SOCIAL ENTREPRENEURSHIP INTENTION DURING COVID-19 CRISIS AMONG UNIVERSITY STUDENTS: A CASE STUDY FROM INDONESIA

The young generation as agents of change needs social entrepreneurship knowledge to create social entrepreneurship intention to solve social issues, especially during the COVID-19 pandemic. This research was conducted to examine social entrepreneurship intention among university students in Batam City, Indonesia based on the Theory of Planned Behavior (TPB) with the addition of empathy, entrepreneurship education, and COVID-19 stress perception variables to the framework. This research used the purposive sampling method through an online questionnaire that was distributed to all students of the Faculty of Economics Universitas Internasional Batam who took social entrepreneurship course. The data collected from 269 respondents were analyzed using SmartPLS version 3.0 to test the proposed hypothesis. The research result confirmed that attitude, perceived behavioral control, and empathy have significant positive relationship with social entrepreneurship intention. Entrepreneurship education and subjective norm have no significant positive relationship on social entrepreneurship intention. Meanwhile, COVID-19 stress perception was not proven to have a significant negative relationship with social entrepreneurship intention. Based on this result, universities are suggested to improve social entrepreneurship learning method so it can generate student’s interest in social entrepreneurship as an attractive career choice.

on Global Entrepreneurship Index (GEI) which assess the entire entrepreneurial ecosystem in the world, The United States leads the world in entrepreneurship (1 st rank), meanwhile, Indonesia is in 75 th rank which is far away from other countries in Asia, such as Hong Kong (13 th ), South Korea (21 th ), Singapore (27 th ), and Malaysia (43 th ) (Ács et al., 2019). Hence, Indonesia needs to improve its entrepreneurship ecosystem to support economic growth. One way to improve the entrepreneurial ecosystem is through entrepreneurial education by which students can develop their skills and competencies to seize business opportunities (Ács et al., 2019;Elnadi et al., 2020). This will help them build resources and create a more positive and good analytical attitude of disadvantages and risks from entrepreneurial activities (Paray & Kumar, 2020).
Socio-economic crises with a high level of uncertainty such as COVID-19 can hinder the promotion of the social entrepreneurship concept through entrepreneurship education (Liguori & Winkler, 2020;Ratten, 2020;Ruiz-Rosa et al., 2020). Youths as an agent of change are important to get the knowledge of social entrepreneurship concept to create a new job and develop an innovative solution to social issues, especially in global crisis (Ruiz-Rosa et al., 2020). However, there are only a few studies on social entrepreneurship intention which also analyze the impact of the COVID-19 crisis. Empirical evidence of the impact of COVID-19 on social entrepreneurship intention is required as this will also give knowledge and best interest to other researchers, practitioners, and academicians.
This research was conducted to fill in the gap by giving more understanding of what factors that have significant relationship with social entrepreneurship intention among university students in Batam City, Indonesia, who were taking social entrepreneurship courses in the COVID-19 crisis time using the modified Theory of Planned Behavior (TPB) framework. This study would contribute to provide literature on social entrepreneurship intention using the TPB framework with the addition of empathy, entrepreneurship education, and COVID-19 stress perception variables. Thus, the research objectives were to find out whether attitude, subjective norm, perceived behavioral control, empathy, and entrepreneurship education have significant positive relationship with social entrepreneurship intention. In addition, this study was also conducted to examine whether COVID-19 stress perception has a significant negative relationship with social entrepreneurship intention. This literature is required as this will also give knowledge and best interest to other researchers, practitioners, and academicians on social entrepreneurship intention during the COVID-19 crisis.
In answering research questions, several literature reviews need to be put forward to support the research. The concept of social entrepreneurship has been defined by many researchers. Social entrepreneurship intention is a person's desire, belief, determination, commitment, and readiness to carry out innovative social activities through the creation of new social businesses (H. M. K. Hassan, 2020). An individual who has an intention towards social entrepreneurship can quickly seize opportunities to create new employment (Ayalew, 2020). Among undergraduate university students, research of social entrepreneurship intention has become widely studied by scholars since it is considered as the gateway to studying the entrepreneurial system as well as the initial phase of a dynamic business process in the long run (Ghatak et al., 2020;H. M. K. Hassan, 2020). Theory of Planned Behavior (Ajzen, 1991) which consists of attitude toward the human behavior, subjective norm, and perceived behavior control which further can predict the actual behavior, is extensively used to explain social entrepreneurship intention (Kruse, 2020;Ma et al., 2020;Paray & Kumar, 2020;Ruiz-Rosa et al., 2020).
