Literature review

View the most recent version.

Archived Content

Information identified as archived is provided for reference, research or recordkeeping purposes. It is not subject to the Government of Canada Web Standards and has not been altered or updated since it was archived. Please "contact us" to request a format other than those available.

Archived

This page has been archived on the Web.

The purpose of the current paper is to examine the link between gender, mathematics and science test scores at age 15, and first program choice in university. Therefore, this literature review will concentrate on those relatively few studies in Canada and elsewhere that address this relationship. The lack of available data on this topic may explain the paucity of research in this area in Canada. Before the Youth in Transition Survey (YITS) and the Programme for International Student Assessment (PISA) were linked in 2000, no suitable data sources comprising measures of both mathematical and science ability (other than marks), in adolescence, existed that could be attributed to a student’s program choice in university.

Many studies (e.g., OECD 2010) have used mathematics or science marks as measures of ability. While these studies are quite closely linked to ability measured via test scores, they may not be true objective indicators of mathematical or science ability. This is because of issues related to student behaviour and performance and other school factors, such as the teacher–student relationship, and curricula (Cornwell, Mustard, and Van Parys 2011). Moreover, important gender differences have been found in terms of performance, as measured by marks or by achievement tests, with girls generally having been found to do better than boys with respect to marks, but not achievement tests (Duckworth and Seligman 2006).

While this paper cannot address why the performance of boys and girls differs in mathematics/science marks and achievement tests, it is important to highlight some of the earlier literature in this area. For example, one potential explanation for the gender discrepancy in marks is that girls are typically more self-disciplined than boys with regard to school work and, as a result, girls may work harder in school (and exhibit better classroom behaviour), resulting in higher marks (Downey and Vogt Yuan 2005). Conversely, however, girls typically exhibit lower mathematics test scores than boys. This work finds that the reasons for the discrepancy range from the type of activities boys and girls do outside the classroom —with boys more involved in activities that promote quantitative skills and girls more involved in activities that promote verbal/reading skills (Downey and Vogt Yuan 2005) — to more biological explanations. These purport that boys outperform girls on most measures of visuospatial abilities (which are more tightly connected with mathematical and science ability) (see Halpern, Benbow, Geary, Gur, Shibley Hyde, and Gernsbacher, 2007 for a review). These issues are complex and beyond the scope of this paper; thus, the focus is placed, instead, on literature looking at university program choice and the intersection between gender and ability in mathematics/science.

The National Educational Longitudinal Study (NELS) from the United States (years encompassing 1988 to 1994) has been used as a source of data for examining the link between academic ability in high school and program choice in university. For example, Goyette and Mullen (2006) used a combined academic proficiency measure that integrates mathematical and reading academic proficiency tests administered to students in grade 12. Major field of study was split into two categories: arts and science (includes humanities, science, mathematics, and social sciences) and vocational (includes business, education, engineering, pre-professional such as law, medicine, architecture and also other occupationally oriented majors such as agriculture, communications, design, protective services).¹ Goyette and Mullen found that their measure of academic proficiency is positively related to entering arts and science programs versus vocational programs. They do not analyze the relationship between academic proficiency and university program choice separately by gender.

Correll (2001) used the 1988 to 1994 waves of NELS, and found that mathematics test scores in grade 8 led to a higher likelihood of a student choosing a major in university that requires ‘quantitative’ skills and knowledge (such as engineering). This effect remained after controlling on verbal test scores, mathematics and English grades, as well as mathematical and verbal ability self-assessments and calculus enrolment in high school. However, while Correll did not assess the interaction effect of mathematics test scores by gender, her study is important because it illustrates the importance of self-assessment. For example, she found that males are more likely than females, with the same mathematics marks and test scores, to perceive that they are competent in mathematics. Thus, her

“results suggest that those who persist on a mathematical career path may not even be the best qualified for careers requiring mathematical proficiency. In other words, boys do not pursue mathematical activities at a higher rate than girls do because they are better at mathematics. They do so, at least partially, because they think they are better.” (p. 1724).

Trusty (2002) echoed this finding on the importance of self-assessment when he found that mathematical self-perception reduced the predictive ability of science and mathematical ability, with those who perceive themselves to be good at mathematics more likely to choose a mathematics/science program.

Additional evidence from the NELS (Trusty, Robinson, Plata and Ng 2000) found that men choose programs such as engineering, medicine and sciences more often, whereas women choose programs such as education, nursing and ethnic studies more often. However, ability in grade eight operated differently by gender. For example, the interaction of mathematics and reading test scores and gender showed that across all postsecondary majors, mathematical ability in grade eight had a stronger influence on men’s program choices, while reading ability had a stronger influence on women’s choices. 

