The article by Shrout and Bolger discusses the importance of mediation analysis in both experimental and non-experimental studies, emphasizing its role in understanding causal mechanisms. They recommend using bootstrap methods for assessing mediation, especially with small to moderate samples, as these methods are powerful in detecting skewed sampling distributions. The authors critique Baron and Kenny's (1986) approach, suggesting that testing the $X \rightarrow Y$ association for statistical significance should not be a prerequisite when there is a priori belief that the effect size is small or suppression is possible. They provide empirical examples and computer setups for bootstrap analyses. Mediation models are valuable for decomposing interesting associations into components that reveal causal mechanisms, aiding in theory development and testing, as well as identifying potential intervention points. The article reviews statistical approaches to mediation analysis, focusing on the framework proposed by Kenny and colleagues. It highlights the limitations of traditional methods, such as the assumption of linear regression and the use of Sobel's test for indirect effects, which can lead to biased estimates and reduced power. The authors introduce the bootstrap framework as an alternative method for analyzing mediated effects, which can be applied to both experimental and non-experimental data. They demonstrate how the bootstrap can provide more accurate confidence intervals and test statistics, especially when the indirect effect is skewed. The article includes numerical examples to illustrate the application of bootstrap methods and discusses the implications for interpreting mediation results, including the detection of partial mediation and suppression effects. Overall, the article advocates for the use of bootstrap methods in mediation analysis, providing a more robust approach to understanding complex causal relationships.The article by Shrout and Bolger discusses the importance of mediation analysis in both experimental and non-experimental studies, emphasizing its role in understanding causal mechanisms. They recommend using bootstrap methods for assessing mediation, especially with small to moderate samples, as these methods are powerful in detecting skewed sampling distributions. The authors critique Baron and Kenny's (1986) approach, suggesting that testing the $X \rightarrow Y$ association for statistical significance should not be a prerequisite when there is a priori belief that the effect size is small or suppression is possible. They provide empirical examples and computer setups for bootstrap analyses. Mediation models are valuable for decomposing interesting associations into components that reveal causal mechanisms, aiding in theory development and testing, as well as identifying potential intervention points. The article reviews statistical approaches to mediation analysis, focusing on the framework proposed by Kenny and colleagues. It highlights the limitations of traditional methods, such as the assumption of linear regression and the use of Sobel's test for indirect effects, which can lead to biased estimates and reduced power. The authors introduce the bootstrap framework as an alternative method for analyzing mediated effects, which can be applied to both experimental and non-experimental data. They demonstrate how the bootstrap can provide more accurate confidence intervals and test statistics, especially when the indirect effect is skewed. The article includes numerical examples to illustrate the application of bootstrap methods and discusses the implications for interpreting mediation results, including the detection of partial mediation and suppression effects. Overall, the article advocates for the use of bootstrap methods in mediation analysis, providing a more robust approach to understanding complex causal relationships.