The labor market is horizontally gender segregated, where work in the care sector (e.g., nursing) is typically performed by women, and work in the technology sector (e.g., IT) is typically performed by men (e.g., European Institute for Gender Equality [EIGE], 2017; Lippa et al., 2014; Organisation for Economic Co-operation and Development [OECD], 2019; World Health Organization [WHO], 2022). This gendered pattern has been explained by the large observed gender differences in work related to “people” (caregiving) versus work related to “things” (technology), which constitute the main dimensions of the horizontal gender segregation observed in the labor market (Barone, 2011; Blackburn et al., 2014; Lippa et al., 2014; Su et al., 2009). The dominance of women in the care sector (or HEED; Healthcare, Early Education, Domestic) has been very slow to change (Block et al., 2018a; Croft et al., 2015). The technology sector, one branch of STEM (Science, Technology, Engineering, and Mathematics), is dominated by men, especially in the Western world, although some STEM-fields (like science) have become less gender skewed over time (Master & Meltzoff, 2020; Nordic Council of Ministers, 2021; Su & Rounds, 2015).

For example, in Sweden, science programs are now balanced in gender proportions, while tech-focused and care-focused educational programs are still strongly gender skewed (Swedish Higher Education Authority, 2020; Swedish National Agency for Education, 2021). When a sector is gender skewed, labor shortages may appear, which is apparent in both the care sector and the tech sector (Nordic Council of Ministers, 2021; OECD, 2019; WHO, 2022). Concern for this has been particularly strong in relation to the tech sector, where increases in digitalization means that the demand for individuals with tech skills will increase dramatically (Manpower Group, 2022; Nordic Council of Ministers, 2021; OECD, 2019; Swedish IT and Telecom Industries, 2021). IT and data-based roles are in top demand globally, where 76% of employers experience a talent shortage (Manpower Group, 2022). It is therefore concerning that girls’ STEM-interest tends to decrease before high school (e.g., see Barth & Masters, 2020). In this study, we investigate gender stereotypes as a key explanation for lower interest in technology among middle school girls compared to boys.

Explanatory Factors for Gender Differences in Career Interest

Why boys and girls tend to develop different educational interests and choose different types of career paths is a hotly debated topic and answering this question has been a focus of a growing social psychological literature. Understanding the mechanisms behind gender differences in career interest is a priority, not the least for employers who are looking to fill their future recruitment needs and want to recruit the best employee, regardless of their gender. Identifying potential barriers for the development of career interest is also important from a gender equality perspective, where one goal is that individual development should not be hindered by societal gender stereotypes (e.g., European Institute for Gender Equality, 2017; Government Offices of Sweden, 2018). It is therefore of interest that several influential psychological career theories connect the horizontal gender segregation in the labor market to gender stereotypes.

Eccles’s well-established expectancy-value theory (EEVT; Eccles, 1987, 1994), recently revised to the situated expectancy-value theory (SEVT; Eccles & Wigfield, 2020), and the stereotypes, motivation, and outcomes model (STEMO; Master & Meltzoff, 2020), all stress that gender stereotypes play an important role in shaping gender differences in psychological factors related to career interest and career choice. In this study, we measure career “interest” as a dependent variable, coherent with the STEMO model. Interest has been described as similar to the concept of “intrinsic value” (or “fun”) in EEVT and SEVT, although there is ongoing work (and debate) concerning construct definitions (Eccles & Wigfield, 2020). Career choice is affected by a multitude of factors (see Eccles, 1987, 1994; Eccles & Wigfield, 2020; Master & Meltzoff, 2020 for reviews) and some studies have shown that interest may be the strongest predictor of career choice (Maltese & Tai, 2011; Rundgren et al., 2019), which motivated our choice to measure interest in this study.

Social role theory (SRT, Eagly, 1987, Koenig & Eagly, 2014) also connects gender stereotypes to gender segregation. It states that we derive the content of gender stereotypes from the roles men and women tend to have in society, including occupational roles. Drawing on SRT, there should be strong gender stereotypes regarding caregiving and technology, since “people” versus “things” represent the major gender divide in the labor market (Barone, 2011; Blackburn et al., 2014; Lippa et al., 2014; Su et al., 2009). Drawing on EEVT, SEVT and STEMO, these stereotypes should then in turn contribute to shape gender differences in self-representations related to career choice.

Although previous research has empirically investigated gender stereotypes that are relevant for gender differences in STEM career identification (see Master & Meltzoff, 2020, for a review), we cannot find studies that have specifically contrasted caregiving and technology (to represent femininity versus masculinity) and connected these stereotype categories to career interest. There is reason to investigate gender stereotypes associated with caregiving versus technology given that this categorical contrast represents the gist of gender differences in occupational choice in the Western world (Barone, 2011; Blackburn et al., 2014; Lippa et al., 2014; Su et al., 2009).

In the current study, we adapted the Implicit Association Test (IAT, Greenwald et al., 1998, 2003) to capture associations of caregiving and technology with gender categories. The IAT measures associations between mental concepts indirectly and is less susceptible to social desirability concerns (such as not wanting to disclose prejudice) as compared to explicit (self-report) measures (Dovidio et al., 2003; Greenwald et al., 1998, 2003; Greenwald & Banaji, 2017). Despite some well-documented limitations, the IAT is the leading implicit measure in psychological research on stereotypes and prejudice, awaiting further test development (e.g., Brownstein et al., 2019; Meissner, et al., 2019). Poor test-retest reliability in individual IAT-scores suggests that it should not be used for assessing the stereotype strength of individuals, but it has been proven robust and effective for assessing stereotyping on the group level and is appropriate to relate to other constructs (like interest), which we aim to do in this study (e.g., Brownstein et al., 2019).

We employ the IAT, along with explicit measures of ability gender stereotypes for technology and caregiving, in a large sample of middle school students and a smaller sample of schoolteachers residing in Sweden. The main aim is to investigate if students endorse stereotypes for technology and caregiving careers according to gender, and if their implicit associations along these dimensions relate to their interest in pursuing a tech-education in the future. We will also explore if explicit technology gender stereotypes relate to students’ tech interest, but due to social desirability concerns associated with explicit measures, predictions for these measures are less straightforward (Dovidio et al., 2003).

