Introduction

Adolescence is a key stage in the development of sexist attitudes and expectations of appropriate behaviors for each gender. This has been shown for adolescents from Spain (de Lemus et al. 2010) and the United States (for reviews, see Glick and Hilt 2000; Maccoby 1998). For women, it can be particularly harmful to endorse benevolent sexist beliefs, according to which they should obtain men’s affection and protection in exchange for maintaining traditional gender roles (Glick and Fiske 1996); this is because the positive affective tone of such beliefs often conceals their discriminatory nature, as was shown by Barreto and Ellemers (2005; and Ellemers & Barreto, 2009), for Dutch college students. The literature suggests that mothers’ attitudes are a central factor influencing how their daughters learn traditional versus more liberal gender roles, as shown for children, teenagers, and parents from the U.S. (Eccles et al. 1990) as well as for mother–daughter pairs from the Netherlands (Ex and Janssens 1998), Israel (Kulik 2004), and the U.S. (Moen et al. 1997; Smith and Self 1980).

The present correlational study focuses on the role played by mothers’ gender ideology in their adolescent daughters’ adoption of sexist beliefs and future goals. To the extent that gender-related attitudes of female adolescents reinforce certain traditional roles, they may have a significant influence on the development of adolescents' future goals and current academic performance. Although the current research was conducted in Spain, its results and conclusions should be of interest for readers from other nations and cultures. Indeed, a meta-analysis of 43 studies conducted in North America, Israel, Europe, and Asia provides cross-cultural evidence that parents’ gender schemas are related to their offspring’s gender attitudes (Tenenbaum and Leaper 2002). Furthermore, previous research demonstrates that ambivalent sexism generalizes across cultures (Glick et al. 2000—data from college students and non-students in 19 countries; Glick et al. 2004—data from college students and non-students in 16 countries), and is related to the Human Development Index as well as national indices of socialization, Human Rights, and family structure (see Moya et al. 2002 for a study with college students from 29 countries).

According to Ambivalent Sexism Theory (AST; Glick and Fiske 1996), apart from traditional sexism, which is clearly hostile (hostile sexism—HS), there also exists a more subtle kind of sexism, called benevolent sexism (BS). BS idealizes and rewards women who correspond to the female stereotype, perpetuating traditionally established gender roles (Glick and Fiske 1996). The idealization of traditional feminine characteristics, together with the positive impression that BS may elicit in the perceiver, can make it difficult to identify BS as sexism, as studies with college students from U.S. clearly demonstrate (Barreto and Ellemers 2005; Ellemers and Barreto 2009; Swim et al. 2005). It may also lead women to accept BS as an attractive, prosocial attitude, which was shown for college students from Germany (Bohner et al. 2010), the U.S. (Kilianski and Rudman 1998), and Spain (Montañés et al. 2011). Both types of sexism function as legitimizing ideologies that complement gender inequality in many countries and cultures, including Spain (see Glick et al. 2000, 2004).

Various theorists argue that subordinate groups often contribute to their own subordination by accepting the ideology of the dominant group as valid (Jackman 1994; Jost and Banaji 1994; Sidanius and Pratto 1999). For example, BS has been shown to increase women’s acceptance of system-justifying beliefs in both college student and non-student samples from the U.S. (Jost and Kay 2005); also, female college students from Spain showed more acceptance of their partners’ sexist discrimination when it was justified in a benevolent way (Moya et al. 2007). Insofar as BS justifies the traditional division of labor conveying the idea of women’s dependence on men to protect and support them (Glick and Fiske 1996), this ideology helps to perpetuate differential gender roles for men and women in today’s society.This justification of roles contributes to people’s understanding of the established status quo between women and men as legitimate (Jost and Hamilton 2005).

The current research was conducted in a Spanish context. Profound social changes have occurred in Spain during the last decades (e.g., Dema-Moreno 2009). However, traditional roles continue to be differentiated for men and women, as shown in a study with married couples (Sánchez and Hall 1999). For instance, women have increased their participation in paid work, but the division of household tasks continues to be highly gender-specific, as Álvarez and Miles (2006) observed in a sample of employed men and women from Spain. This can be explained, at least partly, by the traditional gender division of roles, as indicated by studies with college students and non-students (Garcia-Retamero and López-Zafra 2006), by comparisons between college students from Spain and Germany (Garcia-Retamero and López-Zafra 2009), and by Spanish Ministry of Work and Social Affairs data from 2001–2004 (Goñi- Legaz et al. 2010). Gender segregation at the workplace is still prevalent in Spain: according to national census data from 2001, 59.1% of women have female stereotypical jobs and earn less than men (Ibañez-Pascual 2008); more recent surveys with Spanish employees show that women spend more time working on domestic issues and taking care of children and elderly than men do (Eurostat 2006; Sánchez-Herrero et al. 2009). This gendered division of labor and housework is also mantained in Spanish adolescents (Silván-Ferrero and Bustillos López 2007), indicating that traditional roles continue to affect younger generations.

