The Impact of Gender on Entrepreneurial Intention in a Peripheral Region of Europe: A Multigroup Analysis

: The last two decades were characterized by an increase in attention on entrepreneurship. An emerging trend in the literature is associated with the different contributions that genders may have regarding entrepreneurship and entrepreneurial intentions. This paper assesses gender impact on entrepreneurial intention in a peripheral region of Europe. The methodology used is quantitative and based on a sample of 1114 observations. The obtained results suggest that women’s perceived behavioral control does not have a less direct positive impact on entrepreneurial intention when compared to men’s. Women’s personal attitude has a less direct positive impact on entrepreneurial intention when compared to men’s, as social norms, on the one hand, have a less direct positive impact on women’s personal attitude when compared to men’s, and, on the other hand, have a less direct positive impact on women’s perceived behavioral control when compared to men’s. Finally, it was also veriﬁed that social norms have a less positive indirect impact through personal attitude and perceived behavioral control on women’s entrepreneurial intention when compared to men’s. These ﬁndings contribute to enlighten the literature by strengthening the theoretical framework on women’s entrepreneurial intention in peripheral regions. Regarding practical contributions, suggestions are addressed to public decision makers, universities, and civil society, to adopt practices to increase entrepreneurship in women. This research is original because it is the ﬁrst to perform this study in Portugal, a peripheral region of Europe.


Introduction
Entrepreneurship is widely recognized by all regional actors (government, academia, and industry) as a leading driver of the socio-economic development of countries and their regions; consequently, its promotion has been growing (Lopes et al. 2018;Maes et al. 2014;and Molino et al. 2018). Entrepreneurship is strongly affected by innovation; it is essential that regions, through their policymakers, invest in technology and improve economic structures, which, in turn, causes employment to grow and increases the wealth created by these economies (Mei et al. 2016;Gurel et al. 2010;Nabi and Holden 2008;and Gomes et al. 2021).
In this context, for the last two decades, the interest of researchers to study entrepreneurship has been growing (Lopes et al. 2021a). This growth was accompanied governments and academia have to pay special attention to women's entrepreneurial intentions, as they have specific characteristics (Adamus et al. 2021). Consequently, the analysis of the entrepreneurial intention across the board may not be suitable (Salavou et al. 2021;and Westhead and Solesvik 2016). Nowadays, women are increasingly adopting an important role in entrepreneurship, despite their contributions being lower than those by men (Rubio-Bañón and Esteban-Lloret 2016;and Vracheva and Stoyneva 2020). Moreover, we are witnessing a vigorous promotion by the European Commission on the importance of entrepreneurship throughout the different member states. Female entrepreneurship is emerging as a high-priority topic, as demonstrated in several European Union reports (Salavou et al. 2021;and Miranda et al. 2017).
The literature on entrepreneurship has identified that this is not an exclusively economic phenomenon. Entrepreneurship is also considered a cultural phenomenon; consequently, not all results can be generalized outside the context of their respective study (Adamus et al. 2021;Borges et al. 2021;Gomes et al. 2021;and Lopes et al. 2021b). According to Adamus et al. (2021), economies that have working women are less likely to face economic downturns, that is, families with a working mother are less likely to experience poverty. Thus, it is important to study entrepreneurial intention from a gender perspective, with a significant sample in other contexts to generalize the results (Vamvaka et al. 2020;Mei et al. 2016;Atitsogbe et al. 2019;Maes et al. 2014;and Borges et al. 2021) in similar cultural regions. That said, the main objective of this research is to assess the impact of gender on entrepreneurial intention in a peripheral region of Europe. The methodology used is quantitative and based on a sample of 1114 observations, located in Portugal, a peripheral region of Europe.
The study of peripheral regions is relevant when considering the GDP per capita, as, economically, these regions are considered to be less developed. Generally, these regions have low levels of science and education, which translate into a low level of human capital. This leads to low incomes among the population and a limited amount of local policymakers. Peripheral regions have a low level of infrastructure as well as limited availability of territorial resources (IT and communications) when compared to more developed countries (Lewandowska et al. 2021;and Lopes et al. 2020b).
Based on Worldometer data, Portugal has 10,158,654 inhabitants on 13 October 2021, which corresponds to 0.13% of the world population and 1.36% of the total population of the European Union. About 67% of the population is urban, which corresponds to 6,775,807 inhabitants. Portugal is located on the periphery of Europe and has an area of 91,590 km 2 (Worldometer 2021).
This research is ordered as follows. The first Section introduced the subject and problem under study. In Section 2, an extensive review of the available literature on entrepreneurial intentions and TPB is performed, as well as the formulation of the hypotheses. In Section 3, the methodology and data collection are described. In Section 4, the results are exhibited, discussed, and compared against the current literature. In Section 5, the conclusions are disclosed, revealing the most significant findings and contributions to theory and practice as well as suggestions for forthcoming investigations.