Stereotype Dimensions: Communion vs. Agency or People vs. Things

Agency and communion encapsulate the distinction between masculinity and femininity and the content of gender stereotypes (e.g., see Abele & Wojciske, 2014, for review). However, SRT argues that when societal gender roles change, so will gender stereotypes (Eagly, 1987). We propose that traits associated with the occupational dimensions “people” (caregiving) versus “things” (technology) may perhaps better describe current gender stereotypes in Sweden, which we explain below.

Communion and People (Caregiving)

Communion refers to the desire and motivation to closely relate with and connect to others and is associated with traits central to caregiving, such as warm, caring, and helpful (e.g., Abele & Wojciske, 2014). The stereotype that women are more communal than men holds well over time (Ellemers, 2018; Haines et al., 2016) and also in Sweden (Gustafsson Sendén et al., 2019), and studies show the corresponding gender differences in students’ communal values (Block et al., 2018b; Ely et al., 1998; Tellhed et al., 2017). The stability of this gender stereotype fits well with SRT, which, as mentioned, states that the content of gender stereotypes is based on observations of societal gender roles (Eagly, 1987). If we see more women than men performing caregiving roles, we may conclude that women are better at caregiving than men.

That caregiving roles are performed predominantly by women is apparent around the world (Barone, 2011; Blackburn et al., 2014; Block et al., 2018a; Croft et al., 2015; Lippa et al., 2014; Su et al., 2009; WHO, 2022). In addition to paid care work (e.g., nursing), women also do most of the unpaid domestic caregiving (e.g., childcare), including what has been called emotional labor in relationships (e.g., remembering family birthdays; Hartley, 2018). This is true also in Sweden although it has been rated the most gender equal country in the EU for 16 years in a row (European Institute for Gender Equality, 2021). In Sweden, equal proportions of women and men work outside the home, but the most common occupations for women are in the care sector (Statistics Sweden, 2021a). Swedish women manage most of the caregiving at home and take most of the available parental leave (Försäkringskassan, 2021; Statistics Sweden, 2020). Since Swedish students generally observe more women than men performing caregiving roles, drawing on SRT, we expect that the students in this study would generally hold the belief that women are better at caregiving than men, which corroborates previous research that women are seen as more communal (Ellemers, 2018; Gustafsson Sendén et al., 2019; Haines et al., 2016).

Are women better at caregiving than men? Interestingly, compiled meta-analyses have revealed that most psychological variables, including tests of abilities, show only small gender differences and large individual differences (Hyde et al., 2005, 2019; Zell et al., 2015). Why did gender roles develop in the first place, if gender differences in ability are typically small? SRT states that some biological factors, like the circumstance that females give birth, have likely played a role in the historical development of societal gender roles, and by extension, in the formation of gender stereotypes (Eagly, 1987; Koenig & Eagly, 2014). Social factors also obviously matter for societal gender role construction. For example, it is only in the last century that women in the Western world have been allowed to enroll in higher education and enter the paid labor force on equal terms to men (Nermo, 2000).

Agency and Things (Technology)

Agency refers to the desire and motivation for mastery and independence and is associated with traits such as assertiveness and competence (e.g., Abele & Wojciske, 2014). Demonstrating agency is an important predictor of career success for both men and women (Abele, 2003). Drawing on SRT (Eagly, 1987), since more men than women have historically been enrolled in the labor market (until a century ago, women were excluded from higher education and most occupations in Sweden; Nermo, 2000), it follows that the traits we associate with career behavior (i.e., agency) should be stereotyped as masculine. Congruently, agency has traditionally been regarded as the core tenet of the masculinity stereotype.

However, also drawing on SRT (Eagly, 1987), since equal proportions of men and women now work outside the home in Sweden (Statistics Sweden, 2021a), the association of agency with men should weaken. Correspondingly, research has shown that the stereotype that men are more agentic than women has recently begun to change, in countries where most women are employed (Gustafsson Sendén et al., 2019; Hentschel et al., 2019). In Sweden, people now rate men and women as equally agentic (Gustafsson Sendén et al., 2019). Similarly, US participants recently rated men and women as equal in “instrumental competence” (competent, effective, productive, task oriented), but male participants still rated women lower than men on independence, assertiveness, and leadership competence, which are different aspects of agency (Hentschel et al., 2019). Recent studies in the US show that people still associate extreme levels of agentic competence (i.e., brilliance) more strongly with men than with women, which may relate to the vertical gender segregation in the labor market, where men still are still dominant in leadership and expert positions (Bian et al., 2018; Storage et al., 2020).

The association of agency with employment is also evident in women’s self-perceptions. Meta-analyses have shown that women’s self-reported agentic personality has fluctuated in the US, in parallel with the shifting proportions of women in the paid labor force (Twenge, 1997, 2001). Currently, studies tend to show no gender differences in self-rated agency in Western countries (Diekman et al., 2016; Tellhed et al., 2018; Twenge, 1997, 2001). The same pattern is found for children (4–9 years of age) in the US (Ely et al., 1998) and youth (15-year-olds) in Sweden (Tellhed et al., 2018; but see Block et al. 2018b, where 6-through 14-year-old boys in the US endorsed higher agentic values than same-aged girls).

If agency is becoming less masculine-typed or even gender-neutral in some countries, what is then considered stereotypically masculine in these contexts? We propose that traits associated with occupations related to “things” may be strongly masculine-typed in Sweden, such as technical ability, tech-interest, and behaviors associated with technology, such as choosing a tech-career. Drawing on SRT (Eagly, 1987), the core of current gender stereotypes may thus reflect the typical occupational roles for “people” and “things” associated with women and men in Western labor markets (i.e., the horizontal gender segregation). Sweden prides itself as an “engineering nation” and has been called a mecca for tech startups (after Silicon Valley of course), but less than 0.05% of the recent tech startups were founded by women (Allbright, 2020) and only 16% of employed civil engineers are women (Statistics Sweden, 2019). It is therefore likely that people in Sweden strongly associate technology with men.

The association between technology and men could be limited to Western countries. Although the technology gender stereotype is not well-researched, some previous research has shown that technology is stereotyped as strongly masculine in Western countries, like the US and Sweden, (Master et al., 2017; Mellström, 2004, 2009; Oldenziel, 1999), but not in Malaysia, where many women work in computer science (Mellström, 2009). It is also more common for women to study STEM-subjects in several countries in the Middle East and North Africa (MENA) as compared to Western countries (Huyer, 2015). These patterns imply that technology could perhaps be less masculine-typed in these countries than it is in the West.

