Testing alternative measures of subjective norms in entrepreneurial intention models

Running Head: TESTING ALTERNATIVE MEASURES OF SUBJECTIVE NORM Testing Alternative Measures of Subjective Norms in Entrepreneurial Intention Models TESTING ALTERNATIVE MEASURES OF SUBJECTIVE NORM 2 Abstract Although the Theory of Planned Behavior can be applied to entrepreneurship, study outcomes concerning the role of the subjective norm construct vary substantially. Using data from nine countries exhibiting substantial dissimilarities in history, culture, economy and population size, this study investigates two critical topics held responsible for the conflicting outcomes in research for entrepreneurial intention: (i) measurement issues and (ii) possible indirect influences between the motivational antecedents of intention. It also, at the same time, takes into account the cultural and macroeconomic environment. Regarding the competing methods to measure subjective norms, linear regression and structural equation both suggest that the frequently-applied product of normative beliefs and motivation to comply is not, in the case of entrepreneurial behavior, the best option. A single multi-item subjective-norm measure offers better results, since this type of measurement significantly improves the performance of the subjective norm component. Furthermore, although structural equation modeling reveals a good fit of the original Theory of Planned Behavior model, the internal consistency and explanatory capacity of the constructs is higher when the modified model is tested. This latter model takes into account the indirect influences of subjective norms on attitudes towards entrepreneurship and on perceived behavioral control, suggesting that social pressure would affect personal preferences. These results should serve as a guide for future theoretical and empirical development and put existing study outcomes into perspective. Author Keywords: Entrepreneurial intention, Theory of planned behavior, Subjective norm, Measurement, Structural equation modeling TESTING ALTERNATIVE MEASURES OF SUBJECTIVE NORM 3 Introduction Research on venture creation and on what drives the decision to become an entrepreneur is an eclectic line of work that covers a broad range of subjects. The focus of this field of research has gone through dramatic changes over the past decades. While investigation at the individual level has continued to be at the centre of entrepreneurship research, one of the major developments has been a shift from the traditional focus on the psychological characteristics and personality traits, toward individual behavior and cognitive aspects (Baum & Locke, 2004; Brandstatter, 1997). Gartner (1989) emphasized the need to study individuals before the entrepreneurial event, in order to overcome the disadvantages of ex-post rationalization trends and, thus, to develop good predictors of entrepreneurial behavior. However, one of the major problems of ex-ante approaches in entrepreneurship research is the unpredictable and usually long-time lag between the measurement and the actual observation of the behavior (Reynolds, 1997). Creating a business is a complex decision process, taking place over a certain span of time (Bird, 1992) and is largely under volitional control. Investigating entrepreneurial intentions is a frequently applied method in response to this time-delay problem. Entrepreneurial intentions can be seen as the initial step towards launching a business (S. H. Lee & Wong, 2004) and, in line with this, intentions have been found to be the single best predictor of the subsequent entrepreneurial behavior (L. Kolvereid, 1996; Krueger, 2003; Krueger, Reilly, & Carsrud, 2000). The present paper is based on Ajzen’s Theory of Planned Behavior (“TPB”), one of the most widely used theoretical approaches to understand and foresee a behavior by investigating intentions. Since its inception, a substantial quantity of studies in a wide range TESTING ALTERNATIVE MEASURES OF SUBJECTIVE NORM 4 of domains have applied it to various kinds of behaviors (for an online list of references, see Ajzen, 2010) and an important amount of them have provided broad support for this theory. Entrepreneurship represents one of those behaviors for which the applicability of the TPB has been shown (Autio, Keeley, Klofsten, Parker, & Hay, 2001; L. Kolvereid, 1996), giving the TPB a highly influential role in entrepreneurship research during the last decade. Said theory states that the attitudes towards starting-up (i.e. the individual valuation of the possible outcomes of setting up a venture), the subjective norms (i.e. the perceived social pressure to create a venture or not to start-up) and the perceived behavioral control (i.e. the perceived ease or difficulty of venture creation) are predicting intentions. Although empirical studies based on the TPB are increasingly popular, comparisons between these analyses nevertheless remain difficult. These difficulties especially concern the subjective norm construct, with: (1) The multiple ways applied to measure the construct: While some previous studies in entrepreneurship were using the products of the ‘normative beliefs’ with the ‘motivations to comply’ (L. Kolvereid, 1996; Kolvereid & Isaksen, 2006), others used simplified measures (Autio et al., 2001; Krueger et al., 2000). Other researchers, in turn, simply omitted the subjective norm completely (Peterman & Kennedy, 2003). (2) The varying role played by the subjective norm in the model: Study outcomes concerning the role of the subjective norm in the Theory of Planned Behavior have also been varying substantially. Some meta-analyses (Armitage & Conner, 2001) generally accredited the weak predictive power of the subjective norm in the Theory of Planned Behavior. In the case of entrepreneurship, some researchers (Liñán & Chen, 2009) TESTING ALTERNATIVE MEASURES OF SUBJECTIVE NORM 5 argued that this weakness would be explained by the presence of interactions and indirect effects of the subjective norms on the entrepreneurial intention. (3) The differing environments of the study samples: It has also been argued that the subjective norm reflects the perception the individual has of the prevailing values in its environment. As such, it may be affected by the predominant cultural characteristics of the surrounding society (Ajzen, 1991; Davidsson & Honig, 2003; De Carolis & Saparito, 2006). Cultural norms affect the start-up intention (Brännback, Carsrud, Krueger, & Elfving, 2008). When judging behaviors, the macroeconomic environment is also considered. The impacts of cultural and societal variations at the individual level, however, have largely been ignored, and cross-cultural analyses are still rare. There therefore is a need for research to clarify these discrepancies. While analyses based on intentions significantly increased in number during the last decade, a lot remains to be done in order to overcome contradictions and to achieve a better comprehension of the factors influencing entrepreneurial awareness. In line with these needs, this paper will therefore contribute to a better understanding of three issues: (i) the operationalisation, (ii) the role of the subjective norm and (iii) environmental aspects. First, some of the current measurement practices will be compared, using not simply one but, rather, two different models (an original TPB and a modified TPB) and, by doing so, taking into account possible roles of the subjective norm construct in the Theory of Planned Behavior model applied to the entrepreneurial behavior. This study will thus add to a better TESTING ALTERNATIVE MEASURES OF SUBJECTIVE NORM 6 understanding of the cognitive mechanisms acting in the configuration of entrepreneurial intention. Second, in order to address the environmental aspects, a cross-national study is considered, in which nine culturally different countries are analyzed. The selected countries exhibit substantial dissimilarities in history, culture, economy and population size: European Union countries (Germany, the Netherlands, Spain), a prosperous European non-EU country (Norway), countries linked to each other by a common past in the British Empire (Australia, Canada), transition economies (Romania, Russia) and a representative of emerging Latin America (Mexico). The present paper is structures as follows: following this introduction, the second section will reconsider previous contributions and introduce the hypotheses. A third section will present some methodological issues, and will be followed by the actual empirical analyses. The final two sections will cope, respectively, with the discussion and the conclusion. TESTING ALTERNATIVE MEASURES OF SUBJECTIVE NORM 7 Theoretical Background The Subjective Norm and the Entrepreneurial Intention Model According to Ajzen (1991), the relative importance of the different motivational antecedents – attitudes toward the behavior, the subjective norm and perceived behavioral control – in predicting intention varies. Depending on the type of behavior and the nature of the situation, one or two intentional antecedents might dominate over the other(s) when predicting the intention. Sparks, Hedderley, & Shepherd (1992), for instance, found evidence for the negative relationship between the variability of attitudes and the perceived control, and thus improved the relative predictive power of attitudes. Although the TPB can be applied to entrepreneurial behavior (Autio et al., 2001; Lars Kolvereid, 1996), there are conflicting outcomes between the relative contribution of the attitudes, the subjective norms and the perceived behavioral control. Generally, three critical issues have been held responsible for the conflicting outcomes in research on entrepreneurial intention. The first of them may be linked to measurement issues (Chandler & Lyon, 2001) with respect to the subjective norm, for which contradictions have been especially striking. In this sense, the origin of entrepreneurship is set in social processes (Katz, 1992) and individual