• Keine Ergebnisse gefunden

4.4 Robustness checks

4.4.7 Summary of robustness checks

Figure 26: Model A: Summary of the robustness checks including all interaction effects Source: Own illustration

Figure 26 gives an overview of all five robustness checks that include all interaction effects for Model A. As depicted in Table 6, the results of the hypothesis tests relying on the full data set and the full models show that the negative interaction effects between joint team experience and intra-team trust (H1a), between entrepreneurial experience and intra-team trust (H3a), and between industry experience and intra-team trust (H5a) are significant. Thus, for these three hypotheses, robustness checks support the original results if the coefficients are negative and significant. Otherwise, the robustness checks do not support the original results for these three interaction effects. In contrast, the original results in Table 6 illustrate that H2a, the negative interaction effect between educational experience and intra-team trust, as well as H4a, the negative interaction effect between technological experience and intra-team trust, are not significant. Hence, the robustness checks do not support the original results for these two hypotheses if the coefficient is negative and significant; otherwise, they support the original results. H1a is supported by all robustness checks, with

the exception of the robustness check that only includes start-ups in Munich. As only 39 Munich start-ups participated in our study, the sample for this robustness check is quite small and, consequently, is likely to be underpowered. Thus, the results of this robustness check must be interpreted with care. H3a is supported by all robustness checks, with the exception of the one focused on only one intra-team trust item. At least the coefficient of this robustness check was negative, but it did not meet conventional levels of significance (p < 0.1). However, measuring intra-team trust with only one item is less accurate that measuring it with five items. H5a is clearly supported by three robustness checks. In addition, the result of the robustness check without control variables is negative, but does not reach conventional levels of significance (p < 0.1). The robustness check including only Munich start-ups, with a small sample of 39 teams, is negative but not significant. All robustness checks fully support the rejection of H2a and H4a, as no result is signifi-cantly negative. Despite some isolated robustness check results that do not support the original results, the overall pattern of all robustness checks can be seen as supporting the results of Model A.

The results of the robustness checks that include all five interaction effects for Model B are illustrated in Figure 27.

The results of the comparison between the robustness check and the original findings are similar to those of model A (depicted in Figure 26), as the same three hypotheses are significant. H1b, the negative interaction effect between joint team experience and intra-team trust, is supported by three robustness checks. It is not supported by the test that includes only one intra-team trust item out of five, as the coefficient is negative but does not meet conventional significance levels (p < 0.1). Additionally, H1b is not supported by the robustness check with 39 Munich start-ups. The negative interaction effect between entrepreneurial experience and intra-team trust is found by four robustness checks. Only the robustness check with just one intra-team trust item does not support the original results. H5b is supported by four robustness checks, equally to Model A; only the last robustness check, which focuses on Munich start-ups, does not support the original results. The rejection of H2b and of H4b is found by all robustness checks. In summary, the results of all five robustness checks, including all interaction effects for Model B, can also be seen as supporting the original results.

In addition to the five robustness checks analyzed above in Figure 26 and Figure 27 the following two overviews focus on robustness checks that include individual interaction effects or a combination of interaction effects. Figure 28 gives an overview of these robustness checks for Model A. The first robustness check focuses on individual interaction effects, while the next three robustness checks include the different interaction effects of one category each, namely the team experience, human capital experience, and opportunity-related experience category. In addition, the final three robustness checks concentrate on a combination of categories and include all interaction effects of the respective cate-gories. The results of all seven robustness checks are already described in detail in section 4.4.6 and are shown in Table 13 and Table 14. The robustness check results strongly support H3a, the negative interaction effect between entrepre-neurial experience and intra-team trust (four out of four tests), as well as H5a, the negative interaction effect between industry experience and intra-team trust (three out of four tests).

Figure 27: Model B: Summary of the robustness checks including all interaction effects Source: Own illustration

H1a, the negative interaction effect between joint team experience and intra-team trust, is only supported by the robust-ness check that includes team and opportunity-related experience variables. It is important to mention that robustrobust-ness check 1 and robustness check 2 are identical, as they both solely test the interaction between joint team experience and intra-team trust. Robustness check 1 includes only individual interaction effects, namely between joint team experience and intra-team trust. Robustness check 2 focuses on interaction effects between experience variables of the team expe-rience category and intra-team trust. However, the team expeexpe-rience category consists only of the joint team expeexpe-rience variable. Consequently, robustness check 2 includes only the individual interaction effect between joint team experi-ence and intra-team trust. Thus, H1a is supported at least by one out of three robustness checks. The rejection of H2a and H4a is consistent across all robustness checks that include these interaction effects.

Figure 28: Model A: Summary of the robustness checks for individual interaction effects Source: Own illustration

Figure 29shows the results of the same robustness checks for Model B, which are constructed in a similar way as for Model A in Figure 28. These results are already explained in detail in section 4.4.6and are illustrated in Table 15and Table 16. H1b, the negative interaction effect between joint team experience and intra-team trust, is found by two out of three interaction effects, as robustness checks 1 and 2 are identical for H1b, similar to H1a. The negative interaction effect between entrepreneurial experience and intra-team trust (H3b) is supported by all robustness checks that include this interaction. Two out of four robustness checks support H5b, the negative interaction effect between industry expe-rience and intra-team trust. The rejection of H2b, the negative relationship between educational expeexpe-rience and the selection performance under the condition of intra-team trust, as well as the rejection of H4b, the negative link between technological experience and the selection performance under the condition of intra-team trust, are supported by all robustness checks that include the respective interaction effects.

To summarize for Model B, the results of all seven robustness checks that include only individual interaction effects or combinations of individual interaction effects, but never all five interactions, can be seen as further supporting the original hypothesis tests.

Figure 29: Model B: Summary of the robustness checks for individual interaction effects Source: Own illustration

To test the original results, I ran a total of 24 robustness checks in two steps. First, I conducted five robustness checks including all interaction effects for Model A (cf. Figure 26), as well as the same five robustness checks for Model B (cf. Figure 27). Second, I ran seven robustness checks that focused on individual interaction effects or combinations of interaction effects for each Model (Figure 28andFigure 29). Taking the results of all 24 robustness checks together, the pattern of results for both models is highly consistent with the original results.

5 D

ISCUSSION

In this section, I discuss the implications of the present study's findings based on a short summary of the empirical results. First, I elaborate on the contributions to different literature streams (section 5.1). Second, I outline the implica-tions for practice (section 5.2). Third, I illustrate the limitaimplica-tions of the present study (section 5.3). Fourth, I present potential avenues for future research, followed by a conclusion (section 5.4).

The empirical results outlined in section 4 show that intra-team trust could be a condition that prevents entrepreneurial teams from fully leveraging their available experience, which leads to a lower opportunity recognition performance in both the opportunity recognition (Model A) and the opportunity selection (Model B) phase. In the present study, I identify one experience dimension for each experience category for which intra-team trust negatively influences the relationship between the respective experience dimension and the opportunity recognition performance: The experience dimensions are joint team experience for the team experience category, entrepreneurial experience for the human capital experience category, and industry experience for the opportunity-related experience category. The negative interaction effect between the respective experience dimension and intra-team trust is found for the performance in the opportunity recognition phase as well as in the opportunity selection phase. To put it differently, if there is a negative interaction effect between an experience dimension and intra-team trust, it is supported for both Model A and Model B, not only for one of the two models. Figure 30 gives an overview of the shortened hypotheses, including their status, that are the basis for the following discussion.