Why should technology be strongly associated with men in the West? Oldenziel (1999) describes how technology became a buzzword in the Western world in the early 20th century, associated with modernity and civilization. The engineer was celebrated as a hero figure, but engineering was not an inclusive domain: Oldenziel (1999) describes how the engineering community actively kept women and People of Color out of engineering, which contributed to the contemporary association of technology with White men. To further understand why technology is masculine-typed, we must also consider its contemporary high status in the information society, which is expected to increase even further with developments in digitalization and automation (e.g., OECD, 2019). High status positions and occupations tend to more associated with masculinity in society (e.g., Ridgeway, 2001). Relatedly, in South Korea, where collectivistic traits are perceived to be higher status, communion is more masculine-typed than in the US (Cuddy et al., 2015).

Though we do not know of any studies that have used the IAT to test the associations between gender stereotypes and technology versus caregiving in the literature, there is a substantial amount of IAT-data supporting gendered associations with science and the arts. In support of SRT, data from approximately 350,000 participants in 66 nations revealed that cross-country variation in implicit gender stereotype strength for science versus the arts was related to the national level of gender skewness in STEM (Miller et al., 2015). To conclude, since the care sector and the tech sector are strongly gender skewed in Sweden, we hypothesize that there would be strong gender stereotypes differentially associated with these two sectors.

Gender Stereotypes and Tech-Interest

As previously mentioned, psychological theories on gender differences in career choice, like EEVT (Eccles, 1987, 1994), SEVT (Eccles & Wigfield, 2020) and STEMO (Master & Meltzoff, 2020), state that gender stereotypes affect the formation of psychological self-representations (e.g., expectations of success, ability beliefs, values, and interests), which past research has linked to gender differences in career choice (e.g., Charlesworth & Banaji 2019; Master & Meltzoff, 2020; Nosek et al., 2009). This idea is well integrated with theories about how stereotypes affect self-beliefs through socialization (e.g., see Bosson et al., 2022, for a review), self-stereotyping (Turner et al., 1987), and stereotype threat (e.g., see Spencer et al., 2016, for a review). Past research has also shown that STEM gender stereotypes are linked to career-related attitudes, identity, interest, educational choice, and organizational commitment, which tend to be moderated by participant gender (e.g., Block et al., 2018c; Cvencek et al., 2011; Cundiff et al., 2013; Lane et al., 2012; Nosek et al., 2002; Starr, 2018; Starr & Simpkins, 2021).

In the current study, we examined whether the gender associations with technology and caregiving assessed with the IAT relate to middle school students’ interest in pursuing a tech-focused educational program. We expect that girls who strongly associate technology with boys and caregiving with girls, will report less interest in applying to a tech-focused educational program in the future, as compared to girls with weaker stereotyped associations with these sectors. In contrast, we expect that boys with stronger implicit gender stereotypes will report greater interest in applying to a tech-focused educational program in the future, as compared to boys with weaker stereotyped associations.

Demographic Factors

Lastly, we examined whether the strength of the implicit gender stereotypes for technology and caregiving differ as a function of demographic factors (age, Swedish/foreign background, and gender). To limit the complexity of the results and due to social desirability concerns associated with explicit measures, we only conducted the three-way analysis on the implicit stereotype measure. (Please see the Method section for a link to the online repository hosting the dataset).

For age, extensive data have shown stronger implicit gender-science stereotypes in older generations as compared to younger (see Charlesworth & Banaji, 2019, for a review). These findings could indicate that gender-science stereotypes have weakened in society over time; alternatively, it could reflect a tendency to become more prone to stereotyping with increasing age (Charlesworth & Banaji, 2019). The latter possibility is interesting in relation to previous findings that girls’ interest in STEM weakens before high school, which suggests that it may relate to increases in associations of STEM with men (e.g., see Barth & Masters, 2020). Results vary in the literature regarding what age children tend to learn societal gender stereotypes related to career constructs (see e.g., Master et al., 2017, and Starr & Simpkins, 2021, for overviews). Before high school, most children come to associate math and science more with men (Ambady et al., 2001; Charlesworth & Banaji, 2019; Master & Meltzoff, 2020; Miller et al., 2018; Nosek et al., 2009; Starr & Simpkins, 2021), which coincides with the decline in girls’ ability beliefs and interest in math and science (Ambady et al., 2001; Barth & Masters, 2020; Charlesworth & Banaji, 2019; Hyde, 1990; Shoffner & Dockery, 2015). In a rare study of children’s technology gender stereotypes, Master et al. (2017) found that 6-year-old children in the US believed that boys are better than girls at robots and programming, but not better than girls at science and math. In this study, we explore whether older girls and boys (Grade 8) have stronger implicit technology-caregiving stereotypes compared to younger girls and boys (Grade 6). We also relate age to stereotype strength by comparing the sample of students to a sample of teachers.

For Swedish/foreign background, we explore whether students with a Swedish background differ from students with a foreign background in their implicit gendered associations with technology and caregiving. In Swedish official statistics, a “foreign background” means that a person is either born abroad or that both his/her parents were born abroad. Approximately 20% of the population in Sweden has a foreign background (Migrationsinfo, 2020). Interestingly, it has recently been suggested that gender stereotypes may be stronger in countries high in gender-equality, like Sweden, compared to countries lower in gender-equality (Breda et al., 2020). Since Sweden is the most gender-equal country in the EU (European Institute for Gender Equality, 2021), most students with a foreign background in Sweden are likely to have roots in a less gender-equal country. Perhaps then, students in Sweden with a foreign background have weaker gender stereotypes than students with a Swedish background. This could be due to influence from their parents and relatives, who may have weaker gender stereotypes as compared to people that are only influenced by Swedish culture.

For gender, although both men and women generally show gender stereotyped associations on the IAT (Miller et al., 2015; see also Charlesworth & Banaji, 2019, for a review), some studies find gender differences in these associations, suggesting a tendency to stereotype oneself favorably (Rudman et al., 2001). We therefore also explore gender differences in implicit technology/caregiving gender stereotypes.

Hypotheses

This study investigates implicit and explicit gender stereotypes about technology versus caregiving in middle school students (Grade 6 and 8) and middle school teachers, all residing in Sweden. The hypotheses are as follows:

  1. 1.

    Students and teachers will stereotype technology as masculine and caregiving as feminine, as measured with (a) the implicit association test, (b) an explicit societal stereotype measure, (c) an explicit personal stereotype measure.

  2. 2.

