Running head: ADOLESCENT GIRLS’ IMPLICIT MATH ATTITUDES The Effect of Dating Activity and School Environment on Adolescent Girls’ Explicit and Implicit Math Attitude

This study investigated the effects of dating and school environment on implicit and explicit math attitudes among adolescent girls who applied to the Young Women’s Leadership Charter School (YWLCS), which emphasizes math and science competence. Ninety-three female adolescents, (66 YWLCS and 27 control participants), completed two web-based study sessions. The first session was available beginning the first month of the academic year, and the second beginning in January. Each session included a math attitude Implicit Association Test (IAT) and self-report items gauging explicit academic attitudes and dating behaviors. Contrary to hypotheses, neither attending YWLCS nor dating had a significant effect on students’ implicit math attitudes. These findings support the need for more research on adolescent dating and implicit academic attitudes. Adolescent Girls’ Implicit Math Attitudes 3 The Effect of Dating Activity and School Environment on Adolescent Girls’ Explicit and Implicit Math Attitude A little more than fifteen years after Mattel’s Teen Talk Barbie declared, “Math class is tough!” the stereotype that math and science are masculine domains persists in American culture. However, studies show that there are few significant differences in actual cognitive abilities in math and science between men and women (Hyde, Fennema, & Lamon, 1990; Hyde, Lindberg, Linn, Ellis, & Williams, 2008). Although there has been significant progress in terms of the numbers of women receiving doctoral degrees in science and mathematics, there has not been a comparable improvement in the numbers of women faculty members in these fields (Bleeker & Jacobs, 2004). Given that gender differences in mathematical ability alone cannot account for why women should be underrepresented in math and science careers, it is imperative to investigate what other factors may be involved. Research has shown that women, unlike men, are frequently faced with negative stereotypes regarding their mathematical abilities and that these stereotypes can impair performance (Spencer, Steele, & Quinn, 1999). Despite the similarity in mathematical abilities between males and females, research consistently indicates a pervasive implicit perception that math and science are masculine domains (Nosek, Banaji, & Greenwald, 2002; Nosek, Greenwald, & Banaji, 2006). These attitudes and stereotypes are fostered in girls (and boys) throughout their lives, passed on, perhaps unconsciously, by parents (Parsons, Adler, & Kaczala, 1982), teachers (Garrahy, 2001; Jussim & Eccles, 1992), and the media (Wilgosh, 2001). There is evidence that parental perceptions of children’s capabilities are stronger influences than past academic performance on the child’s own view of his or her abilities (Parsons et al., 1982). Parsons et al. found that mothers and fathers have sex-differentiated views Adolescent Girls’ Implicit Math Attitudes 4 of their children’s math abilities such that daughters were perceived as having to work harder to succeed in math than sons were. Additionally, parents of sons viewed math as more important for their child than did parents of daughters. Such beliefs influence children throughout their development, and have long-term implications for children’s achievement choices and their selfperception (Bleeker & Jacobs, 2004). Aside from parental influences at home, teachers’ expectations of student performance are predictive of how students will actually perform (Rosenthal & Jacobson, 1968, 1992; Jussim & Eccles, 1992). Rosenthal and Jacobson found that teachers’ expectations caused change in students’ academic achievement even when accounting for the students’ past achievement and motivation. Jussim and Eccles conducted a partial replication of Rosenthal and Jacobson’s research in which they investigated teachers’ perceptual biases with regard to math and gender. As predicted, they found that teachers view girls as performing more highly than boys, and as trying harder than boys. However, teachers view boys as having more math talent than girls. According to this view, girls who succeed in math do so because they compensate for their lack of math talent by working hard. Such biased views can unintentionally influence the way teachers treat students. Teachers may treat students differently based on the student’s sex, even when they believe they are acting in a gender-blind manner towards their students (Garrahy, 2001); however, this is not always the case (Helwig, Anderson, & Tindal, 2001). If teachers treat boys and girls differently when teaching math, and parents do not view math as being as important for their daughters as for their sons, young women may be less likely to pursue challenging math classes at the college level, thereby decreasing the likelihood of electing for a career in math or science. Consistent with this, girls are also less likely to report a desire to pursue a mathematics-related career and are more likely to have a lower self-efficacy in math and science than boys, regardless of their actual Adolescent Girls’ Implicit Math Attitudes 5 performance and abilities (Bleeker & Jacobs, 2004). Indeed, boys consistently rate their selfefficacy for male-typed domains (like math and sports) higher than girls, while girls rate their self-efficacy for female-typed domains (like language arts) higher than boys. Such gender differences in self-beliefs may have a role in observed gender differences in certain achievement behaviors (Jacobs, Lanza, Osgood, Eccles, & Wigfield, 2002). Additionally, girls are susceptible to stereotypical media images of women, including portrayals of men disapproving of female success (Wilgosh, 2001). Wilgosh concludes that such images of women in media can lead girls to downplay their intelligence and academics. This in turn can affect their choice of academic pursuits and careers, specifically in math and science (Jacobs & Eccles, 1985). Societal stereotypes are prevalent and difficult to avoid, and research indicates that girls are influenced by the stereotypes about their gender and math at a fairly early point in their education (Steele, 2003). Steele’s experiments show support for a theory of stereotype stratification, where members of a negatively stereotyped group, despite viewing themselves as belonging to the group, view themselves as an exception to the stereotype. Steele found that elementary-school girls rated men as liking math more and being better at it than women; however they viewed girls and boys as liking math equally and being equally talented in math. In another study, elementary-school girls were told either a story about an adult mathematician or about a child mathematician. When told the adult version, the girls were more likely to draw an adult man, but when told the child version, they were more likely to draw a picture of a young girl. This suggests that young girls believe at some level, perhaps an implicit level, that something happens between childhood and adulthood that leads men to be more mathematically inclined than women, even if, in reality, men are not. Adolescent Girls’ Implicit Math Attitudes 6 Awareness of these stereotypes leaves girls vulnerable to stereotype threat, meaning that simply knowing that one’s group is expected to perform less well in a domain will increase the likelihood of poorer performance (Spencer et al., 1999; Steele & Aronson, 1995; Steele, 1997). Research on stereotype threat has shown that it is likely to occur when participants care about doing well on the given task, when the test items are of a high level of difficulty, when participants believe they will be evaluated, and when negative stereotypes about their group are directly applicable to the situation (Steele, 1997). These effects have been specifically demonstrated for women in math and science (Keller, 2002; Shih, Pittinsky, & Ambady, 1999; Spencer et al., 1999). Spencer et al. conducted several studies replicating the method used by Steele and Aronson (1995) to