discrepancies in sociability may boost the predictive power of the subjective norm (Trafimow & Finlay, 1996). However, many studies excluded the subjective norm from their analyses (Peterman & Kennedy, 2003). Therefore, this factor -which reflects a person’s perceived social pressure to perform a specific behavior or notdemands more attention. Ajzen (2005) defined subjective norms as the sum of the products of the ‘normative beliefs’ of individuals about the perceived opinion of reference people (e.g., family, closest TESTING ALTERNATIVE MEASURES OF SUBJECTIVE NORM 8 friends, other important people), with the ‘motivations to comply’ (the extent to which the individual cares about those perceived opinions). Nevertheless, significant dissimilarities can be found in the quantification of this cognitive variable in entrepreneurship research. Amongst those entrepreneurship scholars including subjective norm in their analyses, some of them followed Ajzen’s approach (L. Kolvereid, 1996; Kolvereid & Isaksen, 2006; Tkatchev & Kolvereid, 1999; van Gelderen et al., 2008). Others chose simplified measures of subjective norms, reflecting for instance the environment of students (Autio et al., 2001; Liñán & Chen, 2009) or measuring on a 0 - 100 scale the opinion of the respondents on whether their family and friends would want them to start up (Krueger et al., 2000). In studies using simplified items, the subjective norm tends to be non-significant, while those applying the more complex measure ‘subjective norms x motivations to comply’ confirmed that the subjective norm significantly explains venturing intention. In their metaanalysis, Armitage & Conner (2001) considered the kind of subjective norm measures that were applied (multiple items, single item, normative beliefs x motivations to comply, social support, normative beliefs and other) as a moderating variable of the relationship between the subjective norm and intentions. Multiple item measures of normative beliefs scored generally better than ‘normative beliefs x motivations to comply’ measures. The second critical issue relates to the possible existence of indirect influences between the motivational antecedents of entrepreneurial intention. Ajzen (1991) himself indicated that the specific pattern of relationships between the three motivational antecedents had to be established for every specific behavior. In particular, there are arguments to consider a different pattern. In this sense, the literature on social capital stresses the role of networks in transmitting values and norms which, in turn, would affect motivations (Casson & Della Giusta, 2007; Naphiet & Ghoshal, 1998). TESTING ALTERNATIVE MEASURES OF SUBJECTIVE NORM 9 Thus, the subjective norm, to the extent that it reflects perceived values in the individual’s environment, may be considered as a form of cognitive social capital (De Carolis & Saparito, 2006; R. Lee & Jones, 2008). Therefore, these values –received from others aroundmay partly determine personal attitude and perceived behavioral control (Liñán & Santos, 2007). Since this line of reasoning may be contradictory with the simpler initial TPB model, a comparison of both alternatives may be justified. Consequently, this study will try to gain a better understanding of these indirect effects. Two versions of the intention model will be taken into consideration. a) Model A is the original Theory of Planned Behavior. This model includes possible impacts of personal attitudes (PA), subjective norms (SN) and perceived behavioral control (PBC) on intention, but also considers the correlations between each of the three antecedents of intention. b) Model B is a modified model in which subjective norms (SN) explain personal attitude (PA) and perceived behavioral control (PBC), whereas the three motivational constructs explain the entrepreneurial intention. Based on the literature, we therefore pose this initial set of hypotheses about the alternative models and measures to test the effect of subjective norm on the entrepreneurial intention: H1a. Subjective norms have a positive direct impact on entrepreneurial intentions (models A and B). TESTING ALTERNATIVE MEASURES OF SUBJECTIVE NORM 16 have been validated by (Iakovleva et al., 2011), showing adequate reliability (Cronbach’s alphas over 0.8 for all measures except PBC, for which it was 0.71). Regarding national-level variables, the main economic indicators have been taken from the World Development Indicators on-line database (by the World Bank): Gross National Income per capita (GNIpc) at purchasing power parity current US$ (PPP$, 2006) and time required to start a business (number of days, 2006). Additionally, the unemployment rate in 2006 was obtained from the International Labor Organization statistical database. Cultural dimensions were taken from Hofstede (2003). Results First, a stepwise linear regression analysis was performed, introducing subsequent blocks of variables, as shown in Table 1. The economic variables (Model 2), the cultural variables (Model 