The Role of Sampling in Strategic Management Research on Performance: A Two-Study Analysis
通过分析1980-1999年间437项绩效研究，发现不到五分之一使用随机样本；进一步用四种样本检验CEO二元性与绩效的关系，结果截然不同，表明抽样实践需改进。
Strategic management researchers often note the inability of many phenomena to consistently explain organizational performance. Using a two-study format, we examine the possibility that sampling practices may contribute to this inability. We chronicle sampling practices within 437 studies investigating the determinants of organizational performance published between 1980 and 1999 in the Academy of Management Journal, Administrative Science Quarterly, Journal of Management, Organization Science, and Strategic Management Journal. Our results reveal that fewer than one in five studies rely on a random sample and that researchers offer little a priori acknowledgement of sample limitations. Although the first study suggests that sampling issues have been problematic, it cannot offer direct evidence. Thus, in the second study, we examine the effect of sample selection on empirical results by testing one popular relationship from the strategy literature (chief executive officer duality and performance) using four different samples. Whereas a stratified random sample offers some evidence of a positive relationship, a purposive sample detects a negative relationship and an available sample and simple random sample detect no relationship. Taken together, the studies suggest that sampling practices must improve if strategic management is to approach its objective of explaining performance. Accordingly, several exemplar studies are described and recommendations are offered to guide future sampling practices.