Journal of Vocational Behavior 84 (2014) 109118

Contents lists available at ScienceDirect

Journal of Vocational Behavior

journal homepage: www.elsevier.com/locate/jvb

The role of friends in career compromise: Same-gender friendship intensies gender differences in educational choice
Samantha Sinclair , Rickard Carlsson, Fredrik Bjrklund
Dept. of Psychology, Lund University, Sweden

a r t i c l e

i n f o

a b s t r a c t
We propose a mechanism of how the desire to maintain friendships can intensify gender differences in educational choice. The required conditions for this mechanism would be that (1) adolescent males and females differ in their overall educational preferences, (2) wanting to stay close to friends motivates some adolescents to adjust their educational choice in line with their friends' choices, and (3) adolescents have a higher share of same-gender, than other-gender, friends. Study 1 confirmed that these criteria were met, and Study 2 found an association between friendship priority and gender typed field of study. In conclusion, adjusting educational choices in order to maintain friendships put adolescents at risk of compromising their true career interests, and also becomes an obstacle to a gender balanced labor market. 2013 Elsevier Inc. All rights reserved.

Article history: Received 28 October 2013 Available online 2 January 2014 Keywords: Friendship Gender segregation Gender differences Educational choice Career choice

1. Introduction The answer to why men and women make different educational choices is likely to be multifaceted. For example, previous research suggests that gender atypical career paths may come to be avoided because of poor self-efficacy (Dickerson & Taylor, 2000), fears of being negatively stereotyped (Bergeron, Block, & Echtenkamp, 2006), conformity to gender role norms (Hackett & Betz, 1981; Tokar, Thompson, Plaufcan, & Williams, 2007); and that women avoid math intensive careers because they endorse communal goals (Diekman, Clark, Johnston, Brown, & Steinberg, 2011). Although psychological aspects have often been highlighted in the literature, many theoreticians have claimed that the social forces that affect career choices should receive more attention (Blustein, Schultheiss, & Flum, 2004; Gottfredson, 1981; Vilhjlmsdttir & Arnkelsson, 2013). In the present paper, we suggest that social networks are of importance for gender differences in field of study. We propose a mechanism rooted in the tendency for adolescents to form same-gender friendships. The required conditions for this mechanism to have an impact on gender segregation would be that (1) adolescent males and females differ in their overall educational preferences, (2) wanting to stay close to friends motivates some adolescents to adjust their educational choice in line with their friends' choices, and (3) adolescents have a higher share of same-gender, than other-gender, friends. If these three criteria are met, there will be an effect of increased gender segregation in the field of study due to friendships in adolescence. For example, if a young woman wants to stay close to her friends when making the transition from high school to college, she has a higher probability of adjusting her choice in line with the educational choices of female rather than male friends, simply because she has more of the former than the latter. Because there are gender differences in occupational interests to begin with (Su, Rounds, & Armstrong, 2009), young women who adjust their choice in line with friends will be more likely to choose a female-dominated (compared to male-dominated) field of study, whereas young men will be more likely to choose one that is male-dominated. Because gender segregation in school in the form of gender differences in course enrollment inevitably shapes the division of men and women into different occupations (Sikora & Pokropek, 2012; Smyth & Steinmetz, 2008), and adolescents' aspirations predict their future
Corresponding author at: Box 213, SE-22100 Lund, Sweden. E-mail address: Samantha.Sinclair@psy.lu.se (S. Sinclair). 0001-8791/$ see front matter 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jvb.2013.12.007