3), intentional antecedents - attitudes and PBC - (Model 4) were successively added to the Model 1 that includes the gender control variable. Finally, the two measures of the subjective norm were introduced alternatively (Model 5 and Model 6). Models 2 and 3 are thus dealing with the hypotheses 2 and 3, while models 5 and 6 are used to test the relative contribution of each of the competing measures of subjective norm to explaining the entrepreneurial intentions (hypotheses 1a, 1b and 1c). Multicolinearity tests were performed and this was found to be present. For this reason, some of the control variables had to be left out. The SPSS 17.0 software has been used for the analysis. Several authors have suggested the presence of complex inter-relationships among the variables in the model, starting from Ajzen (1991) himself. Therefore, Structural Equation Modeling (“SEM”) is a more adequate technique to be used in this case. SEM also allows the TESTING ALTERNATIVE MEASURES OF SUBJECTIVE NORM 17 comparison of Ajzen’s original model with the modified version of the TPB by Liñán and Chen (2009), which has also been tested in the present study. Second, thus, following the regression, a more rigorous and adequate comparison - a structural equation analysis - has been performed using AMOS 17.0. This covariance-based software is quite similar to LISREL and others. The main reason for choosing it was its integration with SPSS. The figures 1 and 2 present the results of the SEM . More specifically, figure 1 shows the original TPB with the two alternative measures of subjective norm, while figure 2 presents the results for the modified TPB with the alternative measures of the subjective norms, thus dealing again with hypotheses 1a, 1b and 1c. Finally, the path coefficients of the control variables for the original and for the modified intention models are shown in the tables A1 and A2. Results of the Linear Regression Model 1 includes gender as a control variable. Results indicate that being male increases the start-up intention, but explanatory power is very limited (only 2.1% of explained variance). Originally, age was also included, but as new regressors were added, it presented collinearity with the intercept, and was not significant. Therefore, it was removed from the analysis. Since the age span of the respondents is quite narrow, it is not surprising that no statistically-significant relationship is found. In Model 2 and 3 economic and cultural variables have been successively added. Nevertheless, since only 9 countries were included in the analysis, national variables presented a considerable degree of multicollinearity. Therefore, some of them had to be removed, until the problem was solved in all 6 models. As a result, only three economic variables (the GNI per capita, the unemployment level, and the days to start-up) and two TESTING ALTERNATIVE MEASURES OF SUBJECTIVE NORM 18 cultural dimensions (Individualism and Masculinity) were tested. A larger number of countries would be needed to test the remaining variables or, alternatively, individual level data would be necessary. The inclusion of economic variables raises R2 to 0.157. As may be expected, lower unemployment and fewer days required to start-up result in higher entrepreneurial intention, although this latter variable is not significant in Model 2 (and only marginally in Model 6). Notably, the larger negative effect is that of GNIpc. This is consistent with the fact that three middle-income economies (Mexico, Romania and Russia, with GNIpc below 14000 PPP$) are included together with six high-income ones (with values above 28000 PPP$). The former group presents considerably-higher intention levels. On the contrary, the effect of cultural variables, though significant, is very small (the increase in R2 is only 0.006). In particular, lower intention levels characterize countries with an individualistic culture. This may be explained, the same as with the income level, by the fact that higher income countries are, in general, more individualistic than the rest. In this sense, the inclusion of cultural variables lowers the beta coefficient of GNIpc, suggesting this relationship. Overall, results from models 3 and 4 yield support for hypotheses H2 and H3. INSERT TABLE 1 ABOUT HERE Model 4 includes individual-level motivational variables (PA and PBC). Variance explained increases substantially to 0.651. At the same time, the coefficients of other variables from Model 3 drop a great deal, which would be indicating the presence of some interaction and/or indirect effects. That is, demographic, socioeconomic and cultural TESTING ALTERNATIVE MEASURES OF SUBJECTIVE NORM 19 variables affect motivations and, through them, intention. In this case, Structural Equation Models are a more adequate instrument to be used (see below). In particular, once PA and PBC are included, individualism becomes non-significant. Models 5 and 6 are a direct comparison of the alternative measures proposed for the subjective norm. The beta