Gender Norms and Other Factors Explaining the Gender Gap in Students’ Entrepreneurial Inclination in Spain and Iceland

Entrepreneurship is a Gendered Activity

Research often highlights the fact that entrepreneurship is a gendered process (Bruni, Gherardi, & Poggio, 2004). As evidenced by Eddleston and Powell (2012), the practice of entrepreneurship can best be understood as embedded in the gendering of work. This gendering can be considered in two ways: First, in terms of theories of occupational segregation by gender (Anker 1998), entrepreneurship has traditionally been considered a male-dominated field (Díaz-García & Jiménez-Moreno, 2009; Gupta et al., 2009; Thébaud, 2010). Indeed, masculine stereotypes associated with the business world (lean, hungry, predatory, hostile) are also often associated with entrepreneurship (Gupta et al., 2009). Second, traditional gender norms or roles link woman with domestic work (housework and childcare) and man with the role of breadwinner. These gender norms are changing and, in fact, in the most advanced societies most households are dual-income households. However, this process of change is gradual and lagged, so that increased participation of women in the labor market is not always offset by a greater involvement of men in domestic work (Raley, Bianchi, & Wang, 2012). Thus, many female entrepreneurs have the double responsibility of both work and family, and these gender roles and identities at the micro level influence their behavior and performance compared with male entrepreneurs.

These gender issues appear directly or indirectly in the analysis of the determinants or causes of gender differences in entrepreneurship.

Stereotypes, Self-assessment, Self-confidence, and Status Characteristics Theory

Status Characteristics theory (Berger, Cohen, & Zeldich, 1972; Correll & Ridgeway, 2003) seeks to explain how beliefs about status characteristics get translated into performance expectations about individuals. Initially, the theory was developed to explain how hierarchies are created in small group interactions (working teams, etc.) but Correll and Ridgeway (2003) expanded the scope of this theory. Applying it to the case of gender, they argue that salient beliefs about gender affect the standard individuals use to evaluate their own task ability in different settings. For example, cultural beliefs and stereotypes that men have more mathematical ability prime a status generalization process that causes men to use more lenient standards than women to judge their own mathematical competence (Correll, 2004).

Following the approach of Correll and Ridgeway (2003), Thébaud (2010) suggests that, because of masculine gender-role stereotypes associated with entrepreneurship, women are significantly less likely to perceive themselves as able to be entrepreneurs and they hold themselves to a stricter standard of competence when compared to similarly situated men. Using Global Entrepreneurship Monitor (GEM) data (from the United States), she shows that gender difference in self-assessment accounts for a significant portion of the gender gap in entrepreneurship after controlling for relevant resources. Similar evidence was obtained by Zhao et al. (2005) and Martínez Campo (2011). Additionally, Gupta et al. (2014), following the stereotype threat approach, conducted two controlled experiments to explore differences between men and women in terms of evaluation of new business opportunities. Gender differences in opportunity evaluation were exacerbated when entrepreneurship was linked to masculine stereotypical information, and reversed in favor of women when entrepreneurship was linked to feminine stereotypical information.

Based in these considerations, in relation to self-confidence about abilities at work, the following is proposed:

• Hypothesis 1. Having more self-confidence in job abilities is positively associated with the inclination to entrepreneurship.
• Hypothesis 1b. Since entrepreneurship is primarily associated with male characteristics, it is expected that the above association is higher for male students. According to status characteristics theory, women require (on average) a higher level of self-confidence about their abilities at work to have the same level of inclination to entrepreneurship as men.

Gender Differences in Entrepreneurship and Work–family Balance

Work–family border theory (Clark, 2000) tries to explain how individuals manage and negotiate the work and family spheres and the borders between them in order to attain a work–family balance. Central to this theory is the idea that work and family constitute different domains or spheres which influence each other. The construction of work–family boundaries would be the result of a complex interplay between employees’ strategies and preferences, the social contexts in which they are embedded, and both the idiosyncratic and cultural meanings attached to work and family (Desrochcers & Sargent, 2004). An important concept in this context is that of spillover (Chen, Powell, & Greenhaus, 2009; Greenhaus & Powell, 2006). These spillovers may be positive (work–family enrichment) or negative (work–family conflict).

