An Explanation of Path Analysis and Recommendations for Best Practice
解释了路径分析与OLS或工具变量估计的等价性，指出当前研究常忽视内生性假设或未披露关键假设，并给出使用路径分析的最佳实践建议。
ABSTRACT Path analysis has become increasingly popular, but many studies do not show a deep understanding of how path analysis works or the assumptions on which it relies. In this paper, we explain that path analysis is statistically equivalent to either OLS when the researcher assumes uncorrelated errors, or instrumental variable (IV) estimation when the researcher allows correlated errors and obtains identification using exclusion restrictions. We then identify two problems with the way path analysis is used. First, studies claim that they use path analysis to provide evidence on the causal process, but they assume away endogeneity by imposing the unrealistic assumption of uncorrelated errors. Second, many studies do not explicitly disclose their key assumptions, including the assumptions of uncorrelated errors or exclusion restrictions. This nondisclosure makes it difficult for a reader to determine whether endogeneity is assumed away or whether the study is attempting to address endogeneity. We conclude with detailed guidance for researchers who are considering whether to use path analysis in their research.