The universal nature and stability of traditional gender roles has led to the study of their social transmission in social psychology over the last few decades. Literature on the transmission of gender role attitudes suggests that parents are important socializing agents (Bussey and Bandura 1999; Crouter et al. 1995; Katz and Ksansnak 1994; Leaper 2002; Leaper and Friedman 2006; McHale et al. 1999—all of them in families from U.S; McHale et al. 2003; Serbin et al. 1993—in children from Canada). A meta-analysis of 43 studies conducted in North America, Israel, Europe, and Asia led to the conclusion that parents and their children share gender role beliefs (Tenenbaum and Leaper 2002). Researchers have highlighted the importance of mothers as transmitters of these traditional gender roles to their daughters, providing strong evidence for positive correlations between the gender role attitudes of mothers and daughters (e.g., Eccles et al. 1990; Ex and Janssens 1998; Kulik 2004; Moen et al. 1997; Smith and Self 1980). Other studies have found a relationship between the educational level and occupation status of mothers and less traditional attitudes toward gender roles in their offspring (Affleck et al. 1989—in U.S. college students; Booth and Amato 1994—in married people from the U.S. and their offspring; Corder and Stephan 1984—in U.S. adolescents; Greenberg and Goldberg 1989—in employed fathers and mothers of preschool children from the U.S; Hoffman 1989—in employed mothers from the U.S.; Kulik 2002—in mother-daugther pairs from Israel).

There is evidence of negative correlations between the educational level of women and their acceptance of hostile and benevolent sexist attitudes in Spain (Glick et al. 2002; Moya et al. 2002). In addition, both the educational level of women and their presence in the labor market are related to less traditional attitudes toward gender roles in heterosexual couples from Spain (Kulik 2004; Moya et al. 2000). Such studies suggest that the effect of mothers’ job status and educational level on their daughters’ ideology may be mediated by the mothers’ own ideology. In accordance with this argument, Ex and Janssen (1998) found that Dutch mothers’ educational level predicted their gender role attitudes, which in turn were positively related with their daughters’ gender role attitudes. These authors did not find a direct relationship between the job status and gender role attitudes of daughters; instead, the relationship between mothers’ job status and daughters’ gender role attitudes was mediated by parenting style. In the present research we go one step further by analyzing the relation between job status and education on mothers’ BS, and the influence of mothers’ BS on the benevolent sexist beliefs of daughters (i.e., intergenerational transmission of BS from mothers to daughters), and on daughters' future goals. The apparently positive tone of BS provides a subtle way in which traditional gender roles and ideologies are internalized and possibly transmited within the family. For that reason, we examine the influence of daughter’s BS on their traditional goals and academic goals (getting a degree), as well as on their current academic performance; the study of these relationships promises to add particularly relevant and novel insights to the literature, because, as far as we know, no prior studies have explored them.

Studying the BS of female adolescents is of vital importance to understand the maintenance and justification of traditional gender roles in the present and in future generations. According to Glick and Fiske (2001), BS weakens women’s resistance to male dominance, promising them rewards from men’s structural power. These rewards, which include taking care of women, protecting them, and providing them with resources, may maintain the traditional division of labor, alienating women from academic goals and encouraging them to engage in traditional tasks. Research has confirmed that women who internalize these beliefs maintain traditional gender roles. For example, Spanish female adolescents with benevolent sexist beliefs maintain gender role differences regarding the division of labor at home (Silván-Ferrero and Bustillos López 2007). Similarly, data from U.S. college students indicate that benevolent sexist women tend to rate more highly their potential male partners’ ability to provide for their partners and offspring (Johannesen-Schmidt and Eagly 2002); Spanish college students also tend to accept restrictions on tasks that are not consistent with their traditional role if these restrictions are justified as protective behaviors of their partners toward women (Moya et al. 2007).