Attitude is a determinant of behavioral intention which refers to the extent to which a person has a good or poor judgment of the behavior (Ajzen, 1991). Students may have different attitudes and reactions to entrepreneurial interests. The positive or negative attitude they might show towards this interest in entrepreneurship depends on their background and other traits. They are more likely to become entrepreneurs after graduation if they have a positive attitude towards entrepreneurship and conversely, they may not be self-employed if they form negative attitudes. In general, those who want to be financially independent and get a large income will tend to choose as entrepreneurs (Ayalew, 2020). Many studies evidenced the positive relationship of individual attitude on entrepreneurial intention (Al-Mamary et al., 2020;Ayalew, 2020;Luc, 2020aLuc, , 2020bLukman et al., 2020;Ruiz-Rosa et al., 2020;Tiwari et al., 2017). Therefore, it is proposed that: H1: Attitude toward social entrepreneurship has a significant positive relationship with social entrepreneurship intention. Subjective norm refers to the social pressure a person feels to perform or not to perform a behavior (Ajzen, 1991). Subjective norm on social entrepreneurship refers to the opinions perception of people or environment (family, friends, colleagues) who are close to someone about the implementation of social and/or environmental projects (Ruiz-Rosa et al., 2020). Subjective norm become the main reflection of social and cultural values based on beliefs about how other people think that someone should behave and motivation to obeys the norms of the reference group. This subjective norm can influence a person's interest in doing social entrepreneurship as evidenced by researchers (Luc, 2020a(Luc, , 2020bRuiz-Rosa et al., 2020;Tiwari et al., 2017). Therefore, it is proposed that: H2: Subjective norm has a significant positive relationship with social entrepreneurship intention.
Perceived behavioral control refers to the ease or difficulty a person feels in carrying out behavior and is considered as a reflection of the past and the obstacles that are anticipated. This perceived behavior control varies depending on the situation and action. A person can believe that his achievement depends on his behavior (Ajzen, 1991). Researchers evidenced that perceived behavioral control has a positive relationship to social entrepreneurship intention (Luc, 2020a(Luc, , 2020bRuiz-Rosa et al., 2020;Tiwari et al., 2017). Therefore, it is proposed that: H3: Perceived behavioral control has a significant positive relationship with social entrepreneurship intention. This research added several variables to the Theory of Planned Behavior (TPB) framework to explain more on entrepreneurship intention based on previous studies such as empathy, entrepreneurship education, and COVID-19. Empathy is a feeling of an individual who puts himself in the place of others so that it can influence the individual's behavior to help others. Someone who has empathy can understand the emotional state of others based on the experiences and emotions that other people are experiencing (Tiwari et al., 2020). Most social entrepreneurs are founded because social entrepreneurs are empathetic and willing to help people to solve social problems (Rozar et al., 2020). Empathy has a positive relationship with entrepreneurship intention as evidenced by many scholars (Aure, 2018;Bazan et al., 2020;Garaika, 2020;Ghatak et al., 2020;Hockerts, 2017). Therefore, it is proposed that: H4: Empathy has a significant positive relationship with social entrepreneurship intention. The development of entrepreneurship is supported by the entrepreneurial ecosystem as guidance to system performance. The entrepreneurial ecosystem is a complex structure of a socio-economic system and can be defined as the dynamically inherent interaction between the entrepreneurial attitudes, abilities, and aspirations of an individual, which drives the allocation of resources through the creation and operation of new businesses (Ács et al., 2019). Entrepreneurship education is one of the entrepreneurial ecosystem elements besides other elements such as access to infrastructure, finance, human resources, government support, institutional strengthening, and networking (Ács et al., 2019;Elnadi et al., 2020). Entrepreneurship education must be implemented based on theoretical knowledge of psychological cognition, creative psychological development, and by using specific teaching strategies. This will guide and instill an entrepreneurial spirit in students (Ma et al., 2020).