coefficient for the SNxMC (subjective norms pondered by their respective motives to comply, Model 5) is 0.126, significantly increasing R2 by 0.012. Alternatively, when the simple SN measure is included (Model 6), the resulting beta coefficient is 0.204, which implies a larger increase in R2 of 0.025. Besides, the coefficients for other variables are essentially similar. Therefore, although both coefficients are significant, the contribution of the simple SN measure is greater. Support is thus found for hypotheses H1a and H1c. In particular, the coefficients for PA and PBC are notably lower when an SN measure is included. A first interpretation of this result would be that the path coefficients of PA and PBC in Model 4 would be reflecting not only their own effect on intention, but also the indirect effect of SN. Therefore, as was indicated in the theory section, there are arguments to suggest a more complex pattern of relationships between the motivational antecedents of intention, with the probable existence of indirect and interaction effects between them. This is in accordance with hypothesis H1b. For this reason, structural equations models have been used to achieve a more rigorous comparison of the two alternative SN measures proposed in this study. Figures 1 and 2 summarize the results. Results of the Structural Equations The test of the original TPB model proposed by Ajzen (1991) is presented in Figure 1. The analysis has been performed including control variables (gender, national economic TESTING ALTERNATIVE MEASURES OF SUBJECTIVE NORM 20 variables and cultural dimensions) but, to preserve the clarity of the presentation, only the relationships between the central TPB constructs are presented in the Figure. In the appendix, Table A1 includes all other path coefficients between control variables and the original TPB- model constructs. For every construct, indicator or path relationship, the first figure corresponds to the model with the SNxMC measure, whereas the second one relates to the model with the SN-simple measure. INSERT FIGURE 1 ABOUT HERE As shown in Figure 1, fit indexes are similar (comparative fit index, CFI, goodness of fit index, GFI, or root mean squared error of approximation, RMSEA) for both models. Nevertheless, when the simple SN measure is used, squared multiple correlations of the SN manifest indicators are higher (0.71 for SN1, 0.80 for SN2 and 0.92 for SN3, instead of 0.67, 0.69 and 0.81 for the SNxMC measures, respectively), and standardized regression weights are also higher (0.84, 0.90 and 0.96, instead of 0.82, 0.83 and 0.90 for the SNxMC measures). Thus, internal consistency is greater for the SN-simple measure. Similarly, correlations among latent motivational antecedents are higher (0.61 for the SN-PA measure and 0.47 for the SN-PBC measure). Finally, the path coefficient from SN to Entrepreneurial Intention is also higher (0.14, instead of 0.09 for the SNxMC measure). Hypotheses H1a and H1c are supported in model A (original TPB model). Model B, a modified version of the TPB model (Liñán & Chen, 2009), was also tested. Results are presented in Figure 2. As before, only central constructs are shown, while TESTING ALTERNATIVE MEASURES OF SUBJECTIVE NORM 21 Table A2 presents the corresponding path coefficients from control variables on the modified TPB-model constructs. INSERT FIGURE 2 ABOUT HERE In this modified version, subjective norm is hypothesized to explain PA and PBC (H1b). Results for this model are also satisfactory in terms of model fit and path coefficients. As in the previous model, the simple SN measure seems to offer slightly better results both in terms of the internal consistency of the construct (in relation to its indicators) and also in its explanatory capacity with regards to attitude (a path coefficient of 0.60 for the simple SN measure instead of 0.50 for the SNxMC measure), to PBC (0.47 instead of 0.37) and even to entrepreneurial intention (0.14 instead of 0.08). Besides, variance explained for attitude and PBC is clearly higher when the SN-simple measure is used, reinforcing the support for H1b. Meanwhile, H1a and H1c are also supported in model B. Finally, regarding the control variables, their coefficients are the same for both structural models and for each of the alternative SN measures (Tables A1 and A2), indicating that the effect of these variables is not affected by the way in which SN is operationalized. Overall, hypotheses H2a and H2b are supported, since higher income and unemployment result in lower intention. The negative effect of the unemployment rate on the subjective norm is only found for the SN-simple measure, whereas in the case of income it is found for both measures. The days-to-startup negative effect on intention is only indirect, through PBC. On the other hand, cultural variables exert a weak effect on intentions and the subjective norm (but again, the effect is greater for the SN-simple measure). Countries where TESTING ALTERNATIVE MEASURES OF SUBJECTIVE NORM 22 masculinity is