported their current or most recent occupation in response to the open-ended question, “What is your current occupation?”. Those not in current employment were directed to the question, “What is your most recent occupation?”. Examples of open-ended responses for T1 career aspirations and T2 current or most recent occupations by occupational goal groups are presented in Table 3. The mathematics-relatedness of occupations was coded by four trained research assistants. Two coders coded all occupations; the other two coded portions each such that there were three coders for each occupation. Mathematics-relatedness of occupations reported was quantified in a similar way to T1; coders matched participants’ open-ended responses with occupations listed on the corresponding O*NET™ database (National Center for O*NET Development, 2019), on a 101-point scale that ranged from 0 to 100. Coders matched 79% of responses to O*NET™ occupation titles consistently, indicating a high interrater reliability. The remaining 21% that did not show full initial agreement were discussed and consensus was successfully resolved among coders.

Table 3 Example open-ended responses for occupational goal groups
Full size table

Analytical approach

Gender differences in agentic and communal goals

A multivariate analysis of variance (MANOVA) examined gender differences on agentic and communal goals.

Occupational goal groups and membership

Groups were formed to represent combinations of agentic and communal goals (scored 1–7), each split by median responses (agentic = 5; communal = 4) to form ‘low’ and ‘high’ levels. Exact median scores were categorised into ‘high’ levels. A categorical variable was constructed on the combination of these levels to form four occupational goal categories: (1) Low-Agentic/Low-Communal (LA/LC), (2) Low-Agentic/High-Communal (LA/HC), (3) High-Agentic/Low-Communal (HA/LC), and (4) High-Agentic/High-Communal (HA/HC).

Potential associations of gender and occupational goal groups were explored using a chi-square test of association, followed by configural frequency analysis (von Eye, 1990). Potential differences in mathematical achievement background and occupational goal groups were compared using ANOVA with Tukey post hoc tests.

Change in mathematical career trajectories

Trajectories from aspired to actual mathematical careers for occupational goal groups were explored using repeated-measures ANOVA and Tukey post hoc tests. Aspired and actual mathematical career scores were converted to z-scores to enable comparisons across their different metrics. Follow-up tests were repeated-measures ANOVAs within each occupational goal group to discern potential changes, and between-subjects ANOVAs to compare occupational goal groups at each of T1 and T2. All analyses were conducted using SPSS Statistics 28 software.

Results

Correlations between key study constructs are first presented for women and men, followed by results mapped to the three hypotheses that were advanced. Concerning the first hypothesis, gender differences on agentic and communal occupational goals are presented. Second, gender and background mathematical achievement of the occupational goal groups are shown. Third, we present the mathematical career trajectories based on the occupational goal groups.

Relationships among key study constructs

There was no statistically significant relationship between mathematical career aspiration and actual mathematical career for males (r = .10, p = .271) or females (r = .14, p = .091). There were weak but significant positive relationships for males, between mathematical achievement background and each of mathematical career aspiration (r = .18, p = .044) and actual mathematical career (r = .22, p = .012), but not for females (r = .15, p = .065; r = .09, p = .250, respectively). There was a moderate positive significant relationship between agentic goals and actual mathematical career for males (r = .32, p < .001), but not females (r = .13, p = .119). Weak but negative significant relationships emerged for females, between communal goals and each of mathematical achievement background (r = − .16, p = .045) and actual mathematical career (r = − .21, p = .010), but not for males (r = − .05, p = .559; r = − .05, p = .552, respectively). The correlation matrix between key study constructs for males and females is presented in Table 4.

Table 4 Correlations among key study constructs for males/females
Full size table

Gender differences in agentic and communal occupational goals

There was a significant multivariate effect of gender on agentic and communal goals (F(2, 276) = 5.18, p = .006, Wilk's Λ = 0.964, ηp2 = .036). Consistent with H1, there was no significant univariate effect of gender on agentic goals (F(1, 277) = .31, p = .579, ηp2 = .001), valued similarly highly by men and women (M = 4.65, SD = 1.48 for men; M = 4.74, SD = 1.28 for women). However, there was a significant univariate effect of gender on communal goals (F(1, 277) = 9.74, p = .002, ηp2 = .034), valued more by women (M = 4.38, SD = 1.39) than men (M = 3.86, SD = 1.39).