Theory of Planned Behaviour and Entrepreneurial Intentions
Intention can be understood as a mindful, deliberate, and premeditated mental state that takes place before action and consents to the straight consideration to certain types of behaviors, such as the creation of a business (Esfandiar et al. 2019). In the entrepreneurship framework, entrepreneurial intention can be considered as a "self-recognized conviction" by any individual willing to start a new business venture, thus becoming a crucial element to understand the procedure of forming a new business (Farrukh et al. 2018;Liñán et al. 2011;and Ridha et al. 2017). Entrepreneurial intention can also be understood as a status of mind that leads an individual to choose his/her job instead of working for other people (Karimi et al. 2016;and Lopes et al. 2021b).
The TPB is an intention/behavior model that has been extensively used in entrepreneurship investigation, whereby the effectiveness and capability to envisage entrepreneurial intention was already demonstrated in several entrepreneurship studies (Karimi et al. 2016). The model by Ajzen (1991) enlightens how the social and cultural environment distresses human behavior. According to this model, the individual's intentions are the result of three determinants: personal attitude, perceived behavioral control, and social norms.
Personal attitude is related to the individual's assessment of a certain behavior or action characterized as being beneficial, indicating a favorable or unfavorable personal assessment of the intention to become an entrepreneur (Ajzen 1991;Enkel and Bader 2016;Krueger and Carsrud 1993;and Liñán et al. 2011). Several studies Liñán and Chen 2009;Martínez-González et al. 2019;and Ruiz-Rosa et al. 2020) found a positive relationship between personal attitude and the individual's entrepreneurial intention.
The perceived behavioral control characterizes the tendency to perform and the perceived likelihood of exhibiting a particular behavior is explained as the self-efficacy of the individual (Krueger and Carsrud 1993). This is revealed in the greater or lesser difficulty that an individual feels when performing an action, related to the ability to control their behavior (Ruiz-Rosa et al. 2020). It is considered that the self-perception of an individual capacity to perform a certain action will influence significantly the intention to execute such action, which can be translated into a positive relationship between perceived behavioral control and the individual's entrepreneurial intention (Krueger et al. 2000;Liñán and Chen 2009;Ruiz-Rosa et al. 2020;and Smith and Woodworth 2012).
Social norms refer to the community pressure exerted by individuals' opinions about the projected behavior, depending on the anticipation of support from other important people (Krueger and Carsrud 1993;Lortie et al. 2017;and Lortie and Castogiovanni 2015). Furthermore, it refers to the perception that "reference people" may or may not approve of the individual's decision to become an entrepreneur (Ajzen 2002).
Some authors consider it is sensible to assume a positive relationship between this variable and the entrepreneurial intention as they understand that entrepreneurs are affected by comments of people related to their contiguous environment and related to their entrepreneurial intentions (Ruiz-Rosa et al. 2020;and Tiwari et al. 2017). However, some studies did not find a significant relationship between social norms and the entrepreneurial intention of individuals (Autio et al. 2001;Carsrud and Brännback 2011;Krueger et al. 2000;and Liñán and Chen 2009).

