Across a wide range of countries, males and females differ notably in how often they pursue degrees and careers in certain STEM fields (National Science Foundation, 2011; OECD, 2004). Precursors for such gendered career pathways can be found in secondary school, where males and females already differ in their career aspirations and—given that they have the choice—in the math and science courses they take (Schoon & Eccles, 2014; Watt & Eccles, 2008; Watt et al., 2012). To describe this phenomenon, several authors have used the metaphor of a leaky pipeline towards STEM (e.g., Watt & Eccles, 2008): People drop out of a pathway towards STEM-related careers at various time points and females are generally more likely to drop out than males. Mathematics is seen as the critical filter in this pipeline because math courses in secondary school affect the career options that one has at a later time point (Watt &
Eccles, 2008).
The Eccles et al. expectancy-value theory (1983) is a very prominent approach to explain such gender differences in choices (e.g., Chow & Salmela-Aro, 2011; Nagy et al., 2008; Watt et al., 2012). The most proximal factors that are assumed to explain gender differences in academic choices are expectancy and value beliefs for related subjects—with a particular focus on value beliefs as the driving force of choices (Eccles, 2005, 2009, 2011). These beliefs, in turn, are supposed to be affected through gendered socialization processes. Males and females make different experiences related to their gender role. As long as this gender role is a central part of their identity, these experiences will lead to males and females having different expectancies and values.
For instance, the cultural definition of gender roles can affect the priorities in the long-term goals that males and females pursue. Males’ and females’ expectancies and values are also supposed to be shaped by their experiences with parents, teachers, and peers who might provide them with different feedback on their opportunities.
Whereas expectancy-value theory focuses on socialization processes affecting motivational beliefs in explaining gender differences in choices, several alternative explanatory factors have been proposed (for reviews, see Ceci, Williams, & Barnett, 2009; Wang & Degol, 2013). Ceci et al. (2009) provide a comprehensive review on the factors driving the underrepresentation of women in math-intensive fields, considering
biological as well as sociocultural causes. Based on the evidence considered in this review, they suggest that women’s preferences are the most powerful explanatory factor. In addition, they consider gender differences on gatekeeper tests such as the mathematics section of the Scholastic Assessment Test (SAT) in the US (especially at the right tail of the distribution) as one important factor. Based on the empirical evidence, they conclude that such gender differences in achievement are more likely caused from sociocultural than from biological factors. Gender differences in mathematics achievement (at the mean level as well as in the distribution), however, can only partially explain the female underrepresentation in STEM-related careers (Ceci et al., 2009; Wang & Degol, 2013). In another review on the factors explaining gendered career choices, Wang and Degol (2013) stress the role of occupational preferences and lifestyle values for women’s underrepresentation in STEM and illustrate the sociocultural influences on STEM choices. Expectancy-value theory seems to be an especially powerful explanatory framework for gender differences in choices as it considers a wide range of contributing factors.
According to expectancy-value theory, gendered socialization should result in boys reporting more favorable expectancy and value beliefs in male-typed domains such as mathematics and girls reporting more favorable expectancy and value beliefs in female-typed domains such as languages. The empirical evidence mostly confirms this pattern of gender differences in expectancy and value beliefs. With regards to expectancies, it has been consistently found across diverse samples that boys rate their expectancies in mathematics higher than girls—regardless of their abilities (e.g., Jacobs et al., 2002; Marsh et al., 2005; Nagy et al., 2010). Although similar gender differences can be found across different Western cultures (Nagy et al., 2010), this gender effect seems to be culturally shaped as it varies considerably across nations all over the world (Else-Quest, Hyde, & Linn, 2010). Females, on the other hand, tend to report higher expectancies for more female-typed domains such as language arts and foreign languages, although this female advantage has not always been found (e.g., Durik et al., 2006; Jacobs et al., 2002; Nagy et al., 2008; Watt, 2004). Again, it needs to be noted that such gender stereotypic differences in subjective beliefs cannot fully be explained by boys’ and girls’ achievement. In fact, for grades—which seem to more important for students’ expectancies than achievement tests (see Marsh et al., 2005)—females tend to earn higher grades in almost all school subjects. A recent meta-analysis (Voyer &
Voyer, 2014) reported that this female advantage was largest in language arts (d = 0.374) and smallest for math (d = 0.069). To explain this somewhat paradoxical pattern, it has been argued that females tend to underestimate their abilities in comparison to males (Stetsenko, Little, Gordeeva, Grasshof, & Oettingen, 2000).
For value beliefs, the pattern of results is not straightforward for math. Whereas some studies reported higher values in mathematics for boys (e.g., Marsh et al., 2005;
Steinmayr & Spinath, 2008), others reported no difference between boys and girls (e.g., Jacobs et al., 2002; Meece et al., 1990; Wigfield et al., 1997). These inconsistencies can partly be explained by the operationalization of value beliefs and differences in the value dimensions incorporated (i.e., one general value scale or separate value components). Studies that considered separate value components found that male adolescents reported higher scores on measures of math interest and intrinsic value (Frenzel et al., 2010; Frenzel, Pekrun, & Goetz, 2007; Marsh et al., 2005; Watt, 2004), whereas gender differences in attainment and utility value are more inconsistent and seem to depend on the specific operationalization (Frenzel et al., 2007; Meece et al., 1990; Steinmayr & Spinath, 2010; Watt, 2004; Watt et al., 2012). Few studies, though, included measures of all value components, making direct comparisons of the results across studies difficult. Other moderators such as the age group or the cultural background of the sample are also possible. For the area of languages, on the other hand, girls were consistently found to place higher values on language arts and foreign languages (Durik et al., 2006; Jacobs et al., 2002; Nagy et al., 2008; Watt, 2004).
Several studies addressed the question of how such gender differences in expectancy and value beliefs develop over the school years (Frenzel et al., 2010; Jacobs et al., 2002; Nagy et al., 2010; Watt, 2004). Stereotypic gender differences have been shown to already emerge at the beginning of elementary school and to remain relatively stable throughout the school years. Little evidence, thus, supports an intensification of gender differences as hypothesized by gender role socialization perspectives (Eccles, 1987; Hill & Lynch, 1983). However, socialization processes in the school and at home have been found to contribute to gender differences in children’s beliefs (Eccles, 2007;
Jacobs, Davis-Kean, Bleeker, Eccles, & Malanchuk, 2005; Wang & Degol, 2013):
Parents and teachers show gender-dependent behavior that can shape boys’ and girls’
expectancies and values; they communicate different expectations for boys and girls, provide them with gender-typed experiences, and act as role models.
To summarize, expectancy-value theory is a very powerful approach for explaining gender differences in choices. A particular emphasis in the literature has been put on expectancy and value beliefs in mathematics as a precursor of choosing a career within STEM. Although values are assumed to be the driving force for choices, the gender differences regarding math that have been found in previous research seemed to be more pronounced for expectancies than for value beliefs. More research is therefore needed to explain this seemingly inconsistent pattern of results.