University plays an important role in entrepreneurship to stimulate and support students' entrepreneurial development efforts through entrepreneurship education (Yordanova et al., 2020). Any entrepreneurial intention a person has is usually not enough to take a step forward in creating a venture. This intention needs to be supported by strong education, especially with entrepreneurial features. Individuals need to be fully supported by entrepreneurship education that enables and supports them to carry out their activities professionally and with less risk (Alshebami et al., 2020). Entrepreneurship education can include providing a suitable environment that provides the latest syllabus to meet the requirements of students to become entrepreneurs. Students may have high entrepreneurial intentions when they first start their entrepreneurship course because they have learned many concepts related to the business environment (Boldureanu et al., 2020). Many researchers evidenced that entrepreneurship education has a positive impact on entrepreneurship intention (Alshebami et al., 2020;Ayalew, 2020;Boldureanu et al., 2020;Chen et al., 2020;Elnadi et al., 2020;H. M. K. Hassan, 2020;Noerhartati et al., 2019;Paray & Kumar, 2020;Tiwari et al., 2020;Yordanova et al., 2020). Therefore, it is proposed that: H5: Entrepreneurship education has a significant positive relationship with social entrepreneurship intention. Uncertainty is a condition where there are many unknown possibilities and variables. This uncertainty can be seen by humans with skepticism and is likely to be avoided whenever possible (Kabir, 2019). In a state of economic crisis, there are a lot of uncertainty and disruptions such as the increasing number of unemployed, the difficulty of finding a job, and the increasing number of closed businesses, which make people think about whether opening their own business is the right career choice. The COVID-19 pandemic is not only a socio-economic crisis, but has also become a multi-dimensional crisis with a high level of uncertainty that has changed many things in life. There are many risks and stresses from COVID-19 that affect the future of a person's life which are quite the same or even worse than the risks that occur during other economic crises. The stress caused by COVID-19 can hinder a person's formal education and career choices in the future (Islam et al., 2020). The economic crisis has proven to have a significant negative effect on student's entrepreneurship interests (Arrighetti et al., 2016). Moreover, the research conducted by Ruiz-Rosa et al. (2020) showed that students' interest in social entrepreneurship during the COVID-19 crisis has decreased compared to before the emergence of the crisis. Therefore, it is proposed that: H6: COVID-19 stress perception has a significant negative relationship with social entrepreneurship intention. Based on the description of the hypothesis above, the research framework is illustrated in Figure 1.
The population of this research was all undergraduate students in their fourth semester at Faculty of Economics Universitas Internasional Batam. As this research used the purposive sampling method, online questionnaires were given to all students who took social entrepreneurship courses for at least 1 semester (majoring in management, accounting, and tourism). A filter question was given to confirm this. The sampling frame consists of 516 students from 12 classes. The selection of sample number in this research was based on the analytical method that will be used i.e. PLS-SEM that can predict complex research models using various constructs, variable indicators, and structural paths with no data distribution assumptions, based on a causal relationship approach (Hair et al., 2019). According to Kline (2016), the median sample number used in SEM analysis is 200. While Memon et al. (2020) suggest that suitable sample size for multivariate analysis such as PLS-SEM is between 160 and 300. The determination of the number of samples commonly used is at least 10 times the number of arrows that point at most to a construct (Hair et al., 2014). Based on this requirement, then the minimum number of samples used in this research should be 60 samples because there are 6 arrows connected to 1 dependent construct, namely social entrepreneurship intention. However, a minimum of 200 samples will be used in this research and it will meet the requirements not only stated by Hair et al. (2014), but also as suggested by Kline (2016) and Memon et al. (2020).
Online questionnaires using Google form were distributed through WhatsApp group and Microsoft Teams chat on March 2021 based on the sampling frame. A reminder message to fill out the questionnaire was sent to the students a week after the questionnaire was sent. The data collection timeline was 4 weeks since the questionnaire was distributed. Data were analyzed using SPSS version 23 for demographic information of respondents and Common Method Bias (CMB) analysis. SmartPLS version 3.0 was used for measurement model analysis (reliability, validity, and model fit testing) and structural model analysis (collinearity, hypothesis testing, and R-square testing).