Gender roles may play an important function in the case of work and family experiences. Even in the most advanced societies the entry of women into the labor market seems to have progressed much more than men’s participation in childcare activities (Hook, 2006), and this mismatch suggests that women are more likely to have a greater sense of dual responsibility, thus leading to greater conflict but also to greater enrichment associated with their work and caregiving roles (Fujimoto, Azmat, & Härtel, 2012).

With respect to work–family enrichment, Eddleston and Powell (2012) stress the importance of family to entrepreneurship and suggest that participation in the family role may enrich an entrepreneur’s wellbeing (family-to-business enrichment). Some of these (affective and instrumental) components of family-to-work enrichment would be (Greenhaus & Powell, 2006): Transferring positive affect (positive mood or happiness […]) from the family domain to the work domain; transferring skills (ability to multitask […]); and transferring behaviors acquired or nurtured in the family domain to the work domain (e.g., being supportive). But Eddleston and Powell (2012) also show that there is a gender difference in the way of experiencing work–family enrichment: “Female entrepreneurs tend to take a more holistic approach to their lives than male entrepreneurs do; they tend to view their businesses as cooperative networks of relationships that are integrated with, rather than segmented from, their family and personal relationships.” In fact, they reach an interesting finding in their study of 258 entrepreneurs from the U.S.: Women appeared to nurture their satisfaction with work–family balance by creating instrumental enrichment between their family and business roles, whereas men appeared to nurture their satisfaction with work–family balance by drawing upon instrumental support in the home (family members supporting their ideas concerning the business; helping or liberating them from the routine household tasks).

Based in these considerations about work–family enrichment, the following is proposed:

• Hypothesis 2. Women have a higher average score than men in Positive work–care interaction.
• Hypothesis 2b. Having a positive attitude (enrichment) about work–family relationships is positively related to the inclination to entrepreneurship; this positive relation is more intense among female students.

With respect to work–family conflict, it can be defined as a form of inter- role conflict arising because pressures emanating from one role are incompatible with those from another role (Greenhaus & Beutell, 1985; Shelton, 2006). As a result, obtaining certain rewards in one domain requires foregoing rewards in another (Greenhaus and Beutell (1985) distinguish three different types of work–family conflict: Time-based conflict; strain-based conflict; and behavior-based conflict).

Because of women’ greater sense of dual responsibility, female entrepreneurs are likely to experience greater time-based and strain-based conflict than male entrepreneurs, and this can have negative implications, for instance for the growth of their firms (Jennings & McDougald, 2007).

Jennings and McDougald (2007) and Moen and Yu (2000) identified several strategies for managing work–family conflict at the couple level of analysis: Traditional strategy (only one partner has paid employment while the other assumes the role of stay-at-home spouse); a one-job/one-career strategy (one partner takes on a less demanding job so that the other can pursue his or her career more vigorously); postponing children until the partners’ careers are sufficiently established; hiring a domestic helper; scaling back within the work domain; and scaling back within the non-work domain (lowering expectations for housework; limiting the number of children; reducing time spent on leisure, social, and community pursuits, etc.).

• Hypothesis 3. Because of women’ greater sense of dual responsibility, in the face of a future work–family conflict, there would be more female students than male students who would choose a family model in which they would take on a less demanding job (role reduction) or would be a stay-at- home spouse (role elimination). There is a negative relation with preferring these models of family and the inclination to entrepreneurship.

Hypotheses Related to Other Factors Identified in the Entrepreneurship Literature

There is considerable evidence that offspring from business families are more motivated to start their own firm (their parents can act as positive role models; it is easier for them to obtain managerial human capital and firm-specific human capital; they can benefit from their parents’ networks, etc.) than offspring without this background (Ahmed et al., 2010; Carr & Sequeira, 2006; Fatoki, 2014; Udofia & Essien, 2013; Zellweger et al., 2011).

