The emergence of gender differences in children's perceptions of their academic competence

Abstract

This longitudinal study assessed the emergence of gender differences in over 300 children's academic competency beliefs from kindergarten or first grade through fifth grade. Children, their parents and their teachers rated the children's competencies in math and literacy and their math and literacy skills were also assessed directly. Beginning in third grade, girls rated their math competencies lower than did boys, even though there was no gender difference in math achievement or in teachers' ratings of children's math ability. Parents also rated boys' math competencies higher than girls in both third and fifth grades. Although girls outperformed boys on the literacy achievement measure in third and fifth grade, and teachers' ratings of children's literacy ability reflected this gender difference in performance, gender differences were not found in either parents' or children's ratings of literacy ability. Results of regression analyses indicated that actual achievement and teachers' ratings predicted children's judgments of their literacy skills. Parent ratings of children's competence were a particularly strong predictor of children's judgments of their math skills.

Introduction

Although gender differences in mathematics performance have declined over the last few decades, gender differences in participation in some math courses and professions involving math continue. Studies show, for example, that girls are less likely than boys to take certain advanced courses in high school, and fewer women than men choose to select technical and mathematical college classes and majors and to enter certain technical and mathematical professional fields (AAUW, 2000, NSF, 2002). Because these professions are among the fastest growing and most lucrative careers, being underrepresented in these fields puts women at an economic disadvantage.

Eccles and other theorists (e.g., Eccles, 1984) have proposed that girls' lower participation rates in math courses and professions, but not in courses and professions requiring literacy skills, are explained in part by girls' relatively low perceptions of their competencies in math. If perception of competence affects students' participation, it is important to understand the genesis of gender differences in competency beliefs.

The goals of this longitudinal study were to determine when gender differences in children's perceptions of their abilities emerge, and to explore possible contributing factors, including actual achievement and parents' and teachers' judgments of children's competence which may be conveyed to and internalized by children. The study examines perceptions of competence and actual achievement in literacy as well as mathematics to determine whether girls' underestimation of abilities is confined to male-stereotyped domains, or found also in a subject area in which girls are known to perform relatively well. We were also interested in assessing whether boys develop relatively negative beliefs about their competencies related to literacy, because it is a domain in which girls often outperform boys. Below we summarize relevant evidence from previous research.

Gender differences in performance on math achievement tests have not generally been found in the early elementary school grades (Cole, 1997; see also De Lisi & McGillicuddy-De Lisi, 2002). Studies have reported that by middle school boys outperform girls in some areas of math (Cole, 1997, Davies & Brember, 1999, Garner & Engelhard, 1999, Olszewski-Kubilius & Turner, 2002, Park et al., 1998, Swiatek et al., 2000, Vermeer et al., 2000) and girls outperform boys on grades and on other kinds of math assessments (Cole, 1997, Davies & Brember, 1999, Garner & Engelhard, 1999, Howell, 2001, Hyde et al., 1990, Park et al., 1998, Pomplin & Capps, 1999).

Despite the lack of clear evidence for boys' superior math skills, studies consistently find gender differences in children's perceptions of their ability in math (Burnett, 1996, Eccles et al., 1983, Eccles et al., 1984, Eccles et al., 1993, Manger & Eikeland, 1998, Siegle & Reis, 1998, Skaalvik & Rankin, 1994, Stipek, 1984, Stipek & Gralinksi, 1991, Wigfield et al., 1997). Evidence on when gender differences in perceptions of competence in math emerge is not entirely consistent. Several studies have found gender differences in children's math competency ratings as early as first grade, with some studies finding that differences remain constant across the elementary grades, and other studies finding that the difference increases or decreases through these grades. Wigfield et al. (1997) (see also Eccles et al., 1993) report in their cross-sequential study of children from 1st to 6th grade that boys rated their perceptions of their ability in math significantly higher than girls in first grade; the gender difference did not vary significantly across the elementary grades. Marsh, Craven, and Debus (1998) similarly reported in a cross-sequential study modest but significant gender differences in math self-concept of 5–8 year olds that favored boys and increased over time. Another cross-sequential study found that gender differences emerged by first grade, but the gender difference diminished over time (Fredericks & Eccles, 2002); children's ratings of their math competence, math interest, and math importance declined between 1st and 12th grade, but girls' ratings of their math competence declined at a slower rate than did boys' ratings.

Only one study was found that reported gender differences in perceived competence in math emerging later than early elementary school. In a cross-sequential study of 5th–12th graders, girls' attitudes (perception of ability, perceived task difficulty, and subjective task value) became less positive about math as they got older, even though the standardized math test scores of the older girls were relatively higher than those of younger girls. Not until 10th grade did girls rate their math abilities lower than boys rated their abilities (although there was no gender difference in standardized test scores; Eccles, 1984).

To summarize, the weight of the evidence suggests that a gap between girls' and boys' self-perceptions of competence in math develops in the early elementary grades, as early as first grade, and earlier than or independent of any gap found in actual achievement. We hypothesized, accordingly, that boys would rate their math competencies higher throughout the elementary grades included in this study, but they would not outperform girls on the math achievement test.

