Sampling on the dependent variable is not always that bad: Quantitative case-control designs for strategic organization research

摘要

We are curious about breakthrough innovations, intrigued by megamergers, inquisitive about dramatic collapses. In a nutshell, we are fascinated by outliers. More importantly, our field is fundamentally about the abnormal. Competitive advantage, above-normal returns, it's all about standing out from the crowd. As Starbuck (2006: 150) puts it, 'Studies of exceptional performance are inevitably concerned with extreme cases, and to understand the requirements of exceptional performance, one must study extreme cases, not averages.' Two other reasons push for studying extremes. First, 'extreme cases can expose overlooked causal factors [and] discourage overgener-alization and foster appreciation for individuality and variety' (Starbuck, 2006: 149). Second, extremes might have greater importance than shown by the distribution tail where we confine them. Whereas we're used to think in terms of Gaussian distributions, it seems that many management phenomena exhibit Paretian distributions (Baum and McKelvey, 2006; McKelvey and Andriani, 2005).