    Students’ implicit gender stereotypes for technology will be associated with interest in pursuing a tech-focused education.

  3. 3.

    The link between implicit gender stereotypes for technology and interest in tech-education will be moderated by gender, such that (a) tech gender stereotypes will positively predict tech-interest for boys and (b) tech gender stereotypes will negatively predict tech-interest for girls.

Additionally, we explored Hypothesis 2 and 3 with the explicit stereotype measures (societal and personal) and whether the endorsement of gender stereotypes for technology and caregiving was related to participant demographic variables i.e., (age, gender, background).

Method

The data presented here are part of a larger longitudinal project investigating students’ perceptions of technology, particularly programming. It was funded by the Swedish Research Council for Health, Working Life, and Welfare (grant # 2019 − 00334) and was reviewed and approved by the Swedish Ethical Review Authority (# 2020 − 01606). Participants were informed that their participation was voluntary and informed consent was collected for all participants, including parental consent for the school students. The informed consent material provided to the parents was available in Swedish, English, and Arabic, whereas the survey itself was only available in Swedish. The information included details of the study’s aims (e.g., investigating student/teacher perceptions of technology and programming) but did not mention gender comparisons explicitly, to avoid priming effects.

Participants

A total of 886 Swedish students in 24 different schools in the south of Sweden were recruited. Thirteen did not finish the IAT or the survey questions, which left a sample size of 873 school students: 369 in Grade 6 (Mean age = 11.86, SD = 0.44) and 504 in Grade 8 (Mean age = 13.87, SD = 0.37). The vast majority (97%, n = 846) self-identified with their legal gender (which in Sweden is woman or man) which we used for gender difference analyses. The gender balance was nearly even, with 51% boys (n = 449) and 49% girls (n = 424). 27% (n = 97) of the students in Grade 6 and 19% (n = 94) in Grade 8 were categorized as having a foreign background, which is defined as being born outside of Sweden or that both of one’s biological parents were born outside of Sweden (Statistics Sweden, 2002).

We also recruited 86 teachers to the study (Mean age = 47.44, SD = 10.77). All teachers completed the explicit measures and 71 of the teachers also completed the IAT. The vast majority (98%, n = 84) identified with their legal gender, and 63% (n = 54) were women, which is less gender-skewed that the national average in Sweden of 74% (Statistics Sweden, 2019). Since the study was part of a larger project investigating student perception of programming, the majority (68%, n = 58) of the teachers had programming as one part of their teaching. A total of 29% (n = 25) taught only Grade 4–6, 60% (n = 52) only Grade 7–9, and 11% (n = 9) taught Grade 4–9.

Measures and Procedure

The online survey was hosted on the Qualtrics (Qualtrics, Provo, UT) platform. The school students were tested during school hours and the teachers participated in their free time. The IAT was presented after the demographic information but before the explicit measures in the survey. The measures presented in this article represent part of a larger project (see see Project Description and Compliance with Ethical Standards).

Implicit Association Test

We adapted the Implicit Association Test (IAT; Greenwald et al., 1998, 2003) to measure implicit associations among gender, technology, and caregiving. We administered the IAT in Qualtrics using code provided by Carpenter et al. (2019). The original versions of the Swedish and English IAT are available in the open science repository (https://osf.io/nrjzb).

We used the target categories “Technology” and “Caregiving” to reflect the two major sectors in the gender-segregated labor market. The stimulus words (translated from Swedish) for the attribute category “Technology” included: program, code, debug, write code, think logically, troubleshoot, create apps, fix computer. The stimulus words for “Caregiving” included: take care of, care for, support emotionally, comfort, listen sincerely, nurse, show empathy, tend to. The target categories for gender were “Boy” and “Girl” and the stimulus words were person names collected from the top 100 most common Swedish names for the specific age group of the students in the study. The names for “Boy” were Lucas, Oscar, William, Elias, Filip, and Mohammed, and the names for “Girl” were Emma, Maja, Agnes, Julia, Alva, and Jasmine (Statistics Sweden, 2021b).

The IAT required participants to sort a set of stimulus words into the categories “Technology” and “Caregiving” and “Boy” and “Girl” across a total of seven blocks (120 practice trials + 80 critical trials; see Carpenter et al., 2019 for details of the trial order). Participants responded using two keys on a computer keyboard: “E” (left key) or “I” (right key). Block 4 and 7 represented the critical blocks where all four target categories were shown simultaneously. In one of the critical blocks (randomized for each participant) “Technology” and “Boy” shared one response key and “Caregiving” and “Girl” shared the other response key, reflecting congruent stereotype combinations. In the other critical block. “Technology” and “Girl” shared one response key and “Caregiving” and “Boy” shared the other response key, reflecting incongruent stereotype combinations. Differences in response latency between these two blocks were translated into D scores (see Data Analysis below). Positive scores reflect stronger gender stereotype congruent associations, and negative scores reflect stronger gender stereotype incongruent associations. Scores at zero or close to zero reflect no implicit gender associations with technology and caregiving.

Explicit Measures

Gender-Skewness Awareness

This measure was created for the study and captured awareness of the current gender skewness in occupations in the care sector and tech sector in Sweden. It was only presented to the students, since we presumed that adult teachers in Sweden are generally aware that these sectors are gender skewed. The instructions read:

In some occupations, more women than men work, and in others, more men than women work. In some occupations an equal number of men versus women work. What do you believe is the case in the following occupations in Sweden?

They were then asked to rate each item (i.e., “Work in the care sector” and “Work in the technology sector”) on a scale ranging from 1 (Many more men) to 3 (Equal numbers of men and women) to 5 (Many more women). Means that significantly differ from the midpoint in the expected direction for each item was interpreted as awareness of gender skewness in this sector.

Ability Gender Stereotypes

We measured explicit gender-stereotypical beliefs about abilities related to technology and caregiving in two ways. We constructed two items that measured awareness of societal stereotypes, inspired by Fiske et al. (2002), who suggested probing for societal stereotypes to minimize social desirability associated with admitting to personal stereotypes. The instructions read: “How do you think most Swedes view men’s versus women’s abilities (what they are good at)? We are not asking what you think, but what you believe that others think.” The first item asked about technical ability and the second item asked about caregiving ability (i.e., What do you believe that Swedes in general think of men’s and women’s technical/caregiving ability (how good they are with technology/caregiving)?” The scale ranged from 1 (Men are generally better) to 3 (There is no gender difference) to 5 (Women are generally better).