investigate whether stereotype threat is implicated in women’s math performance. The study demonstrated that women underperform on math tests when stereotype threat is high. Additionally, when stereotype threat is lowered by telling participants that the test produces no gender differences in performance, women perform better. Despite the unsettling reality of the negative effects of stereotype threat, recent studies have suggested that it can be reduced through interventions (Cohen, Garcia, Apfel, & Master, 2007; Dar-Nimrod & Heine, 2006). Dar-Nimrod and Heine investigated how women’s mathematical performance is affected by the source to which gender differences in math performance are attributed. Dar-Nimrod and Heine concluded that stereotype threat in this domain can be reduced by presenting experiential causes, instead of genetic causes, of gender differences in math performance. Thus, by exposing women to scientific theories that suggest that math performance is the result of experience and not genetics, women’s performance in math can be improved. Adolescent Girls’ Implicit Math Attitudes 7 Furthermore, stereotype threat can be reduced by protecting women’s math performance through the presence of female role models in a mathematics domain (Marx & Roman, 2002). Marx and Roman found that women perform better when a math test is administered by a female, and that this effect is heightened when the female administrator is highly math-competent, compared to when she is not. Additionally, studies have shown that being in the presence of a female role model can lead to women having higher career aspirations, by demonstrating multiple-role self-efficacy, thereby communicating that it is possible to be successful in both career and home life (Nauta, Epperson, Kahn, 1998). Having such a role model within the classroom can change academic attitudes (Evans, Whigham, & Wang, 1995), and Nosek and Banaji (2001) demonstrated in the laboratory a similar effect on implicit math attitudes. Building on this literature, studies have also indicated that at-risk girls who have consistent contact with a female mentor in a technology-rich, all-girls mathematics environment can dramatically improve their confidence and performance in mathematics (Reid & Roberts, 2006). These benefits may occur in large part because of operations at an implicit level, where girls are not aware that their performance and confidence are improving, yet it appears in measures. Thus, the current research aims to more directly measure implicit associations within the context of an all-girls learning environment. There is evidence that simply being in a single-sex learning environment can enhance girls’ academic achievement (Inzlicht & Ben-Zeev, 2000) and reduce adolescent delinquency (Caspi, Lynam, Moffitt, & Silva, 1993). Inzlicht and Ben-Zeev (2000) studied the effects of being a numerical minority or majority during difficult math and verbal tests on women’s test performance. Female participants in the minority condition displayed impaired performance for the math test only. The researchers also found that female participants’ performance was slightly Adolescent Girls’ Implicit Math Attitudes 8 impaired when they were in a mixed-sex group with one other female and one male where females were still the majority, suggesting that performance is proportional to the number of males present in the group. The researchers conclude that girls benefit from being placed in single-sex classrooms. Riordan (1998) cites a reduction in gender differences in opportunities, in gender differences in curriculum, and in gender bias in student-teacher and peer interactions as potential benefits of a same-sex learning environment. Additionally, the culture of same-sex schools emphasizes academics over youth-culture values, creating an environment that fosters feelings of empowerment within the student (Riordan, 1998). However, contrary to Inzlicht and Ben-Zeev (2000) and Caspi et al. (1993), Riordan asserts that single-sex schools work only for those from a low socio-economic status or who are otherwise disadvantaged in society by racial or gender factors. Despite encouraging evidence for the benefits of same-sex education for girls, the data is mixed, with some studies showing no difference in academic performance for girls from same-sex versus coeducational schools (Haag, 1998). There is even evidence that all-girls schools perpetuate sexism through academic dependence and less rigorous instruction more than co-educational schools that promote gender equality in the student and faculty bodies (Lee, Marks, & Byrd, 1994). A potential drawback to most of the research on girls’ attitudes about academics is that it relies on self-report measures that, while capturing explicit attitudes, are not sensitive to the implicit attitudes that affect behavior. Humans have the ability to introspect; however, introspection is limited when it comes to accessing information that people are either unwilling or unable to report (Nisbett & Wilson, 1977). Individuals may be unwilling to report information that is in opposition to other values or to acceptable societal norms, and, even if an individual is willing to report such information, it may simply be inaccessible to them (Nosek, Greenwald, & Adolescent Girls’ Implicit Math Attitudes 9 Banaji, 2006). Implicit constructs, according to Greenwald and Banaji (1995), are “introspectively unidentified (or inaccurately identified) trace[s] of past experience” that mediate a response (p. 5). They proposed that studies of implicit cognition could provide researchers with access to information that, via self-report, has formerly been inaccessible. In an effort to access such unconscious attitudes and beliefs, researchers have developed implicit measures and applied them to various areas of research, including racial attitudes, political attitudes, and gender attitudes (Nosek et al., 2006). Despite the unconscious nature of these attitudes, they have been shown to predict certain kinds of behavior (Greenwald, Poehlman, Uhlmann, & Banaji, in press). The Implicit Association Test (IAT; Greenwald, McGhee, & Schwartz, 1998) has proved an effective measure the strength of associations among concepts (Nosek et al., 2006). The IAT is a within-participants experimental measure that compares the time a participant takes to sort words or images into categories under two counterbalanced conditions, which, theoretically, indicates the relative strength of the cognitive associations between concepts (Lane, Banaji, Nosek, & Greenwald, 2007). Nosek et al. (2006) reviewed the Implicit Association Test in terms of its internal validity, addressing the characteristics of the stimuli used, the procedure, and possible extraneous influences. They also examined its construct validity, convergent and discriminant validity, predictive validity, as well as issues of interpretation of IAT effects. They concluded that there is strong evidence for the IAT’s internal, construct, and predictive validity. The IAT has been applied to the study of men and women’s implicit attitudes about math and science (Nosek et al., 2002). They found that implicit math-gender stereotypes favoring men were equally strong for among men and women, and these stereotypes are related to one’s performance in, liking of, and identification with math. Individuals with a strong implicit female gender identity tended to have greater implicit negativity towards math and weaker selfAdolescent Girls’ Implicit Math Attitudes 10 identification with math. The researchers concluded that groups (such as male and female) communicate certain expectations (in this case, regarding math) of its members that can influence preferences and performance. Research by Kiefer and Sekaquaptewa (2007) has shown that implicit, but not explicit, stereotyping interacts with gender identification to influence not only women’s will to pursue a career in mathematics, but also women’s performance on math exams, indicating that implicit attitudes influence behavioral outcomes even outside of an individual’s awareness. Their research also indicated that a math=male