higher tend to exhibit lower intention among their individuals while, at the same time, this cultural dimension is also associated with a more favorable subjective norm. Alternatively, individualism is associated with a less favorable PA. Discussion The originality of this study lies in its attempt to determine the specific role of subjective norms in a setting of a nine-country sample. In this context, two competing measures of the subjective norm have been tested on two alternative intention-model typologies. Until now, the application of intention models to entrepreneurial behavior has resulted in conflicting outcomes concerning subjective norms, to the extreme of simply neglecting this intentional antecedent completely, despite recent evidence of its significant contribution in multi-country analyses (Engle et al., 2010). The present study sheds some light on this problem. The results clearly indicate that studies neglecting subjective norms (Peterman & Kennedy, 2003), indeed fail to take into account a significant parameter. The subjective norm has a positive effect not only directly on entrepreneurial intention (H1a), but also on PBC and PA towards venture creation (H1b). If SN is not explicitly included in the analysis, the coefficients of PA and PBC on intention will be biased, as is derived from the results in Table 1. Based on the support found for Hypotheses H1a and, especially, H1b, it may be concluded that a modified TPB model (in which the subjective norm explains PA and PBC) offers a better fit than the traditional one. Of course, further investigation is needed before this statement is confirmed. In essence, though, what it means is that the opinion of others TESTING ALTERNATIVE MEASURES OF SUBJECTIVE NORM 23 about a behavior has a direct effect on personal attitudinal and control beliefs, and this relationship is unidirectional, and not bidirectional as traditionally assumed. Regarding the competing subjective norm measures, from the results of the linear regression models, it may be argued that a simple measure of the subjective norm offers a better fit than a more elaborated measure based on a product of the subjective norm and motivation to comply. Further, the results from the structural model confirmed those of the linear regression, allowing the conclusion that the type of measurement used has a significant impact on the performance of the subjective norm component. Therefore, it may be concluded that a multiple-item simple subjective norm measure, when compared to subjective norms pondered with the motivation to comply, should be deemed superior in at least three aspects: parsimony, internal consistency and explanatory capacity. That is, the SN simple measure is more parsimonious, it has a higher Cronbach’s alpha, and a higher and more significant path coefficient, together with a better overall model fit. This is in line with indications derived from general meta-analyses of the theory of planned behavior (Armitage & Conner, 2001). This superiority of the simple measure still holds when taking into account the more complex configuration of relationships between the antecedents of entrepreneurial intention (model B) and comparing this more thoroughly by means of structural equations. The results regarding socioeconomic variables seem to confirm those of previous studies. According to the literature, individuals in high-income countries should present a lower start-up intention but, mainly, because of lower necessity motivations. In these countries there will be less people willing to become entrepreneurs, but a higher proportion of them will do so because of an opportunity motivation (van Stel et al., 2003). TESTING ALTERNATIVE MEASURES OF SUBJECTIVE NORM 24 Nevertheless, since this is one of the first research works in which the effect of those variables on the motivational antecedents of intention has been tested, some new interpretations may be proposed. Thus, socioeconomic variables affect all antecedents but, especially perceived control. In this sense, if the market is declining (higher unemployment) or regulation is restrictive (more days to start-up), less people feel able to successfully start a venture. The strongest relationships, though, were established between national per capita income and both entrepreneurial intention and subjective norms. While the first of these relationships has already been well documented, the second is notably under-researched. It would be indicating that a lower wealth level of the population not only directly impacts the personal “cost-benefits” analysis of the potential entrepreneur, but also acts indirectly through the perceived opinions of those with close ties, who are probably doing the same kind of analysis themselves. The results also suggest that cultural variables exert only a modest effect on entrepreneurial intentions and on subjective norms (H3a and H3b). This could be an indication that behind the intuitively often-expected cultural impact on entrepreneurial intention, what actually are