Lindquist, Sol, and Van Praag (2012) and Bosma, Hessels, Schutjens, and Van Praag (2012) argue that the effect of role models is such that the transmission of entrepreneurship from mothers to daughters is significantly stronger than that from fathers to daughters, and for sons the effect of entrepreneurial fathers is significantly stronger than the effect of entrepreneurial mothers. Following this line:

• Hypothesis 4. Family precedents in entrepreneurship are a positive determinant of EI.
• Hypothesis 4b. According to the role modeling approach, having an entrepreneur mother influences the daughters’ EI more, and having an entrepreneur father influences the sons’ more.

Access to finance is an important constraint to entrepreneurship, particularly for women (they report disadvantageous access to finance; they are subject to higher charges for loans; they demand less funding because they are relatively concentrated in smaller, part-time, and/or home-based enterprises in low-cost sectors with associated growth constraints, etc.). The literature on this subject is also considerable (Marlow & Patton, 2005; Muravyev, Talavera, & Schaefer, 2009; Saridakis, Marlow, & Storey, 2014).

This article uses a sample of university students, most of whom have not yet had experience in creating a business, so in this case we expect that both male and female students will have the same perception about the existence of a financial constraint. Thus, the following is proposed:

• Hypothesis 5. When the level of parental income is high, the student has a perception of being less subject to financial constraint, and this is related positively to his/her EI.

In reference to the relation between tolerance of risk and EI, Douglas and Shepherd (2002) found that the higher the individual’s tolerance for risk, and the stronger his/her preference for decision-making autonomy, the stronger his/her intention to be self-employed. On the other hand, several studies in the fields of economics and psychology show that women appear to be, on average, more financially risk averse than men (see Booth and Nolen 2012; Croson and Gneezy 2009). According to Thébaud (2010), some explanations for this finding may include aspects like men’s less emotional reactions to uncertain situations; but there can also be a component of prescriptive stereotypes about agentic, masculine behavior. In this same line, for example, Booth and Nolen (2012), using an experimental setting, show that women and men may differ in their propensity to choose a risky outcome (in part) because of pressure to conform to gender stereotypes.

Based in these considerations, the following is proposed:

• Hypothesis 6. Having a higher risk tolerance is positively associated with EI.
• Hypothesis 6b. Female students, on average, have a lower tolerance for risk than male students do.

Finally, there are some authors who consider that entrepreneurship research can be integrated into the traditions of leadership and management research. In fact, many of the skills and attitudes identified in these two fields are very similar (Painoli, 2012; Vecchio, 2003). Since many of the attitudes that characterize the area of entrepreneurship are common to those corresponding to the area of leadership, the following is proposed:

Based in these considerations, the following is proposed:

• Hypothesis 7. Greater leadership is positively correlated to EI.

Data

Data were collected from 1321 university students (748 women, 573 men) who attended the Complutense University of Madrid, and 501 university students (338 women, 163 men) who attended the University of Iceland. Sampling was performed during the period November 2013–April 2014. The questionnaire was administered in randomly selected classrooms during class time. A paper version of the questionnaire was completed by 724 participants in Spain and 112 in Iceland, and 597 in Spain and 389 in Iceland completed the on-line version. Of the total number, 1156 students in Spain and 353 in Iceland were studying for their bachelor’s degree; the rest were studying for their master’s degree. They were distributed throughout three gender-integrated fields of study, Business Administration (469 in Spain, 163 in Iceland), Law (356, 170), and Economics (244, 52); and in one feminized field, Social Work (234, 110); 18 and 6 students were in other fields. The average age of participants was 22.2 years in Spain and 26.7 years in Iceland, and 22.7% of students in Spain and 49.3% in Iceland were combining their studies with a full- or part-time job. After excluding some cases that presented missing values, the final sample was 1145 respondents for Spain and 426 for Iceland.