In subsequent work using the NELS, Trusty (2002) looked at the factors determining the choice of science/mathematics majors versus all others. He observed that, in models for men that included science and mathematics test scores, the science score was positive and significantly related to a person choosing a science/mathematics major. The influence of mathematical ability was not significant in these models, but was in those that did not include science, which suggests a relatively high degree of overlap between mathematics and science measures. Moreover, course-taking behaviour appeared to matter. For example, for women, mathematics scores appeared to be more important than science scores, but the inclusion of variables related to course-taking behaviour (i.e., especially important was whether the respondent took a mathematics-related course such as geometry, trigonometry and calculus in high school) completely removed the positive effect of mathematics test scores on postsecondary program choice. For men, the effect of science test scores on choosing a mathematics or science major remained after including course-taking behaviour.

Using a more selective American sample of entrance test scores, from 12 academically selective colleges and universities from the 1951, 1976, and 1989 entering cohorts, Turner and Bowen (1999), showed that men with high mathematics Scholastic Aptitude Test (SAT) scores were more likely to enter the fields of mathematics/physics and engineering than their female counterparts with similar mathematics SAT scores. In contrast, females with high mathematics SAT scores were more likely to enter the humanities, other social sciences, and bio-life sciences than their male counterparts with high mathematics SAT scores. However, Turner and Bowen found that achievement scores were only one, small factor that helped predict gender differences in program choice at university. In fact, factors related to differences in preferences and labour market expectations were even more important in explaining the gender gap in program choice.² Similarly, Ware and Lee (1988) discovered that, in high school, different things influence men’s and women’s decisions on program choice. For example, women who went into non-scientific fields placed the greatest importance on how their education decisions would affect future family and personal life.³

The research that examines the impact of mathematical/science ability, measured via test scores instead of marks, on postsecondary program choice is scant internationally, and is even less common in Canada. In fact, in Canada there appears to be no single study that actually uses mathematics or science test scores to estimate a youth’s first program type in university. The two closest studies at the national level also use the YITS-PISA file (see Finnie and Childs 2010; and OECD 2010). However, neither of these studies uses mathematics or science PISA test scores, but rather use the PISA reading test score in combination with reading, mathematics, and science marks.

Finnie and Childs (2010) found that high PISA reading scores increase the probability of entering a STEM (science, technology, engineering and mathematics) field in university, but the effect is somewhat greater for males than it is for females. Conversely, the PISA reading score appears to be related more strongly to non-STEM participation for females than males.  For both sexes, mathematics and science marks in high school also indicate a greater propensity for entering STEM programs in university.  The OECD (2010) results also generally support these findings, but split out university fields of study beyond the STEM/non-STEM dichotomy.  For instance, greater reading proficiency measured via the PISA reading scores, at age 15, indicate a greater likelihood of a student choosing a pure sciences or life science program versus one in the human sciences, arts, or communication fields. Marks in mathematics or science achieve the same effect, while marks in reading show the opposite; namely, that youth with higher reading marks are more likely to choose the human sciences, arts or communication fields versus the pure or life sciences. The link between high marks in high school and choosing science programs in university is further exemplified in recent work from Burrow, Dooley, Wright and DeClou (2012). Using administrative data from Ontario, they found that high school graduates with the highest marks (the top 5%) are most likely to enrol in science courses. For example, in 2008, approximately half of these high achievers chose science programs in university, while less than 20% chose arts programs.

This review provided an overview of recent research that studied the relationship between relatively objective measures of mathematical and science ability and the eventual choice of program in university.  Across different countries, examples were found that used different mathematics/science tests, as well as different categorizations of subject area in university.  Even though the type of mathematical and science ability measures as well as the university program categorization varied across studies, their findings showed a great deal of consistency in:  there is a link between mathematics or science test scores (and also reading scores) and university program choice. Youth with stronger mathematical and science ability are more likely to choose programs where these skills are essential (such as science and engineering). However, these studies also illustrate that many other factors are also important. Most notable are mathematics or science marks, as well as subjective self-assessments of one’s mathematical skills. In fact, a recent OECD (2012) study suggests that “gender disparities in subjects chosen appear to be related more to student attitudes (such as motivation and interest) towards a particular subject rather than to ability and performance at school” (p. 104).

---

Notes

1. Differences in categorization between Goyette and Mullen and the current study make comparisons difficult; the categories completely overlap and are not mutually exclusive.
2. They suggest that women may prefer fields where skills do not deteriorate or become obsolete as quickly. They may be more likely to stay away from fields related to highly technical subjects such as computer science, so that time away from work to raise a family is not as harmful for one’s career.
3. Using a birth cohort in the United Kingdom from 1958, Van de Werfhorst, Sullivan and Cheung (2003) also found that a large gender discrepancy in program choice (particularly into the sciences) cannot be fully explained via mathematical ability at age 11: other factors such as social class, family background and cultural capital in the home also remain very important.