TPB Factors Mediating the Effect of Gender on Entrepreneurial Intentions
Numerous studies (Davidsson and Honig 2003;Ferri et al. 2018;Karimi et al. 2013;Krueger et al. 2000;Liñán and Chen 2009;Lortie et al. 2017;Maes et al. 2014;Miranda et al. 2017;Veciana et al. 2005;and Wang and Wong 2004) analyzed entrepreneurial intentions according to gender. Krueger et al. (2000) state that gender roles enhance the understanding of entrepreneurial intentions, applied to the differences between men and women in this context. Moreover, the study of Wang and Wong (2004) concluded that gender, experience, and level of education are significant factors to explain the individuals' entrepreneurial intentions. Ferri et al. (2018) referred that, in general, women face gender-related obstacles to obtain trustworthiness and resources in the business world. This is not because of lack of skills, but due to reasons associated with their personal and professional life (e.g., breaks for maternity leave, difficulty in obtaining loans, and less capital to invest). Social recognition is also very important to them.
An analysis of the TPB factors in a detailed manner reveals that the study of Karimi et al. (2013) demonstrates that males are driven by instrumental factors (since personal attitude is very important for men), while females are more motivated by social factors (social norms are significant for women). In this study, perceived behavioral control was the factor that contributed the most to the prediction of the entrepreneurial intentions of both genders.
As mentioned above, several studies have shown that personal attitude is positively related to entrepreneurial intentions, both in women (Ferri et al. 2018) and men (Karimi et al. 2013). The perceived behavioral control factor also has a positive effect on women's entrepreneurial intentions, as it explains that they will translate their intentions into actions (Ferri et al. 2018). The study of Yang (2013) disclosed that, when compared to men, women are negatively correlated with personal attitude, perceived behavioral control, and social norms.
Concerning social norms, and since women's role in entrepreneurial intentions is inconclusive, Liñán and Chen (2009) proposed that social norms exert their influence on entrepreneurial intentions through personal attitude and perceived behavioral control. These authors perceived a positive and noteworthy influence between male gender and personal attitude and perceived behavioral control; that is, according to their study, men are more captivated with entrepreneurial behavior and consider themselves more competent than women. Maes et al. (2014) also verified that social norms indirectly influence the entrepreneurial behavior of individuals, through a personal attitude and perceived behavioral control. The study of these authors showed that women are less attracted to an entrepreneurial career and consider themselves less suitable for entrepreneurship, concluding that the mediating role of personal attitude and perceived behavioral control factors can explain the fact that women have less entrepreneurial intentions than men.

Entrepreneurship in Peripheral Regions
Institutional contexts influence "entrepreneurial outcomes", as the formal and informal institutions inherent to each region influence the decisions of individuals to pursue entrepreneurial activity (Acs et al. 2008;Williams and Vorley 2014;Estrin et al. 2016;and Gherhes et al. 2018).
Formal institutions include the norms, rules, and legislation inherent to the economic activity and the legal framework of society, while informal institutions refer to the customs, principles, and values rooted in society (Acs et al. 2008;andVorley 2014, 2017).
As Martin and Sunley (2006), Dennis (2011), and Martin (2011) mentioned, formal institutions promote entrepreneurship, resilience, and economic growth directly, while informal institutions, by highlighting "history matters" and "carriers of history", promote an "institutional hysteresis" that inhibits change and promotes "self-reproduction". The low innovation that affects peripheral regions, characterized by low dynamism, influences their economic potential, which, in turn, is determined by the low entrepreneurial potential of these regions. Asheim et al. (2011) characterized peripheral regions as having a low-level innovative activity, due to the lack of dynamic companies and knowledge-generating organizations, a "thin" and less specialized structure of knowledge providers, educational institutions, and the poorly developed networks of these providers. Leick (2020) also conducted an investigation applied to peripheral rural regions and concluded that the various agents that operate in these regions (including entrepreneurs) can promote change and provide economic development in the region.
In a study applied to the Canary Islands, in Spain, based on the theory of planned behavior and which considered motivation, opportunity, and ability theory, García-Rodríguez et al. (2017) concluded that policies to promote entrepreneurship and education programs for entrepreneurship should focus on improving entrepreneurial motivation using intensive pedagogical strategies that promote creativity, with a focus on recognizing opportunities and solving problems (problem-solving).
With a focus on peripheral places, Xu and Dobson (2019) tried to understand the challenges associated with the creation of entrepreneurial ecosystems in these regions, including their small size, the distance from large urban centers, and the scarcity of resources. They concluded that peripheral areas can benefit from creating their entrepreneurial ecosystem Soc. Sci. 2021, 10, 415 6 of 20 if a holistic collaborative approach, integrating finance, talent, socio-cultural environment, infrastructure, markets, and policies, is taken into account.

Theoretical Model and Hypotheses
Since the main objective of this study is to evaluate the impact of gender on entrepreneurial intention in a peripheral region of Europe, we started our approach using the existing model of Liñán and Chen (2009) on the differences in entrepreneurial intention at the level of the factors, and including the sociodemographic variable gender as performed by Maes et al. (2014). Taking into account the TPB, personal attitude, perceived behavioral control, and social norms that are used as determinants of entrepreneurial intention, we analyzed how and why these three factors arbitrate the gender impact on entrepreneurial intention, that is, how gender affects entrepreneurial intention, directly and indirectly. Considering the referred objective, the following hypotheses were formulated: Hypothesis 1a (H1a). Being a woman has a negative impact on the personal attitude of being an entrepreneur.