Composition of occupational goal groups

There was a significant association between occupational goal group and gender (χ2 [N = 279, df = 3] = 12.81, p = .005). H2 was supported by configural frequency analysis (von Eye, 1990), which identified significantly more women (n = 53) than men (n = 30) in the LA/HC group (p = .024). In addition, significantly more men (n = 34) than women (n = 19) were represented in the LA/LC group (p = .001). There were no significant gender differences in membership of HA/LC and HA/HC groups. Figure 1 illustrates the proportions for gender composition within each occupational goal group. There was no significant difference between occupational goal group and mathematics achievement background (F(3, 275) = .01, p = .999, ηp2 < .001), indicating no significant dependency on mathematics achievement background.

Fig. 1
figure 1

Gender composition of occupational goal groups

Note. Frequencies are indicated above each bar. * configural frequency analysis indicated a significant difference between gender groups (p < .05). LA/LC: Low agentic, low communal; LA/HC: Low agentic, high communal; HA/LC: High agentic, low communal; HA/HC: High agentic, high communal

Full size image

The role of occupational goals in mathematical career choice trajectories

According to H3, we had hypothesised that the mathematical career choice trajectory of the HA/LC goal group would be sustained or increased, whereas the LA/HC goal group would decrease. Descriptive statistics for occupational goal groups are displayed in Table 5. The repeated-measures ANOVA revealed a significant two-way interaction between mathematical career trajectory and occupational goal group (F(3, 275) = 3.36, p = .019, Wilk's Λ = 0.965, ηp2 = .035; see Fig. 2). Tukey post hoc tests showed lower mathematical career choice by LA/HC than HA/LC overall (p = .019), but no other significant differences between groups.

Table 5 Descriptive statistics for occupational goal groups
Full size table
Fig. 2
figure 2

Mathematical career choice trajectories for occupational goal groups

Note. Standardised means of aspired and actual mathematics-related careers. * p < .05. LA/LC: Low agentic, low communal; LA/HC: Low agentic, high communal; HA/LC: High agentic, low communal; HA/HC: High agentic, high communal

Full size image

Subsequent repeated-measures ANOVAs within each occupational goal group revealed significant changes between aspired and actual mathematical careers only for LA/HC (F(1, 82) = 7.17, p = .009, ηp2 = .080; aspired: M = − .02, SD = .96; actual: M = − .39, SD = .91; see Fig. 2). No significant changes were detected for the other groups. These findings supported H3.

Participants in the four occupational goal groups reported similarly mathematics-related career aspirations at T1 (F(3, 275) = .50, p = .680, ηp2 = .005); however, by T2 there was a significant effect of group on actual mathematical careers (F(3, 275) = 6.20, p < .001, ηp2 = .063). This was accounted for by significant differences between the LA/HC group (M = − .39, SD = .91) and each other group (LA/LC: p = .009, M = .16, SD = .91; HA/LC: p = .002, M = .19, SD = 1.05; HA/HC: p = .004, M = .14, SD = 1.01).

Discussion

The current study contributes to existing literature by testing three hypotheses using long-term longitudinal data. Each of our three hypotheses was supported in our investigation of whether, and the extent to which, different combinations of communal and agentic occupational goals moderated the relationship between aspired mathematical careers in adolescence and actual careers in adulthood, and how this differed for women and men.