With respect to a person’s academic development, adolescence is of vital importance as a transition stage between childhood and adulthood. Indeed, academic choices and performance during this stage determine the academic possibilities and job opportunities in adulthood. For instance, in Spain women form a clear majority (74.2%) of students in the Health Sciences, but a clear minority (27.3%) in Engineering (Instituto de la Mujer 2009, data from 2005-2006). Academic choices are determined in adolescence when Spanish students must select among different high-school modalities (Science and Technology vs. Social Sciences and Humanities) (Instituto Nacional de Estadística 2009). That such choices may be related to ideological factors was shown by Rudman and Heppen (2003), who found an association between the idea of chivalry (present in BS beliefs) and less interest in career development in U.S. college women. In that research, young women who associated their male partner with the idea of “chivalry” (e.g., “my partner should protect me”) showed less interest in high status jobs and the financial independence they imply.

Women’s dependence on men, subtly transmitted by BS through “courteous and chivalrous” norms and behavior, can be particularly dangerous during adolescence, given that this idea is consistent with the romantic scripts of female childhood (Rudman and Heppen 2003). These scripts imply that men take care of women and support them, and can ultimately convey that academic goals are not a priority issue for them. This can lead female adolescents to become less academically involved, or to be less interested in studies associated with traditionally male domains (e.g., Science, Technology, Engineering, and Mathematics). Lack of motivation of female adolescents regarding academic performance contributes to perpetuating gender inequalities in adulthood. Lower academic grades can reduce adolescents' chances of having a successful career and securing the financial resources this implies. The hypothesis of the present study is that female adolescents who internalize benevolent sexist beliefs may limit their academic and educational goals (e.g., get a degree) to focus on more traditional goals (get married and have children, look smart and pretty all the time). Goals, as future objectives, guide the motivations and daily behaviors of female adolescents. Their academic goals can be related to better current academic performance, whereas traditional goals seem to be related to lower academic grades. Thus, BS seem to contribute to justifying the current inequality and perpetuate gender inequality in new generations of women in subtle and concealed ways.

In short, this study explored the intergenerational transmission of BS from mothers to daughters, considering the influence of educational level and job status on mothers’ BS. Such variables have been related empirically to women’s attitudes toward gender roles. The study also analyzed the relationship between female adolescents’ BS and their academic and traditional goals, and how such goals influence their current academic performance. The specific predictions are represented in the path model in Fig. 1.

Fig. 1
figure 1

Path model proposed for the variables of interest

Full size image

Method

Participants

The survey was kept anonymous by a coding system that enabled only the match between adolescents and their mothers to be identified and linked in a data set. A data set was considered to be complete if both the adolescent and her mother returned the survey. Altogether, we collected 192 mother-daughter pairs recruited from 10 different Spanish Secondary Schools, some of them were located in rural areas and others in medium or large size cities. Therefore, the sample includes a large and varied range of families from different backgrounds and SES level. Pairs in which either the mother or the daughter did not participate were excluded from further analyses, leaving us with 164 complete mother-daughter data sets (N = 328 in total). Adolescents’ age ranged between 11 and 18 years (M = 13.37; SD = 1.09). Mothers' age ranged between 31 and 57 years (M = 43.07; SD = 4.08). Regarding their nationality, 94.8% were Spanish, 1.9% were English, 1.3% Bolivian, 1.3% were Moroccan and 0.6% were Uruguayan. Regarding their level of education, 5.6% of them had not completed studies, 32.3% had completed primary education, 43.5% Secondary education and 18.6% University studies. Regarding their professional status, 56.5% were currently employed (had a paid job). Participation of mothers and adolescents was voluntary.

Materials

The adolescents who participated in the study completed a questionnaire that included the following sections:

Future Goals Scale

Seven items were developed to measure the future goals of the adolescents. The scale had a response format ranging from 1 (not at all) to 7 (totally). Bartlett’s test of sphericity (χ 2(21) = 264.77, p < .001) and the KMO index of .66 both indicated that the covariance structure was suitable for factor analyis. A principal components analysis with promax rotation suggested a 2-factor solution that explained 58.97% of the variance.

