RESULTS

The policy capturing experiment provides many interesting insights. In general, VCs have a difficult time introspecting about their decision process. As such, past research needs to be considered in a cautious manner. The following paragraphs further explore the results.

All 51 VCs demonstrate statistically significant policy equations at the .01 level or better. Adjusted R-squares vary from .35 to .85 for the VCs' policy equations. Table 2 summarizes the results. Specifically, Table 2 details how important each information cue is to the VC decision for both her/his actual and stated decision policies. The next section discusses these results further as they relate to each of the hypotheses.

The results of Hypothesis 1 which suggests that VCs do not understand their decision process can be answered by first focusing on treatment 1. By visually examining the weights and respective ranks for each cue, it appears that VCs in the base treatment have a strong understanding of their decision process. The rank order is generally the same, except VCs believe market growth is more important than product proprietary (stated decision policy) but actually weigh the proprietary product factor higher (actual decision policy). The rank between these two factors is reversed. Thus, at initial glance, Hypothesis 1 is not supported. However, a more rigorous test of understanding can be achieved by studying whether the actual decision policy or the stated decision policy explains more variance of the VC's actual decision (Summers, Taliaferro & Fletcher, 1970 --see Table 3). Within the cognitive cues treatment, the actual decision policy explains 13 percent more of the variance than the stated decision policy. This difference means that the actual decision policy captures the VC's true decision policy better than the VC does. In other words, VC understanding isn't perfect. Nonetheless, the strong rank order correlation between the actual and stated decision policies suggests good understanding. Thus, Hypothesis 1 receives mixed support.

Hypothesis 2 which suggests that more information decreases VC understanding is supported. A comparison of each VC's actual decision policy with her/his stated decision policy (see Table 5 -- Treatment 2) indicates that understanding may be low. Note that in Table 2, VCs are relying heavily on number of competitors and competitor strength (actual decision policy), yet they don't believe that to be the case (number of competitors and competitor strength are believed to be the sixth and seventh most important out of the eight factors presented in the VC's stated decision policy). VCs in Treatment 2 also believe that they are using leadership much more so than they are (ranked as most important in stated decision policy but is actually seventh out of eight in actual decision policy). Perhaps even more striking is that the VCs within Treatment 2 seem to understand their actual decision policy even less then their peers within Treatment 1. In Treatment 2, the VCs' stated decision policy explains 25 percent less variance than their actual decision policy (versus only a 13 percent drop in Treatment 1). When there is more information, VCs' understanding of their actual decision policy is greatly diminished.

Hypothesis 3 which suggests that VCs using an optimal set of information factors can better introspect about their decision policy than those using an intuitive set of cues receives conditional support. It appears that VCs don't appreciate the importance of product superiority (see Table 2). Although product superiority is ranked as most important in both the stated and actual decision policies, the magnitude of that importance isn't fully appreciated. In their stated decision policy, VCs typically rate product superiority 2.3 to 3 times more important than the other factors. However, in their actual decision policy, product superiority is 3.5 to 5.5 times more important than the other factors. Within Treatment 3, the decrease in explained variance is 19 percent which is also greater than the decrease in Treatment 1. However, VC understanding is greater than those VCs in Treatment 2 (19 versus 25 percent). Thus, VCs are more aware of their thought process using optimal cues conditioned upon the number of intuitive cues they use. Since VCs typically rely on far more than the four to eight cues provided in this experiment, whereas optimal models rarely exceed three to seven cues (Stewart, 1988), it can be posited that VCs using optimal cues have better understanding. Thus, Hypothesis 3 receives conditional support.

Hypothesis 4 which suggests that VCs are consistent in applying their decision policies even if they do not consciously understand that policy is supported. Multiple R from the regression analysis of each individual gauges consistency (see Table 4). It appears that VCs across treatments are very consistent in applying their actual decision policies. The very high Multiple R within Treatment 3 is likely a function of the overwhelming reliance on the product superiority cue. Consistency is also tested via a series of repeated cases (see Table 4). If the VCs are within +/- 1 of their initial success assessment on the Likert Scale, then they are considered to be consistent. As Table 4 illustrates, VCs are consistent on approximately four out of five repeated cases on average. Again, these results suggest that VCs are very consistent in applying their actual decision policies. Thus, Hypothesis 4 is supported.

Top of Page
Previous Page | Main Page | Next Page


1997 Babson College All Rights Reserved
Last Updated 4/2/97 by Cheryl Ann Lopez

To sign-up for the Center for Entrepreneurial Studies' publication lists,
please register with the
Entrepreneurship WebTeam.