stereotype impairs math performance for women. Kiefer and Sekaquaptewa measured implicit gender-stereotyping of math among female college students enrolled in a calculus class. Participants also completed explicit surveys on gender identification, explicit gender stereotyping, and career goals. Findings indicate that implicit gender-math stereotyping interacted with gender identification to predict women’s calculus performance as well as their desire to pursue a career in math. Women who did not identify as strongly with their gender showed less implicit gender stereotyping regarding math, performed better in the class, and were more likely to report a desire to pursue a career in math. The authors conclude that the underrepresentation of women in mathematics may be in part due to implicit beliefs about women’s mathematical abilities. Dasgupta and Asgari’s (2004) research applied implicit methodology to the study of same-sex learning environments. They found that an all-girls educational setting reduces implicit gender-stereotypic attitudes. When exposed to female leaders, the female participants were less likely to express automatic stereotypic beliefs about women. Additionally, Dasgupta and Asgari found that male-dominated university classes, such as courses in math and science, produced an increase in automatic stereotypic beliefs, but only among students attending coed college, and Adolescent Girls’ Implicit Math Attitudes 11 not for those attending a women’s college. This research indicates that local environments influence women’s implicit beliefs about their ingroup. The present research was conducted through the Full Potential Initiative (FPI), a program of research the University of Virginia, which focuses on girls in a single-sex charter school, the Young Women’s Leadership Charter School (YWLCS) where math and science competence is emphasized. The YWLCS is the only all-girls public school in the city of Chicago. The goal of the school is to offer urban girls in grades 7-12 a college preparatory curriculum emphasizing math, science and technology, in hopes of equipping students with the skills necessary to pursue degrees and careers in these fields where women have been historically underrepresented. Applicants are admitted to the YWLCS by an annual lottery, and the student population mirrors that of Chicago Public Schools as a whole, with 78 percent African American, 15 percent Latina, 6 percent Caucasian, 1 percent Mixed Race and 1 percent Asian. Additionally, 80 percent of the students are eligible for either free or reduced price lunch. FPI applies implicit methodology to the study of adolescent girls’ attitudes and stereotypes concerning math and science and gender identities and compares their attitude and stereotype development with that of girls who were not selected in YWLCS’s annual random lottery for admission. Much research on the effects of same-sex environments is flawed by a selection bias in terms of who attends these schools. Girls who apply and are accepted on basis of academic merit to all-girls schools are unique from those in public schools to begin with. In such instances it would be difficult to determine with confidence that differences exhibited in the two groups over time are attributable to the affects of the school environment and not to preexisting differences. For example, some research indicates that differences in outcomes between co-educational and same-sex schools disappear when socio-economic status is taken into account (Haag, 1998). FPI’s quasi-experimental research Adolescent Girls’ Implicit Math Attitudes 12 design capitalizes on YWLCS’s random admission lottery by recruiting enrolled and nonaccepted students into treatment and control groups, respectively. To the extent that representative samples of each group are recruited, one can assume that any individual differences among the girls are randomly distributed between the two groups; thus, the composition of the two groups should be similar on all aspects aside from YWLCS attendance. Participating girls complete IATs on math attitudes, stereotypes, and gender-role identification every few months. Preliminary findings indicate that girls attending YWLCS show more implicit and explicit preference for mathematics, show a weaker math=male stereotype, and identify less with the stereotypically female identity. Girls’ academic performance is by no means solely based on the students’ educational environments. Individual factors, such as whether the student matures earlier or later relative to her peers, are also related to academic outcomes. Early maturing girls are at greater risk for delinquency, however this association is only found for girls in a mixed-sex school setting (Caspi et al., 1993). Caspi et al. propose that the onset of puberty triggers a stimulus to others in the environment, creating a press for more adult-like behavior among adolescent girls. They note that in our age-graded society, girls become biologically mature long before they are expected to be socially and financially independent. In an effort to acquire adult privileges, girls may participate in norm-defying behavior. Caspi et al. cite their finding that early-maturing girls in mixed-sex settings have more contact with deviant peers and have participated in more deviant behaviors themselves than their peers in same-sex schools. Early maturing girls also tend to start dating at a younger age than late maturing girls (Phinney, Jensen, Olsen, & Cundick, 1990; Lam, Shi, Ho, Stewert, & Fan, 2002). This is particularly problematic because early dating is correlated with lower academic achievement for Adolescent Girls’ Implicit Math Attitudes 13 girls (Brendgen, Vitaro, Doyle, Markiewicz, & Bukowski, 2002; Neeman, Hubbard, and Masten, 1995; Scott, Stewart, & Wolfe, 2005). Brendgen et al. (2002) administered surveys to early adolescents which asked about their social acceptance among same-sex peers, whether or not they had a boyfriend or girlfriend, and their self-esteem. Additionally, teachers completed measures on the students’ antisocial behavior and academic performance. Brendgen et al. hypothesized that only those adolescents who had healthy relationships with same-sex peers would benefit from a romantic relationship during early adolescence. Those in unhealthy relationships with same-sex peers were predicted to be negatively affected by the presence of a romantic relationship, because they would not have yet developed the skills necessary to handle relationships in a same-sex context. Researchers found that the level of same-sex peer acceptance moderated the relationship between having a romantic boyfriend or girlfriend and the students’ self-esteem and problem behavior. However, among these students there was no indication that being in a relationship is related to specifically positive emotional and behavioral adjustment. For early adolescents who were unpopular with same-sex peers, having a romantic relationship was correlated with lower levels of self-esteem and higher levels of antisocial behavior. Additionally, regardless of whether or not the student was popular with same-sex peers, being in a relationship during early adolescence was negatively related to academic performance, however only for girls. The researchers concluded that being in a romantic relationship is related to self-esteem and antisocial behavior, but only for those students who are rejected by their same-sex peers. By contrast, same-sex peer acceptance did not moderate a negative relationship between being in a romantic relationship and academic performance for girls. The current research focuses on the effects of dating activity in the context of quasiexperimental groups in the FPI study. Consistent with FPI’s preliminary findings, I predict that Adolescent Girls’ Implicit Math Attitudes 14 girls attending YWLCS will show an increase in positive attitudes towards math from the fall to the spring, but less so for the girls who have dated. I also predict that girls in the non-YWLCS schools will show a decrease in positive attitudes towards math from the fall to the spring, consistent with previous findings. Additionally, this decrease will be greatest in girls who have dated. See Figure 1 for a graphical representation of hypotheses. Method Participants Female adolescents between the ages of 12 and 19 attending schools in the Chicago, Illinois area who had applied to attend the Young Women’s Leadership Charter School (YWLCS) were eligible to participate in this study. Of these students, 208 completed a web-based study session in September 2008. 