110

S. Sinclair et al. / Journal of Vocational Behavior 84 (2014) 109118

career development (Cochran, Wang, Stevenson, Johnson, & Crews, 2011; Schoon, 2001), this effect will in the long run serve to maintain and intensify gender segregation on the labor market. Although not all adolescents should be expected to adjust their educational choices in order to be with friends, those who have a high fear of isolation or consider themselves to be low-achievers at school may be likely to do so. 2. Friends as barriers to nontraditional choice Several career development theories emphasize that career-related interests and choice behaviors are affected by contextual factors. For example, social cognitive career theory (SCCT; Lent, Brown, & Hackett, 1994, 2000) proposes that several cognitive-person variables (e.g., self-efficacy and goals) interact in reciprocal ways with other aspects of the person (e.g., gender, ethnicity) and with aspects of his or her environment (e.g., social supports, barriers), in shaping career development. Barriers are often defined as events or conditions, either within the person or in his or her environment, that make career progress difficult (Swanson & Woitke, 1997, p. 434). Qualitative work has indeed indicated that contextual supports and barriers are more frequently mentioned by adolescents as relevant to their career choice implementation compared to person factors (Lent et al., 2002). Similarly, according to Gottfredson's (1981, 2005) theory of circumscription and compromise, people's perceptions of opportunities and barriers to obtain a certain type of job will be weighted with their interests to form into aspirations. If perceptions of external barriers leave a certain career choice inaccessible, a career compromise occurs where individuals adjust their occupational aspirations (Tsaousides & Jome, 2008). Perceived opportunities and barriers could concern, for example, the perceived risk of unemployment or ease of obtaining training for the job within the surrounding geographic area, but may also consist of social factors. In the present paper, we focus on the possibility that the fear of losing friends can motivate some adolescents to make a career compromise. The need to belong hypothesis (Baumeister & Leary, 1995) states that people are strongly motivated by a) the need for frequent personal interactions with another person, and b) the need to perceive that the relationship is marked by stability, affective concern, and continuation into the foreseeable future. When a valued relationship is threatened, the individual should experience negative affect and strive to secure the safety of the relationship (Leary, Tambor, Terdal, & Downs, 1995). For example, an adolescent whose best friend is in his class may experience concern or even anxiety when he realizes that he could end up in a different school, separated from his friend. We propose that when this situation creates a threat of separation, it may be perceived as an external barrier to obtain a certain career, resulting in a compromise with interests (Gottfredson, 1981). In other words, the adolescent adjusts his educational choice as a means of remaining close to his friend. Among vocational scholars there has recently been an upswing in the emphasis on relational aspects of career development (Blustein, 2011; Blustein, Prezioso, & Schultheiss, 1995; Blustein et al., 2004; Flum, 2001; Phillips, Christopher-Sisk, & Gravino, 2001; Schultheiss, 2003, among others). In contrast to the more classical approach to career development where the individual is sometimes viewed as reaching career related decisions in an autonomous manner, this relational approach is based on assumptions of human beings as motivated by striving to develop and maintain close interpersonal relationships throughout the lifespan. These studies thus tend to focus on the quality dimension of close relationships and how these relationships can provide a secure base for exploration of career alternatives. Indeed, empirical findings suggest that attachment to peers can be of importance for adolescent career development (Kracke, 2002; Schultheiss, Kress, Manzi, & Glasscock, 2001), even above and beyond parental attachment (Felsman & Blustein, 1999). It is well established that we tend to befriend people who are similar to us (Hafen, Laursen, Burk, Kerr, & Stattin, 2011). One aspect of similarity is that of gender. Indeed, people tend to have more same- compared to other-gender friends across the lifespan (Mehta & Strough, 2009), and adolescence is no exception. Although children tend to prefer same-gender friends from an early age, this tendency typically peaks between middle childhood and early adolescence (Maccoby, 1990), and the proportion of same-gender friends tends to be at least 70% during adolescence (Mehta & Strough, 2010; Poulin & Pedersen, 2007). Furthermore, even when adolescents form other-gender friendships, they are often in the form of secondary as opposed to best friends (Poulin & Pedersen, 2007). This coincides with the time of entry into upper secondary education (the final stage of secondary education) in most OECD (Organization for Economic Co-operation and Development) countries which typically occurs around the age of 15 or 16 (OECD, 2013). Even though the duration of upper secondary education differs between countries, a common feature is that some programs are terminal in the sense of preparing the students for entry directly into working life, while others are preparatory; i.e., preparing students for tertiary education. In Sweden where the present research was conducted, 15 year olds have a choice of 16 national programs to apply to, that specialize in for example social or natural sciences (preparatory), or construction or restaurant work (terminal). The Swedish system is thus similar to many European countries (e.g., Denmark, Germany, Belgium, Czech Republic, Italy, and Spain), and there is also resemblance to the U.S. high school system in that some courses are mandatory while others allow for specialization. In Sweden, the choice of high school program in part determines future access to higher education. The more opportunity for choice on behalf of the student in the education system, the larger the effect of prioritizing friends should be for gender segregation on the labor market. In adolescence, best friends often go to the same school (Calv-Armengol, Patacchini, & Zenou, 2009), and individuals continuously obtain academic as well as social attitudes and behaviors from their friends (Ryan, 2000). As Jones, Audley-Piotrowski, and Kiefer (2012) note, the transmission of such attitudes and beliefs may be especially important in adolescence, which is a period when spending time with friends and valuing friendships increase (Larson & Richards, 1991). However, a distinction should be made between general peer socialization and the mechanism proposed in the present study. First, having mostly same-gender friends may over time result in an increase in gender typical behaviors and attitudes through reinforcement processes in interaction, (i.e., peer socialization). For example, biological sex differences or tendencies for adults to encourage gender stereotyped behavior may be