The first factor pertained to traditional goals (four items: “have a boyfriend,” “being attractive to boys,” “get married and have a family,” and “look nice and pretty all the time”). The second factor referred to academic goals (three items: “go to University to get a degree,” “be acknowledged as a good professional,” and “get good grades”). The psychometric properties of the traditional goals subscale scale were adequate (Cronbach’s alpha = .72). The reliability of the career goals subscale was lower than expected (Cronbach’s alpha = .61); therefore for further analyses a single item was used (“go to University to get a degree”), because it best represented the intended construct of academic goals.

Ambivalent Sexism Inventory for Adolescents (de Lemus et al. 2008)

This scale has 20 items clustered into two subscales measuring HS and BS. Ten items address hostile sexist content (e.g., “boys should control who their girlfriends hang out with,” “girls are easily offended,” “girls are better at doing household tasks, whereas boys are better at fixing things”), and 10 items address benevolent sexist content (“for boys it is important to find a girl to date,” “girls should be loved and protected by boys,” “a boy can feel incomplete if he’s not dating a girl”). Internal consistencies obtained in the present study were good (HS: α = .80 ; BS: α = .79).

Adolescents’ Sociodemographic Variables

At the end of the questionnaire, participants were asked to provide information about their age, nationality, course level, and academic grades (“How many subjects did you pass in the last assessment?”).

The adolescents’ mothers who participated in the study completed a questionnaire that included the following sections:

Ambivalent Sexism Inventory (ASI; Glick and Fiske 1996)

The Spanish version, translated and adapted by Expósito et al. (1998), was used. This scale has 22 items clustered into two subscales measuring HS and BS in adults. The internal consistencies obtained in this study were very α = .89 for the HS subscale and α = .89 for the BS subscale.

Mothers’ Sociodemographic Variables

At the end of the questionnaire, participants were asked to provide information about their age, nationality, educational level (primary education, secondary education, vocational training, university studies), and job status (responses obtained were classified as 0 = no paid job or 1 = paid job outside the home).

Procedure

The study was conducted between October 2009 and May 2010. Adolescents participated in the study in classrooms at their schools; the mothers’ study was self-administered, and we assumed that it took place at home. Adolescents were seen on a predetermined date at their schools and allowed to participate with parental consent only. They were given sufficient time to familiarize themselves with the content of the survey, to read through it carefully, and, when they felt ready, to answer the questions. Completing the questionnaire took about 20 to 30 min. The instruments were presented in the same order as described above. After completion, all surveys were collected. Adolescents were then introduced to the study for their mothers. We had prepared sealed envelopes containing instructions for the study and a survey. Adolescents were asked to take the envelope home to their mothers and to bring the completed materials back to school within a predetermined period of time. As an incentive, a 15-Euro voucher from department stores was given to those who returned the mothers’ survey.

Results and Discussion

The scores obtained for BS and HS were similar to those of earlier studies performed with similar samples of both adolescents (de Lemus et al. 2008; 2010) and adult women (Expósito et al. 1998). Higher scores were obtained in BS than HS for both mothers and daugthers; these differences were marginally significant for mothers, t(147) = 1.95, p = .05, and significant for daughters, t(159) = 15.01, p < .001 (see Table 1). The apparently large differences in the BS and HS scores between mothers and daughters shown in Table 1 might be due to the different, age-specific, content of the items included in the ASI and ISA and are thus difficult to interpret. Nonetheless, they might as well reflect developmental changes. An analysis of a large Spanish female sample (N = 5067) aged from 13 to 57 who completed the ASI (Expósito et al. 1998) showed a significant decrease with age in the endorsement of both benevolent and hostile attitudes (Moya 2011).

Table 1 Descriptive statistics of main constructs
Full size table

Correlational Analyses

The relationship between the variables studied was first analyzed by means of a bivariate correlation analysis. The relationships found are shown in Table 2. A strong correlation between HS and BS was found in both samples (adolescents and adult women). This result is consistent with earlier studies (e.g., de Lemus et al. 2008; 2010; Expósito et al. 1998; Glick et al. 2000; Glick and Fiske 1996). Mothers’ HS correlated negatively with mothers’ educational level and occupation. No relationship was observed between mothers’ HS and daughters’ HS; however, mothers’ HS correlated positively with daughters’ BS and negatively with the number of subjects passed by adolescents. Furthermore, mothers’ BS was positively correlated with daughters’ BS and traditional goals, but negatively correlated with the number of subjects passed by daughters. Adolescents’ HS was positively related to their traditional goals, but was not related to their academic goal (getting a degree) or the number of subjects passed.