Girls typically outperform boys in literacy skills as early as kindergarten and first grade (Denton & West, 2002). In the most recent National Assessment of Educational Progress (NAEP, 2003) data, for example, girls outperformed boys in reading at both the fourth and eighth grades. A variety of smaller studies also show girls outperforming boys in assessments of literacy skills (Frome & Eccles, 1998, 6th and 7th grade; Pajares, Miller & Johnson, 1999, 3rd–5th grade; Swiatek et al., 2000, 3rd–6th grade).

The evidence on whether and when gender differences in children's perceptions of their abilities in literacy emerge is inconsistent. No mean-level gender differences in perceptions of competence were found by Eccles et al. (1993) for 1st–4th graders in reading or by Stipek, (1984) for 5th graders in spelling. In contrast to these studies, Wigfield et al. (1997) found that girls rated their reading skills higher than boys as early as first grade and throughout elementary school. Similarly, in a study of 5th- and 6th-grade children, girls had higher self-concepts of ability than boys in English (Frome & Eccles, 1998).

Pajares et al. (1999) have examined gender differences in writing achievement and motivation in a number of cross-sectional studies. Taken together, their findings suggest that gender differences favoring females emerge by the end of elementary school, continue through middle school, and decline in high school. For example, in one study of 3rd, 4th, and 5th graders, even though girls performed better on an in-class essay than boys, they did not rate themselves higher than boys on a measure of writing self-efficacy, when controlling for aptitude (see also Swiatek et al., 2000). Eccles' (1984) study of 5th–12th graders showed that girls' attitudes (perception of ability, perceived task difficulty, and subjective task value) about English became more positive as they got older, whereas boys' attitudes remained stable over time; by 8th grade, girls earned higher grades in English than boys, although there were no gender differences in English standardized test scores (Eccles, 1984).

In another study, although fifth-grade girls did not outperform boys on a timed, in-class essay, they reported higher self-efficacy beliefs in writing (Pajares & Valiante, 1997). In a similar study of sixth-, seventh- and eighth-grade students, teachers rated girls' writing competence higher than boys', and although girls' and boys' writing self-efficacy did not differ, the proportion of girls who claimed they were better at writing than the boys in their class was higher than the proportion of boys who claimed they were better at writing than the girls in their class (Pajares & Valiante, 1999). Pajares and Johnson (1996) found that ninth-grade boys' and girls' writing performance was similar, but boys reported higher self-efficacy in writing than did girls. Comparing this study of high school students with those of younger students, Pajares and Valiante (1997) suggest: “it may be that, at the elementary school levels, girls have more positive self-beliefs about writing and express greater confidence in themselves as writers but that this confidence erodes as they progress in school” (p. 358).

In summary, the findings for literacy-related skills are inconsistent, with some studies finding gender differences as early as first grade, many in late elementary school and early middle school, and a few others not until high school. In all but one case, when gender differences were found, they favored girls. Previous research led us, therefore, to hypothesize that if gender differences are found in the present study, girls would rate their literacy competencies higher than boys; it does not suggest a specific hypothesis regarding the grade at which such a gender difference would emerge.

One possible explanation for the fairly consistent finding that girls have lower perceptions of their math competency despite equivalent performance is that significant adults—parents and teachers—convey gender stereotypic views that girls are not as good in math as boys, which girls then internalize. Consistent with this interpretation, parents have been shown to rate girls' and boys' abilities differently in accordance with stereotypic beliefs about gender roles.

In one study of children in kindergarten through third grade, parents rated daughters more capable in reading at all grades (Eccles, 1993). In another study, parents' ratings of their fifth- and sixth-grade children's English abilities corresponded with children's actual grades in English; girls performed better than boys in English and parents rated daughters' abilities in English higher than they rated sons' abilities (Frome & Eccles, 1998). Parents in this study also claimed that daughters required less effort to get high grades in English.

In contrast, parents' ratings of fifth- and sixth-grade girls' and boys' math ability was the same even though girls were performing better in math (Frome & Eccles, 1998). Eccles found in another study that whereas parents of children in kindergarten through third grade reported sons as more capable in sports than daughters, on average they did not rate sons and daughters differently in math in these early elementary grades (Eccles, 1993). In contrast to the two studies above, in which parents did not rate boys' math abilities higher than girls, other studies did find gender differences in parents' ratings, favoring boys, for parents who held stereotyped beliefs about gender roles (Eccles, 1993, for children in kindergarten through third grade; Jacobs & Eccles, 1992, 6th–7th grade; Tiedemann, 2000, 3rd–4th grade).

Taken together, these studies suggest that parents may accurately perceive daughters to be better in literacy-related competencies, but may underestimate girls' performance (or overestimate boys' performance) in math, especially if they hold stereotypic gender-role beliefs. Findings for literacy were found as early as kindergarten in at least one previous study, leading us to hypothesize that parents in the present study would rate girls' literacy skills higher than boys' throughout the elementary grades studied. If differences in parent ratings are found for math, they were expected to favor boys.