S. Sinclair et al. / Journal of Vocational Behavior 84 (2014) 109118

111

reinforced in peer interactions as spending time with peers increases. Indeed, research suggests that friends tend to become more similar over time (Hafen et al., 2011; Kandel, 1978). In contrast to this peer socialization process, which describes a reciprocal exchange between friends with attitudes and beliefs changing gradually over time, the mechanism proposed in this paper concerns a more direct tradeoff between an individual's social needs and actual career preferences. Specifically, this effect could reflect a basic need of wanting to stay close to one's friends (going to the same school or the same class), which may cause the individual to adjust his or her choice in line with the choice of a friend. Adolescents with strong interpersonal needs may be particularly likely to make such adjustments. It is important to note that our proposed effect on educational (and occupational) gender segregation would not manifest itself if there were no overall gender differences in preferences to begin with. Hence, the main contribution of our proposed mechanisms is not in explaining how occupational segregation was created in the first place, but rather how it is currently maintained and reinforced. Put differently, it can explain why gender differences in field of study (and hence career paths) are resistant to change. As such, this mechanism does not challenge any of the previously proposed explanations behind gender differences in vocational preferences (e.g., self-efficacy and stereotype threat) but may rather add to the segregation above and beyond these effects. In a sense, our proposed mechanism would act as a rubber band where individuals with atypical preferences are likely to be pulled back into more gender typical behavior because of their friends, resulting in overall gender mainstream tendencies. For example, if a young woman would like to pursue a career in plumbing she may have to leave her female friends behind; in other words, she would face a high social barrier. This implies that she would need to be quite strongly motivated to become a plumber, as this career choice comes with the risk of sacrificing friendships. In contrast, a young man with identical level of interest in becoming a plumber would be much more likely to have a friend who is also interested in this occupation. It should also be emphasized that at the individual level, the consequence of adjusting one's educational preferences to be with friends may very well be the opposite of increased gender typicality in some cases. This would be true for an individual who has primarily other-gender friends with preferences that are typical for their gender, or who has primarily same-gender friends with preferences that are atypical of their gender group. Our proposed mechanism thus appears to be conceptually distinct from processes that involve conformity to gender stereotypes and norms (when individuals internalize prescriptive gender role norms into their self-concept; Hackett & Betz, 1981). In sum, our proposed mechanism adds to the literature by introducing an important contributor to gender segregation in education and, in the long run, on the labor market. 3. Overview of empirical studies In the present paper we claim that if people adjust their field of study in line with their friends' choices, and friendship networks are gender segregated, the tendency of friendship priority can become an obstacle to gender equality. We conducted two studies to estimate the magnitude of this education compromise effect on gender typical choice. In Study 1, we examined adolescents' prospective behavioral intentions: are they willing to adjust their choice of high school program depending on what their friends will choose? Further, we investigated four factors that may affect such friendship priorities. First, although the need to belong seems to be a universal human need, there is also considerable variation among individuals in the general need to belong and in fears about being socially excluded (Baumeister & Leary, 1995; Gere & MacDonald, 2010; Pickett, Gardner, & Knowles, 2004). Based on previous research showing that people who experience threats to belonging tend to directly focus on their social connections (Gardner, Pickett, & Brewer, 2000; Knowles, Lucas, Molden, Gardner, & Dean, 2010), we suggest that a high fear of isolation increases the tendency to compromise one's field of study in line with friends. Second, considering that the motivation to secure the safety of a relationship should depend on how valued the relationship in question is (Leary et al., 1995), we expected that the tendency to adjust one's field of study would be stronger for closer friends (e.g., best friends). Third, we reasoned that low academic self-concept (i.e., feeling like a low-achiever in school) may increase the probability of making choices that depend on friends. For example, remaining close to friends when changing schools may be a means of getting help with homework or other achievement related resources (Crosnoe, Cavanagh, & Elder, 2003). Finally, feeling uncertain about which field to choose may translate into a stronger tendency to adjust one's choice in line with friends. In our second study, we turned to the question of whether the effect of friendship priority on gender segregation is present in actual choices, and further if this effect goes beyond that of other people such as parents and teachers. For example, even though parents may exert substantial influence on their children's choices, they may be more motivated to encourage advanced careers in general rather than gender typical careers per se. In Study 2 we thus hypothesized that perceived education compromise (measured post-choice) in line with friends' choices relates to actual gender typical choice, compared to the perceived importance of parents, teachers, career counselors, interests, and career priorities. In both studies we obtained verbal informed consent and follow both international (APA) and local ethical guidelines. 4. Study 1 4.1. Method 4.1.1. Participants The initial sample consisted of 86 students in the eighth grade, who were soon to apply for high school programs in Sweden. One of them did not report gender and as this is crucial to our research question, we omitted this participant. Further, two