Table 2 Intercorrelations of main constructs
Full size table

Because HS and BS were strongly correlated in both samples (r = .69, p < .001, for mothers and r = .46, p < .001, for daughters), partial correlations were performed, controlling for the effects of each of these variables on one another. The results of the partial correlations controling for HS continued to show significant relationships between mothers’ BS and (a) daughters’ BS (r = .29, p < .001), (b) daughters’ traditional goals (r = .19, p < .05) and (c) subjects passed by daughters (r = -.32, p < .001). When BS was controled, however, the relationships between mothers’ HS and daughters’ BS (r = -.06, n.s.) as well as subjects passed were no longer significant (r = -.15, n.s.).

Path Model Analyses

In order to test the complete predictive model including Hypotheses 1 to 4 as depicted in Fig. 1, a path analysis was performed with AMOS 18.0 statistical software. The goodness of fit of the model proposed was assessed with a series of indices: χ 2, CFI (Comparative Fit Index), and NFI (Bentler-Bonett Normed Fit Index). If the model is an adequate representation of the data observed, the χ 2 test is not significant and therefore the associated p value should be greater than .05. Moreover, values lower than .95 in the NFI and CFI indices indicate that the model can be improved. The RMSEA (Root Mean Square Error of Approximation; Byrne 2001; Steiger 1990) value was also calculated, considering that any value equal to or lower than .06 indicates good fit (Hu and Bentler 1999). Because of the sensitivity of the χ 2 statistic to the sample size and the deviations with regard to assumptions of linearity, multivariate normality, and additivity (Joreskog and Sorbom 1989), the ratio χ 2/df was used, considering that values equal to or lower than 3.0 indicate good fit (Carmines and McIver 1981).

Results obtained for the initial model proposed in the introduction (Model A) showed that Hypothesis 1 was partially confirmed as mothers’ educational level significantly predicted their endorsement of BS, however job status did not. Mothers’ BS was a significant predictor of daughters’ BS, but not of daughters’ future academic or traditional goals, partially confirming Hypothesis 2. Daughters’ BS negatively predicted their academic goal and positively predicted their traditional goals (Hypothesis 3). Finally, academic performance was positively predicted by their academic goal and negatively predicted by traditional goals (Hypothesis 4). The overall fit of Model A could be improved, the chi square is significant, the RMSEA is above the .06 cut off, and the CFI and NFI values are below the .95, χ 2 = 20.04, df = 11, p = .04; χ 2 /df = 1.81; RMSEA = .07 (PCLOSE = .22), CFI = .94, NFI = .88.

Modification indices suggested the existence of a direct relationship between mothers’ BS and the number of subjects passed by adolescent daughters that is not mediated by the transmission of sexist beliefs. It is theoretically possible that less sexist mothers value their daughters’ academic performance more as a means to obtain greater independence in adulthood. As a result, these mothers may supervise their daughters’ academic performance more thoroughly and push them to obtain better academic results. On the other hand, mothers’ BS could be related to parenting style or other variables not assessed that may explain the direct influence of mothers’ BS on their daughters’ academic performance. For all these reasons, the suggestion of including the BS (mother) → Academic performance (daughter) relationship in the model was accepted. When this link is included in Model B the effect of traditional goals on academic performance becomes non-significant, although it is still showing the same negative tendency as proposed in hypothesis 4. Results are shown in Fig. 2. Model B presented approtiate fit indexes, χ 2 = 5.52, df = 9, p = .79; χ 2 /df = .61; RMSEA = .00 (PCLOSE = .93), CFI = 1.00, NFI = .97. A chi square different test, ∆χ 2 = 14.52, p < .001, show that Model B improved the goodness-of-fit indices of the Model A. Both models were also compared by using the Akaike information criterion (Akaike 1974) (AIC). Model B shows a smaller AIC than Model A, which implies a better fit of the former (see Table 3).

