بررسی شکل گیری سرمایه انسانی در کارآفرینی : متا تجزیه و تحلیل نتایج آموزش کارآفرینی

Effective human capital formation through the medium of entrepreneurship education and training (EET) is of increasing concern for governments, as EET is growing rapidly across the world. Unfortunately, there is a lack of consistent evidence showing that EET helps to create more or better entrepreneurs. We undertake the first quantitative review of the literature and, in the context of human capital theory, find that there is indeed support for the value of EET. Based on 42 independent samples (N = 16,657), we find a significant relationship between EET and entrepreneurship-related human capital assets (rw = .217) and entrepreneurship outcomes (rw = .159). The relationship between EET and entrepreneurship outcomes is stronger for academic-focused EET interventions (rw = .238) than for training-focused EET interventions (rw = .151). We find evidence of heterogeneity in many of our correlations, and recommend that future studies examine potential moderators to more clearly delineate EET effect sizes. We also find a number of methodological weaknesses among the studies analyzed and that those studies with lower methodological rigor are overstating the effect of EET. Recommendations to improve the quality of future work in the field are provided.

Entrepreneurship education and training (EET) is growing rapidly in universities and colleges throughout the world (Katz, 2003 and Kuratko, 2005) and governments are supporting it directly and indirectly through funding major investments to would-be entrepreneurs and existing small businesses. This trend is fuelled by a recognition that entrepreneurship can play an important (even critical) role in economic growth and employment (Schumpeter, 1934, Shane and Venkataraman, 2000 and Kuratko, 2005), and assertions that entrepreneurship education can play a vital role in developing more and/or more able entrepreneurs (e.g., Gorman et al., 1997, Katz, 2007 and Pittaway and Cope, 2007). Unfortunately, as several scholars have noted (e.g., Weaver et al., 2006 and Peterman and Kennedy, 2003), there is little consistent evidence to support these claims. This study contributes to the entrepreneurship literature by providing the first meta-analytic review of extant EET studies linking EET-specific interventions with entrepreneurship outcomes. Our analysis is grounded in human capital theory (Becker, 1964 and Mincer, 1958), as it is well suited to the examination of educational outcomes. The use of human capital theory to explain aspects of entrepreneurial success is also well established in the entrepreneurship literature (e.g. Pfeffer, 1994), but almost exclusively as a static model where accumulated education and experience is related to various forms of success (e.g. Chandler and Hanks, 1998, Rauch et al., 2005, Cassar, 2006, Dyke et al., 1992 and van der Sluis et al., 2005). In their recent meta-analysis, Unger et al. (2011) make the case for why human capital theory must be considered in less static terms, at least as it relates to the field of entrepreneurship: Given the dynamics in entrepreneurship and the constant need to learn and to adapt, it may prove useful to look beyond the static concept of human capital and to examine outcomes of actual learning activities… (p. 3) Our work also addresses the need for a more dynamic depiction of human capital in the entrepreneurship field, by examining the outcomes of educational activities that are specific to entrepreneurship. This dynamic view of human capital for the entrepreneurship field examines the relationships between human capital investments, which are inputs, such as the time and money spent taking a course in entrepreneurship; human capital assets, which represent the capability that may be garnered from the investments, such as knowledge and skills; and entrepreneurship outcomes, such as starting or growing a new business. The distinction that we are making between human capital investments, human capital assets, and entrepreneurship outcomes is important to note, because, although our terminology is in line with several researchers (e.g., Hitt et al., 2001 and Lepak and Snell, 1999), there is little consistency in terminology in the literature generally (cf. Reuber and Fischer, 1999 and Unger et al., 2011). Investigating a more dynamic approach to human capital in the context of the potential impact of entrepreneurship education and training requires a clear delineation of these three terms. Our work also contributes to a growing body of evidence that human capital assets specific to entrepreneurship have a stronger link to positive new venture performance than more general human capital assets (Unger et al., 2011). We take this specificity further to elucidate the differences between two main types of EET—training-focused educational interventions and academic-focused educational interventions—which we posit will influence outcomes differentially. Finally, we provide recommendations to future researchers in this field regarding specific methods that should be used in order to improve the quality and the utility of the information generated. This is consistent with concerns raised by several scholars regarding the quality of research conducted in the entrepreneurship education field (e.g. Béchard and Grégoire, 2005, Kailer, 2005 and Weaver et al., 2006).