112

S. Sinclair et al. / Journal of Vocational Behavior 84 (2014) 109118

participants were omitted because they skipped several questions, leaving a sample of 83 participants (48 females and 35 males; Mage = 14.33, SD = .52). We decided on the sample size based on power analyses using the G*Power software. We expected gender segregation in friendship networks to be large and as our goal was to identify main contributors to education compromise due to friends (e.g., fear of isolation), we were interested in effects that were expected to be strong or moderate. To this end, we decided on a sample size in excess of 80% power for moderate effect sizes (r N .30). 4.1.2. Materials and procedure The participants were first asked to list the first names of their five best friends who were about the same age as themselves (not including romantic partners and siblings), and to rank them from the best down to the fifth best friend. This served as an unobtrusive measure of friendship gender segregation: In this way, we avoided activating gender concepts, which might have influenced our data. To assess the degree of gender typicality in educational preferences, the participants were then provided with a list of all the Swedish national high school programs and asked to select the first, second, and down to the fifth choice, had they applied that same day. They also rated how certain they were of which program they would pick as their first choice. Next, the key question of whether the participants were willing to compromise their educational choice to be with friends was investigated by asking them to imagine that their very best friend (previously reported as friend number 1) chose a program that they had not intended to apply to, but that they could nevertheless consider applying for. The participants indicated whether they would choose the same program as their best friend (1 = definitely not, 7 = definitely). This question was then repeated for each of the remaining four friends. Academic self-concept was captured with three items (Marsh & O'Neill, 1984): I learn things quickly in most school subjects, I do well on tests in most school subjects, and I'm good at most school subjects (1 = do not agree at all, 7 = completely agree; = .83). We assessed fear of isolation with the following items (based on a scale by Hayes, Matthes, & Eveland, 2011): One of the worst things I can think of is if my friends would hang out and not invite me to join them, It is important to me to fit in with the group that I am with, It would bother me if nobody wanted to hang out with me, and It would bother me if I wasn't invited to parties or other gatherings (1 = do not agree at all, 7 = completely agree). We also included a more specific item where the participants rated the importance of choosing the same program as one of their best friends (1 = not at all important, 7 = very important). This scale had poor reliability ( = .65) which could not be improved by excluding items. Two variables that we chose not to analyze in detail were included. First, we asked the participants about their parents' occupation in order to confirm that our sample was representative with respect to socioeconomic diversity, which it was. The participants were also asked to guess which programs their friends would choose. It turned out that they had a hard time reporting this (as suggested by missing values and creative answers) and as such we found this variable hard to interpret. Finally, the participants answered demographic questions. 4.1.3. Data preparation It turned out that several of the participants had trouble ranking more than four friends, presumably because they did not have that many close friends. We therefore omitted the fifth friend variable from further analysis, leaving us with a total of 332 friends' names. Four friends had gender ambiguous names (e.g., Kim). In these cases, we randomly assigned the value of male or female, based on the assumption that males and females have gender ambiguous names to a similar extent (because ambiguous names were so few, the decision of how to handle them had no impact on the results; even so, we chose the most conservative option). For each selected high school program (e.g., natural science) we coded the percentage of females enrolled in the program based on national statistics. Two participants failed to answer whether they would change their choice for friend number four. We replaced these missing values with the value of friend number three, as the answers for friend number three and four correlated highly. 4.2. Results and discussion 4.2.1. Educational preferences Female participants chose high school programs with on average 62% female students, whereas male participants preferred programs with 33% female students, confirming that our sample was representative of the population (compared to national statistics). 4.2.2. Friendship gender segregation To estimate the degree of gender segregation in the participants' friendship networks, we created two variables for the number of male and female friends, respectively. As expected, nearly all of the male participants' friends were male (M = 3.83, SD = .45) whereas the female participants had almost no male friends (M = .31, SD = .75), and thus had a clear majority of female friends (M = 3.68, SD = .75). Naturally, this difference between female and male participants in the gender distribution of friends was significant, t(81) = 24.68, p b .001.