143 were students enrolled in the YWLCS, and the remaining 65 participants were students enrolled in other schools. In January 2009, 139 participants completed a study session, 104 of whom were enrolled in the YWLCS and 35 of whom were enrolled in other schools. At the completion of the study, 93 participants had completed both sessions, 66 from YWLCS and 27 from the comparison group. YWLCS students were, on average, about a year older (M=15.5, SD= 1.7) than control students (M= 14.4, SD= 1.1). All participants were recruited from yearly lists of applicants to the YWLCS. Those enrolled in YWLCS are recruited through their teachers to participate in studies through the Full Potential Initiative. Non-YWLCS students are recruited through presentations at open houses prior to the admission lottery or by direct mail from lists provided by YWLCS. As an encouragement to seriously considering registering with FPI, students were offered a $5.00 gift card to their choice among Subway restaurants, Target stores, or Barnes & Noble booksellers for returning signed consent materials, Adolescent Girls’ Implicit Math Attitudes 15 even if they decided not to register. An additional $5.00 gift card is awarded each time the student completes a study session. The Implicit Association Test The Implicit Association Test (IAT; Greenwald, McGhee, & Schwartz, 1998) was used to measure participants’ implicit math attitude. By measuring the time participants take to sort stimuli from four categories (e.g., a person smiling or scowling; math words or language arts words) into two response options, the IAT provides a relative measure of the strength of automatic associations among concepts. For this study, I used a math attitude IAT developed by the Full Potential Initiative to measure participants’ implicit attitudes towards language arts and math. The IAT was modeled after the recommendations set forth by Greenwald, Nosek, and Banaji (2003), which suggests seven blocks. See Table 1 for the sequence of trial blocks for the math attitude IAT. For the purposes of this study, we reduced the number of trials in each block by one-third. To measure positive and negative associations with math and language arts, we used pictures of an individual either smiling or scowling. Thus, positive, or good, associations were operationalized by a smiling face, and negative, or bad, with a scowling face. The IAT measured the degree to which participants associate good with language arts and bad with math compared to the opposite combination. In two counter-balanced conditions, participants sort stimuli from these four concepts (good, bad, math, and language arts) into two combination categories (either good/math and bad/language arts or bad/math and good/language arts). Response options are represented by the “E” key or “I” key on the computer keyboard. Participants complete a practice phase to learn how to take the IAT before proceeding the critical phases. In the first critical phase of the IAT, words or images representing two concepts (e.g., scowling faces and words such as Adolescent Girls’ Implicit Math Attitudes 16 count, divide) are classified with one key. Items representing the remaining two concepts (smiling faces and words such as letter, sentence) are classified with the other key. In the second critical phase, the concepts are sorted in the reverse manner, such that good expressions are associated with math words, and bad expressions are associated with language arts words. The face images were counterbalanced by race and gender. Researchers assume that if the participant implicitly associates positive feelings (represented by a smiling face) with language arts, then associating these concepts together should be easier than the reverse. Ease of association is determined by the speed at which the participant responds (with slower responses indicating weaker associations) and by the frequency of errors (with more errors indicating weaker associations). An IAT effect for an individual is determined by taking the difference between the mean response latencies of the two critical conditions (i.e., math-bad/language arts-good and mathgood/language arts-bad). Response latencies were measured from the beginning of a trial until the time of a correct response and were calculated using the D algorithm to standardize the IAT effect for each individual. The effect was calculated such that positive and higher D scores indicate a stronger implicit positive attitude towards math (Greenwald et al., 2003). For a complete list of math/language arts words and good/bad expression images used in the IAT, see Appendix A. Procedure Participants in this study completed two 15-minute experimental sessions online. Students were informed that new sessions were available via email messages, and YWLCS science teachers made announcements in their classes. Participants completed the sessions at their leisure during the time span during which the session was available, in locations and with Adolescent Girls’ Implicit Math Attitudes 17 computers of their choosing. We began administering the first session in September, about one month into the academic school year for YWLCS students; the second session administration began in January, at the mid-point of the academic year. The first session was open to nonYWLCS students from September 19, 2008 to January 7, 2009 and to YWLCS students from September 19 to October 15, 2008. The second session was available from January 13, 2009 to February 8, 2009. Using web-based accounts on the Full Potential Initiative website, participants logged into their account using a user name, thereby completing the sessions confidentially. Participants did not need to have completed the first session to participate in the second. The first session included three distinct IATs: math attitude, math-gender stereotype, and gender identity. The second session included only the math attitude and the math-gender stereotype IATs. Both sessions included several self-report survey items. Explicit and implicit measures were counterbalanced such that participants either completed all IATs first, followed by the survey items, or the reverse. Survey questions concerned participants’ self-efficacy and attitudes towards language arts and math, beliefs about gender differences in academic abilities in language arts and math, their career/academic aspirations. All participants in the second session were presented with the dating items as the last section of the session. This study focused on the information gathered from the questions regarding dating and romantic relationships, specifically, whether the participant reported ever having had a romantic or dating relationship. Appendix B contains all explicit and self-report items included in the sessions. Results Group Comparison at First Measurement Occasion We first tested whether baseline (Time 1) explicit and implicit attitudes differed by groups (YWLCS and control). Descriptive statistics and correlations for these and other variables Adolescent Girls’ Implicit Math Attitudes 18 are listed in Table 2 for YWLCS and Table 3 for control students. Since we found that age differed for the two groups, in these and all other analyses we included age and the age X group interaction term as covariates to account for possible effects of age. By regressing each dependent variable on the grouping variable “YWLCS student” (where controls were coded “0” and YWLCS coded “1”), we found that groups did not differ significantly on any of the attitude measures at Time 1. With age held constant, the estimated mean implicit math attitude for YWLCS and control participants were -.04 and -.05, respectively, p=.95. Because