RESULT AND DISCUSSION
A total of 269 completed responses (out of 516 sent questionnaires) were returned, thus indicate a 52.13 percent of response rate. There were 72 male (26.8 percent) and 197 female (73.2 percent) respondents. The majority of respondents (83.6 percent) were between 18-20 years old, followed by 14.5 percent aged 21-23 years and 1.9 percent aged more than 23 years. The profile of students studying at Universitas Internasional Batam are diverse. Some are purely students and some have jobs other than studying because the study programs are held in morning or evening classes. As many as 47.6 percent of respondents were pure students, 46.8 percent of respondents were students who also worked as employees, and 5.6 percent of respondents were students who were also entrepreneurs. There are three study programs at the Faculty of Economics Universitas Internasional Batam, namely management, accounting, and tourism. Most respondents were from management study programs (66.2 percent), followed by accounting (30.5 percent), and tourism (3.3 percent). The majority of respondents (98.1 percent) did not have personal experience in running social entrepreneurship and there were 1.9 percent of respondents who had an experience of running social entrepreneurship.
The use of questionnaires in research can raise the issue of common method variance originating from the measurement method, respondent selection, and survey context and will then lead to Common Method Bias (CMB) (Hulland et al., 2018). Several ways to overcome this are by conducting pre-test questionnaires, using multi-scale measurements on different constructs, developing appropriate research designs, and using statistical analysis to determine the presence of CMB (Hulland et al., 2018). Harman's single-factor test is one of the statistical analysis techniques to detect the presence of CMB by using factor analysis of all items of the research construct. If there is no single factor that has a variance value of more than 50 percent, then there is no CMB in the study. Based on SPSS software analysis, the total value of variance in this research was 32.750 percent, so it can be concluded that there was no CMB in this research data. The initial measurement using SmartPLS for this research was the evaluation of the measurement model or outer model. At this stage, reliability and validity testing was carried out. Reliability testing was analyzed from the value of outer loading, Cronbach's alpha, and Composite Reliability (CR). Outer loading shows how big the contribution of the indicator in measuring a construct. The greater the value of the outer loading of an indicator, the greater the reliability of the measurement model. The recommended minimum outer loading value for the reflective measurement model is 0.7 (Garson, 2016). However, Hair (Hair et al., 2019) suggests that the minimum value for outer loading is 0.708 because at this value, a construct can explain more than 50 percent of the variance in the indicators. In this study, the minimum outer loading value used was 0.7.
Several indicators had outer loading values below 0.7 so they were deleted and not included in the next evaluation, namely ATT1, COV 1, COV 2, COV 7, COV 12 (Table 2). In addition to outer loading, the reliability of a construct can also be seen from the value of Cronbach's alpha and Composite Reliability (CR). Cronbach's alpha value above 0.8 indicates good reliability, a value of 0.7 is quite acceptable, and a value of 0.6 is used for exploratory research scales (Garson, 2016). All constructs in this study had good reliability, as it can be seen from the value of Cronbach's alpha above 0.8, namely ATT (0.805), SN (0.847), PBC (0.903), EMP (0.877), EE (0.921), COV (0.902), and SEI (0.912). Composite reliability (CR) shows the internal consistency of a construct that is more often used by researchers because it can estimate reliability more precisely (Garson, 2016;Hair et al., 2014Hair et al., , 2019. The CR values between 0.7 to 0.9 are satisfactory to good values. CR values of 0.6 to 0.7 are values that can be used for exploratory research (Hair et al., 2019). All constructs had CR values above 0.9 which indicated good internal consistency reliability, i.e ATT (0.911), SN (0.907), PBC (0.926), EMP (0.924), EE (0.944), COV (0.918), and SEI (0.932).
Measurement of validity was done by looking at the value of Average Variance Extracted (AVE) which is a measurement of convergent validity or the extent to which a construct converges to explain the variance of the items or indicators.