6.1. Contributions Although EET programs are growing rapidly around the world, extant qualitative reviews have been equivocal about their impact on entrepreneurship-related human capital assets and entrepreneurship outcomes (e.g. Weaver et al., 2006). This is due in part to the fact that, although most studies report positive relationships, a number of important studies have shown negative results for EET on entrepreneurship-related human capital assets and entrepreneurship outcomes. Thus, it is not clear what impact EET might be having on its students. At the same time recent entrepreneurship literature has highlighted the need to better understand the dynamic nature of human capital development in the highly dynamic entrepreneurship field (Unger et al., 2011). Our study addresses these gaps in the entrepreneurship literature in an important way. We have provided a quantitative assessment of the EET literature showing that EET has positive, significant relationships with a number of entrepreneurship-related human capital assets and entrepreneurship outcomes. This builds on previous work, such as that of Unger et al. (2011) to show the specific impact of entrepreneurship education and training on human capital development and entrepreneurship development. Overall, our results provided full support for the first two hypotheses associated with this study and partial support for the third hypothesis. Evidence in support of Hypothesis 1 demonstrated that EET is associated with higher levels of (a) total entrepreneurship-related human capital assets, (b) entrepreneurship-related knowledge and skills, (c) positive perceptions of entrepreneurship, and d) intentions to become an entrepreneur (see Table 3). To the best of our knowledge, this is the first time that the transfer of entrepreneurship-related human capital investments to entrepreneurship-related human capital assets in the field of entrepreneurship has been demonstrated via meta-analytic examination. Evidence in support of Hypothesis 2 showed that EET was positively associated with (a) entrepreneurship outcomes in general (b) start-up, and (c) entrepreneurship performance (see Table 3). Again, to our knowledge this is the first time this link has been demonstrated in a quantitative review of the extant entrepreneurship literature. We believe these findings provide some empirical indication that EET is positively related to important benefits in a number of areas. Evidence showing partial support for our third hypothesis indicates that the focus of EET interventions does moderate the relationship between EET and entrepreneurship outcomes, but not the relationship between EET and entrepreneurship-related human capital assets. Specifically, academic-focused EET was found to have a significantly stronger relationship with entrepreneurship outcomes than training-focused EET, as we hypothesized, but training-focused EET was not found to have a significantly stronger relationship with entrepreneurship-related human capital assets than academic-focused EET, contrary to our hypothesis (see Table 3). Borrowing from the field of educational psychology, and in particular transfer of learning perspective (Thorndike and Woodworth, 1901), we argued that the near-far distinction (Haskell, 2001 and Barnett and Ceci, 2002) applied to the two broad types of EET and the differing content and contexts that were required to develop entrepreneurship-related human capital assets and entrepreneurship outcomes. Specifically, we expected that training-focused EET would be more likely to allow students to demonstrate the core entrepreneurship knowledge and skills required to start a particular business in a particular setting, because the learning and application context and content are more “near” than with academic-focused EET. This aspect of our argument was not supported, and although the reported correlations follow the directions that we expected, the lack of significance in the differences is difficult to interpret without further study. Future research that more specifically examines the development of entrepreneurship-related human capital assets in both training-focused and academic-focused EET is needed to help bring clarity to this issue. However, support for the moderation of EET to entrepreneurship outcome relationship is encouraging. It suggests that academic-focused EET, with its broader conceptual and theoretical content may be more likely to allow students to make decisions in the highly ambiguous and dynamic contexts that are required to achieve financial success and maintain a business over an extended period of time. We found indications of homogeneity in the correlations for academic-focused EET and for training-focused EET, suggesting that it is unlikely that moderators are significantly affecting the results. This was not the case for all of our other analyses, where we found strong indications of heterogeneity in our correlations, suggesting that there are moderators, which might help to better explain the relationships. Future research should explore these moderators. Our findings related to study rigor help to quantify a concern that has been raised by scholars, such as Weaver et al. (2006), that the EET literature suffers from low quality studies. We have noted that a large number of studies were not viable for inclusion in our meta-analysis because of methodological and/or reporting issues, and that among the 42 studies that were viable for inclusion, 31 did not meet a high standard of methodological rigor (incorporating both pre- and post-measures of participant responses and comparisons of treatment