Mathematical career aspirations in secondary school are considered important because they shape subsequent choices such as post-secondary education and career entry (Riegle-Crumb & King, 2010). As a result, many studies have focused on factors predicting mathematics enrolment and career aspirations in secondary school settings (e.g., Eccles, 2009; Lazarides et al., 2020; Mau & Li, 2018; Watt et al., 2012, 2017). However, across an approximately 20-year timeframe, we did not find evidence that secondary school mathematical aspirations translated to mathematical careers for men or women, similar to a U.S. study (Wegemer & Eccles, 2019) across six waves of data spanning 13 years. Career choice has been described as a ‘moving target’, such that different factors influence career choice at different times in the career trajectory (Watt, 2010). Therefore, it is imperative that studies examine occupational goals (and other influences) at various timepoints due to changing priorities and circumstance. Our focus was women’s and men’s potentially different combinations of agentic and communal occupational goals.

Gender differences in occupational goals

Our first hypothesis, that women would value communal goals more highly than men, but both gender groups would value agentic goals similarly, was supported in line with previous studies. Due to the cultural similarity between the USA and Australia, we had expected similar patterns for gendered occupational goals. Women endorsed communal goals more than men (Block et al., 2018; Diekman et al., 2010; Ferriman et al., 2009; Twenge, 1997), but there was no gender difference in the endorsement of agentic goals (Diekman et al., 2010; Twenge, 1997, 2001).

Since individuals are likely to make career decisions based on a combination of agentic and communal goals, the current study extended the literature by constructing groups that combined agentic and communal goals. We found that there were more women than men in the LA/HC group, supporting our second hypothesis, in line with gender role stereotypes that associate women with high communal but not agentic goals (Eagly, 1987; Eagly & Wood, 2012). In this sample, we also found that the LA/LC group consisted of more men than women. One possible explanation could be due to our median-split approach which categorised low agentic goals by ratings up to 4.5 on the 1–7 point scale, thus including some individuals who rated agentic goals just above the scale midpoint, which might partly explain the overrepresentation of men in this group.

How do occupational goal combinations impact mathematical career choice trajectories?

Consistent with hypothesis three, only the LA/HC goal group had a decline in their aspired versus actual mathematics-related career choice. As women were more likely to be in this group than men, this finding supports the communal goal congruity perspective, that women were less likely to seek mathematical careers due to their higher endorsement of communal than agentic goals (Diekman et al., 2010, 2015). Supporting the hypothesis as well is the finding that mathematics-related career choice was sustained for the HA/LC goal group, whose goals were consistent with what mathematical careers are perceived to afford (Diekman et al., 2010).

While the current findings point to a similar direction as previous studies on the effects of occupational goals on mathematical careers, we extend this literature by detecting this effect for individuals with a combination of LA/HC goals. It was not the case that individuals with a combination of HA/HC goals decreased their mathematical career choice, or contained more women. Previous findings have observed a similar pattern where communal goals negatively predicted interest in mathematical careers, and agentic goals positively predicted interest in mathematical careers (Diekman et al., 2010), and occupational altruistic values predicted preference away from a mathematical career (Wegemer & Eccles, 2019). From the previous studies one could have concluded that endorsing communal goals alone would reduce mathematical career interest/choice—however our findings showed this effect only appeared among participants with high communal goals and low agentic goals, but not when participants held high communal and agentic goals. It seems that the agentic goal affordance of mathematical careers may appeal to individuals who hold both agentic and communal goals, who are not deterred by a possible lack of communal goal affordances in mathematical careers.

Limitations and future directions

As with all research studies, limitations to the current study should be considered when interpreting our findings. First, despite the long timeframe needed to gather rich contemporary longitudinal data, the current study showed an attrition rate similar to other long-term studies (e.g., 25.7% after 30 years in MSALT; see Eccles et al., 1989; Dicke et al., 2019). However, future studies could benefit from a larger sample size to examine more nuanced effects such as low, moderate and high occupational goal groups. Second, information on occupational goals was only available in adulthood. Future studies utilising large datasets with such information at several timepoints can examine at which timepoints along the career choice pathway (e.g., secondary and tertiary education, early career stages) that occupational goals most influence career choice or change for different kinds of individuals. Third, although construct reliability was marginal for the agentic goals, given the unique longitudinal dataset, this study presents the first long-term assessment of how occupational agentic and communal goals influence long-term mathematical career trajectories using a normative sample, enriching the goal congruity literature which is so far predominated by cross-sectional studies among U.S. psychology students. Current findings raise interesting priorities for future research in this vein, which should include stronger measures of the goal constructs.