The recommended AVE value is at least 0.5 which indicates that a construct can explain at least 50 percent of the variance of the items (Hair et al., 2019). The AVE value of each construct in this research varied, but all of them were above 0.5. The highest value was AAT (0.837) and the smallest was COV (0.583) ( Table 2). Other validity measurements were performed using the Fornell-Larcker Criterion Discriminant Validity. The square root of AVE value of a construct must be greater than the value of other constructs. In other words, the variance value of a construct with its indicators must be greater than the variance between the construct and other construct indicators (Garson, 2016). Based on Table 3, all constructs met the discriminant validity requirements where all values of the square root of AVE were greater in the construct itself compared to other constructs. The next test carried out was Heterotrait-Monotrait Ratio (HTMT) analysis to confirm discriminant validity. HTMT ratio refers to the geometric mean of the relationship between indicators between different constructs divided by the average mean of the relationship between indicators in the same construct (Garson, 2016;Hair et al., 2019). A well-fitting model is said to have an HTMT ratio below 1.0 (Garson, 2016). All constructs in this research met the discriminant validity requirements with the HTMT ratio ranging from 0.201 to 0.754 (Table 4). Standardized Root Mean Square Residual (SRMR) was used to see the model fit of this model. SRMR shows how far the difference between the observed correlation matrix and the model-implied correlation matrix. The smaller the SRMR value, the better the model. The ideal SRMR value used to show a good model is below 0.8 (Garson, 2016). The SRMR value was obtained at 0.069 which indicated that this research model was good.
The next step after the measurement model evaluation stage had been carried out was the structural model or inner model evaluation. Collinearity testing was done to ensure that there was no bias in the regression measurement of the relationship between constructs. This can be shown from Variance Inflator Factor (VIF) value. The ideal VIF value is below or close to 3.0 which indicates that there is no collinearity problem in the research model (Hair et al., 2019). All constructs in this research met the requirements of the collinearity test because they had VIF values between 1.112 (COV) and 1.948 (ATT) as shown in Table 5. Based on the structural or the inner model evaluation, the path coefficient, tvalue, and p-value were obtained from the analysis using the bootstrapping technique. The path coefficient (standard β), refers to the strength of the relationship in a path that ranges from -1 to 1. If the value is close to 1 or -1, the stronger the positive or negative relationship on that path (Garson, 2016). In Table 6, H3 (PBC → SEI) had the largest path coefficient, i.e 0.470 which means that there was the strongest positive relationship among other paths. While H2 had the smallest path coefficient, which was -0.074 which indicated that there was a negative relationship on the path. The hypothesized negative relationship on H6 (COV → SI) was not proven. In fact, a positive relationship was found on that path with a path coefficient of 0.037. To interpret the statistical significance of the relationship between constructs, it can be seen from the t-value. If the t-value above 1.96 (based on a 95% confidence interval), it is said that the relationship between constructs is significant (Garson, 2016;Hair et al., 2014Hair et al., , 2019. Based on Table 6, there were 3 relationships between constructs that have been proven to be significant and the proposed hypothesis can be proven true, namely H1 (ATT → SEI), H3 (PBC → SEI), and H4 (EMP → SEI). While H2 (SN → SEI), H5 (EE → SEI), and H6 (COV → SEI) were not confirmed.
Based on the description above, H1 (ATT → SEI) was proven and is in line with research conducted by Al-Mamary et al. (2020) Attitude showed a significant positive relationship towards social entrepreneurship intention, which was the second strongest among the three positive significant relationships in the model (β = 0.236, t = 3.770, ρ = 0). The perception that social entrepreneurship is considered to have many advantages (by helping other people besides doing business), more attractive and satisfying, is a positive attitude that needs to be instilled in the younger generation (Ruiz-Rosa et al., 2020). In the context of entrepreneurship in general, entrepreneurship education and financial support can foster a positive attitude and interest in entrepreneurship in students in Malaysia . With the increasing exposure of entrepreneurship knowledge that students get from an early age from school, it will further increase the enthusiasm to create their own jobs. Likewise, good financial support from family, friends, or other parties will increasingly build a positive attitude in starting a new business, especially if they see that entrepreneurship is a way to be financially independent and gain a large income (Ayalew, 2020;. The second hypothesis, i.e H2 (SN → SEI) was not proven in this study. The result showed the opposite, i.e there was a negative and insignificant relationship (β = -0.074, t = 1.337, ρ = 0.182). It means that subjective norm which refers to opinions or perceptions from people around about the implementation of social entrepreneurship or social projects, did not even cause students' interest in making social entrepreneurship projects. This might be because people around them did not have a good understanding of the concept of social entrepreneurship itself. This has led to a lack of support and motivation for people who will open businesses and also have a social mission. This result does not support research conducted by several previous researchers (Luc, 2020a(Luc, , 2020bRuiz-Rosa et al., 2020;Tiwari et al., 2017).