Hypothesis 1b (H1b).
Being a woman has a negative impact on the perceived behavioral control of being an entrepreneur.

Hypothesis 1c (H1c).
Being a woman has a negative impact on the social norms related to entrepreneurial activity.
Hypothesis 2 (H2). Women's personal attitudes have less direct positive impact on entrepreneurial intention when compared to men's.

Hypothesis 3 (H3).
Women's perceived behavioral control has less direct positive impact on entrepreneurial intent when compared to men's. including their small size, the distance from large urban centers, and the scarcity of resources. They concluded that peripheral areas can benefit from creating their entrepreneurial ecosystem if a holistic collaborative approach, integrating finance, talent, sociocultural environment, infrastructure, markets, and policies, is taken into account.

Theoretical Model and Hypotheses
Since the main objective of this study is to evaluate the impact of gender on entrepreneurial intention in a peripheral region of Europe, we started our approach using the existing model of Liñán and Chen (2009) on the differences in entrepreneurial intention at the level of the factors, and including the sociodemographic variable gender as performed by Maes et al. (2014). Taking into account the TPB, personal attitude, perceived behavioral control, and social norms that are used as determinants of entrepreneurial intention, we analyzed how and why these three factors arbitrate the gender impact on entrepreneurial intention, that is, how gender affects entrepreneurial intention, directly and indirectly. Considering the referred objective, the following hypotheses were formulated: Hypothesis 1a (H1a). Being a woman has a negative impact on the personal attitude of being an entrepreneur.

Hypothesis 1b (H1b). Being a woman has a negative impact on the perceived behavioral control of being an entrepreneur.
Hypothesis 1c (H1c). Being a woman has a negative impact on the social norms related to entrepreneurial activity.

Hypothesis 3 (H3). Women's perceived behavioral control has less direct positive impact on entrepreneurial intent when compared to men's.
Hypothesis 4 (H4). Social norms have less direct positive impact on the personal attitude of women when compared to men's.

Hypothesis 6b (H6b). Social norms have less indirect positive impact through perceived behavioral control of women's entrepreneurial intention when compared to men's.
The structural model and hypotheses of this research are described in Figure 1.

Methodology
This research uses a quantitative methodology, similarly to most studies on the subject of entrepreneurship, and according to the bibliometric review performed by Hlady-Rispal and Jouison-Laffitte (2014). This type of methodology has some advantages, such as allowing the exploration of relationships between variables, generalizing the results, and applying analysis methods and techniques to other samples, amongst others (Queirós et al. 2017).
The sample used contained 1114 observations. It was a convenience sample, which was collected through an online questionnaire on Twitter and Facebook social networks that was distributed among the population of Portugal between April 2017 and December 2020. The applied questionnaire was previously used in the study of Lopes et al. (2020b) as well as in Liñán et al. (2011). The questionnaire consisted of five groups of questions: (G1) entrepreneurial intention with four questions; (G2) perceived behavioral control with four questions; (G3) personal attitude with three questions; (G4) social norms with two questions; and (G5) sociodemographic characteristics with four questions related to age, sex, residence, and employment experience. The groups of questions G1 to G4 use a 7-point Likert scale, in which 1-I totally disagree and 7-I totally agree. The sociodemographic variable of gender took the values of 0 for men and 1 for women. Table 1 shows the main statistics of the group of questions of sociodemographic characteristics (G5). Briefly, 65.6% of respondents were women and 34.4% were men. The average age was 26.53 years, with 76.8% of respondents being under 30 years old. They were mostly residents in mainland Portugal (83.8%) and were studying toward or had an undergraduate degree (67.6%). Regarding professional experience, most have had at least one professional experience (68.9%). Concerning the groups of questions G1 to G4, the mean and standard deviation of the answers are described in Table 2.
In the entrepreneurial intention group, on average, the issues with the highest agreement amongst respondents were the commitment of all efforts to start and manage a new business (A06-4.89) and the determination to create an enterprise in the future (A13-4.47); in the perceived behavior control questions, the questions related to the perception of success (A14-4.64) and the perception of the ability to control the process of creating a new business (A07-4.6) were the ones that generated the highest agreement, and the Soc. Sci. 2021, 10, 415 8 of 20 respondents demonstrated, on average, that they have little knowledge of all the practical details to start a new business (A20-2.93). In terms of personal attitude, most respondents agree that they would like to open a new business (A10-5.74), and that being an entrepreneur would give them great satisfaction (A15-5.07). In terms of social norms, the respondents showed greater agreement with the fact that entrepreneurial activity is valued in the country, despite the risks it involves (4.76).