Practical implications

The current study extends the literature on how combinations of occupational goals together influence mathematical career choice from adolescence to adulthood, among a sample of youth recruited during secondary school who had not yet self-selected into particular career training pathways. Our findings show robust effects for the communal goal congruity perspective across different life stages, and extend empirical support beyond the USA, and psychology undergraduates.

Our key finding that individuals who hold the combination of low agentic and high communal goals (LA/HC) were more likely to be deterred from their adolescent mathematical career aspirations, and that this group consisted more of women than men, highlights the importance of making communal goal opportunities salient in efforts to reduce the gender disparity in mathematical STEM. Highlighting communal opportunities in science careers can increase women’s interest. A U.S. study found that when scientists were depicted as undertaking communal work in their daily activities (e.g., collaborating with and helping people), introductory psychology college students’ beliefs about communal goal affordance in science increased, along with positivity towards science careers in general (Clark et al., 2016). In mathematics-intensive fields such as engineering, subdisciplines perceived to afford for more communally oriented goals (e.g., environmental, biomedical and civil engineering) graduated more women than mechanical and nuclear engineering (Gibbons, 2009).

Given fluctuations in the ‘moving target’ of mathematical career choice in our findings and by Wegemer and Eccles (2019) in the USA, policymakers and practitioners aiming to increase the representation of women in mathematical STEM are recommended to highlight communal opportunities when promoting these careers at different stages of the career trajectory.

A few examples could be considered at the secondary school level. First, initiatives such as ‘Girls’ Day’ in countries including the Netherlands (Jansen & Joukes, 2013) and Norway (Jensen & Bøe, 2013) could be adopted and implemented in Australia. On these days, technical companies and STEM organisations promote interest and knowledge in these careers to girls aged 10–15 years old. Adolescent girls are given a tour, including hands on activities and talks with women STEM professionals to gain a more comprehensive understanding of their careers. Second, school leaders could plan enrichment activities linking mathematics solving problems which have a communal focus. A recent study found evidence to support such an intervention, where students studied and internalised the communal utility of science in a range of popular professions, leading to greater interest and confidence in science, greater appreciation for the utility of science in their future careers, and increased aspirations towards science-related careers—especially for girls (Shin et al., 2022). This offers a promising approach to attract individuals who highly value communal goals in their choice of occupation—particularly girls and women, who were more likely to be in this group. Third, schools could hold information sessions for students and parents including information about choosing matriculation subjects, assessing what occupational goals they value, and the types of occupations that would fulfil these goals. To support women’s mathematical career aspirations and pathways, it is particularly important to highlight a range of different mathematical career options for those who endorse LA/HC goals. Various initiatives should be promoted at every stage possible to protect against interest losses: through secondary school, tertiary education, career fairs and job advertisements for entry level and more experienced roles.

Conclusion

Of the many reasons put forward for why women are severely underrepresented in mathematical STEM, the communal goal congruity perspective posits that women’s greater communal goals deter them, as mathematical STEM careers are perceived to not afford communal goal opportunities. While previous studies have focused on the effects of each of agentic and communal goals on concurrent mathematical career choice, less has been known about how combinations of goals can work together to influence changes in mathematical career choice. The current study demonstrated how occupational goal combinations can affect mathematical career choice trajectories across a timespan of approximately 20 years. Individuals endorsing low agentic and high communal goals, who were mainly women, were more likely to end up in less mathematical careers relative to their adolescent aspirations. To support women’s mathematical aspirations and create a more gender diverse workforce, policymakers and practitioners could fruitfully promote the social contributions afforded by a range of mathematical careers along the career choice pathway.