Fig. 2
figure 2

Model proposed for the variables of interest. χ 2 = 5.52, df = 9, p = .79; χ 2 /df = .61; RMSEA = .000 (PCLOSE = .93), CFI = 1.00, NFI = .97. (Model B)

Full size image
Table 3 Fit indices for path models
Full size table

Mediation Analysis of the Effect of Mothers’ Educational Level on Daughters’ BS Through Mothers’ BS

We examined the conditional indirect effect of mothers’ educational level on daughters’ BS (through mothers’ BS) using bias corrected and accelerated (BCa) bootstrapping methods with 2000 resamples (Efron 1987; Efron and Tibshirani 1993; MacKinnon et al. 2004; Preacher and Hayes 2004, 2008). The total effect of mothers educational level on daughters’ BS (B  = -.16, p < .01), the direct effect of mothers’ BS on daughters’ BS (B = .26, p < .001) and the indirect effects of mothers’ BS on daughters’ BS with a point estimate of -.11 (p < .001) and 95% BCa bootstrap CI of .20 to .05 were significant. The indirect effects of mothers’ BS exerted a complete mediation of the effects of mothers’ educational level on daughters’ BS, since the significant total effect was reduced to a non-significant direct effect (B = -.49, p < 1, ns) when controlling for the mediator.

Mediation Analysis of the Effect of Mothers’ BS on Daughters’ Academic Performance Through Daughters’ BS

The analysis of the conditional indirect effect of mothers’ BS on daughters’ academic performance (through daughters’ BS) show that the total effect (B = -.55, p < .001), the direct effect (B  = -.45, p < .001) and the indirect effects of mothers’ BS on academic performance (through daughters’ BS), is not zero by a 95% bias-corrected bootstrap confidence interval based on 2000 bootstrap samples (- .23 to—.01, with a point estimate of—.10). These results suggest the presence of partial mediation. Although the significant total effect was reduced, it remained significant when controlling for the mediator. Hence, direct effects of mothers’ BS and indirect effects (through daughters’ BS) predicts daughters’ academic performance.

Alternative Models

Finally, given that this was a correlational study and it was not possible to be certain about the direction of causality between the variables, two new models were tested inverting the direction of the predictive relationships for mothers’ and daughters’ BS (Model C) as well as the variables related to the adolescents’ goals and academic performance (Model D). In Model C it was hypothesized upward intergenerational transmission of attitudes instead of downward. That is, whether daughters traditional gender attitudes and future goals, influence their mothers' beliefs. The results of the analysis, χ 2 = 27.72, df = 10, p = .002; χ 2 /df = 2.77; RMSEA = .10 (PCLOSE = .03), CFI = .87, NFI = .84, a chi square different test, ∆χ 2 = 22.20, p < .001, showed that Model B improved the goodness-of-fit indices of Model C.

In Model D it was hypothesized that the adolescents’ grades could predict their traditional and academic goals, given that academic performance can lead to greater or lesser motivation for and interest in a future academic and traditional goals. The results of the analysis, χ 2 = 31.86, df = 11, p = .001; χ 2 /df = 2.89; RMSEA = .11 (PCLOSE = .02), CFI = .85, NFI = .81, showed that Model D had poorer fit than Model B. The significance of a chi square different test, ∆χ 2 = 26.34, p < .001, and the comparison of AIC (Akaike 1974) between the models, confirms this conclussion (see Table 3).

Conclusions and Outlook

The main objective of this study was to examine the intergenerational transmission of benevolent sexist beliefs from mothers to adolescent daughters; another goal was to explore the contribution of BS to maintaining the traditional division of social roles through its influence on future traditional and academic goals and the academic performance of adolescent females.

Over the last few decades, many studies have explored the transmission of gender role attitudes from mothers to daughters (e.g. Eccles et al. 1990; Ex and Janssens 1998; Kulik 2004; Moen et al. 1997; Smith and Self 1980); others have looked into the influence of different maternal variables such as presence in the labor market or education on children’s attitudes toward gender roles (Booth and Amato 1994; Corder and Stephan 1984; Hoffman 1989; Kulik 2002). Based on earlier evidence on the relationship between the educational level and job performance of women with less traditional gender role attitudes (Affleck et al. 1989; Kulik 2004), it was hypothesized that the relationship between mothers’ job status and education and daughters’ benevolent sexist ideology would be mediated by mothers’ own benevolent sexist ideology. Data obtained in the present study support this hypothesis and the previous results found by Ex and Janssens (1998). In the present research, mothers’ educational level negatively predicted daughters’ benevolent sexist ideology; this effect was mediated by mothers’ own BS. However, no relationship was found between mothers’ presence/absence in the labor market and their benevolent sexist beliefs. Therefore, the current study corroborates earlier studies in which one factor found to negatively predict women’s benevolent sexist attitudes was education (Moya et al. 2002). Our results further suggest that mothers’ BS influences the development of their daughters’ BS. Moreover, we found that mothers’ BS negatively predicts the academic performance of their daughters during adolescence. BS rewards women who endorse traditional gender roles and conveys the idea that men are supposed to support and protect women (Glick and Fiske 1996). Mothers that do not adopt benevolent sexist beliefs may push their daughters to have a good academic performance as a means for them to be independent in the future.