neither mean is significantly different from zero, these data indicate that neither YWLCS nor control participants had a preferential bias towards math or language arts at Time 1; that is, both groups seemed to have equally positive or negative implicit attitudes towards math and language arts. Likewise, explicit math attitudes (a 7-point scale, coded -3, hate it, to 3, love it) between YWLCS and control participants did not differ significantly, with estimated means of .79 and .86, respectively, p= .82. Nor did explicit language arts attitudes differ between YWLCS and control participants (a 7-point scale, coded -3, hate it, to 3, love it), with estimated means of .80 and 1.18, respectively, p=.11. For math and language arts explicit attitudes, both groups evidenced attitudes slightly (between 0= “it’s okay” and 1= “like it”) in the positive direction. Observed means for math attitudes differed significantly from zero for both YWLCS and control students (t= 4.95 and 4.56, respectively, p< .0001), as did observed means for language arts attitudes (t= 7.19 and 6.85, respectively, p< .001). Group Comparison at Second Measurement Occasion As with Time 1, group means on implicit attitudes measured by the Math Attitude IAT did not significantly differ from one another at Time 2. Means for YWLCS and control participants, respectively, were M=-.09 (SD= .42 and .47), p= .94. Both groups exhibited a Adolescent Girls’ Implicit Math Attitudes 19 slightly negative attitude towards math, although only significantly for YWLCS participants (t= -2.17, p=.03). Explicit math attitudes between YWLCS (M=1.1, SD= 1.6) and control (M= 1.4, SD= 1.5) did not significantly differ, p= .32, with both groups’ observed means indicating a statistically significant slightly positive explicit math attitude (t= 6.4 and 5.6, respectively, p<.0001). Explicit language arts attitudes also did not significantly differ between YWLCS (M= 1.8, SD=1.3) and control participants (M= 1.2, SD= 1.2), p= .90. As with Time 1, both YWLCS and control groups evidenced slightly positive attitudes (between 1=“like it” and 2= “love it”) towards both math and language arts, and observed means significantly differed from zero (t= 8.9 and 5.9, respectively, p< .0001). In addition to explicit and implicit attitude measures, dating behavior measures were administered at Time 2. Among all participants, both YWLCS and control, 70 percent reported having had at least one romantic or dating relationship (n=89). Of those, most (61 percent) reported being in a relationship at the time they participated in the session, and most (79 percent) reported having had their first dating relationship at or before age 14, M= 13.0, SD= 1.8. Since the primary dating variable is dichotomous, a yes or no response, we used logistic regression to test whether YWLCS and control participants differed in their likelihood of having ever dated and found that they did not (p= .16). Controlling for age, among those who had dated, and again using standard multiple regression, we found no group differences for YWLCS and control for the age at which they had their first dating relationship (M=13.1, SD= 1.9; M= 12.9, SD=1.1, respectively, p=.23), or the length of their longest relationships, p=.16, (see Figure 1). Based on the manner in which we coded the length of longest relationship variable, these data indicate that the mean length of longest relationship for YWLCS students was between six months and one and a half years. For control students, the mean length of longest relationship was between five Adolescent Girls’ Implicit Math Attitudes 20 months and one year. There were, however, significant group differences in whether participants were currently in a dating relationship and the number of dating relationships participants have had (see Figure 2). YWLCS participants were more likely to report currently being in a relationship than control participants (p=.01) and to report having had more relationships than control participants, p=.0098. Based on the coding scale used for the number of relationships variable, YWLCS participants had a mean of between three and four dating relationships, while control participants had a mean of between two and three. These group differences should be interpreted cautiously, however, due to the significant attrition that occurred between Time 1 and Time 2, which resulted in a low number of participants used in this study. Of the 208 participants at Time 1, only 93 also participated in Time 2. Due to the low number of participants who completed both Time 1 and Time 2, we were unable to further break down participants who reported having had a dating relationship into groups of those who had began dating before age 14 and those who began at or after. Thus, we decided to investigate the effect of having had at least one dating relationship in order to maximize our potential statistical power. Change in Attitudes Implicit Attitude Change Our primary research interest concerned whether attendance of YWLCS would increase positive implicit math attitudes and whether dating activity would decrease them. We hypothesized a main positive effect of attending YWLCS on implicit attitude change and a main negative effect of dating on attitude change. We also predicted an interaction of school and dating such that attending YWLCS would dampen the negative effects of dating. Change in implicit attitude was indexed by the difference between the attitudes at Time 1 and 2 (IAT D score at Time 2 minus D score at Time 1). For a graphical representation of group change, see Adolescent Girls’ Implicit Math Attitudes 21 Figure 4. Although none of the groups evidenced significant implicit attitude change from Time 1 to Time 2, control students who reported not having had a dating relationship showed the greatest change in implicit attitudes, becoming more negative over time (M= -.39). Thus, although not significant, this change was nearly a full standard deviation different from the Time 1 attitude (M=-.04, SD= .41). Control students who have dated (M= -.02, SD= .36) and YWLCS students who had not dated (M=-.05, SD=.58) became more negative in implicit attitudes over time, but given that the change was not significant, this small difference over time is negligible. YWLCS students who reported dating showed no change from Time 1 to Time 2 (M=.00, SD=.50). We tested the effects of school and dating on attitude change by estimating a series of multiple regression models. With age and the age X group interaction term included as covariates, we found the estimated average change score for YWLCS students was about zero (b=.00) and .14 for control students, a trend partly in line with our prediction, though non-significant (p=.29). That is, we expected negative change for the control participants’ implicit math attitudes over time, but we did not expect to find no change for YWLCS participants. Next, we tested the effect of dating and its possible interaction with school by adding the dichotomous dating variable and the dating X school interaction term to the model. Neither the main effect of dating (p=.13) nor the interaction of dating with school was significant (p=.31). We report the following estimates anyway, because they are sizeable in some cases and would be meaningful if they held for a larger sample. For a control participant at the mean age who has not had a dating or romantic relationship, the predicted IAT attitude change was -.40, compared to -.01 for a control participant who had dated. For an YWLCS participant at the mean age who has not had a dating relationship, the predicted IAT attitude change was -.07, compared to .01 for an YWLCS Adolescent Girls’ Implicit Math Attitudes 22 participant who has dated. Thus, a hypothetical control participant who has dated evidenced a trend contrary to our hypotheses; estimated implicit math attitude change was less negative than for a control participant who has dated. Likewise, for a hypothetical YWLCS participant who has not dated, predicted implicit math attitude change was more negative than for an YWLCS participant who has dated. Figure 5 shows estimated