and control groups; see Table 1). Although there is a directional indication toward increased methodological rigor, newer studies are not significantly more rigorous than older studies. We go further, however, to show that studies that do not meet our criterion for high rigor report stronger relationships than those that do meet the rigor criterion (see Table 3). In other words, our results suggest that poorer quality studies tend to overestimate the impact of EET. We hope that by demonstrating this exaggerating influence of low methodological rigor on the impact of EET we will help to encourage future researchers to conduct more rigorous studies. Our results suggest that the effect sizes of EET are small using Cohen's (1992) guidelines for categorizing effect sizes. We compared these results to those found in meta-analyses examining management training in general (e.g., Burke and Day, 1986, Morrow et al., 1997 and Collins and Holton, 2004), which showed medium effect sizes. Reasons for the weaker performance of EET are difficult to determine. It may be unrealistic to expect outcomes of university courses, many of which may be designed to introduce students to the subject of entrepreneurship for the first time, to compare favorably with outcomes of management training courses, many of which are designed and sponsored by companies to improve their current managers' performance. However, this does not account for the relative low effect sizes that we found among the training-focused courses that were examined by studies in our analysis (see Table 3). It may be that, when comparing behavioral outcomes, the broad set of knowledge, skills and competencies that one must put into practice in order to become a successful entrepreneur is of a much greater magnitude than the more specific sets of skills that are required of a manager to demonstrate transfer of learning from a particular training course to a current management position. Yet this does not account for the relative low effect sizes we found for knowledge and skills specifically (see Table 3), which are most comparable to the learning outcomes often studied in management training (cf. Burke and Day, 1986). It may be that EET is simply not developed enough at this point. Further studies that examine the impact of course design and teaching methods may help to explain these relative weak effect sizes, and, by identifying important moderators, show that certain types of EET garner larger effect sizes. Such learning, if incorporated into future EET interventions may lead to improvements that will make EET more effective generally. 6.2. Potential limitations Contributions aside, this paper is subject to at least three potential limitations. First, our meta-analysis included a variety of studies that ranged from low to high methodological rigor. Meta-analyses are sometimes criticized for mixing good and bad studies (Rosenthal and DiMatteo, 2001). This criticism, known as the “garbage in and garbage out” issue (Hunt, 1997: 42), is relevant to this study. For example, as can be seen in Table 1, some studies did not include a design that compared a treatment group, whose members received entrepreneurship education or training, with a control group, whose members did not receive entrepreneurship education or training (e.g., Tam and Hansen, 2009 and von Graevenitz et al., 2010). Some designs did not include measures of variables at both pre- and post-education times (e.g., Charney and Libecap, 2000 and Lee et al., 2005). Although we decided to include these studies, because they represent much of the literature in this field and do provide some indication of EET outcomes, we acknowledge that the validity of our results would be stronger if all analyzed studies were conducted with a high standard of methodological rigor. Importantly, our study helps to quantify the impact that study rigor is having on research results, showing that lower rigor studies may be overestimating the impact of EET. Second, all but six of the studies we analyzed did not involve randomized assignment to treatment and control groups, which raises a concern for sampling bias. It is possible that those who have not chosen to take an EET course may demonstrate much weaker and possibly negative results in terms of entrepreneurship-related human capital assets and entrepreneurship outcomes. This was the case with participants in the Oosterbeek et al. (2010) study, which was one of only six studies that we categorized as random assignment. However, when we meta-analyzed all six studies that included randomized assignment we found that results were still positive (r = .156) and not significantly lower than those studies that did not employ randomization (r = .212; z = .909, ns). This suggests that sampling bias did not have a significant impact on the results. A third potential limitation is that our meta-analysis corrected only for sampling error. This is because we could not find information relating to other artifacts (e.g., range restriction, reliability estimates, etc.) for the majority of the studies we identified as having the potential to be included in our meta-analysis. As such, we were left with making a judgment call on their excludability. These types of judgment calls are common in the majority of meta-analytic research (Guzzo et al., 1987), and our study is no exception. The exact magnitude of the relationships remains to be confirmed once a broader sample of methodologically rigorous studies is available. Moreover, this limitation to our study is ameliorated by research that has found the majority of artifactual variance to be due to sampling error alone (Koslowsky and Sagie, 1994). 