Methods of Analysis
To test the relationships between the variables available in the structural model and the hypotheses (Figure 1), a partial least squares (PLS) model in Smart PLS 3.0 software was used  to perform a multigroup analysis. Multigroup analysis allows the comparison of different models between different groups that were created according to the gender of the respondents. The PLS is a variance-based method and the objective is to maximize the explained variance of endogenous latent variables (constructs), assuming the non-normal distribution of data. As the data was collected by questionnaire, there is no normality in the data. Since the PLS is a structural equation model, it is applied in two phases (McDonald and Ho 2002;and Maes et al. 2014): (1) check the adequacy of the measuring instruments, and (2) test the direct and/or indirect relationships between the latent variables.
The first step of the PLS method is to apply the PLS algorithm to the structural model, resulting in the PLS model as shown in Figure 2 for model 1. The squares represent the indicators of each latent variable and the circles represent the latent variables. Outer loadings, that is, the contribution of each indicator to explain the latent variable, are represented in the lines that make the connection between the latent variables and the indicators; according to Hair et al. (2019), they must be greater than 0.50. Inside the circles are the R-Square values.
Soc. Sci. 2021, 10, x FOR PEER REVIEW 9 of 21 sented in the lines that make the connection between the latent variables and the indicators; according to Hair et al. (2019), they must be greater than 0.50. Inside the circles are the R-Square values. To test the adequacy of the model for the data, a confirmatory analysis was performed on the latent variables: entrepreneurial intention, personal attitude, social norms, and perceived behavioral control. Smart PLS 3.0 provides three evaluation measures of the fit model: the SRMR (standardized root mean of approximation), the NFI (normed fit index), the RMS theta (root mean squared residual covariance matrix of the outer model residuals). We started with a model (model 1) for all the observations (1114), that is, without differentiating the gender (male/female). The structural model has a good fit for data because: (1) SRMR measures the difference between the observed correlation and the implicit correlation matrix in the model, allowing the assessment of mean discrepancies between observed and expected correlations (Hair et al. 2019) where the reference value is 0.08 (Hu and Bentler 1998). In model 1, the SRMR is 0.071 below the cut-off value (0.08).
(2) NFI results from the calculation of the Chi² value from the proposed model and compares it with a meaningful benchmark. The value obtained must be between 0 and 1; model 1 has an NFI of 0.806, being in the referred range of values. (3) RMS theta evaluates the degree of correlation of the model's residuals, and the closer to zero, the better the model's fit. According to Henseler et al. (2014), RMS theta values below 0.12 indicate a well-fitted model, while higher values indicate a lack of fit. In model 1, the RMS theta has the value of 0.113 and, therefore, the model is considered to be adjusted.
For models 2 and 3, which refer, respectively, to the men's and women's models, the PLS algorithm was also applied to the structural model, resulting in the PLS models shown in Figures 3 and 4. To test the adequacy of the model for the data, a confirmatory analysis was performed on the latent variables: entrepreneurial intention, personal attitude, social norms, and perceived behavioral control. Smart PLS 3.0 provides three evaluation measures of the fit model: the SRMR (standardized root mean of approximation), the NFI (normed fit index), the RMS theta (root mean squared residual covariance matrix of the outer model residuals). We started with a model (model 1) for all the observations (1114), that is, without differentiating the gender (male/female). The structural model has a good fit for data because: (1) SRMR measures the difference between the observed correlation and the implicit correlation matrix in the model, allowing the assessment of mean discrepancies between observed and expected correlations (Hair et al. 2019) where the reference value is 0.08 (Hu and Bentler 1998). In model 1, the SRMR is 0.071 below the cut-off value (0.08).