mean change in implicit math attitude by group. Explicit Attitude Change We conducted a similar progression of multiple regression analyses to examine explicit attitude change. For explicit math attitude change, controlling for age, there was no significant effect of school. The estimated average change score for YWLCS participants was .3, compared to .2 for control participants, p=.60. Next, we tested the effect of dating and its possible interaction with school, and, as with the implicit attitudes, there was neither a significant main effect for dating (p= .47) nor a significant interaction of dating with school (p=.72). For both control, nondating and for control, dating students, mean explicit math attitude change was .1 (SD=1.1). For YWLCS, nondating students, mean change was .2 (SD=.9), while for YWLCS, dating students, mean change was .3 (SD= .9). Accounting for age and whether or not participants had ever been in a dating relationship, the predicted explicit math attitude change was b= -.9 for a control participant who has not had a dating relationship, compared to -.6 for a control girl who has. For an YWLCS participant who has not had a dating relationship, predicted explicit math attitude change was -.7, compared to -.5 for an YWLCS participant who has. Thus, for all groups, predicted change in explicit math attitudes is in the direction of increasingly negative attitudes from Time 1 to Time 2; however this is less so for YWLCS and control participants who report having dated. Adolescent Girls’ Implicit Math Attitudes 23 For explicit language arts attitude change, analyses indicate no significant effect of school for YWLCS and control students, controlling for age. The estimated average change score for YWLCS participants was zero, and .4 for control participants, p= .24. Tests for the effect of dating revealed no significant differences in language arts attitude change (p=.29). Analyses did reveal, however, a significant interaction of dating and attending YWLCS, such that those students who attend YWLCS and have dated have greater decrease in explicit language arts positivity from Time 1 to Time 2 (p= .02). For control students who had not dated, mean change was -.4 (SD= .8), compared to .1 (SD= 1.1) for control girls who had dated. For YWLCS students who had not dated, mean change was 1.1 (SD= 1.7), compared to .1 (SD= 1.0). Accounting for age and whether participants had ever been in a dating relationship, the predicted explicit language arts attitude change was -.4 for a control participant who has not had a dating relationship, compared to .2 for one who has. For an YWLCS participant who has not had a dating relationship, predicted explicit language arts attitude change was 1.2, compared to .1 for one who has. Thus, predicted explicit language arts attitude increases more for a control participant who has dated than for an YWLCS participant who has dated. Additionally, predicted explicit language arts attitude increases less for an YWLCS participant who has dated compared to one who has not, while having dated for a control girl is associated with increased explicit language arts attitude compared to a decreased explicit language arts attitude for one who has not. Discussion Because of the YWLCS’s strong emphasis on math and science education for adolescent girls, we hypothesized that attending this school would result in an increase in positivity in implicit and explicit math attitudes over time. Furthermore, we predicted that girls who do not attend YWLCS would evidence a decrease in positivity in implicit and explicit attitudes. We Adolescent Girls’ Implicit Math Attitudes 24 were specifically interested in the potential effect of dating activity on attitude change. We hypothesized that YWLCS participants who reported having had a dating relationship would show less improvement in math attitudes, while control participants who reported having had a dating relationship would show even more worsening math attitudes. None of our hypotheses were supported by the study results. We found that school type (YWLCS vs. control) did not predict attitude change, either implicitly or explicitly. Furthermore, dating activity was not related to attitude change for either group. Results also failed to support the predicted interaction of school and dating,i.e., that attending YWLCS would dampen the negative effects of dating. Overall, there was no significant implicit or explicit attitude change as a function of school or dating, except for explicit language arts attitude change. YWLCS students who have dated were more likely to show decrease in explicit language arts positivity. This is partly in line with our hypothesis that dating would negatively affect academic attitudes; however, the other results are not consistent with our hypotheses. Thus, the study found no evidence that attendance of YWLCS or that dating activity affected the implicit and explicit math attitudes of the participants. These results are seemingly inconsistent with FPI’s preliminary findings that girls attending YWLCS evidence more implicit and explicit preference for mathematics. Our results also do not correspond with prior research indicating that a same-sex learning environment can reduce implicit gender-stereotypic attitudes (Dasgupta & Asgari, 2004) and that dating can have a detrimental effect on girls’ academic achievement (Brendgen et al., 2002; Neeman et al., 1995; Scott et al., 2005). A failure to find change in implicit math attitudes over time could be related to the finding that neither the YWLCS nor the control group evidenced a differential preference for math or language arts; in other words, the participants evidenced about equal implicit Adolescent Girls’ Implicit Math Attitudes 25 attitudes towards math and language arts and did not significantly prefer one domain over the other. One possibility for why participants did not indicate the stereotypically negative attitude towards math is that all participants in the study liked math at least enough to encourage them to apply to YWLCS, where math is emphasized. Since all potential participants were randomly assigned to YWLCS and control groups by the school’s random admissions lottery, it is logical to expect that groups would not differ in attitude at the beginning of the school year. Additionally, we could still expect relatively neutral preferences for math versus language arts in both groups, because it is reasonable to assume that the kind of student who would apply to attend a math/science charter school is a more motivated student generally. Therefore, the student could likely have no preference for one academic domain over another. This might be especially true for participants in the control group, who self-select to participate far more than YWLCS participants, who have teacher encouragement to participate (about half of all enrolled YWLCS students participate, but only about 10% of potential control students). Another factor contributing to our finding that attitudes did not appear to differ at Time 1, Time 2, or over time is the possibility that YWLCS math courses may be more challenging than those in the control schools; thus, the YWLCS students may not show an increase in positive math attitude because they perceive their courses to be difficult. It is initially intuitive to propose that the reason we did not find any group or individual change in implicit attitudes from Time 1 to Time 2 is that we were measuring attitude change over a relatively short time. After all, since implicit attitudes are related to stereotypes that are pervasive in an individual’s life from an early age, it would indeed be very impressive that a lifetime of implicit bias could be reversed over a period of a few months. However, Dasgupta and Asgari (2004) observed implicit attitude change over a year. Adolescent Girls’ Implicit Math Attitudes 26 A more plausible explanation for why there was no difference between YWLCS and control at Time 1 and Time 2, and between the two times, is the self-selection of control participants. Because significantly fewer control students than YWLCS students participate, it