Sampling Issues

Summary and Conclusions

I opened this chapter by arguing that theoretical relevance is the most important criterion for sampling. The composition of units in the sample should match the theory used. Statistical representativeness is desirable when possible to achieve at all, but secondary to theoretical relevance. We have further observed that entrepreneurship research can be conducted on many different levels of analysis, and that each level has its problems that have to be dealt with. It is not always possible to overcome those problems—there is no such thing as “perfect” research—but it is certainly worth trying to solve as many as possible and to be aware of the remaining shortcomings of one’s sample.

Importantly, the most conventional levels of analysis in entrepreneurship research—the individual and the firm—are not markedly less problematic than are other alternatives. This insight should provide incentive for researchers to consider leaving the most trodden paths and apply other levels of analyses than those that first come to mind.