We also constructed two items to measure personal gender stereotypes, where we expected lower ratings due to social desirability (Fiske et al., 2002; Greenwald & Banaji, 1995). The instruction read: “What do you personally think about men’s and women’s abilities (what they are good at)?” Similar to above, the first item asked about technical ability and the second item asked about caregiving ability (i.e., What do you think about men’s and women’s technical/caregiving ability (how good they are with technology/ ability to care for other people)?). The scale ranged from 1 (Men are generally better) to 3 (There is no gender difference) to 5 (Women are generally better). Means that significantly differ from the midpoint in the expected direction was interpreted as indicating gender stereotypical awareness on the explicit stereotype measures.

Tech Interest

Lastly, we asked the school students (but not the teachers) to rate their interest in choosing a tech-focused educational program in the future on a scale from 1 (not at all interested) to 5 (extremely interested). This measure was adapted from the social cognitive career theory (Lent et al., 1994) and STEMO paradigms (Master & Meltzoff, 2020). In Sweden, school students choose a specific program orientation for high school, where “technology” is one choice. Students make this choice in Grade 9, and we therefore specifically asked for interest in this program orientation for the students in Grade 8 (“How interested are you in enrolling in the technology program in high school”). Since we believed the students in Grade 6 to have generally less awareness of the different program orientations that they later get to choose from for high school, we more generally asked them to rate interest in choosing “a tech-focused educational program in the future.”

Data Analysis

The data were analyzed in R version 4.0.5 (R Core Team, 2021). The dataset and code are provided in the open science repository (https://osf.io/nrjzb). All categorical variables used in the analyses were coded using Helmert contrasts (e.g., Gender coded as −1 for boys, 1 for girls). We processed the IAT data according to the D-score data-cleaning and scoring algorithm (Greenwald et al., 2003; Lane et al., 2007) and using the iatgen package (Carpenter et al., 2019). Moderation analyses to test Hypothesis 2 were conducted using multiple regression, and robust multiple regression was employed when assumptions of the original models were violated. Because the same patterns of results were found in both the original models and in the robust analyses, we present the results of the original models.

Preliminary analyses of the IAT data are provided in Table 1. Participants were dropped from the IAT analysis if they either failed to complete the IAT (e.g., because of a technical error) or if more than 10% of their responses were faster than 300 ms (Greenwald et al., 2003). Moreover, individual trials with responses above 10,000 ms were dropped (Greenwald et al., 2003). The dropped-trial rate was low for all three samples and the error rate was acceptable. Internal consistency was assessed using both a split-half procedure (De Houwer & De Bruycker, 2007) and using a variant of Cronbach’s alpha adapted for IATs (Schnabel et al., 2008), and both measures were relatively high across all three samples.

Table 1 Preliminary Analyses of the IAT Data
Full size table

Results

Gender Stereotyping of Technology and Caregiving

Descriptive statistics for the IAT and explicit measures are presented in Table 2 for the students and Table 3 for the teachers. To investigate Hypothesis 1a-c (i.e., whether students and teachers stereotype technology and caregiving according to gender), the tables also present the results from one-sample t-tests to determine whether the means differed from the scale’s midpoint (or zero for the IAT). As the scale midpoint (or zero) represents a lack of stereotyping, a significant difference would suggest gender stereotyping and thus support our predictions.

Table 2 Descriptive Statistics for the Implicit and Explicit Measures
Full size table
Table 3 Descriptive Statistics for Teachers and Teachers vs. Students Comparisons
Full size table

The results in Table 2 showed stereotype-consistent associations as measured by the IAT: Boys and girls in both Grade 6 and Grade 8 had mean D-scores in the hypothesized direction that differed from zero. The effect sizes were strong for all groups except for girls in Grade 6, who had a medium-sized effect, d = 0.80–1.05 and d = 0.50, respectively. That is, in line with Hypothesis 1a, the students tended to strongly implicitly associate technology more with men and caregiving more with women.

Next, the results on the explicit measures showed that the students, regardless of gender and school grade, were aware that both work in the care and tech sectors are gender-skewed in Sweden. All means largely differed from the scale midpoint, d = 0.86–1.60 (see Table 2). Another consistent pattern across gender and school grade was the students’ awareness of societal gender stereotypes in relation to the gender-segregated labor market. The means on these measures also largely differed from the scale midpoint in the expected direction, d = 0.75–1.79, indicating that the students believe that people in Sweden think that men have stronger technical ability than women and women have stronger caregiving ability than men, which supported Hypothesis 1b. Lastly, the students also tended to report personal endorsement of these stereotypes, which supported Hypothesis 1c. As expected, (possibly due to social desirability), the effect sizes were smaller for the personal stereotypes, but still significant with small to medium-sized effects, d = 0.32–0.71.

As with the students, the teachers showed large stereotype-consistent associations on the IAT (d = 1.48; see Table 3). The teachers also reported strong societal gender stereotypes (d = 1.23–1.57) and small to medium-sized personal stereotypes (d = 0.32–0.74), reflecting the belief that men have stronger technical ability than women and women have better caregiving ability than men. These results further supported Hypothesis 1a, b and c among teachers.

To assess whether teachers and students differ across the stereotype measures, we conducted two independent samples Welch’s t-tests for each stereotype measure, contrasting teachers with Grade 6 and Grade 8 students. The results revealed that the teachers had significantly stronger implicit stereotypes than the students in Grade 6 and Grade 8, d = 0.91 and d = 0.75, respectively (see Table 3). We also explored the explicit measures for differences, which showed more varied results. Specifically, the teachers reported stronger societal gender stereotypes than the students, except for the nonsignificant difference between teachers’ and Grade 8 students’ societal tech stereotypes (see Table 3). The teachers also reported stronger personal gender stereotypes about caregiving compared to Grade 8 students, but not compared to Grade 6 students. In contrast, the teachers reported significantly weaker personal gender stereotypes regarding tech ability compared to both Grade 6 and Grade 8 students.