Similar to the case with H1, H3 (PBC → SEI) was also proven true (β = 0.470, t = 8.091, ρ = 0). It can be seen that perceived behavioral control which refers to the ease or difficulty a person feels in carrying out behavior, had the strongest relationship to social entrepreneurship intention compared to other relationships in the model. This result supports the research conducted by Luc (2020aLuc ( , 2020b, Ruiz-Rosa et al. (2020), and Tiwari et al. (2017). Thus, this shows that the greater the perceived behavioral control a person has, the more interest in creating social entrepreneurship projects. The perceived behavioral control in this context can be seen as the ease of identifying and creating new business opportunities, ability to be creative, able to manage a social project, and become a leader.
The next hypothesis, i.e H4 (EMP → SEI) was confirmed and it had the weakest relationship compared to other significant positive relationships (β = 0.139, t = 2.047, ρ = 0.041). This result is supported by previous research (Aure, 2018;Bazan et al., 2020;Garaika, 2020;Ghatak et al., 2020;Hockerts, 2017). This result indicates that people have an interest in creating a social business project if they can place themselves in a person who has a disadvantage which then create a sense of willingness to help, find new business opportunities, create a new product/service with their creativity, and become leaders in solving social problems. This is in line with the fact that many social entrepreneurs start their business from social concern for the people around them and want to solve the social problem (Rozar et al., 2020).
Similar to H2, H5 (EE → SEI) was also not proven in this study. Although the path coefficient showed that it had a positive relationship, however, it was not significant (β = 0.152, t = 1.935, ρ = 0.054). Therefore, H5 was not confirmed. This result is not in line with previous research (Alshebami et al., 2020;Ayalew, 2020;Boldureanu et al., 2020;Chen et al., 2020;Elnadi et al., 2020;H. M. K. Hassan, 2020;Noerhartati et al., 2019;Paray & Kumar, 2020;Tiwari et al., 2020;Yordanova et al., 2020). The COVID-19 pandemic is a socio-economic crisis with a high level of uncertainty that has changed many things in life around the world, one of which is a change in education. This change requires a rapid migration from the real educational environment to the online environment due to social distancing policy, where before the crisis, online education is considered as something is optional and expensive. As such, COVID-19 has resulted in a major disruption to the education system which is still largely understood due to its severe effects (Ratten, 2020). This online learning condition that has limitations may result in the entrepreneurship teaching process not being as optimal as face-to-face teaching. Therefore, students might not absorb and understand the concept of entrepreneurship well which further reduces their interest in running a social business. This condition makes entrepreneurship educators and universities must have innovation and uniqueness in providing digital lessons that can be accessed in a timely and practical manner (Ratten, 2020). In addition, they can strive to maintain the education quality and produce the same output as the face-to-face teaching process. With this pandemic condition, in the future students are expected to be more adaptable, creative, and innovative in seizing opportunities through entrepreneurship education (Highfield et al., 2020).
