Propensity score modeling for business marketing research
本文为商业营销研究者提供倾向得分建模（PSM）的实用指南，澄清其适用条件、常见误解和报告要点，并附Stata操作示例与核查清单，帮助提升研究严谨性。
Propensity score modeling (PSM) is a powerful statistical technique that, in the appropriate data contexts, addresses biases from confounding and selection, which can otherwise distort results and lead to erroneous inferences. However, while the number of PSM applications in business marketing research is growing, many studies mistakenly assume that PSM is a universal solution for all endogeneity issues. Often, studies lack sufficient detail about the specific endogeneity problem they aim to address, which is a critical issue, as PSM is appropriate only for certain types of endogeneity. Additionally, essential tests to confirm the validity and robustness of PSM results are frequently overlooked or insufficiently reported, raising concerns about the reliability of findings. This article aims to enhance the rigor of PSM applications in business marketing research by offering updated practical guidance on its appropriate use, key aspects to report, and common misconceptions and errors to avoid. A practical example of PSM implementation in Stata is included, along with a comprehensive checklist of justifications and best practices to guide business marketing researchers in their future PSM-based studies. • PSM is a powerful statistical technique that, in the appropriate contexts, addresses endogeneity. • However, it is not a universal solution for all endogeneity issues. • Our review of prior applications in business marketing reveals several misconceptions. • We clarify in which data scenarios PSM is appropriate and which aspects to report. • A practical example of PSM implementation in Stata is provided.