Stereotyping and Tech Interest Link: Moderation by Gender

Implicit Stereotypes

Next, we were interested in the association between the student’s implicit stereotypes and their interest in enrolling in a tech-focused educational program (i.e., tech interest). We hypothesized that implicit gender stereotypes about technology would predict tech interest (Hypothesis 2), and that gender would moderate this relationship (Hypotheses 3a and 3b). The hypotheses were tested using multiple linear regression in a model with tech interest as the dependent variable and IAT D-score and gender as predictor variables, as well as the interaction term between gender and IAT D-score. IAT D-score was standardized for this analysis. The regression model (Model 1) was significantly better at predicting tech interest than the null model, F(3, 797) = 40.27, p < .001, R2adj = .128. Gender significantly predicted tech interest, β = −0.33, 95% CI [–0.41, −0.25], p < .001, ηp2 = .107, whereas IAT D-score did not, β = 0.03, 95% CI [–0.05, 0.11], p = .358, ηp2 = .001. Notably and as predicted, the interaction between gender and IAT D-score was significant, β = −0.14, 95% CI [–0.22, −0.06], p < .001, ηp2 = .021 (see Fig. 1). Specifically, IAT D-score positively predicted tech interest for boys (b = 0.63, 95% CI [0.28, 0.97]), whereas IAT D-score negatively predicted tech interest for girls (b = −0.40, 95% CI [–0.74, −0.06]), which supports both Hypothesis 2, 3a and 3b.

Fig. 1
figure 1

Gender × IAT D-Score Interaction Plot for Model 1 (With 95% CIs)

Full size image

Explicit Stereotypes

To explore if tech-interest also relates to explicit tech gender stereotypes (societal and personal), and if the relationship is moderated by gender, we created a new linear model (Model 2) with tech interest as the dependent variable and gender, societal technology stereotypes, and personal technology stereotypes as predictor variables. We also included a Gender × Societal Technology Stereotype and a Gender × Personal technology stereotype interaction, and we standardized the variables for both societal and personal technology stereotypes. Preliminary analyses identified six multivariate outliers based on Mahalanobis distances, whereupon the analyses were conducted both with and without these participants. The same pattern of results was found in both models; therefore, we present the findings below based on the full sample.

Model 2 was found to significantly outperform the null model, F(5, 863) = 24.78, p < .001, R2adj = .121. As in Model 1, gender was identified as a significant predictor of tech interest, β = −0.31, 95% CI [–0.39, −0.24], p < .001, ηp2 = .094. Societal technology stereotypes also significantly predicted tech interest, β = 0.11, 95% CI [0.03, 0.20], p = .001, ηp2 = .012, whereas personal technology stereotypes did not, β = 0.03, 95% CI [–0.06, 0.11], p = 434, ηp2 = .001.

Further, gender was found to moderate the association between societal technology stereotypes and tech interest, as indicated by the significant Gender × Societal Technology Stereotype interaction, β = 0.10, 95% CI [0.02, 0.18], p = .005, ηp2 = .009. While there was no association between societal technology stereotypes and tech interest for boys (b = 0.02, 95% CI [–0.11, 0.16]), there was a significant positive association for girls (b = 0.29, 95% CI [0.16, 0.42]): girls who endorsed stronger societal gender stereotypes about tech (i.e., that Swedish people find men to be better than women at technology) reported lower tech interest, whereas tech interest was higher for girls with weaker societal gender stereotypes. No interaction effect was found for gender and personal gender stereotypes about tech, β = −0.01, 95% CI [–0.10, 0.07], p = .750, ηp2 < .001.

Gender × Background × School Grade on Implicit Associations

Lastly, we explored the interaction effects of gender, background, and school grade on the students’ implicit gender associations for technology/caregiving. We used the Anova() function in the car package (Fox & Weisenberg, 2019) to conduct a 2 × 2 × 2 ANOVA with IAT D-score as the dependent variable and gender, school grade, and background as the independent variables. We used Type III sums of squares and conducted Tukey’s HSD post hoc tests to interpret interaction effects. Descriptive statistics for each subsample are presented in Table 4. Preliminary analyses revealed no univariate outliers and no violations of the assumptions of the ANOVA. There was a significant main effect of gender, F(1, 786) = 16.08, p < .001, ηp2 = .020, where boys demonstrated slightly stronger implicit stereotype-congruent associations than girls. There was also a significant main effect of school grade, F(1, 786) = 6.60, p = .010, ηp2 = .008, where Grade 8 students demonstrated somewhat stronger stereotype-congruent associations than Grade 6 students. No significant main effect of Swedish/foreign background was found, F(1, 786) = 3.41, p = .065, ηp2 = .004, and no significant two-way interactions including background, Fs < 1.60.

Table 4 Descriptive Statistics for the Students’ IAT D-Scores
Full size table

There was, however, a small significant interaction effect between gender and school grade, F(1, 786) = 8.52, p = .004, ηp2 = .011, indicating an effect of school grade for girls (p = .009) but not boys (p = .995). The interaction showed that while the boys’ IAT-scores did not differ between the age groups, the girls in Grade 6 had weaker implicit gender stereotypes than the girls in Grade 8. Notably, the Gender × School Grade interaction was superseded by a significant three-way Gender × School Grade × Background interaction, F(1, 786) = 4.08, p = .044, ηp2 = .005 (see Fig. 2). For students with a Swedish background, the pattern for the Gender × School Grade interaction was almost identical to the corresponding two-way interaction outlined above. By contrast, for students with a foreign background, there was no significant gender difference in Grade 8 (p = .998), but girls in Grade 6 showed significantly weaker implicit stereotypes than boys (p = .019). Interestingly, girls in Grade 6 with a foreign background (M = 0.06, SD = 0.32) did not show any implicit stereotypes, as their IAT score was not significantly different from zero, t(40) = 1.25, p = .217, 95% CI [–0.04, 0.16], d = 0.20.

Fig. 2
figure 2

Gender × School Grade × Background Interaction Effect on IAT Scores (Standard Error Bars Included)

Full size image

Discussion

The main aim of this study was to investigate students’ gender stereotyping of technology and caregiving and whether these stereotypes relate to their interest in pursuing a tech-focused education. We administered both implicit (IAT) and explicit measures of gender stereotypes for technology and caregiving in a sample of middle school students in Grade 6 and Grade 8 in Sweden. For a comparison, we also investigated these gender stereotypes in a sample of middle school teachers.

Students demonstrated strong implicit gender stereotypes and strong endorsement of society’s explicit gender stereotypes of men being better at technology than women and women better at caregiving than men, but weaker personally held explicit gender stereotypes for these dimensions. Further, stronger endorsement of implicit tech gender stereotypes was associated with less interest in pursuing a tech education for girls, and stronger interest in tech education for boys. Similar results were found for explicit societal tech gender stereotypes, but the students’ personally held explicit technology gender stereotypes did not relate to their tech interest. This could be due to social desirability concerns, whereby people tend to hesitate to report personally held stereotypes, which is why implicit and societal gender stereotype measures may be better indicators (Crandall et al., 2002; Dovidio et al., 2003; Fiske et al., 2002; Greenwald et al., 1998, 2003; Greenwald & Banaji, 2017).