S. Sinclair et al. / Journal of Vocational Behavior 84 (2014) 109118

113

4.2.3. Education compromise related to friendship gender segregation The hypothesis that the willingness to adjust one's field of study would be stronger for closer friends was confirmed with a planned linear contrast (using the repeated function in SPSS), F(1, 82) = 30.70, p b .001, partial 2 = .27. However, the reported willingness to switch to another program for the sake of a friend was still highly correlated for the four friends, and we could thus average (based on the mean) the results within-participants to a single variable ( = .94). We found an unexpected gender difference in that female participants were more willing (M = 3.51, SD = 1.34) than male participants (M = 2.57, SD = 1.39) to adjust their choice for the sake of a friend, t(81) = 3.11, p = .003. So far we have seen that there is large gender segregation among friends, and that a portion of adolescents would consider compromising their educational choice in order to be with friends. This suggests a stronger total influence exerted by same-gender (compared to other-gender) friends, but the question of how large this differential influence is remains. Thus, the next step was to estimate the size of this differential influence. To this end, we multiplied the willingness to change program scale with the number of female and male friends, respectively. This yielded a scale of the total impact of male friends and female friends, respectively. We were specifically interested in the interaction between participant gender and friends' gender on total influence. To this end, we conducted a 2 (compromise for female and male friends; within subjects) 2 (participant gender; between subjects) ANOVA. The predicted interaction was significant in support of our hypothesis, F(1, 81) = 159.71, p b .001, partial 2 = .66. The means in Fig. 1 reveal that female adolescents were much more likely to make a compromise in line with females than males, t(47) = 9.51, p b .001, whereas male adolescents were clearly more influenced by males than females, t(34) = 9.36, p b .001. Of less interest to our research question, there was no significant main effect of gender of the friends, F(1, 81) = 1.49, p = .23, partial 2 = .018, but a significant main effect of participant gender, F(1, 81) = 9.67, p = .03, partial 2 = .11. We would caution against drawing any conclusions about these main effects because of the presence of the substantial interaction effect. Because of the low share of other-gender friends, resulting in skewed variables, we also conducted a series of non-parametric tests (MannWhitney-U and Related-Samples Wilcoxon Signed Rank test), which confirmed our findings. 4.2.4. Predicting education compromise We proceeded by conducting a multiple regression analysis in order to investigate which factors best explain the willingness to change programs in line with friends (with the willingness to switch program scale as the criterion variable). We decided on a multiple regression approach since our focus was on the unique contribution of our predictor variables while controlling for the others. For example, will the gender difference found in the univariate analyses remain even when controlling for fear of isolation? Because we had quite few variables, we decided to enter them simultaneously rather than testing and comparing multiple models (i.e., stepwise regression). Further, because the variables were not clearly normally distributed we used robust standard errors (using STATA 12; OLS standard errors yielded similar results). We also used standardized variables to allow for direct interpretations of the beta coefficients. The model turned out significant, F(4, 77) = 8.76, p b .001, with R2 = .24. As can be seen in Table 1, fear of isolation was a significant predictor of willingness to change programs for the sake of friends, whereas the effect of academic self-concept was marginally significant, and certainty of choice was in the expected direction but nonsignificant. Descriptive statistics for these variables are presented in Table 2. The gender difference of female participants being more willing to adjust their choice remained when controlling for the other variables. To conclude, it appears that some adolescents (in particular those who have a high fear of isolation) will indeed consider making a career compromise by adjusting their field of study according to their friends' choices. As friendship networks are gender unbalanced, this means that compromise is significantly more likely to occur in line with same-gender, compared to other-gender, friends.

Influence score

10

15

Female participants
Influence of male friends

Male participants
Influence of female friends

Fig. 1. The tendency for adolescent males and females to adjust their educational choice in line with male and female friends. Influence score = willingness to adjust choice of education in line with friends multiplied with the number of male and female friends, respectively.

114

S. Sinclair et al. / Journal of Vocational Behavior 84 (2014) 109118 Table 1 Predictors of willingness to adjust choice due to friends (Study 1). Variable Gender Fear of isolation Uncertainty of choice Academic self-concept Constant B 0.73 0.34 0.22 0.27 4.59 SE B 0.32 0.14 0.16 0.15 0.18 0.25 0.27 0.15 0.17+

Note. N = 83. Standard errors are set as robust. Gender is coded as 0 = females, 1 = males. + p b .10. p b .05.

5. Study 2 The findings of Study 1 showed that the required conditions for the proposed effect exist in prospective behavioral intentions. But what about real educational choices? If we can demonstrate that the actual gender-typed choices that men and women make can be related to education compromise in line with friends, this would lend further credibility to our proposed mechanism. In Study 2, we therefore turned to investigate whether high school students in retrospect perceive that they have adjusted their actual educational choice in line with friends' choices. Furthermore, we examined compromise in line with friends in relation to the perceived importance of other choice related factors. Even though a study consisting of retrospective self-reports would be of limited value on its own, together with Study 1 it serves as a complementary piece of evidence to evaluate our proposed mechanism. 5.1. Method 5.1.1. Participants The participants were 150 students (85 males and 65 females; Mage = 16.95, SD = .80). As in Study 1, this was a sample of convenience but we tried to make sure that the participants came from different Swedish high school programs with varying gender distributions. Each participant received a score according to the percentage of students enrolled in the program (based on national statistics) of the same gender as the participant. This variable served as an indicator of gender typical choice, with a positive value representing gender typical choice (i.e., more than 50% of students enrolled in the program being of the same gender as the participant), and a negative value indicating atypical choice (less than 50% of students being of same gender as the participant). Statistical power was calculated in the same way as in Study 1. However, because we had multiple predictors, and we wanted to have precision to distinguish between them, we used a larger sample compared to Study 1. 5.1.2. Materials and procedure The participants first reported which program they attended. Next, they were asked to think back on the time when they had applied to high school, and indicate the importance of the following factors for their choice of program: friends' educational choices; and the following control variables: own interests and career concerns; and advice from career counselors, teachers, and parents (1 = no importance, 7 = high importance). They were then asked to write down the first names of the friends who had been of importance for their educational choice. As in Study 1, this allowed for unobtrusive measurement of friends' gender, with minimal risk of revealing the hypothesis. Next, the participants responded to four more specific questions about the role of friends in their educational choice, e.g., How important was it to choose the same program as one of your friends from school? (1 = not at all important, 7 = very important). We combined the in total five items concerning compromise of educational choice due to friends into a scale ( = .81). Of the five items included in this perceived compromise scale only three specifically asked about prioritizing a program or school to be together with friends. To confirm that the predictive relationship was not due to only the two more general items, we conducted sensitivity analyses which showed that running regressions with each item in isolation produced nearly identical results as the scale as a whole.