(2) NFI results from the calculation of the Chi 2 value from the proposed model and compares it with a meaningful benchmark. The value obtained must be between 0 and 1; model 1 has an NFI of 0.806, being in the referred range of values. (3) RMS theta evaluates the degree of correlation of the model's residuals, and the closer to zero, the better the model's fit. According to Henseler et al. (2014), RMS theta values below 0.12 indicate a well-fitted model, while higher values indicate a lack of fit. In model 1, the RMS theta has the value of 0.113 and, therefore, the model is considered to be adjusted.
For models 2 and 3, which refer, respectively, to the men's and women's models, the PLS algorithm was also applied to the structural model, resulting in the PLS models shown in Figures 3 and 4.
Subsequently, the same tests were performed for the group of men and women, separately. The model in which the respondents were only men, model 2 (SRMR = 0.074; NFI: 0.802; RMS theta: 0.115), and the model in which the respondents were only women, model 3 (SRMR = 0.072; NFI: 0.792; RMS theta: 0.115), proved to be good fit for the data.  Subsequently, the same tests were performed for the group of men and women, separately. The model in which the respondents were only men, model 2 (SRMR = 0.074; NFI: 0.802; RMS theta: 0.115), and the model in which the respondents were only women, model 3 (SRMR = 0.072; NFI: 0.792; RMS theta: 0.115), proved to be good fit for the data.
Internal consistency is evaluated by composite reliability and Cronbach's Alpha (Table 3). In all models, these indicators have values greater than 0.70 (reference values according to Hair et al. (2019)) and, as such, internal convergence is "satisfactory to good". The AVE is the sum of the square external loads of the indicators associated with the latent variables divided by the number of indicators, the reference value being 0.50. The AVEs of the latent variables in all models is greater than 0.50, that is, on average, the latent variable explains more than half of the variance of its indicators.   Subsequently, the same tests were performed for the group of men and women, separately. The model in which the respondents were only men, model 2 (SRMR = 0.074; NFI: 0.802; RMS theta: 0.115), and the model in which the respondents were only women, model 3 (SRMR = 0.072; NFI: 0.792; RMS theta: 0.115), proved to be good fit for the data.
Internal consistency is evaluated by composite reliability and Cronbach's Alpha (Table 3). In all models, these indicators have values greater than 0.70 (reference values according to Hair et al. (2019)) and, as such, internal convergence is "satisfactory to good". The AVE is the sum of the square external loads of the indicators associated with the latent variables divided by the number of indicators, the reference value being 0.50. The AVEs of the latent variables in all models is greater than 0.50, that is, on average, the latent variable explains more than half of the variance of its indicators. Internal consistency is evaluated by composite reliability and Cronbach's Alpha (Table 3). In all models, these indicators have values greater than 0.70 (reference values according to Hair et al. (2019)) and, as such, internal convergence is "satisfactory to good". The AVE is the sum of the square external loads of the indicators associated with the latent variables divided by the number of indicators, the reference value being 0.50. The AVEs of the latent variables in all models is greater than 0.50, that is, on average, the latent variable explains more than half of the variance of its indicators.
To assess the discriminant validity of the model, the Fornell-Larcker criterion was used (Table 4). According to this criterion, each AVE of the latent variables (elements on the main diagonal that are in bold) must be greater than all square correlations of the latent variables (off-diagonal elements), thus establishing the discriminant validity of each of the four latent variables (Henseler et al. 2015). Once all models meet these conditions, then the model ensures the discriminant validity of the constructs. Once the good fit, internal consistency, and discriminant validity of the models were validated, the next step was to test direct and/or indirect relationships between the latent variables, according to the structural model and the formulated hypotheses (Figure 1).