Specifically, the girl’s tech interest was weaker if they had stronger implicit gendered associations for technology and caregiving and if they perceived that society endorses the belief that men are better at technology. In contrast, the boys were more interested in tech-education if they endorsed stronger implicit stereotypes, but perceptions of societal beliefs were unrelated to their interest in tech gender stereotypes. The finding that implicit and explicit societal tech gender stereotypes related to weaker tech interest in girls corroborates EEVT, SEVT and STEMO, which state that gender stereotypes may serve as a barrier to developing an interest in tech-focused careers for girls and women, such as IT and engineering (Eccles, 1987, 1994; Eccles & Wigfield, 2020; Master & Meltzoff, 2020), and expands previous research linking other implicit stereotype measures and career interest (e.g., Cvencek et al., 2011; Cundiff et al., 2013; Lane et al., 2012; Nosek et al., 2002; Starr, 2018; Starr & Simpkins, 2021).

Demographic Factors

We also explored if the strength of the implicit gender stereotyping of technology and caregiving differed in relation to participant age, gender and Swedish versus foreign background, and interactions between these variables. To reduce complexity and due to the social desirability concerns associated with explicit measures (Fiske et al., 2002), we mainly focused our comparisons on the implicit stereotype measure in this study.

Age Comparisons: Teachers Versus Students

To explore if stereotype strength related to age, we compared the student mean values to a sample of middle school teachers. The results showed that the teachers had much stronger implicit technology/caregiving gender stereotypes than the students, and the youngest students had the weakest associations. This result corresponds with previous research that has shown stronger implicit gender stereotypes in older generations as compared to younger (Charlesworth & Banaji, 2019). Drawing on SRT, this could perhaps relate to people increasingly consolidating these implicit stereotypical associations over time, when continuously encountering examples that fit into the expected gender roles, women in caregiving roles and men in technology roles.

The results were more complex when exploring age differences on the explicit gender stereotypes, which may relate to social desirability concerns. First, like the IAT effect, the teachers reported stronger societal gender stereotypes as compared to the students (i.e. the belief that Swedes in general think that men are better at technology and women are better at caregiving). However, for the personal beliefs, the teachers rated men’s superior technical skills less extremely than the students, but rated women’s superior caregiving skills more extremely than the older students (and similarly to the younger students). This pattern of results perhaps reflects adults’ stronger awareness of gender-power relations, which should make expressing personally held prejudice towards lower-status groups (“women are not good at technology”) less socially acceptable than expressing prejudice toward higher-status groups (“men are not good at caregiving”, e.g., Crandall et al., 2002; Ridgeway, 2001).

That the teachers had very strong implicit technology-caregiving stereotypes is worrying, since previous research has shown that teachers’ stereotypes influence students’ self-concepts, academic achievement, and educational choices, often through unintentional stereotype expressions, such as in feedback (Dickhäuser & Meyer, 2006; Starr & Simpkins, 2021; see also Master & Meltzoff, 2020 for a review). Since most of the teachers in our participant sample taught programming (among other subjects), it is particularly problematic that they associated technology so strongly with men and explicitly reported a personal belief that men have better technical abilities than women. Studies of gender differences in ability tend to show gender similarity or small gender differences, and large individual differences (Hyde et al., 2005, 2019; Zell et al., 2015). This appears to be true also for programming ability, although very little research has tested for it (Lin, 2016). Programming was recently implemented as mandatory in Swedish schools, to meet the future recruitment demands of an increasingly digitalized labor market (Ministry of Education, 2017). However, if teachers unintentionally convey the belief to students that technology is for men, gender differences in interest in tech-careers may persist despite this reform. Future research should investigate if teachers’ level of tech gender stereotyping relates to the size of gender differences in their students’ tech interest.

Gender Differences in Implicit Stereotyping

We also tested for gender differences in students’ implicit gender stereotyping. The results demonstrated that the boys had stronger implicit gender stereotypes than the girls. Further, while the boys’ IAT-scores did not differ between the age groups, the girls in Grade 6 had weaker gender stereotypes than the girls in Grade 8. When previous research has found a gender difference in implicit stereotyping, there has been a tendency to stereotype one’s own group more favorably (Rudman et al., 2001). Since the tech sector has higher status than the care sector (e.g., Block et al., 2018a; Croft et al., 2015; Oldenziel, 1999), our result may reflect girls need more exposure to the horizontal gender segregation in the labor market, before forming the association of technology with men and caregiving with women, since this implies internalizing a lower status for their gender ingroup.

Gender, Age, and Background

Since a quarter of the students were classified as having a foreign background according to Swedish official criteria (being born abroad or both parents being born abroad), we lastly explored interactions between gender, age, and background on the implicit gender stereotyping measure. Interestingly, the results showed that although the students overall demonstrated strong implicit gendered associations with technology and caregiving, the young girls (Grade 6) with a foreign background did not demonstrate implicit gender stereotypes, as their mean IAT-scores did not differ from zero. Due to the low number of girls in this category (n = 41), the result needs to be replicated before drawing conclusions, but it raises interesting prospects for future research.

We do not know what countries the girls (or their parents) with a foreign background in this study originated from. However, the region in Sweden where the data was collected has recently had many immigrants from countries in the MENA region (Migrationsinfo, 2020). Interestingly, it is more common for women to study engineering and computer science in several countries in this region (e.g., United Arab Emirates and Bahrain ) as compared to Sweden (Huyer, 2015). It has previously been shown that cross-country variation in gender-skewness in the STEM-sector relates to variation in the strength of implicit science-gender stereotypes (Miller et al., 2015). This raises the possibility that technology gender stereotypes may be stronger in Sweden as compared to countries with less gender-skewed tech-educations. We therefore suggest that future research should replicate and expand our result and especially encourage cross-country comparisons of our technology-caregiving IAT.