Table 2 Descriptive statistics for factors affecting willingness to adjust choice due to friends (Study 1). Variable Fear of isolation Uncertainty of choice Academic self-concept M 4.38 4.60 5.05 SD 1.13 1.84 1.19

S. Sinclair et al. / Journal of Vocational Behavior 84 (2014) 109118

115

5.2. Results and discussion 5.2.1. Friendship gender segregation One hundred and twenty-six participants reported names of friends who had an impact on their choice of program. As expected, the participants reported significantly more names of friends belonging to the same gender group as themselves (M = 2.82, SD = 2.18) as compared to friends of the other gender (M = .44, SD = .79), t(125) = 15.84, p b .001. As in Study 1, male participants reported more same-gender friends (M = 3.10, SD = 1.11) compared to female participants (M = 2.45, SD =3.18), t(124) = 3.14, p = .002. However, there was no significant gender difference in the number of other-gender friends reported, t(124) = .90, p = .37. 5.2.2. Perceived education compromise Interests (M = 6.09, SD = 1.26) and career priorities (M = 5.43, SD = 1.73) were rated as more important than the importance of other people (parents: M = 3.43, SD = 1.73; teachers: M = 2.53, SD = 1.61; career counselors: M = 2.64, SD = 1.65; and the compromise due to friends scale: M = 2.60, SD = 1.31). However, one can suspect some resistance in admitting to having based one's educational choice on other people, relative to the impact of interests and career aspects. Hence, we believe that the reported influence of other people might be underestimated in this study due to social desirability concerns. There was no difference between male and female participants in to what extent they felt that their choice had been compromised due to friends, t(148) = .01, p = .99. 5.2.3. Compromise related to gender segregation in eld of study Having confirmed that our participants were more influenced by same-gender than other-gender friends, we moved on to test our hypothesis that perceived education compromise in line with friends leads to more gender typical choices. We decided to test this by means of multiple regression analysis in order to be able to examine the independent contribution of the perceived influence of friends, when the effects of the other perceived sources of influence are controlled for. Because this was our focus (rather than finding the optimal regression model), we decided on a single model and to enter all effects simultaneously. To this end, we ran a regression analysis with the compromise due to friends scale, and perceived importance of parents, teachers, career counselors, interests, and career priorities on gender typical educational choice (participant gender was not included because its effect would merely be a product of the sampling). Because some of our variables were not normally distributed, we used robust standard errors (STATA 12, OLS standard errors yielded highly similar results). The results of this regression can be found in Table 3. The model turned out significant, F(6, 142) = 3.03, p = .002, with R2 = .10. The average participant attended a program with 62% of students being same gender as the participant. Supporting our predictions, having adjusted one's choice according to friends' choices meant having made a more gender typical educational choice. Specifically, an increase by one standard deviation on the compromise due to friends scale corresponded to an increase of four percentage units in gender typical educational choice. That is, compromising one's choice to be with friends meant selecting a program with on average 66% (compared to 62%) same-gender students. If we look at the effect of compromise in line with only same-gender friends, the effect is (as expected) even more pronounced; in fact, this corresponded to attending a program with on average 69% same-gender students. Somewhat surprisingly, higher perceived importance of career counselors also corresponded to a more gender typical field of study. None of the remaining variables predicted gender typical choice. It is important to keep in mind that friends were not rated as the top contributor to educational choice; for example, interests were perceived as more important overall. Nevertheless, to the extent that choice was adjusted in line with friends, this had consequences for gender typical field of study, whereas there was no such relationship for interests. It is also important to note, however, that robust standard errors in the regression analysis only helps to protect the analysis from spurious statistical findings (Type I error) and do not remedy that a skewed variable may be underpowered. A closer look at the variables revealed a possible ceiling effect on the interest variable. Although this is not surprising since people are likely to feel that their interests are an

Table 3 Predictors of gender typical educational choice (Study 2). Variable Friends Teachers Parents Career counselors Interests Career priorities Constant B 4.17 1.82 1.02 3.53 1.50 2.63 62.37 SE B 1.60 1.73 2.12 1.70 1.70 1.96 1.49 0.21 0.09 0.05 0.18 0.07 0.14

Note. N = 150. Standard errors are set as robust. p b .05. p b .01.

