“The mediating effect of entrepreneurial self-efficacy in entrepreneurial intention – a study in Saudi Arabian context”

The purpose of the present study is to test how do the four important psychological antecedents, namely need for achievement (nAch), propensity to risk (PtR), self-con-fidence (SeC), internal locus of control (IlC) are mediated by the entrepreneurial self-efficacy (ESE) to predict the entrepreneurial intention (EI) in a traditionally oil-based Saudi economy striving for economic diversification. Hypotheses are tested by apply-ing the partial least square (PLS) structural equation modeling (SEM) to a sample of 282 undergraduate business students (male and female) collected from a public university of Kingdom of Saudi Arabia (KSA). The results suggest that psychological variables and self-efficacy play a fundamental role in enhancing entrepreneurial intention (EI) of Saudi students. Personality-related variables have direct influence over EI in other context, but in Saudi context, psychological variables are necessary but not sufficient to develop entrepreneurial intention unless combined with self-efficacy. Self-efficacy has a strong mediating effect between psychological variables and entrepreneurial intentions in Saudi context.


INTRODUCTION
Entrepreneurship is one of the significant contributors to the economic efficiencies, employment generation, innovation and welfare of the society, etc. (Wennekers et al., 2005;Baumol, 2002;Shane & Venkataraman, 2000).However, "What persuades people to become entrepreneurs?" is still a debatable issue among researchers (Zhao et al., 2005;Drost, 2010).Though a galaxy of researchers agreed that no other than intentions could better explain the behavior of people to be the future entrepreneurs (Bird, 1988 Kautonen et al., 2015).TPB which was pioneered by Ajzen (1985) originally derived from "theory of reasoned action".It explains that when more favorably combined together with attitude, perceived behavioral control and subjective norm strengthen the intention to perform a particular behavior (Ajzen, 1985).Similarly to the constructs mentioned in Ajzen's (1985) model, self-efficacy is al-so considered as a phenomenal antecedent of entrepreneurial intention (Zhao et al., 2005).The concept of self-efficacy was first conceived by Bandura (1986).Self-efficacy in entrepreneurial context is one's strong belief in performing successfully the task and role of entrepreneurship (Chen et al., 1998;Miao et al., 2016).It is considered as a motivational construct in general (Zhao et al., 2005).It influences the choice of activities of individuals, their level of goals, perseverance, and performance in wide arrays including entrepreneurship (Zhao et al., 2005;Pihie & Bagheri, 2013).The degree of individual's capability or motivation to perform a task up to a great extent depends upon the level of self-efficacy (Forbes, 2005).Self-efficacy is linked with both task and outcome (Drnovšek et al., 2010).However, in both the cases, it is helpful for entrepreneurship.As a task-based focus, it solves the problem of ambiguity in earlier entrepreneurial personality-related research (Chen et al., 1998).Conversely, as an outcome construct, it serves as a force to generate the entrepreneurial intention that results in startup or new venture creation (McGee et al., 2009;Chen et al., 1998;Lüthje & Franke, 2003)_ENREF_7.Because of its broad usefulness and coverage, it is the most researched area of the field of entrepreneurship (Liñán & Fayolle, 2015).
Thus, the focus of the current study is to investigate that combining with the antecedents of TPB, how does self-efficacy mediate the personality-related variables or psychological variables to predict the entrepreneurial intention in Saudi Arabian context.

REVIEW OF LITERATURE AND HYPOTHESES SETTING
The meaning and definitions of ESE are welldefined and widely available in the literature e.g.Risk-taking one of the key elements connected with entrepreneur's personality (Zhao et al., 2010).
Those who prefer to take more risk are expected to be high in the level of self-efficacy (Barbosa et al., 2007).Though, the risk propensity is more relevant to entrepreneurial intention, not the performance (Zhao et al., 2010).The meta-analysis carried out by Stewart and Roth (2001) outlined the importance of risk propensity in explaining the self-efficacy.

H2: Propensity to risk (PtR) is positively related to entrepreneurial self-efficacy (ESE).
H2a: Propensity to risk (PtR) is positively related to entrepreneurial intention.
Entrepreneurial self-efficacy (ESE) itself is termed as the task-specific self-confidence (Boyd & Vozikis, 1994).Self-confidence is considered as one of the entrepreneurial personality-related characteristics.Those who are involved in entrepreneurship are found to be more confident than non-entrepreneurs (Kickul et al., 2009).
Internal locus of control (IlC) is also one of the psychological constructs associated with the psychological aspect of the individual.It is termed as the belief of individuals that largely their success depends upon their actions/tasks, not by external forces (Luthans & Ibrayeva, 2006).Internal locus of control pushes towards a progressive entrepreneurial attitude, and those who have it are to be found with a higher level of self-efficacy (Ajzen, 2002).Though, it does not poise to show the direct impact on the intention of the people.

H4a: Internal locus of control (IlC) is not positively related to entrepreneurial self-efficacy (ESE).
As per Ajzen (2002), "perceived behavioral control (PBC) is the perceived ease or difficulty of performing the behavior".The doctrine of PBC is based on the notion that an individual usually prefers such behaviors which can be easily mastered and controlled (Moriano et

H5: Entrepreneurial self-efficacy (ESE) is positively related to entrepreneurial intention (EI).
Therefore, the hypothesized model emerged after the hypotheses testing can be seen in Figure 1 below.