Questionnaire

First, the questionnaire contained a series of questions concerning respondents’ demographic data, family background, and gender attitudes. Then there were several sets of questions using a 5-point Likert scale ranging from 1 = strongly disagree to 5 = strongly agree. These groups of questions were related to a) career aspirations, based in part on the “Career Aspirations Scale” of O’Brien (1996) and Greenhaus’ (1973) “Work Role Salience Scale”; and b) family aspirations, based in part on the “Career Family Attitude Measure” of Sanders, Lengnick-Hall, Lengnick-Hall, and Steele-Clapp (1998) and Swanson, Daniels, and Tokar’s (1996) “Career Barriers Inventory.” There were also two questions using 0–10-point Likert scales, one of them concerning entrepreneurship.

Dependent Variable

The dependent variable is Entrepreneurial inclination (EI), which is a 0–10 Likert item variable that serves as an indicator of the students’ inclination to entrepreneurship. The following question was asked: “On the following scale of ‘entrepreneurship,’ ranging from 0 (‘I consider myself very little entrepreneurial’) to 10 (‘I consider myself very entrepreneurial’), in which box would you place yourself? (Entrepreneur: A person with a tendency to start business ventures and innovation, taking financial risks).”

Independent Variables

Parental Income, a 0–10 Likert item variable (0 = very low income; 10 = very high income), that comes from this question: “On the following scale of ‘level of household income,’ which ranges from 0 (very low income) to 10 (very high income), in which box would you situate the level of income of your parents or legal guardians? (Now or when you were younger)”

Mother entrepreneur or independent worker, a dummy coded variable (1 = The mother was an independent worker or entrepreneur [during most of the student’s childhood]; 0 = the mother was a wage earner or did not do paid work).

Father entrepreneur or independent worker, a dummy coded variable (1 = The father was an independent worker or entrepreneur; 0 = the father was a wage earner or did not do paid work).

After conducting a Varimax factor analysis with the 16 questions included in the group of questions “thinking about your future career,” four factors were obtained, two of which were used for constructing the following scales:

Leadership aspirations (Cronbach’s alpha = .764) involves seven items (e.g., “I hope to become a leader in my career field”). The range of values is from 1.14 to 5. The higher the score, the higher the aspirations of the respondent to lead organizations and teams.

Willingness to Risk (Cronbach’s alpha = .627). It includes these four items: “I would prefer a less secure and stable job with a net monthly salary of $4,000 to a secure and stable job with a net monthly salary of $2,000”; “I would prefer a less secure and stable job with significant career development opportunities, to a secure and stable job with limited opportunities of professional promotion”; “I would prefer working in the public sector to the private sector (reversed scores)”; “I would prefer being an entrepreneur or self-employed to being an employee”. The range of values is from 1 to 5. The higher the score, the higher the participant’s willingness to take risks (in the labor market).

Confidence about work abilities is a 1–5 Likert item variable (1 = strongly disagree; 5 = strongly agree) corresponding to “I don’t feel confident about my abilities in the world of work.” After reversing the scores, this variable will serve as an indicator of higher self-confidence in job abilities.

Positive work–care interaction is a 1–5 Likert item variable (1 = strongly disagree; 5 = strongly agree) corresponding to “I think I will be able to find positive interaction between my time at work and my time caring for my family.” This variable will serve as an indicator of having a positive attitude (enrichment) to the relationship between work and family.

Home centered is a dummy coded variable, in which, to the standard question about the ideal family (“There are many ways to distribute the work and family responsibilities between a couple. If money were not a problem for you, your ideal household would be closer to a family where [...]”), the respondent prefers a family model in which he/she would be a stay-at-home partner or would take on a less demanding job (and be responsible to a greater extent for family responsibilities and childcare).

Finally, in the regression analysis we controlled for the effect of Age (age in years), and several coded dummy variables (1 = yes; 0 = no): Female; Immigrant; Foreign student; Business administration; Law; Economics; Other fields; and Final year bachelor’s or master’s. We also considered as a control variable Mobility, obtained from these two items: “I would move to another part of the country if it would help me progress in my career” and “I would move anywhere in the world if it would help me progress in my career” (Cronbach’s alpha = .742).