116

S. Sinclair et al. / Journal of Vocational Behavior 84 (2014) 109118

important foundation for their career choices, it limits this variable's statistical power and may mask a true relationship. The (lacking) contribution of this control variable should thus be interpreted with some caution. 6. General discussion 6.1. Support for the proposed mechanism We proposed a mechanism for intensified gender segregation in field of study. Our empirical findings confirm that its three main assumptions are fulfilled: There are gender differences in educational choices, friends tend to be of the same gender, and some adolescents appear willing to commit trade-offs with their interests when it comes to educational choices. Specifically, Studies 1 (behavioral intentions) and 2 (perceived compromise related to actual choice) provide two complementary pieces of evidence for our proposed mechanism. Together, the findings suggest that the fear of losing friends can constitute a barrier to obtain a desired career (Gottfredson, 1981; Lent et al., 2000), which may result in a career compromise. This compromise will not be random, but clearly gender-typed, and will thus have consequences for gender segregation on the labor market. The friendship mechanism thus adds to the literature on the link between gender and career development by emphasizing an additional route, besides gender socialization processes (e.g., Hackett & Betz, 1981; Tokar et al., 2007), to gender-typed career choice. The present research further investigated the psychological underpinnings of the proposed friendship mechanism. First, as expected, the willingness to adjust one's educational choice was greater for closer friends, confirming that the motivation to secure the safety of a relationship should depend on how valued the relationship in question is (Leary et al., 1995). Second, fear of isolation moderated the effect. In other words, concern of being alone is associated with an increased propensity to trade educational interests for staying close to one's friends. And third, there was a tendency for academic self-concept to moderate the effect, suggesting that confidence in one's academic abilities may buffer against trading interests for friends. 6.2. Potential generalization of the mechanism The mechanism demonstrated here could likely be applicable for other groups and in other contexts where segregation exists. Essentially, when there are systematic differences in preferences between groups, and predominantly in-group members in the circle of friends, our proposed mechanism may be in play. First, friendship is not restricted to gender similarity but may also concern other social groups. To the extent that one's friends belong to a particular social group, and the members of this social group have similar career preferences, an effect akin to the one demonstrated here may occur. For example, people with predominantly blue-collar friends may be more likely to trade their actual job preferences for such an occupation instead, particularly if the risk of being alone looms large. The reverse should be true for individuals in well-off circles who would like to work with their hands but lack friends who share this interest. Second, the effect may be of relevance in other contexts than field of study, such as employee turnover on the labor market: When people consider switching to a new workplace, our mechanism suggests that friendship priorities should matter in the form of self-selection into workplaces where perceived social opportunities are high and social barriers are low. Although there is arguably less room for a gender-typed choice to arise in this context (because specialization has already taken place, and since adults may not be as peer focused as adolescents), it may still exist, particularly for individuals with a high fear of isolation. To conclude, our proposed mechanism may have considerably larger scope than educational choice, ranging from young children choosing to play with dolls versus trucks to grown-ups switching jobs; its strength being that it pin-points the influence friends will have whenever there is a potential tradeoff between one's own, and friends', preferences. Although its relative importance is likely to vary across context and further research is warranted, we would not be surprised if there are other specific situations where this effect will be of high importance to the point where it affects life changing decisions. 6.3. Limitations The empirical results reported here are limited to self-reports, a type of data which may be subject to social desirability concerns. However, as it is hardly desirable to report trading one's actual career interests for the choices one's friends make, we may arguably have underestimated the effect. Furthermore, as we estimated the role of same-gender vs. other-gender friends in an indirect manner, social desirable responding related to friends' gender was minimized. If one is concerned about internal validity, i.e. with direct testing of the mechanism itself, an experimental approach would probably be preferable. One way to go would be to manipulate the need to belong (e.g., through priming or false feedback about peer rejection) to see if this produces more gender-typed preferences; akin to what gender identity threat does (Sinclair & Carlsson, 2013). Admittedly, there is also room for further specification of the mediating and moderating social-cognitive mechanisms of the effect, as the main purpose of the present study was restricted to introducing the basic mechanism and complement it with relevant data. Furthermore, the willingness to adjust one's educational choices may depend on additional friendship factors, besides closeness level of the friendship. As described in the Introduction section, there are differences in educational structure that may put limits to the tradeoff between friendship and own educational preferences. In a sense, friendship priority could be seen as a luxury rendered possible by societies that allow for free choice. When one's profession is dictated by family, cast systems and the like, the influence of friendship on educational choice should be of little concern.

S. Sinclair et al. / Journal of Vocational Behavior 84 (2014) 109118

117

Previous research indicates that individual differences in collectivism are in some cases of importance for work values (Hartung, Fouad, Leong, & Hardin, 2010). It is also possible that the tendency to prioritize friendships in educational choice may be particularly pertinent in more collectivist cultures, where social relations tend to constitute a more central part of an individual's identity (Markus & Kitayama, 1991). On the other hand, in collectivist cultures it may be the case that family members, rather than friends, are given main priority when it comes to career choices. A recent cross-cultural study with American, Chinese, and Hong Kong samples indicates that, at least when it comes to the role of perceived support from peers for career exploration and commitment, these cultures do not differ (Cheung, Wan, Fan, Leong, & Mok, 2013). 6.4. Implications Our findings attest to the importance of considering interpersonal relationships in career development research. It is important, however, to keep in mind that we propose a specific mechanism of how friendship priority leads to maintained educational gender segregation. We do not attempt to provide a complete analysis of career choice as do for example relational theorists (e.g., Blustein, 2011), and SCCT (Lent et al., 1994), but rather a specific and unique contribution to the literature. The fact that it is a specific mechanism implies that it cannot be expected to cover all possible varieties of influence between friends. We of course realize that peer influence can also be of a more dynamic nature than is captured by our mechanism, for example, one possibility may be that some individuals have the personal agency to persuade their friends to adjust their choice instead. Our finding that some adolescents factor in friendship in their educational choices has implications for related models, such as Gottfredson's (1981, 2005) circumscription and compromise model. This model also focuses on sex-type of the job, but emphasizes that gender atypical occupations are rejected already in childhood because of perceived incompatibility with gender identity. In terms of Gottfredson's concepts, there appears to be no connection between our proposed friendship mechanism and the circumscription process. The compromise process, however, is similar in that perceived barriers to obtain a certain type of career may be weighted with one's interests to form into career choice. Our findings are also consistent with social cognitive career theory (Lent et al., 1994, 2000) in that individuals not only actively influence their situation but also may be less aware of certain external barriers to their career outcomes. In SCCT, however, gender is primarily considered as a person factor, while our friendship mechanism emphasizes how the role of gender as a person factor interacts with the role of gender (in friendship networks) as a contextual variable. SCCT further proposes that proximal contextual influences can affect career implementation either in a direct way, or in an indirect way, moderating the relation between interests and goals or between goals and actions (Lent et al., 2000). Our friendship mechanism could be interpreted as an example of the latter, showing how a barrier (the perceived risk of losing friends) can moderate the link between interests and educational choice. As such, our mechanism is consistent with the general line of thought in both SCCT (Lent et al., 2000) and in Gottfredson's (2005) theory of circumscription and compromise, but cannot be directly inferred from either theory. If some individuals are willing to trade their true interests for the opportunity to be with friends, this may create negative consequences for society as well as for the individual. Consider for example a young man who, because of friends, pursues a career path in engineering, despite being more interested in health care. Even if he does not suffer any personal consequences from his career compromise, his choice may mean that he occupies a position that a more motivated female engineer could have filled and that the labor market misses out on a well-needed male nurse. That is, because this mechanism adds to the occupational gender segregation on the labor market, it may contribute to the gender gap in wages (Petersen & Morgan, 1995) and may possibly hinder organizational performance (Krishnan & Park, 2005). Career choices are among the most important decisions in life, and previous research suggests that compromise can result in reduced well-being and work-related satisfaction (Carr, 1997; Meir, Melamed, & Dinur, 1995; Tsaousides & Jome, 2008). Indeed, it has been shown that inactions are associated with more regret than actions (Gilovich & Medvec, 1995), suggesting that choosing to settle for a career path that the individual considers less interesting may have negative psychological consequences later in life. However, even though the decision to trade one's educational preferences for friends appears irrational at first glance, it should be noted that reduced self-esteem and well-being related to the loss of friends may also be very painful to the individual. The perceived costs of losing company may loom large in the short term whereas the costs associated with career compromise may not. In the light of such cost calculations a compromise appears less irrational. On the whole, regrets about social relationships are more intense than several other types of regrets (e.g., health related issues), but career regrets are among top regrets too (Morrison, Epstude, & Roese, 2012; Roese & Summerville, 2005). Regardless of how reasonable it is to weigh friendship into one's career related decisions, our findings certainly have implications for Baumeister and Leary's (1995) belongingness hypothesis. The finding that the wish to be among friends appears strong enough to affect career related choices lends credibility to the claim that the need to belong is a crucial psychological motive. In conclusion, the effect of friendship priority on educational choice is an obstacle to the labor market becoming more gender balanced. Although it is not possible (or desirable) to eliminate social needs, individuals and society would benefit from interests, rather than friends' choices, as the main priority when it comes to career related decisions. References
Baumeister, R. F., & Leary, M. R. (1995). The need to belong: Desire for interpersonal attachments as a fundamental human motivation. Psychological Bulletin, 117, 497529. Bergeron, D. M., Block, C. J., & Echtenkamp, B. A. (2006). Disabling the able: Stereotype threat and women's work performance. Human Performance, 19, 133158. Blustein, D. L. (2011). A relational theory of working. Journal of Vocational Behavior, 79, 117. Blustein, D. L., Prezioso, M. S., & Schultheiss, D. P. (1995). Attachment theory and career development: Current status and future directions. The Counseling Psychologist, 23, 416432.