6.3. Future research Although our findings did not show a statistically significant improvement in study rigor over time, we did find evidence of a positive directional relationship. Further, the fact that all of the 11 studies that met our rigor criterion were produced since 2003 (see Table 1) provides some reason for optimism that the quality of EET research is improving. Nevertheless, many recent studies have been produced that do not meet a high standard of methodological rigor. In order to improve the literature so that future meta-analyses can provide even more valuable findings for academics and practitioners, EET researchers must include pre- and post-EET interventions (ideally at several points in time post-intervention), and should include treatment and control groups. Souitaris et al. (2007) is an example of a good quasi-experimental design with both pre- and post-intervention surveying and use of treatment and control groups. Where possible, random assignment to treatment and control groups should be carried out, although this is often difficult to implement for both practical and ethical reasons. Oosterbeek et al. (2010) show how it may be possible to reflect random assignment in certain circumstances if researchers are alert to opportunities. In this case, Oosterbeek and his colleagues identified that the creation of a required course in entrepreneurship for a business program at one campus of a school and lack of any similar course offering at another campus of the same school created an opportunity to examine the outcomes of the new entrepreneurship course in a manner that is very close to that of a randomized experiment. Although the participants were not truly randomized, the authors confirmed through extensive post hoc analysis that the only notable difference between the two groups was their physical proximity to each campus. In terms of data reporting, researchers should include correlation tables and estimates of reliabilities. Also, at a minimum, researchers should report the correlations amongst all study variables in a manner that is consistent with data reporting in other fields of management research (e.g., providing a zero-order correlation matrix). Such reporting improvements will greatly increase the ability of future researchers to make accurate claims about the effect sizes of EET on its outcomes. It will also increase the ease of conducting future meta-analytic reviews of the EET literature. In terms of potential moderators, where possible, future research should include measures of age, education level, academic institution, academic program, course type (e.g., required versus elective), ethnicity, nationality, gender, previous entrepreneurship experience, previous employment experience, and participants' perception of course goal, level, and content. We also recommend obtaining and reporting the main elements of the course syllabus, so that future research can control for the potential impact of course content and structure. This type of information might provide valuable insight into whether such things as course content (e.g. lecture material, guest speakers, online resources, modes of delivery, etc.), and course goals (e.g., learning introductory concepts and theory compared to learning specific skills) influence the outcomes of EET interventions. Examining different methods of employing experiential exercises, such as use of online venture creation simulations versus actual venture creation projects may also help explain variation in EET outcomes. Other examples of the types of moderation elements that may be fruitful to examine can be found in several studies included in this meta-analysis. Hanke et al. (2010) compared two undergraduate university entrepreneurship courses; a new course that employed problem based learning with a more traditional lecture-style course. Cooper and Lucas (2007) compared two training programs that incorporated different levels of interactive and experiential learning. The literature would benefit greatly from more examination of this type of course content variation, especially in studies that employ full quasi-experimental methods and examine both entrepreneurship-related human capital assets as well as entrepreneurship outcomes. Future research might also examine differences in course instructors, such as the skill and/or background of course instructors (e.g. experienced entrepreneur versus academic), or teaching methods employed. For instance it may be that instructors who do not engage their students via class discussion or fair procedures, for example, are less motivating than instructors who are perceived to be more engaging or fair, even if the same course materials are used in both cases. Future meta-analyses of EET outcomes would provide greater insight if these types of moderator variables were examined and reported in all new studies. 6.4. Concluding remarks We have provided what we believe to be the first meta-analytic examination of the relationship between EET and both entrepreneurship-related human capital assets and entrepreneurship outcomes. Our results were supportive of the notion that entrepreneurship-specific human capital formation can be influenced by entrepreneurship-specific education. This is important, given the heretofore equivocation of the narrative literature reviews on the subject, particularly in light of the immense growth and investment in entrepreneurship education on a global scale. We have also provided recommendations for improving the literature to enhance the value of future EET research results.