Descriptive Analysis

Table 1 shows the average scores of the different variables. In the first place, it appears that in many cases there is a statistically significant difference between the average values for female and male students.

Table 1.

Differences by Gender in the Variables Used in the Anaysis

The dependent variable, EI, shows higher values among Spanish students than among Icelandic students. Additionally, in the two cases there is a noticeable gap between the mean scores of males and females, the latter being larger in the case of Iceland (the average score of male students is 13.3% higher than that of female students in the case of Spain and 19.9% higher in the case of Iceland). These results (statistically significant) are in line with those obtained in the literature on entrepreneurial intentions and by Global Entrepreneurship Monitor (Kelley et al., 2012).

There are no gender differences in Confidence about work abilities.

The two variables related to the work–family balance present gender differences. In the case of Positive work–care interaction, the mean scores of female students are slightly higher than those for male students in both Spain and Iceland (and those differences are statistically significant), which appears to support hypothesis 2 (more female than male students have a positive attitude (enrichment) to the relationship between work and family). However, there is a much greater gender difference in the variable Home centered. Indeed, 8.5% of female Spanish students and 10.5% of female Icelandic students would prefer a family model in which they would be a stay-at-home partner or would take on a less demanding job, whereas for male Spanish students this figure was 2.9% and for Icelandic 2.8%. This last result supports the first part of hypothesis 3.

The indicator Leadership aspirations is slightly higher (and statistically significant) for Spanish males rather than for Spanish females. In Iceland, there is no gender difference in this variable.

As for the correlations among the different variables, in Tables 2–3 a comparison is made between those obtained for the subsamples of female and male students. Many significant relationships between the variables were the same for both males and females. For example, as posited in hypothesis 5, in the case of Spain there is a positive correlation between EI and Parental Income, both for female (r = .163; p = .000) and male students (r = .187; p = .000); in the case of Iceland these correlations are r = .118 for female students and r = .098 for male. Something similar happens with the positive correlations obtained between EI and Willingness risk (hypothesis 6). The positive correlations between EI and “Leadership” (r = .251 for females and r = .261 for males in the case of Spain; r = .270 for females and r = .300 for males in the case of Iceland) also would appear to confirm hypothesis 7 (many of the skills and attitudes identified in entrepreneurship and leadership are very similar, so greater leadership should be positively related to the inclination to entrepreneurship).

Table 2.

Intercorrelations of Variables (Females)

Table 3.

Intercorrelations of Variables (Males)

However, some interesting gender differences are also to be found. For example, in the Spanish sample, correlation between EI and Mother entrepreneur is higher for female students (r = .110; p = .001) than for male students (r = .076; p = .055) and something similar happens with Father entrepreneur and male students (hypotheses 4 and 4b).

Also in the case of the Spanish sample, the correlations between EI and Confidence about work abilities are positive and statistically significant (hypothesis 1), but there are gender differences between these coefficients (.150 for females and .214 for males) pointing to hypothesis 1b.

Moreover, with respect to the variables related to work–family balance, in both Spain and Iceland the correlation between EI and Positive work–care interaction is positive and significant in both sexes (female and male students), but the coefficient is higher for female students (hypothesis 2b); in both Spain and Iceland, the correlation between EI and Home centered is negative and significant only in the case of females (hypothesis 3).

Regression Analysis

In Table 4 (Spanish sample) and Table 5 (Icelandic sample) three stages were followed. In the first stage, the regression of the dependent variable was calculated on the basis of all independent variables except Positive work–care interaction and Home centered. In the second stage, a new regression model was obtained as a result of incorporating these two independent variables (related to work–family balance) into the first one, with the aim of evaluating to what degree their addition to the model improved the first adjustment carried out. And in the third stage, a series of statistically significant interaction effects were added, combining some of the variables with the dummy variable Women.