Results and Discussion
The impact of gender (male = 0 or female = 1) on the explanatory factors of the entrepreneurial intention was tested for the entire sample through model 1: personal attitude, perceived behavioral control, and social norms with the application of the PLS model obtained in Figure 2 from a bootstrap analysis (Hair et al. 2019), as shown in Figure 5.
With a statistical significance of 95% (Table 5), we conclude that gender has a negative impact (β = −0.148) on personal attitude, meaning that being a woman has a negative influence on the attitude towards entrepreneurship, thus confirming H1a. We also conclude that gender has a negative impact (β = −0.204) in the perceived behavioral control, that is, being a woman has a negative influence on the perceived behavior to be an entrepreneur, therefore confirming H1b. Gender is not significant to explain the social norms, although there is a positive impact of being a woman on the social norms and, therefore, H1c is rejected.
In this context, we can observe that, in this research, the findings are in line with the contributions of  Krueger and Carsrud (1993), Lortie and Castogiovanni (2015), Ruiz-Rosa et al. (2020), and Tiwari et al. (2017) are not confirmed regarding the influence of social norms on entrepreneurial intention. These authors assume that it is sensible to anticipate a positive relationship between this variable and entrepreneurial intention, as they consider that entrepreneurs are affected by comments from people related to their closest environment on their entrepreneurial intentions. Moreover, the non-confirmation of H1c is in line with the contributions of Autio et al. (2001), Brännback (2011), Krueger et al. (2000), and Liñán and Chen (2009), who did not find a significant relationship between the social norms and the entrepreneurial intention of individuals. With a statistical significance of 95% (Table 5), we conclude that gender has a negative impact (β = −0.148) on personal attitude, meaning that being a woman has a negative influence on the attitude towards entrepreneurship, thus confirming H1a. We also conclude that gender has a negative impact (β = −0.204) in the perceived behavioral control, that is, being a woman has a negative influence on the perceived behavior to be an entrepreneur, therefore confirming H1b. Gender is not significant to explain the social norms, although there is a positive impact of being a woman on the social norms and, therefore, H1c is rejected.
In this context, we can observe that, in this research, the findings are in line with the contributions of  Smith and Woodworth (2012) tell us that there is a positive relationship between perceived behavioral control and the individual's entrepreneurial intention. Nonetheless, the contributions of Krueger and Carsrud (1993), Lortie and Castogiovanni (2015), Ruiz-Rosa et al. (2020), and Tiwari et al. (2017) are not confirmed regarding the influence of social norms on entrepreneurial intention. These authors assume that it is sensible to anticipate a positive relationship between this variable and entrepreneurial intention, as they consider that entrepreneurs are affected by comments from people related to their closest environment on their entrepreneurial intentions. Moreover, the non-confirmation of H1c is in line with the contributions of Autio et al. (2001), Carsrud and Brännback (2011), Krueger et al. (2000), and Liñán and Chen (2009), who did not find a significant relationship between the social norms and the entrepreneurial intention of individuals.  Having assessed the impact of gender on the explanatory factors of entrepreneurial intention, a multigroup analysis was used to search for gender differences (male or female) in these explanatory factors. That is, we sought to understand how gender influences, through personal attitude, perceived behavioral control, and social norms, entrepreneurial intention. To achieve this, a bootstrap analysis was performed again for the structural model referring to men (model 2) and women (model 3), as shown in Figures 6 and 7. intention, a multigroup analysis was used to search for gender differences (male or female) in these explanatory factors. That is, we sought to understand how gender influences, through personal attitude, perceived behavioral control, and social norms, entrepreneurial intention. To achieve this, a bootstrap analysis was performed again for the structural model referring to men (model 2) and women (model 3), as shown in Figures 6 and 7.  All latent variables are significant (ρ = 0.000) for a statistical significance of 95%, and the results are show in Table 6.  All latent variables are significant (ρ = 0.000) for a statistical significance of 95%, and the results are show in Table 6. We can conclude that the personal attitude of women has a less direct positive impact (β = 0.579) in entrepreneurial intention when compared to men's (β = 0.636), thus confirming H2. These findings are in line with the contributions of Ferri et al. (2018) and Karimi et al. (2013), through which it is possible to verify that male individuals are driven by instrumental factors (since personal attitude is very important for men), while females are more motivated by social factors.
Women's perceived behavioral control has a direct positive impact (β = 0.352), stronger in entrepreneurial intention when compared to men's (β = 0.293), therefore rejecting H3. Hence, this research is in line with the contributions of Ferri et al. (2018), which tells us that the perceived behavioral control factor also has a positive effect on women's entrepreneurial intentions.
Social norms have a less direct positive impact (β = 0.176) on the personal attitude of women when compared to men's (β = 0.230), consequently confirming H4. Liñán and Chen (2009) suggest that the role of social norms influences entrepreneurial intentions through personal attitude and perceived behavioral control.
Social norms have a less direct positive impact (β = 0.260) in the perceived behavioral control of women when compared to men's (β = 0.194), therefore confirming H5. Social norms have a less positive indirect impact through personal attitude on women's entrepreneurial intention when compared to men's (β = 0.102 and β = 0.146, respectively), thus confirming H6a. Social norms have a less positive indirect impact through perceived behavioral control on women's entrepreneurial intention when compared to men's (β = 0.068 and β = 0.076, respectively), thus confirming H6b.
The study of Maes et al. (2014) stresses that women are less tempted to follow an entrepreneurial career and consider themselves less suitable for entrepreneurship, concluding that the arbitrating role of personal attitude and perceived behavioral control factors can explain the fact that women have less entrepreneurial intentions than men, which is in line with the present research and with the confirmation of hypotheses H4, H5, H6a, and H6b. Figure 8 shows the coefficients obtained in the estimation of models 1, 2, and 3 to test each of the hypotheses, and Table 7 exhibits a summary of the hypotheses discussed in this research.  Social norms have a less direct positive impact (β = 0.176) on the personal attitude of women when compared to men's (β = 0.230), consequently confirming H4. Liñán and Chen (2009) suggest that the role of social norms influences entrepreneurial intentions through personal attitude and perceived behavioral control.
Social norms have a less direct positive impact (β = 0.260) in the perceived behavioral control of women when compared to men's (β = 0.194), therefore confirming H5. Social norms have a less positive indirect impact through personal attitude on women's entrepreneurial intention when compared to men's (β = 0.102 and β = 0.146, respectively), thus confirming H6a. Social norms have a less positive indirect impact through perceived behavioral control on women's entrepreneurial intention when compared to men's (β = 0.068 and β = 0.076, respectively), thus confirming H6b.
The study of Maes et al. (2014) stresses that women are less tempted to follow an entrepreneurial career and consider themselves less suitable for entrepreneurship, concluding that the arbitrating role of personal attitude and perceived behavioral control factors can explain the fact that women have less entrepreneurial intentions than men, which is in line with the present research and with the confirmation of hypotheses H4, H5, H6a, and H6b. Figure 8 shows the coefficients obtained in the estimation of models 1, 2, and 3 to test each of the hypotheses, and Table 7 exhibits a summary of the hypotheses discussed in this research.