The last hypothesis, i.e H6 (COV → SEI) was also not proven. The results of the analysis on H6 showed that COVID-19 did not have a significant negative relationship with social entrepreneurship intention as previously proposed, but instead had an insignificant positive relationship (β = 0.037, t = 0.854, ρ = 0.394). Previous study conducted by Ruiz-Rosa et al. (2020) showed that social entrepreneurship intention decreases during the COVID-19 period. Meanwhile this study shows that with the COVID-19 crisis, students still have the intention to create social entrepreneurship even though it is not significant. It might be due to the stress level of the younger generation is not as high as the older generation and they could also still manage their lives so they did not feel too much tension from the COVID-19 pandemic. Moreover, this is supported by the fact that the respondents in this study were unmarried students, so they did not have family responsibilities. Therefore, the COVID-19 pandemic might not have a severe impact on their lives and they could still see their future optimistically. Even though they feel a little stressed due to COVID-19, but this did not reduce their interest in creating a social business. They might notice that many people around them were more severely affected by this pandemic condition, so they were actually interested in doing social missions and also doing business at the same time. Therefore, the pressure and stress from COVID-19 did not necessarily make them reluctant or reduce their interest in doing social entrepreneurship. This is supported by research conducted by Arrighetti et al. (2016) that in time of economic crisis, there is indeed a negative effect on the economic environment and significant negative effect on possibility in creating new business. However, it does not have an impact on the tendency of someone to decide to become an entrepreneur. The high unemployment rate can also create a new spirit to create the second-best alternative as an entrepreneur rather than being unemployed. However, Ruiz-Rosa et al. (2020) suggest that due to the specificity of the type of social entrepreneurship that provides solutions to both social and environmental problems, the barriers that arise from crises such as COVID-19 on social entrepreneurship intention can be anticipated with support from the educational environment. The university can provide good educational methods and support for social entrepreneurship so that even in times of crisis, it will make the younger generation still be able to identify social business opportunities. This will certainly help a country not only in creating jobs, but also in solving social and environmental problems. The variance in the endogenous construct can be seen from the value of Rsquare. The R-square values have different categories, where 0.75 is a substantial category, 0.5 is a moderate category and 0.25 is a weak category (Hair et al., 2019). The value obtained in this study was 0.601 which is in the category range between substantial and moderate. It was also indicated that the exogenous variables in this study contributed to explaining the endogenous variable, i.e social entrepreneurship intention by 60.1 percent. Other variables as examined in previous research would contribute to the remaining variances such as entrepreneurship experience, moral obligation, self-efficacy, social support, business ties, sharing achievement, perceived social worth, environment support, emotional intelligence, and social activity (Darmanto & Pujiarti, 2020;Garaika, 2020;Hockerts, 2017;Latif & Ali, 2020;Noerhartati et al., 2019;Wu et al., 2020). The structural model of this research is shown in Figure 2.

CONCLUSION AND SUGGESTION
This research aims to examine what factors that have significant relationship with social entrepreneurship intention by using a modified Theory of Planned Behavior (TPB) on students who were taking social entrepreneurship courses. The results confirmed previous literature that attitude, perceived behavioral control, and empathy have significant positive relationship with social entrepreneurship intention. Perceived behavioral control has the strongest relationship with social entrepreneurship intention compared to others. Subjective norm, and entrepreneurship education do not have significant positive relationship with social entrepreneurship intention. Meanwhile, the COVID-19 stress perception which was hypothesized to have a significant negative relationship with social entrepreneurship intention is also not supported.
This research provides not only a contribution to the literature of social entrepreneurship intention using Theory of Planned Behavior (TPB) with the addition of empathy, entrepreneurship education, and COVID-19 stress perception variables to the framework, but also provides practical implications. Universities that play an important role in providing social entrepreneurship education should be able to ensure that the teaching system provided during the COVID-19 period remains good and of the same quality as before the pandemic occurred. However, this may be understandable because at the time this research was conducted, education was conducted online due to the COVID-19 pandemic. This online education could not produce optimal learning compared to the face-to-face teaching system. This what makes students might not understand the concept of social entrepreneurship well so their interest in doing social entrepreneurship decreases. Therefore, universities should be able to develop innovative and creative teaching systems that can increase students' learning engagement. In addition, it is important to provide students with not only soft skills, hard skills, technical skills, but also creative thinking skills. Various ways can be done by providing many group discussions, case studies, making business projects, organizing seminars and workshops with practitioners, and conducting business competitions. By generating interest in social entrepreneurship, it will increasingly open up new job opportunities and increase the resolution of social and environmental problems. Therefore, social entrepreneurship will be an attractive career choice for students in the future.
The results of this research certainly have limitations such as the findings cannot be generalized as the scope of the research was only carried out at one university and used limited variables. Further research is recommended to use a larger research scope, diverse variables, different theoretical frameworks, use mediators, and also examine the moderating effect of COVID-19 impact on social entrepreneurship intention.