The evidence on the effects of gender on teachers' evaluations of children's math competence is mixed, although there is some evidence that teachers may perceive boys to be more able in math, even though they do not perform better than girls. In Tiedemann's (2000) study, teachers rated 3rd- and 4th-grade boys slightly higher in mathematical ability than girls, although these girls' and boys' grades from the previous year were not significantly different. Jussim and Eccles (1992) also found that teachers rated boys' talent in math higher than they rated girls' talent, even though they rated girls' classroom performance higher than boys' and even though boys did not perform better on standardized achievement tests. In contrast, Helwig, Anderson, and Tindal (2001) found that gender was not a significant contributor to teachers' ratings of 3rd- and 5th-grade girls' and boys' math ability.

Few studies have assessed teachers' ratings of children's literacy skills, although those found did favor girls. In a study of primary school classrooms of children aged 8–11 years, teachers rated girls' performance in reading and writing higher than boys' (Fergusson, Lloyd, & Horwood, 1991). In a study of 4th–8th grade gifted students, teachers rated girls' ability in language arts higher than boys' ability, although teachers did not rate girls' and boys' abilities in other subjects (mathematics, social studies, science) differently (Siegle & Reis, 1998).

Too few studies have been conducted with young children to make specific hypotheses about when gender differences in teachers' ratings are likely to emerge, but previous research led us to hypothesize that gender differences would be found by the fifth grade, with teachers rating girls higher in literacy and boys higher in math.

If parents' or teachers' gender-related beliefs about children's competencies are somehow conveyed to children and are thus in part responsible for gender differences in children's own beliefs, there should be an association between children's and their significant others' beliefs. We turn now to this literature.

Extant research has shown that parents' beliefs about their children's ability are associated with children's actual achievement and children's own beliefs about their academic ability in both literacy and math (Eccles, 1993, Fredericks & Eccles, 2002, Jacobs & Eccles, 1992, Parsons et al., 1982, Shaver & Walls, 1998). In one study, 1st- through 5th-grade children's perceptions of their academic ability were more strongly correlated with their parents' perceptions of their academic ability than with their teachers' perceptions (Wigfield et al., 1997), and in some studies of elementary school children, more than with their grades and/or actual performance (Eccles, 1993, Frome & Eccles, 1998, Jacobs & Weisz, 1994).

There is some evidence that teachers' views of children's ability are also correlated with children's perceptions of their ability as well as with children's actual achievement. Jussim and Eccles (1992), for example, found that teachers' expectations predicted 6th-grade students' math achievement beyond the effects of children's previous achievement (see also Wigfield et al., 1997).

Although previous evidence is sparse, it suggests that parents' and teachers' judgments of children's competencies may have some influence over children's own perceptions. We hypothesized, therefore that parents' and teachers' ratings would predict children's self-ratings, even after actual achievement is controlled statistically, throughout the grades included in the present study.

Most previous studies have been cross-sectional, limiting opportunities to identify the emergence of gender differences and to assess different factors that might explain children's own judgments. The present study followed the same group of children from kindergarten or first through fifth grade, assessing actual skills and perceptions of competencies at least three times over this five- or six-year period. Teachers' and parents' perceptions as well as children's own perceptions of their competencies in both literacy and mathematics were assessed at each time period, allowing us to compare the predictive power of these three variables. Actual achievement was expected to be the strongest predictor of children's judgments about their math and literacy competencies, followed by parents' and then by teachers' judgments, in part because teachers' judgments were expected to be more redundant than parents' judgments with actual achievement.

We predicted that gender might continue to predict children's judgments of the mathematics competencies even after these other possible influences were taken into account because children receive messages about gender stereotypes through other sources, such as the media and from peers. For example, children may see more men than women in professions involving math—in real life or in the media. Whether or not their parents hold gender stereotype beliefs about math competencies, children may be more likely to observe men in professions requiring math (e.g., engineers, accountants), or they may have older siblings or cousins who convey gender stereotypes (e.g., a sister who does not like math, a brother who does). These observations may affect their own beliefs about their math competencies regardless of the beliefs conveyed to them directly by teachers and parents.

Previous studies have included primarily middle-class students. This study's sample consists of economically disadvantaged students, mostly students of color, who are at risk of poor academic achievement. Low-income students were studied because they have been neglected in past research and because girls from low-income families might face gender-related as well as poverty-related challenges to the development of their academic competencies, especially in mathematics. The present study, therefore, attempted to compensate for previous neglect of a population of unusually vulnerable students.

In brief, the present study seeks to expand on the current literature on gender differences in self-perceptions by analyzing longitudinal data, studying a different and more vulnerable sample than that included in most previous studies, and by comparing assessments of performance and self-perceptions in two important academic domains, literacy and math. The two primary questions addressed were: (1) Are there gender differences in children's, parents' and teachers' ratings of competence in literacy and math? (2) Do parents' and teachers' judgments of children's competencies predict children's self-perceptions after children's actual achievement is held constant? We investigated age differences with regard to both of these sets of questions.