Table 4.

Three Models of Linear Regression and the Oaxaca Decomposition Technique (Spanish sample) (Women and Men)

Additionally, for each of the two samples two more regressions were carried out (not shown for reasons of space), one for the subsample of females and the other for the subsample of males, including in both cases all the independent variables. From this we were able to do an exercise with the Oaxaca decomposition technique (the results of which are presented at the end of Tables 4 and 5).

Table 5.

Three Models of Linear Regression and the Oaxaca Decomposition Technique (Icelandic sample: Women and Men)

Hypothesis 1 held that having more self-confidence in job abilities would be positively associated with the inclination to entrepreneurship. In this context, according to regressions 1, 2, and 3 in Table 4, among Spanish students the variable Confidence about work abilities has a positive and statistically significant relationship with EI, after controlling for the other independent variables, so this hypothesis seems to be confirmed in the case of the Spanish sample. In turn, when in regression 3 an interaction between Confidence about abilities work and Women is introduced, a negative and statistically significant effect is obtained (the variable Women seems to play a moderating effect). Indeed, in the regression of Spanish males (not presented in the article), the positive coefficient corresponding to Confidence about work abilities is much higher (B = .357; p = .000) than the one obtained in the regression of Spanish females (B = .120; p = .103). The EI of Spanish male students seems to be more sensitive to the effect of having higher self-confidence. Put another way, in the Spanish sample (as we showed in Table 1), the levels of self-assessment of job abilities were similar for female and male students, but since entrepreneurship is primarily associated with male characteristics, according to a double standards approach, Spanish female students would require (on average) a higher level of self-confidence about their abilities in the world of work in order to consider themselves as entrepreneurial (hypothesis 1b).

In the second regression in Tables 4 and 5, two independent variables related to work–family balance are added. In both cases, the explanatory capacity of the model rises, although it is higher in the case of the Spanish sample.

In the Spanish sample Positive work–care interaction has a positive and statistically significant effect on EI. This may confirm that having a positive attitude (enrichment) about work–family relationships is positively related to the inclination to entrepreneurship. However, when an interaction between Positive work–care interaction and Women is introduced in regression 3, a positive and statistically significant effect is obtained (the variable “Women” again seems to play a moderating effect, in this case positive). In fact, in the Spanish regression (not offered) corresponding to female students, a positive relation is obtained between Positive work–care interaction and EI (B = .319; p = .002), whereas in the regression corresponding to male students no such relationship exists (B = .013; p = .908). This result seems to confirm hypothesis 2b for the Spanish sample (female students are more sensitive to the enrichment perspective of work–family relationships, and so the positive relation with this attitude and the inclination to entrepreneurship is more intense among them).

Moreover (in terms of a work–family conflict perspective), the variable Home centered (preferring a family model implying that the respondent be a stay-at-home spouse or take on a less demanding job) shows a clear negative association with EI (hypothesis 3) only in the case of female students (as was previously stated, this is probably due to the fact that the number of males answering this option was very small). Indeed, in the case of the Spanish sample, regressions 2 and 3 in Table 4 show negative and statistically significant estimated coefficients for Home centered, whereas in the regression corresponding to females (not shown) a significant association is obtained (B = -.603; p = .032), but this is not the case in males (B = -.734; p = .169). Icelandic sample regressions 2 and 3 in Table 5 show non-significant estimated coefficients for Home centered, but when in regression 3 an interaction between Home centered and Women is introduced, a negative and statistically significant effect is obtained (in fact, in the regression corresponding to female Icelandic students, a significant association is obtained, B = -.905; p = .041, while in the one for males this does not occur, B = 1.395; p = .173).