118

S. Sinclair et al. / Journal of Vocational Behavior 84 (2014) 109118

Blustein, D. L., Schultheiss, D. E. P., & Flum, H. (2004). Toward a relational perspective of the psychology of careers and working: A social constructionist analysis. Journal of Vocational Behavior, 64, 423440. Calv-Armengol, A., Patacchini, E., & Zenou, Y. (2009). Peer effects and social networks in education. Review of Economic Studies, 76, 12391267. Carr, D. (1997). The fulfillment of career dreams at midlife: Does it matter for women's mental health? Journal of Health and Social Behavior, 38, 331344. Cheung, F. M., Wan, S. L. Y., Fan, W., Leong, F., & Mok, P. C. H. (2013). Collective contributions to career efficacy in adolescents: A cross-cultural study. Journal of Vocational Behavior, 83, 237244. Cochran, D. B., Wang, E. W., Stevenson, S. J., Johnson, L. E., & Crews, C. (2011). Adolescent occupational aspirations: test of Gottfredson's theory of circumscription and compromise. The Career Development Quarterly, 59, 412427. Crosnoe, R., Cavanagh, S., & Elder, G. H., Jr. (2003). Adolescent friendships as academic resources: The intersection of friendship, race, and school disadvantage. Sociological Perspectives, 46, 331352. Dickerson, A., & Taylor, M. A. (2000). Self-limiting behavior in women: Self-esteem and self-efficacy as predictors. Group and Organization Management, 25, 191210. Diekman, A. B., Clark, E. K., Johnston, A. M., Brown, E. R., & Steinberg, M. (2011). Malleability in communal goals and beliefs influences attraction to STEM careers: Evidence for a goal congruity perspective. Journal of Personality and Social Psychology, 101, 902918. Felsman, D. E., & Blustein, D. L. (1999). The role of peer relatedness in late adolescent career development. Journal of Vocational Behavior, 54, 279295. Flum, H. (2001). Relational dimensions in career development. Journal of Vocational Behavior, 59, 116. Gardner, W. L., Pickett, C. L., & Brewer, M. B. (2000). Social exclusion and selective memory. How the need to belong influences memory for social events. Personality and Social Psychology Bulletin, 26, 486496. Gere, J., & MacDonald, G. (2010). An update of the empirical case for the need to belong. The Journal of Individual Psychology, 66, 93115. Gilovich, T., & Medvec, V. H. (1995). The experience of regret: What, when, and why. Psychological Review, 102, 379395. Gottfredson, L. S. (1981). Circumscription and compromise: A developmental theory of occupational aspirations. Journal of Counseling Psychology Monograph, 28, 545579. Gottfredson, L. S. (2005). Applying Gottfredson's theory of circumscription and compromise in career guidance and counseling. In S. D. Brown, & R. W. Lent (Eds.), Career development and counseling: Putting theory and research to work (pp. 71100). New York: Wiley. Hackett, G., & Betz, N. E. (1981). A self-efficacy approach to the career development of women. Journal of Vocational Behavior, 18, 326339. Hafen, C. A., Laursen, B., Burk, W. J., Kerr, M., & Stattin, H. (2011). Homophily in stable and unstable adolescent friendships: Similarity breeds constancy. Personality and Individual Differences, 51, 607612. Hartung, P. J., Fouad, N. A., Leong, F. T. L., & Hardin, E. E. (2010). Individualismcollectivism: Links to occupational plans and work values. Journal of Career Assessment, 18, 3445. Hayes, A. F., Matthes, J., & Eveland, W. P. (2011). Stimulating the quasi-statistical organ: Fear of social isolation motivates the quest for knowledge of the opinion climate. Communication Research. http://dx.doi.org/10.1177/0093650211428608 (Advance online publication). Jones, M. H., Audley-Piotrowski, S. R., & Kiefer, S. M. (2012). Relationships among adolescents' perceptions of friends' behaviors, academic self-concept, and performance. Journal of Educational Psychology, 104, 1941. Kandel, D. B. (1978). Homophily, selection, and socialization in adolescent friendships. American Journal of Sociology, 84, 427436. Knowles, M. L., Lucas, G. M., Molden, D. C., Gardner, W. L., & Dean, K. K. (2010). There's no substitute for belonging: Self-affirmation following social and nonsocial threats. Personality and Social Psychology Bulletin, 36, 173186. Kracke, B. (2002). The role of personality, parents, and peers in adolescents' career exploration. Journal of Adolescence, 25, 1930. Krishnan, H. A., & Park, D. (2005). A few good women on top management teams. Journal of Business Research, 58, 17121720. Larson, R., & Richards, M. H. (1991). Daily companionship in late childhood and early adolescence: Changing developmental contexts. Child Development, 62, 284300. Leary, M. R., Tambor, E. S., Terdal, S. K., & Downs, D. J. (1995). Self-esteem as an interpersonal monitor: The sociometer hypothesis. Journal of Personality and Social Psychology, 68, 518530. Lent, R. W., Brown, S. D., & Hackett, G. (1994). Toward a unifying social cognitive theory of career and academic interest, choice, and performance. Journal of Vocational Behavior, 45, 79122. Lent, R. W., Brown, S. D., & Hackett, G. (2000). Contextual supports and barriers to career choice: A social cognitive analysis. Journal of Counseling Psychology, 47, 3649. Lent, R. W., Brown, S. D., Talleyrand, R., McPartland, E. B., Davis, T., Chopra, S. B., et al. (2002). Career choice barriers, support, and coping strategies: College students experiences. Journal of Vocational Behavior, 60, 6172. Maccoby, E. E. (1990). Gender and relationships: A developmental account. American Psychologist, 45, 513520. Markus, H. R., & Kitayama, S. (1991). Culture and the self: Implications for cognition, emotion, and motivation. Psychological Review, 98, 224253. Marsh, H. W., & O'Neill, R. (1984). Self description questionnaire III (SDQ III): The construct validity of multidimensional self-concept ratings by late adolescents. Journal of Educational Measurement, 21, 153174. Mehta, C. M., & Strough, J. (2009). Gender segregation in friendships and normative contexts across the life span. Developmental Review, 29, 201220. Mehta, C. M., & Strough, J. (2010). Gender segregation and gender-typing in adolescence. Sex Roles, 63, 251263. Meir, E. I., Melamed, S., & Dinur, C. (1995). The benefits of congruence. Career Development Quarterly, 43, 257266. Morrison, K. R., Epstude, K., & Roese, N. J. (2012). Life regrets and the need to belong. Social Psychological and Personality Science, 3, 675681. OECD (2013). Education at a Glance 2013: OECD Indicators. Retrieved from. : OECD Publishing. http://dx.doi.org/10.1787/eag-2013-en. Petersen, T., & Morgan, L. A. (1995). Separate and unequal: Occupation-establishment sex segregation and the gender wage gap. American Journal of Sociology, 101, 329365. Phillips, S. D., Christopher-Sisk, E., & Gravino, K. (2001). Making career decisions in a relational context. The Counseling Psychologist, 29, 193213. Pickett, C. L., Gardner, W. L., & Knowles, M. (2004). Getting a cue: The need to belong and enhanced sensitivity to social cues. Personality and Social Psychology Bulletin, 30, 10951107. Poulin, F., & Pedersen, S. (2007). Developmental changes in gender composition of friendship networks in adolescent girls and boys. Developmental Psychology, 43, 14841496. Roese, N. J., & Summerville, A. (2005). What we regret mostand why. Personality and Social Psychology Bulletin, 31, 12731285. Ryan, A. M. (2000). Peer groups as a context for the socialization of adolescents' motivation, engagement, and achievement in school. Educational Psychologist, 35, 101111. Schoon, I. (2001). Teenage job aspirations and career attainment in adulthood: A 17-year follow-up study of teenagers who aspired to become scientists, health professionals, or engineers. International Journal of Behavioral Development, 25, 124132. Schultheiss, D. E. P. (2003). A relational approach to career counseling: Theoretical integration and practical application. Journal of Counseling and Development, 81, 301310. Schultheiss, D. E. P., Kress, H. M., Manzi, A. J., & Glasscock, J. M. J. (2001). Relational influences in career development: A qualitative inquiry. The Counseling Psychologist, 29, 214239. Sikora, J., & Pokropek, A. (2012). Gender segregation of adolescent science career plans in 50 countries. Science Education, 96, 234264. Sinclair, S., & Carlsson, R. (2013). What will I be when I grow up? The impact of gender identity threat on adolescents' occupational preferences. Journal of Adolescence, 36, 465474. Smyth, E., & Steinmetz, S. (2008). Field of study and gender segregation in European labour markets. International Journal of Comparative Sociology, 49, 257281. Su, R., Rounds, J., & Armstrong, P. I. (2009). Men and things, women and people: A meta-analysis of sex differences in interests. Psychological Bulletin, 135, 859884. Swanson, J. L., & Woitke, M. B. (1997). Theory into practice in career assessment for women: Assessment and interventions regarding perceived career barriers. Journal of Career Assessment, 5, 431450. Tokar, D. M., Thompson, M. N., Plaufcan, M. R., & Williams, C. M. (2007). Precursors of learning experiences in social cognitive career theory. Journal of Vocational Behavior, 71, 319339. Tsaousides, T., & Jome, L. (2008). Perceived career compromise, affect and work-related satisfaction in college students. Journal of Vocational Behavior, 73, 185194. Vilhjlmsdttir, G., & Arnkelsson, G. B. (2013). Social aspects of career choice from the perspective of habitus theory. Journal of Vocational Behavior, 83, 581590.