Data sample
The data for the study were collected from the undergraduate business students (male and female) enrolled in Bachelor of Science and Business Administration (BSBA) program, consisting of eight levels from a Saudi public university.The questionnaire was constructed as per Liñán and Chen (2009) for TPB variables, and the personality-related variables were conceived from Zhao et al. (2010).Three hundred questionnaires were distributed in hard copy to the level five and above, the sumpling of two hundred and eighty-three students who completed the survey resulted in a 94 percent effective response rate.Only two hundred and sixty-one (i.e.92 percent) were usable for the study.Among the respondents, 55 (overall 21 percent) were female, and rest were male students.The average age of the students was 25 years and they already spent two and a half years in the university system (see Table 1 for data characteristics).

RESULTS
Before delving into the factor analysis and structural modelling, descriptive analysis was performed (see Table 2).It can be noticed that self-confidence has the lowest mean out of the six constructs.While entrepreneurial intention (EI) has the highest mean and standard deviation among all constructs, it is indicated that the group is heterogeneous in regard to EI (Ferreira et al., 2012).This construct has the largest range along with entrepreneurial self-efficacy and internal locus of control.None of the constructs has the mean below 3 by the students who are highly enthusiastic towards the entrepreneurship, and its other constructs.
The exploratory factor analysis (EFA) using principal axis factoring and non-orthogonal promax rotation with Kaiser normalization (eigenvalues ≥ 1) was conducted as a first step to confirm the loadings for measurement model and factor structure (hypothezed) for the constructs.To determine the sufficiency of data for EFA, missing items and multicollinearity check were done, and the variables with problems were dropped.The items with a threshold limit of ≤ 0.50 were eliminated (Hair et al., 2010).Among six items on ESE, represented by PBC, one was eliminated because of its low loading to the factor.Four items from internal locus of control (IlC), IlC1, IlC2 IlC3, and IlC7, were also dropped because of low and crossloadings.For need for achievement (nAch) construct, only two items qualified for final loadings and rest of the four items were dropped, while in propensity to take risk (PtR), only half of the items qualified for final loadings.Moreover, for the last factor, i.e. self-confidence, one item has been dropped.But, notably, in self-confidence, half of the total items loaded to a separate construct that possibly is because of the reverse statements used in the questionnaire.Table 3 presents the refined and summarized pattern matrix for final loadings.
In order to draw the mediating role of entrepreneurial self-efficacy (ESE), PLS-SEM was applied in two stages.In stage one, we assess the measurement or outer model based on Cronbach's alpha, average variance extracted (AVE), composite reliability (CR) and discriminant validity (DV) as shown in Table 4 below.
Reliability of the constructs used in the model were evaluated by Cronbach's alpha ( ).
α The alpha value for all constructs is found to be significant that is above the threshold limit, i.e. 0.7 (Santos, 1999).As regards to CR, we found that except NACH and PTR constructs, all the other were found to be well above the suggested value of 0.7 (Fornell & Larcker, 1981).
Convergent validity as measured by AVE was also found to be well above the recommended value of 0.5.(Carlson & Herdman, 2012).The discriminant validity was established using the Fornell-Larcker criterion, and thus, the square root of constructs AVE was greater than its correlations with other constructs.
In stage two, the bootstrapping with 5000 subsamples was used to examine the predictive power of the structural model.The table below gives the direct, indirect and total effects along with the final column for the mediation results.Table 5 depicts the results for the nine hypotheses along with the significant or not significant path and meditation results.

CONCLUSION
The study confirms that only one antecedent, namely internal locus of control (IlC), directly predicts the entrepreneurial intention of Saudi undergraduate students.Moreover, the study indicated that selfefficacy mediated the relationship between propensity to risk, self-confidence, internal locus of control and entrepreneurial intention.
These results suggest that psychological variables and self-efficacy play a fundamental role in enhancing entrepreneurial intention (EI) of Saudi undergraduate students.Personality-related variables have direct influence over EI in other context, e.g.Dinis et al. (2013), but in Saudi context, psychological variables are necessary but not sufficient to develop entrepreneurial intention unless combined with selfefficacy.Self-efficacy has a strong mediating effect between psychological variables and entrepreneurial intention among Saudi undergraduate students.
The study is one of the kinds of premier study regarding measuring self-efficacy is not free from limitation.One limitation of the study is excluding the education variable as the dependent variable.There are ample evidence that education translates into the actual entrepreneurial behavior and one of the significant contributors to the entrepreneurial intention (Trivedi, 2016).One of the valid reasons of exclusion of this variable is that the target audience of the survey did not have the formal entrepreneurship education as modern Saudi undergraduate education system is still in its infantile stage.Another limitation of the study can be considered, as for the self-efficacy construct, no separate items were used.Six items of perceived behavioural control (PBC) of Ajzen (1985) model has been utilized, because it resemblance was mentioned in the literature.

Table 1 .
Data characteristics ESE measures were taken as the PBC as mentioned by Zhao et al. (2005) and Moriano et al. (2012) that PBC has more resemblance to the self-efficacy construct.Though, few studies like Manstead and Eekelen (1998), Terry and O'Leary (1995) assume self-efficacy as a separate construct than PBC, but due to its nature similarity, we have taken PBC items (i.e. six in number) adopted from (Chin, 1998), 2011;Babin et al., 2008)nce-based structural equation modeling (SEM) technique applied broadly in social sciences and business(Henseler et al., 2016).It is considered as one of the most powerful tools to evaluate the measurement and in-between relationships of latent variables(Hair et al., 2011;Babin et al., 2008).The model successfully works on the concepts which are difficult to observe directly(Chin, 1998).The technique is extensively used by Zhao et al. (2005), Tsai et al. (2014), Chen and He (2011), Bagheri and Lope Pihie (2014) to access the mediating effect over several constructs.

Table 4 .
Summary reliability and validity statistics