Another hypothesis of this study was that benevolent sexist beliefs may contribute negatively to the academic goals of female adolescents; this is because the idea of chivalry, consistent with the romantic scripts of female childhood (Rudman and Heppen 2003; Walkerdine 1984), may lead adolescent females to show less interest in their career, as a consequence of the underlying belief that men “are supposed to” take care of them and support them. BS may elicit a greater interest among adolescents in traditional feminine roles which allow them to be compensated with the benevolence of men. Our results confirm that adolescents who adopt benevolent sexist beliefs show lower academic goal and tend to focus more on traditional goals. These results confirm the prescriptive influence exerted by BS on gender roles during adolescence. They show that the academic and traditional goals of female adolescents predicted by their BS limit their future possibilities in practice by determining their academic grades in the present. Thus, female adolescents’ acceptance of BS contributes to the maintenance of traditional gender roles, in that they accept an ideology that justifies their own inferiority to men (Glick and Fiske 2001). Acceptance of these beliefs by female adolescents reduces their academic or career interest, which is related to their academic performance. Career development would give them the future possibility of obtaining financial resources, considered by previous literature as an important power base in gender relationships (Pratto and Walker 2004). Thus, adolescents’ internalization of BS may contribute to maintaining the power asymmetry between the sexes and may perpetuate the dominance of males in the future generation.

Mothers’ BS was not directly related to daughters’ academic or traditional goals, but through daughters’ own endorsement of benevolent sexist beliefs. This result might be regarded as positive with respect to possible interventions fostering gender equality, as it suggests that only those adolescents that internalize the sexist beliefs endorsed by their mothers will take up more traditional goals and less academic goals. Still, mothers’ BS directly influence daugthers’ academic performance, which might limit their access to future career opportunities. The underlying process that explains this direct influence of mothers’ BS on academic performance remains unexplained. This relationship may be mediated or explained by several variables not addressed here, such as mothers’ parenting style, or the assignment of more household task to daughters (vs. sons), which might affect their respective studying time. These hypotheses should be tested in further research.

We are aware that the present study has its limitations. First of all, future studies should explore the effect of the ideology of both parents on the sexist ideology of female adolescents. Secondly, to be able to draw more certain conclusions about the transmission of benevolent sexist ideology from mothers to daughters, it is necessary to use longitudinal methodology and test causal hypotheses over time. Thirdly, the sample we studied consisted only of Spanish adolescents and their mothers; therefore, it would be useful to replicate this research with an adolescent population in other countries. Another limitation of this study is that we used short scales to measure adolescents’ traditional goals, academic performance and mothers’ education level and job status and a single-item measure of adolescents’ academic goals Although this allowed us to keep our study short and perhaps more feasible in a school context, it would be useful in future studies to include more valid and differentiated measures for these important constructs. Finally, future studies should further analyze the direct relationship found between mothers’ BS and the academic performance of their adolescent daughters, as mentioned above. Despite these limitations, we conclude that the present study provides significant empirical evidence of the transmission patterns of sexist beliefs from mothers to daughters, and of the subtle way benevolent sexism operates. Indeed, this type of sexism perpetuates traditional gender roles through its influence on the goals and academic performance of female adolescents.

The findings of this study highlight the need to identify in the early stages of adolescence benevolent sexist behaviors, both in female adolescents and their parents. BS beliefs and behaviors are rarely identified as sexist and dangerous by women in general, more often these attitudes are perceived as prosocial (Glick and Fiske 2001) or even valued as attractive (e.g., Bohner et al 2010). Therefore it is important to make adolescent and adult women aware of the sexist nature that underlies those benevolent attitudes and their role in the perpetuation of gender inequality, in order to allow women to challenge and reject them. For that purpose, it seems important to clearly underline two major aspects: first, the influence of the internalization of benevolent sexist beliefs on the future academic and career oportunities of female adolescents; secondly, the contribution of benevolent sexism to perpetuating gender inequality over generations by means of the subtle and concealed mechanisms of traditional gender roles (Appendix).