Hypotheses
Supported | Not Supported H1a: Being a woman has a negative impact on the personal attitude of being an entrepreneur Supported H1b: Being a woman has a negative impact on the perceived behavioral control of being an entrepreneur Supported H1c: Being a woman has a negative impact on the social norms related to entrepreneurial activity Not Supported

Conclusions
This article assessed the impact of gender on entrepreneurial intention in a peripheral region of Europe. It was possible to conclude that women's perceived behavioral control does not have less direct positive impact on entrepreneurial intention when compared to men's; women's personal attitude has less direct positive impact on entrepreneurial intention when compared to men's; social norms have less direct positive impact on women's personal attitude when compared to men's; and social norms have less direct positive impact on women's perceived behavioral control when compared to men's. Additionally, it is also concluded that social norms have less positive indirect impact through the personal attitude and perceived behavioral control of women's entrepreneurial intention when compared to men's.
On the theoretical side, this research contributes to the reinforcement of the theoretical framework on which the studies on entrepreneurial intention of gender are based, expanding the contributions on entrepreneurial intention to peripheral regions.
Although women constitute more than half of the world's population, they own fewer businesses than men (Kim 2007), are less inclined to follow an entrepreneurial career, and consider themselves less suitable for entrepreneurship (Maes et al. 2014). As we already explained, there are inequalities between genders that are related to the financial inclusion of women, time availability, education gap, and other legal issues. This raises the question of reducing gender inequalities (House et al. 2021) in entrepreneurship. Does this introduce entrepreneurial opportunities to women or are they a gender barrier?
To overcome this possible barrier, and as practical contributions, our research can serve as a base for policymakers, universities, further civil society entities, and other community stakeholders to develop and strengthen the necessary conditions based on gender equality to help improve entrepreneurship amongst women. Our research serves both genders commonly when addressing entrepreneurship, for example by providing adequate legislation, access to financial resources, and market opportunities.
This paper is original as it studies a large sample with 1114 observations. It is not easy to collect so many observations in peripheral countries. The sample can be larger on the majority of studies on the subject under investigation, but it is widely recognized that studies in peripheral regions are scarce.
As for suggestions for future investigations, it would be motivating to integrate other variables in the research, in line with the relevant literature, that may influence the entrepreneurial intention of both genders and in other peripheral regions of the globe. It would also be important to compare the obtained results concerning Portugal and other peripheral countries, where entrepreneurship is more interesting and captivating, and where the rates of business creation by entrepreneurs are higher. From this analysis, it will be possible to draw inferences on the strategies and policies to be adopted in the promotion of entrepreneurship in Portugal and the creation of employment through nascent entrepreneurship, taking into account the gender factor. Another possibility for future research is to perform longitudinal research, which has the advantage of showing the indicator's evolution over time and promote chronological comparisons. It would also be pertinent and important to study entrepreneurial intention regarding sustainability in general and specifically in women. Funding: This work was supported by the Foundation for Science and Technology "UIDB/04630/2020".

Institutional Review Board Statement:
The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Institutional Review Board.
Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.