is logical to suspect that the control students who self-select to participate are more motivated or have more supportive-than-average parents. Unlike YWLCS students, control students do not have encouragement from their teachers to participate. Thus, control students make the decision to continue completing sessions through FPI far more independently than do YWLCS students. It is likely that these control students have an intrinsic interest in the research FPI conducts and enjoy thinking and answering questions about their academic preferences and attitudes. Thus, our groups may have differed in a fundamental way that could have affected our results. Our data on dating behaviors support the hypothesis that the two groups are actually different. Controlling for age, YWLCS students were significantly more likely to report currently being in a relationship than control students; additionally, YWLCS students report having had significantly more dating relationships than control students. Thus, YWLCS students seem to be more active in their dating behaviors than control students. This may be indicative of other important differences that could potentially explain why control students in this sample do not exhibit the predicted decrease in positive math attitudes over time in high school. Given the pervasiveness and consistency of the research evidencing that early dating (age 14 or younger) is associated with lower academic achievement, lower self-esteem, and higher rates of delinquency, among other negative outcomes (Brendgen et al., 2002; Neeman et al., 1995; Scott et al., 2005), it is surprising that this study found no evidence for this. Furthermore, in most other studies (Brendgen et al., 2002), approximately 30 percent of adolescent girls begin dating at age 14 or younger. Because 70 percent of the participants in our sample who have had a Adolescent Girls’ Implicit Math Attitudes 27 dating relationship began dating at age 14 or younger, it would be logical to think that negative academic outcomes, and negative academic attitudes, could be particularly pronounced in our sample. The present study did not find any evidence for this, however. Because “early” dating is the norm in this particular sample, perhaps the normalcy of this dating behavior in this cohort mitigates the negative effects typically associated with early dating (Dornbusch, Carlsmith, Gross, Martin, Jennings, Rosenberg, & Duke, 1981). Thus, beginning to date at age 14 may not actually be early relative to their peers in this particular sample. Furthermore, all participants in this study live in the Chicago, Illinois area, and many come from, or at least are exposed to, lower socioeconomic status backgrounds, which is a demographic associated with higher rates of teen pregnancy (Miller & Moore, 1990). Presuming that sexually active teens are more likely to be active daters, early dating may be more normative among the participants of the present study. Most research indicates negative outcomes for girls who date early relative to their peers, but most of the girls in this study were already dating by age 14. This may be why dating was not associated with negative academic attitudes in this study. A significant limitation that likely accounts for the dearth of statistically significant differences by group in attitude change is the overall lack of statistical power resulting from a low number of participants, especially in the control group. Given that our groups had a very low number of participants who had completed both sessions, particularly within the nondating groups, low statistical power is an important drawback to this study. Thus, any trends reported in this study must be interpreted cautiously. Additionally, a methodological limitation to this study is that, although we controlled for age, we did not control for year in school. Because participants had been in either YWLCS or a control school for varying amounts of time, it is possible that our first measurement in September was not truly a baseline measurement. Adolescent Girls’ Implicit Math Attitudes 28 However, the impacts of this on our results are likely fairly small, considering that controlling for age should approximate controlling for year in school. Although this study did not find significant results, it is important that future research continue to investigate possible obstructions and aids to young women’s math attitudes. A future project should investigate the effect of dating on implicit and explicit math attitudes among young women in a representative sample in order to gain a better understanding of the affects of dating in the general population. It would also be interesting to investigate adolescent boys’ implicit academic attitudes in order to draw comparisons to those of girls. This will allow researchers to determine if implicit attitudes are gendered, and, if they are, how to improve the attitudes of the disadvantaged group. Additionally, future work should address the question of how various aspects of dating affect academic attitudes. For instance, does the length of the romantic relationship have a mediating effect on negative outcomes in academic attitudes, as it does in other domains (Furman, Ho, & Low, 2007)? Is there a difference in attitudes between girls who have had several relationships as opposed to a few? Relationship quality and seriousness should also be taken into account. These investigations could also address affects on other academic subjects in addition to math. It would be particularly interesting to see if dating affects math and science attitudes similarly. Continuing research in this field is vital to addressing the disparity between the numbers of men and of women in math and science career fields. Understanding the environmental and societal influences on achievement and attitudes in this domain will have implications for current teaching practices and educational environments, as well as for current gender norms and stereotypes. Despite the significant progress made over the past century for women’s equality, women remain disadvantaged in their careers in terms of salaries and prestige. In order to change Adolescent Girls’ Implicit Math Attitudes 29 this, it is imperative that women be given every equal opportunity to pursue their interests, regardless of what society’s stereotypes deem as appropriate feminine behavior. By investigating how to improve girls’ math attitudes we can come closer to eradicating the math=male stereotype, finally giving young women the equal playing field they deserve in math and science. And perhaps, someday, Mattel will release a Mathematician Barbie. Adolescent Girls’ Implicit Math Attitudes 30 ReferencesBleeker, M. M. & Jacobs, J. E. (2004). Achievement in math and science: Do mothers’ beliefsmatter 12 years later? Journal of Educational Psychology, 96(1), 97-109.Brendgen, M., Vitaro, F., Doyle, A. B., Markiewicz, D., & Bukowski, W. M. (2002). Same-sexpeer relations and romantic relationships during early adolescence: Interactive links toemotional, behavioral, and academic adjustment. Merrill-Palmer Quarterly, 48(1), 77 103.Caspi, A., Lynam, D., Moffitt, T. E., & Silva, P. (1993). Unraveling girls’ delinquency: Biological, dispositional, and contextual contributions to adolescent misbehavior.DevelopmentalPsychology, 29(1), 19-30.Cohen, G. L., Garcia, J., Apfel, N., Master, A. (2007). Reducing the Racial Achievement Gap: ASocial-Psychological Intervention. Science, 313, 1307-1310.Dar-Nimrod, I. & Heine, S. J. (2006). Exposure to scientific theories affects women’s math performance. Science, 314, 435.Dasgupta, N. & Asgari, S. (2004). Seeing is believing: Exposure to counterstereotypic women leaders and its effect on the malleability of automatic gender stereotyping. Journal ofExperimental Social Psychology, 40, 642-658.Dornbusch, S., Carlsmith, J., Gross, R., Martin, J., Jennings, D., Rosenberg, A., & Duke, P.