hidden are more objective factors, such as population wealth. On this issue, however, the data set has its limitations, as the range of cultures that were tested, even if broad in some aspects, does not really include the more extreme cases. Additional research might here be desirable. Overall, while some of the cultural dimensions weakly influence attitudes, others do so with subjective norms. Meanwhile, none of them has any effect on perceived control. It may be the case that people determine their perceived control based on relatively objective factors (or, better said, their perceptions about the actual socioeconomic situation), whereas culture (values) affects mainly more subjective TESTING ALTERNATIVE MEASURES OF SUBJECTIVE NORM 25 perceptions such as those regarding established rules (subjective norms) and personal preferences (attitudes). Finally, a number of limitations have to be acknowledged in this research. In the first place, the sample is made up of students. This has the advantage of facilitating comparisons, but may not be completely representative of the adult population. Secondly, the number of countries analyzed is limited. Some meaningful results have been found regarding the influence of socioeconomic and cultural variables, but this may be the consequence of the specific countries selected. A generalization of these results should be made with extreme caution. Besides, multicollinearity of some national-level variables is another consequence of the limited number of countries being studied. Therefore, a complete test of the socioeconomic and cultural hypotheses has not been possible. Conclusions This paper has tested two alternative measures of subjective norms in two different models of entrepreneurial intention. The results have favored parsimony, since the simpler subjective norm measure offers better results than the more complex ‘normative beliefs x motivations to comply’ measures. The other result that emerges from this study is that the subjective norm has an influence on PA and PBC. That is, values and beliefs shared by important people around the individual do affect personal preferences towards performing entrepreneurial behaviors and the sense of being able to carry them out. A second objective of the paper was to gain generalizability by testing the hypotheses in an international sample. This would also allow the testing of the influence of socioeconomic and cultural variables. In this sense, the results have been as expected, even TESTING ALTERNATIVE MEASURES OF SUBJECTIVE NORM 32 Lee, S. H., & Wong, P. K. (2004). An exploratory study of technopreneurial intentions: a career anchor perspective. Journal of Business Venturing, 19(1), 7-28. Liñán, F., & Chen, Y. W. (2009). Development and cross-cultural application of a specific instrument to measure entrepreneurial intentions. Entrepreneurship Theory and Practice, 33(3), 593-617. Liñán, F., & Santos, F. J. (2007). Does social capital affect entrepreneurial intentions? International Advances in Economic Research, 13(4), 443-453. Mcgrath, R. G., & MacMillan, I. C. (1992). More Like Each Other Than Anyone Else - a Cross-Cultural Study of Entrepreneurial Perceptions. Journal of Business Venturing, 7(5), 419-429. Meager, N. (1992). Does Unemployment Lead to Self- Employment ? Small Business Economics, 4(2), 87-103. Minniti, M., & Bygrave, W. (1999). The Microfoundations of Entrepreneurship. Entrepreneurship: Theory and Practice, 23(4), 41 - 52. Mitchell, R. K., Smith, B., Seawright, K. W., & Morse, E. A. (2000). Cross-cultural cognitions and the venture creation decision. Academy of Management Journal, 43(5), 974- 993. Mueller, S. L., & Thomas, A. S. (2001). Culture and entrepreneurial potential: A nine country study of locus of control and innovativeness. Journal of Business Venturing, 16(1), 51-75. Mueller, S. L., Thomas, A. S., & Jaeger, A. M. (2002). National entrepreneurial potential: the role of culture, economic development and political history. In M. A. Hitt & J. L. C. Cheng (Eds.), Managing Transnational Firms: Resources, Market Entry and Strategic Alliances (Vol. 14, pp. 221-257). Amsterdam: JAI Press. Naphiet, J., & Ghoshal, S. (1998). Social Capital, Intellectual Capital, and Organizational Advantage. Academy of Management Review, 23, 242-266. TESTING ALTERNATIVE MEASURES OF SUBJECTIVE NORM 33 Noorderhaven, N., Thurik, R., Wennekers, S., & van Stel, A. (2004). The Role of Dissatisfaction and Per Capita Income in Explaining Self- Employment across 15 European Countries, SCALES-paper (Vol. SCALES). Zoetermeer: EIM Business and Policy Research. Ovaska, T., & Sobel, R. S. (2004). Entrepreneurship in Post-Socialist Economies: West Virginia University, Department of Economics. Parker, S. C., & Robson, M. T. (2004). Explaining International Variations in Self- Employment: Evidence from a Panel of OECD Countries. Southern Economic Journal, 71(2), 287-301. Peterman, N. E., & Kennedy, J. (2003). Enterprise Education: Influencing Students’ Perceptions of Entrepreneurship. Entrepreneurship: Theory and Practice, 28(2), 129-144. Prantl, S., & Spitz-Oener, A. (2009). How does entry regulation influence entry into selfemployment and occupational mobility? Economics of Transition, 17(4), 769-802. Reynolds, P. D. (1997). Who starts new firms? - Preliminary explorations of firms-in- gestation. Small Business Economics, 9(5), 449-462. Reynolds, P. D., Storey, D. J., & Westhead, P. (1994). Cross-national comparison of the variation in new firm rates. Regional Studies, 28, 443-456. Ritsilä, J., & Tervo, H. (2002). Effects of unemployment on new firm formaton: micro-level panel data evidence from Finland. Small Business Economics, 19(1), 31-52. Shane, S. (2003). A general theory of entrepreneurship : the individual-opportunity nexus. Cheltenham: E. Elgar. Shane, S., Kolvereid, L., & Westhead, P. (1991). An Exploratory Examination of the Reasons Leading to New Firm Formation across Country and Gender. Journal of Business Venturing, 6(6), 431-446. Shaver, K. G., & Scott, L. R. (1991). Person, process, choice: the psychology of new venture creation. Entrepreneurship Theory and Practice, 16(2), 23-45. TESTING ALTERNATIVE MEASURES OF SUBJECTIVE NORM 34 Sparks, P., Hedderley, D., & Shepherd, R. (1992). An investigation into the relationship between perceived control, attitude variability and the consumption of two common foods. European Journal of Social Psychology, 22(1), 55-71. Thurik, R. A., Carree, M. A., van Stel, A., & Audretsch, D. B. (2008). Does self-employment reduce unemployment. Journal of Business Venturing, 23(6), 673 - 686. Tkatchev, A., & Kolvereid, L. (1999). Self-employment intentions among Russian students. Entrepreneurship & Regional Development, 11(3), 269-280. Trafimow, D., & Finlay, K. A. (1996). The importance of subjective norms for a minority of people: Between-subjects and within-subjects analyses. Personality and Social Psychology Bulletin, 22(8), 820-828. van Gelderen, M., Brand, M., van Praag, M., Bodewes, W., Poutsma, E., & van Gils, A. (2008). Explaining entrepreneurial intentions by means of the theory of planned behavior. Career Development International, 13(6), 538 - 559. van Stel, A., Wennekers, S., Turik, R., Reynolds, P. D., & de Wit, G. (2003). Explaining nascent entrepreneurship across countries, SCALES-paper (Vol. SCALES). Zoetermeer: EIM Business and Policy Research. Verheul, I., van Stel, A., & Thurik, R. (2006). Explaining female and male entrepreneurship at the country level, SCALES-paper (Vol. SCALES). Zoetermeer: EIM Business and Policy Research. Table 1. Linear regression on entrepreneurial intention (standardized coefficients) Variables Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Gender 0.144*** 0.160*** 0.168*** 0.065*** 0.072*** 0.069*** GNI per capita Unemployment Days to Start-up --- --- --- -0.361*** -0.192*** -0.035 ns -0.295*** -0.176*** -0.053 ns -0.127 ** -0.052 * -0.028 ns -0.121 ** -0.057 ** -0.032 ns -0.122*** -0.053 * -0.039 † Individualism Masculinity --- --- --- --- -0.107 ** -0.023 ns -0.015 ns -0.031 ns -0.015 ns -0.040 * -0.008 ns -0.049 * TESTING ALTERNATIVE MEASURES OF SUBJECTIVE NORM 35 Personal Attitude P. Behav. Control --- --- --- --- --- --- 0.683*** 0.146*** 0.631*** 0.124*** 0.582*** 0.099*** Subj.NormxMC Subj.Norm simple --- --- --- --- --- --- --- --- 0.126*** --- --- 0.204*** R2 Adjusted R2 Change in R2 Change in F 0.021 · 0.020 · 0.021 · 26.619*** 0.157 · 0.155 · 0.137 · 68.102*** 0.163 · 0.159 · 0.006 · 4.306 * 0.651 · 0.649 · 0.488 · 878.302*** 0.663 · 0.660 · 0.012 · 43.227*** 0.676 · 0.673 · 0.025 · 95.184*** Significance level: ns = not significant; † = p<0.1; * = p<0.05; ** = p<0.01, *** = p<0.001 TESTING ALTERNATIVE MEASURES OF SUBJECTIVE NORM 36 Figure 1: Original TPB model with alternative SN measures (standardized coeffs.) Attitude Subj. Norm P.Beh. Control Entrep. Intent AT1 AT2 AT3 AT4 AT5 SN1 PBC1 PBC2 PBC3 PBC4 SN2 SN3 EI1 EI2 EI3 EI4 EI5 EI6 EI7 SNxMC/ SNsimple CFI = .947/.950 GFI = .925/.925 RMSEA = .058/.058 .67/.71 .69/.80 .81/.92 .82/.84 .83/.90 .90/.96 .67/.67 .78/.79.45/.46 .80/.80 .63/.63 .31/.32.59/.59 .29/.30 .31/.31 .79/.79.82/.82 .88/.89 .89/.89 .67/.68 .51/.61 .41/.40 .38/.47 .55/.56 .56/.56 .54/.54 .77/.77 .78/.78 .69/.69 .92/.92 .90/.90 .91/.91 .76/.76 .66/.66 .60/.60 .48/.48.85/.85 .82/.82 .82/.82.58/.58.43/.43 .12/.13 .05/.08 .14/.14 .71/.72 .68/.65 .09/.14 .12/.10 TESTING ALTERNATIVE MEASURES OF SUBJECTIVE NORM 37 Figure 2: Modified TPB model with alternative SN measures (standardized coeffs.) Attitude Subj. Norm P.Beh. Control Entrep. Intent AT1 AT2 AT3 AT4 AT5 SN1 PBC1 PBC2 PBC3 PBC4 SN2 SN3 EI1 EI2 EI3 EI4 EI5 EI6 EI7 SNxMC/ SNsimple CFI = .944/.949 GFI = .922/.924 RMSEA = .059/.059 .67/.71 .69/.80 .80/.92 .82/.84 .83/.90 .90/.96 .67/.67 .78/.79.45/.45 .79/.79 .63/.63 .32/.32.61/.60 .27/.29 .31/.31 .80/.79.82/.82 .89/.89 .89/.89 .67/.67 .55/.55 .57/.56 .52/.54 .78/.78 .77/.78 .69/.69 .92/.92 .90/.90 .91/.91 .76/.76 .65/.66 .60/.60 .47/.48.85/.85 .81/.81 .82/.82.57/.58.43/.43 .36/.45 .05/.09 .27/.33 .71/.72 .70/.66 .08/.14 .14/.12 .50/.60 .37/.47