If technology gender stereotypes were shown to be unusually strong in Sweden, it may also be relevant in relation to the “gender-equality”-paradox. This pattern suggests that gender differences in interest in “people” versus “things” may be larger in countries that score high on gender-equality indexes, which Sweden does (European Institute for Gender Equality, 2021; Stoet & Geary, 2018). Recently, Breda et al. (2020) suggested that the gender-equality paradox might relate to stronger gender stereotypes in countries which score high on gender-equality indexes. Drawing on EEVT, SEVT and STEMO, if technology gender stereotypes are stronger in more gender-equal countries, this may hinder tech-interest (“things”-interest) development for girls, and subsequently lead to fewer young women studying tech-educations, as compared to less gender-equal countries (Eccles, 1987, 1994; Eccles & Wigfield, 2020; Master & Meltzoff, 2020).

Although strong gender stereotypes in gender-equal countries may sound counter-intuitive, previous research has shown that strong gender-equity values tend to coexist with gender-essentialist thinking in European countries (Knight & Brinton, 2017). Historical events may have played a role in shaping a gender-segregated labor market which led to the development of strong gender stereotypes despite reductions in male-primacy associations (i.e., attitudes privileging men over women, Knight & Brinton, 2017). The development of the welfare state in Scandinavian countries in the 1960s meant that many Swedish women entered the labor force, but the majority were employed by the new publicly financed care sector (Nermo, 2000). Further, although Sweden has long had a self-image as an “engineering nation,” it is only relatively recently that the tech sector welcomed women. One telling example is that the industry recruited school students with a campaign called “The men of the future” in the 1960s (Leidi, 2011).

Lastly, only the younger girls (Grade 6 as compared to Grade 8) with a foreign background did not demonstrate implicit gender stereotypes in this study. Perhaps the two-year age gap between these girls meant longer exposure to the gender stereotypes that appear to exist in Sweden, and subsequent internalization.

Limitations and Future Research Directions

Although the sample of school students was relatively large in this study (n = 873), the external validity of our study is limited. We encourage replications, especially outside of Scandinavia, to allow for cross-country comparisons. Also, the effect size for the relationship between gender stereotypes and tech-interest was weak. To further understand how gender stereotypes may hinder career interest developments, we recommend investigating the relationship between tech gender stereotypes and factors that impact career interest. EEVT (Eccles, 1987, 1994), SEVT (Eccles & Wigfield, 2020), STEMO (Master & Meltzoff, 2020) and other influential theories like social cognitive career theory (SCCT; Lent et al., 1994) state that career interest is affected by a multitude of factors, such as self-efficacy, various outcome goals (e.g., salary expectations), utility value, and social belongingness expectations.

We did not ask the participants with a foreign background what specific country of origin they had in this study. We encourage replications with larger participant samples and more nuanced measures of background, including nationality and ethnicity, to learn more about potential interaction effects of gender and nationality/ethnicity on technology and caregiving stereotypes. We also encourage qualitative research with in-depth interviews with young girls from ethnic minority groups in Sweden, to get a fuller understanding of their perception of technology and gender. We also recommend testing their parents’ level of gender stereotyping, to investigate if it relates to the students’ level of stereotyping.

Also, since the teacher sample was small in this study, it should be replicated with larger samples of teachers, and other adults, to test if adults generally have stronger implicit technology and caregiving gender stereotypes than children and adolescents, which research on implicit science-arts gender stereotypes has shown (Miller et al., 2015).

Further, the internal validity of the IAT has been questioned. In this study, one concern is whether the IAT is measuring personally endorsed stereotypes (i.e., “Men have stronger technical ability”) or awareness of these stereotypes and the gender segregation in the labor market (i.e., “More men do tech work”; Arkes, 2004; Nosek, 2005; Olson & Fazio, 2009). We argue that regardless of whether it is personal endorsement or awareness, perceptions of gender roles should relate to perceptions of who fits these roles and could form a barrier to developing career interest based on individual person-job fit rather than on gender role fit. Due to other limitations of the IAT, described in the introduction, we encourage researchers to consider complementary implicit measures (Meissner et al., 2019). The IAT was developed in part to reduce social desirability associated with explicit stereotype measures (Greenwald et al., 1998), which was also the rationale for measuring societal explicit stereotypes rather than personally held stereotypes (Fiske et al., 2002).

Lastly, our career-interest measure only captured interest in a future tech-focused educational program, since it was part of a larger project that longitudinally investigates students’ perceptions of technology. We suggest adding a measure of interest in the care sector in future replications, to test whether students’ implicit associations of gender with technology versus caregiving also relate to gender differences in interest in HEED-occupations (like nursing). What shapes HEED-interest is much less understood (Block et al., 2018a).

Practice Implications

We believe that the results of this study have important implications for the tech industry, which sees rapidly growing recruitment needs (Manpower Group, 2022; Nordic Council of Ministers, 2021; OECD, 2019). It suggests that to attract women to the tech industry, we need to counteract the stereotype that men are better at technology. To break this stereotype, we recommend educating the public about the research which has shown large individual differences in abilities and overall minor gender differences (Hyde et al., 2005, 2019; Zell et al., 2015), including in programming ability (Lin, 2016). Educating about gender similarity appears to be the most effective remedy for stereotype threat (Liu et al., 2021). The tech industry has a unique opportunity to help counteract tech-related gender stereotypes, including negative stereotypes of women’s abilities, which currently plagues social media and internet forums (e.g., Bates, 2020; Opp & Lagunas, 2022).

This study showed stronger implicit technology and caregiving gender stereotypes among teachers than students. It is important for teachers to realize that they may, unintentionally and without awareness, convey the stereotypes they have internalized to the students they teach (Dickhäuser & Meyer, 2006; Starr & Simpkins, 2021; see also Master & Meltzoff, 2020 for a review). The United Nations Development Programme (UNDP, 2020) reports that close to 90% of the world population hold negative stereotypes about women, and in some countries, the bias is increasing. The EU has stated that gender stereotypes are the greatest obstacle to achieving gender equality (European Institute for Gender Equality, 2017), so teaching about gender similarity could increase gender equality, in addition to helping the tech industry meet its recruitment needs.

Conclusion

Students and teachers in Sweden strongly associate technology with men and caregiving with women, and girls with stronger stereotyped associations are less interested in tech-focused educations. The insight that gender stereotypes relate to students’ career interest has important implications for policy makers invested in gender equality. As career fit is better predicted by individual traits than by gender affiliation (Ceci et al., 2009; Hyde et al., 2019), counteracting gender stereotypes should allow for better person-job fit in the labor market. Many more women could thrive in the tech sector and many more men in the care sector, than is currently the case.