(1981). Sexual development, age, and dating: A comparison of biological and socialinfluences upon one set of behaviors. Child Development, 52, 179-185.Evans, M. A., Whigham, M., & Wang, M. C. (1995). The effect of a role model project upon theattitudes of ninth-grade science students. Journal of Research in Science Teaching, 32(2),195-204. Adolescent Girls’ Implicit Math Attitudes 31 Furman, W., Ho, M. J., & Low, S. M. (2007). The rocky road of adolescent romantic experience:Dating and adjustment. In Engles, R. C. M. E., Kerr, M., & Stattin, H. (Eds). Friends,Lovers, and Groups: Key Relationships in Adolescence (pp. 61-80). West Sussex,England: John Wiley & Sons Ltd.Garrahy, .D.A. (2001). Three third grade teachers' gender-related beliefs and behavior. TheElementary School Journal, 102(1), 81-94. Greenwald, A. G., & Banaji, M. R. (1995). Implicit social cognition: Attitudes, self-esteem, andstereotypes. Psychological Review, 102(1), 3-25. Greenwald, A. G., McGhee, D. E., & Shwartz, J. L. K. (1998). Measuring individual differencesin implicit cognition: The Implicit Association Test. Journal of Personality and SocialPsychology, 16, 270-280.Greenwald, A. G., Nosek, B. A., & Banaji, M. R. (2003). Understanding and using the ImplicitAssociation Test I: An improved scoring algorithm. Journal of Personality and Social Psychology, 85(2), 197-216.Greenwald, A. G., Poehlman, T. A., Uhlmann, E., & Banaji, M. R. (in press). Understanding and using the Implicit Association Test: III. Meta-analysis of predictive validity. Journal ofPersonality and Social Psychology.Haag, P. (1998). Single-sex education in grades K-12: What does the research tell us? In S.Morse (Ed.), Separated by sex: A critical look at single-sex education for girls (pp. 1338). Washington, DC: American Association of University Women EducationalFoundation. Adolescent Girls’ Implicit Math Attitudes 32 Helwig, R., Anderson, L., & Tindal, G. (2001). Influence of elementary student gender onteachers’ perception of mathematics achievement. The Journal of Educational Research,95(2), 93-102.Hyde, J. S., Fennema, E., & Lamon, S. J. (1990). Gender differences in mathematicsperformance: A meta-analysis. Psychological Bulletin, 107(2), 139-155.Hyde, J. S., Lindberg, S. M., Linn, M. C., Ellis, A. B., & Williams, C. C. (2008). Gender similarities characterize math performance. Science, 321, 494-495.Inzlicht, M. & Ben-Zeev, T. (2000). A threatening intellectual environment: Why females are susceptible to experiencing problem-solving deficits in the presence of males.Psychological Science, 11(5), 365-371.Jacobs J. E. & Eccles, J. S. (1985). Gender differences in math ability: The impact of mediareports on parents. Educational Researcher 14, 20–25.Jacobs, J. E., Lanza, S., Osgood, D. W., Eccles, J. S., & Wigfield, A. (2002). Changes in children’s self-competence and values: Gender and domain differences across grades onethrough twelve. Child Development, 73(2), 509-527. Jussim, L. & Eccles, J. S. (1992). Teacher expectations II: Construction and reflection of studentachievement. Journal of Personality and Social Psychology, 63(6), 947-961.Keller, J. (2002). Blatant stereotype threat and women’s math performance: Self-handicapping asa strategic means to cope with obtrusive negative performance expectations. Sex Roles,47(3/4), 193-198.Kiefer, A. K. & Sekaquaptewa, D. (2007). Implicit stereotypes, gender identification, and mathrelated outcomes: A prospective study of female college students. Psychological Science,18(1), 13-18. Adolescent Girls’ Implicit Math Attitudes 33 Lam, T. H., Shi, H. J., Ho, L. M., Stewart, S. M., & Fan, S. (2002). Timing in pubertalmaturation and heterosexual behavior among Hong Kong Chinese adolescents. Archivesof Sexual Behavior, 31, 359–366Lane, K. A., Banaji, M. R., Nosek, B. A., & Greenwald, A. G. (2007). Understanding and usingthe Implicit Association Test: IV. What we know (so far) about the method. In B.Wittenbrink & N. Schwarz (Eds.) Implicit Measures of Attitudes (pp. 59-102). The Guilford Press, New York.Lee V. E., Marks H. M., & Byrd, T. (1994). Sexism in single-sex and coeducational independent secondary school classrooms. Sociology of Education, 67(2), 92-120.Marx, D. M. & Roman, J. S. (2002). Female role models: Protecting women’s math testperformance. Personality and Social Psychology Bulletin, 28, 1183-1193.Miller, B. & Moore, K. (1990). Adolescent sexual behavior, pregnancy, and parenting: Researchthrough the 1980s. Journal of Marriage and the Family, 52, 1025-1044. Nauta, M. M., Epperson, D. L., & Kahn, J. H. (1998). A multiple-groups analysis of predictors ofhigher level career aspirations among women in mathematics, science, and engineering majors. Journal of Counseling Psychology, 45(4), 483-496.Neeman, J., Hubbard, J., & Masten, A. S. (1995). The changing importance of romanticrelationship involvement to competence from late childhood to late adolescence.Development and Psychopathology, 7(4), 727-750.Nisbett, R. E. & Wilson, T. D. (1977). Telling more than we can know: Verbal reports on mentalprocesses. Psychological Review, 84(3), 231-259.Nosek, B. A., & Banaji, M. R. (2001). The go/no-go association task. Social Cognition. 19(6),625-666. Adolescent Girls’ Implicit Math Attitudes 34 Nosek, B. A., Banaji, M. R., Greenwald, A. G. (2002) Math= male, me= female, therefore math≠ me. Journal of Personality and Social Psychology, 83, 44-59.Nosek, B. A., Greenwald, A. G., & Banaji, M. R. (2006). The Implicit Association Test at age7: A methodological and conceptual review. In J. A. Bargh (Ed.), Automatic Processesin Social Thinking and Behavior (pp. 265-293). New York: Psychology Press.Parsons, J. E., Adler, T. F., & Kaczala, C. M. (1982). Socialization of achievement attitudes and beliefs: Parental influences. Child Development, 53(2), 310-321.Phinney, V. G., Jensen, L. C., Olsen, J. A. & Cundick, B. (1990). The relationship between early development and psychosexual behaviors in adolescent females. Adolescence, 25(98),321-332.Quatman, T., Sampson, K., Robinson, C., & Watson, C. M. (2001). Academic, motivational, andemotional correlates of adolescent dating. Genetic, Social, and General PsychologyMonographs, 127(2), 211-234. Rosenthal, R., & Jacobson, L. (1968). Pygmalion in the classroom: teacher expectation andpupils' intellectual development. New York, Holt, Rinehart and Winston. Rosenthal, R. & Jacobson, L. (1992). Pygmalion in the classroom. Expanded edition. NewYork: IrvingtonReid, P. T. & Roberts, S. K. (2006). Gaining options: A mathematics program for potentially atrisk adolescent girls. Merrill-Palmer Quarterly, 52(2), 288-304.Riordan, C. (1998). The future of single-sex schools. In S. Morse (Ed.), Separated by sex: Acritical look at single-sex education for girls (pp. 53-62). Washington, DC: AmericanAssociation of University Women Educational Foundation. Adolescent Girls’ Implicit Math Attitudes 35 Scott, K., Stewart, L. L., & Wolfe, D. (2005). Dating relationships among at-risk adolescents: Anopportunity for resilience?. In R. DeV. Peters, B. Leadbeater, and R. J. McMahon (Eds).Resilience in Children, Families and Communities: Linking Context to Practice andPolicy (pp. 83-100). New York: Kluwar Academic/Plenum Publishers.Shih, M., Pittinsky, T., & Ambady, N. (1999). Stereotype susceptibility: Identity salience andshifts in quantitative performance. Psychological Science, 10(1), 80-83. Spencer, S. J., Steele, C. M., & Quinn, D. M. (1999). Stereotype threat and women’s mathperformance. Journal of Experimental Social Psychology, 35, 4-28. Steele, C. M. (1997). A threat in the air: How stereotypes shape intellectual identity andperformance. American Psychologist, 52, 613-629.Steele, C. M. & Aronson, J. (1995). Stereotype threat and the intellectual test performance ofAfrican Americans. Journal of Personality and Social Psychology, 69(5), 797-811.Steele, J. (2003). Children’s gender stereotypes about math: The role of stereotype stratification. Journal of Applied Social Psychology, 33(12), 2587-2606.Wilgosh, L. (2003). Examining gender images, expectations, and competence as perceived impediments to personal, academic and career development. International Journal for theAdvancement of Counseling, 24, 239-260. Adolescent Girls’ Implicit Math Attitudes 36 Appendix AMath Attitude IAT Exemplars Adolescent Girls’ Implicit Math Attitudes 37 Appendix BExplicit Questionnaire Items