Concerning role models, hypothesis 4 stated that family precedents in entrepreneurship could be a positive determinant of the inclination to entrepreneurship. For the Spanish sample, in regressions 2 and 3 of Table 4 the results obtained for the variables Mother entrepreneur and Father entrepreneur point in this direction (although only the results for Father entrepreneur were significant). But more interestingly, when the regressions are made for the subsamples of females and males (not shown), some evidence is provided in favor of hypothesis 4b (having an entrepreneur mother has a stronger influence on daughters’ inclination to entrepreneurship while having an entrepreneur father influences sons more). Indeed, in the regression corresponding to females, a positive and statistically significant coefficient was obtained for Mother entrepreneur (B = .507; p = .034) and not for Father entrepreneur (B = .059; p = .734), and something equivalent happened with Father entrepreneur in the corresponding regression of male students.

In all the regressions in Table 4 (Spanish sample) and in one regression in Table 5 (Icelandic sample) a positive and statistically significant coefficient is obtained for Parental income (this result was very similar for the female and male subsamples), confirming hypothesis 5 (when the parental income level is high, the student has a perception of being less subject to financial constraint, and this is related positively to the inclination to entrepreneurship).

According also to all regressions in Tables 4 and 5, there is an important and positive relation between tolerance for risk and the inclination to entrepreneurship (hypothesis 6 is confirmed). Furthermore, in Spain the estimated coefficient (not shown) for women (B = .666; p = .000) is quite similar to that of men (B = .738; p = .000), whereas in Iceland the coefficient for women (B = 1.116; p = .000) is higher than that of men (B = .477; p = .103). A different point is that, as shown in Table 1, in our samples female students, on average, have a lower tolerance for risk than male students (hypothesis 6b).

Concerning the relation (or the similarities) between entrepreneurship and leadership, Tables 4 and 5 show that in all the regressions a positive and statistically significant coefficient was obtained for leadership (hypothesis 7). Nonetheless, in that case the estimated coefficient for Spanish women (B = .801; p = .000) is higher than that of Spanish men (B = .435; p = .006), whereas in the Icelandic sample the opposite occurs.

Two final considerations are made about the dummy variable Female and the Oaxaca decomposition technique.

The dummy variable Female has a negative and significant effect in the three stages (regressions) of Table 4 (Spain) and Table 5 (Iceland). This means that despite being controlled by a series of determinants of EI, female students are significantly less likely than male students to consider themselves as very entrepreneurial.

As was shown in Table 1, in our sample the male–female gap (male–female difference in terms of female score) in Entrepreneurship inclination was 13.3 percentage points in Spain and 19.9 in Iceland. In terms of the two estimated models in each country for female and male students (not offered), a question of interest is to discern how much of these differentials or gender gaps in entrepreneurship inclination can be attributed to differences between female and male students in the independent variables (determinants) included in the models (and listed in Table 1 as control variables and main independent variables) and how much can be attributed to differences in the estimated regression coefficients (including the intercepts).

In this sense, after estimating two models for two different groups of analysis (in our case, male and female students), the Oaxaca decomposition technique (Oaxaca, 1973; Blinder, 1973; Neumark, 1988) decomposes the dependent variable differential (usually the wage differential; in our case the entrepreneurial male–female differential) into two components: A portion that arises because the two groups, on average, have different characteristics (e.g., average differences in age, in the fields they are studying, in Willingness to risk scores, etc.) when both groups have the same regression coefficients (explained component); and a portion that arises because the two groups have different regression coefficients (termed unexplained component in the wage gap literature).

The results obtained for Spain and Iceland with the Oaxaca decomposition are shown at the ends of Tables 4 and 5. For the Spanish sample, the difference between the mean scores corresponding to male and female students in EI is .776 points (the score of males is 13.3% higher than that of females); from this difference, .384 points (49.5%) are attributed to the difference in the characteristics of female and male students in the independent variables listed in Table 1; and .391 points (50.5%) are attributed to the effect of differences in the estimated coefficients (including the intercept). For the Icelandic sample, the male–female difference in EI is .970 points (the score of males is 19.9% higher than that of females); 24.6% of this difference is attributed to the difference in the characteristics of female and male students in the independent variables and 75.4% to the effect of differences in the estimated coefficients (unexplained component).