The article provides a comprehensive overview of mediation analysis in psychological research. It begins by defining mediating variables and distinguishing them from confounders, moderators, and covariates. The authors discuss the historical and methodological importance of mediation in psychology, highlighting its role in understanding causal mechanisms and improving the design of interventions. The article then delves into statistical methods for assessing mediation, including the single-mediator model and its extensions. It covers various approaches such as causal steps, difference in coefficients, and product of coefficients, along with their assumptions and limitations. The authors also discuss the importance of standard errors, confidence intervals, and significance testing in mediation analysis. The review explores extensions of the single-mediator model, including multilevel mediation, mediation with categorical outcomes, multiple mediators, and longitudinal mediation. It addresses the challenges and considerations in each of these extensions. Additionally, the article discusses the relationship between moderation and mediation, presenting models such as moderated mediation, mediated moderation, and mediated baseline by treatment moderation. These models help in understanding how moderators influence the mediated effects. Finally, the authors address the causal inference aspect of mediation, critiquing traditional regression-based methods and introducing new approaches based on principal stratifications and latent class models. They emphasize the importance of causal interpretation in mediation analysis and suggest future research directions to improve the understanding and application of mediation in psychological research.The article provides a comprehensive overview of mediation analysis in psychological research. It begins by defining mediating variables and distinguishing them from confounders, moderators, and covariates. The authors discuss the historical and methodological importance of mediation in psychology, highlighting its role in understanding causal mechanisms and improving the design of interventions. The article then delves into statistical methods for assessing mediation, including the single-mediator model and its extensions. It covers various approaches such as causal steps, difference in coefficients, and product of coefficients, along with their assumptions and limitations. The authors also discuss the importance of standard errors, confidence intervals, and significance testing in mediation analysis. The review explores extensions of the single-mediator model, including multilevel mediation, mediation with categorical outcomes, multiple mediators, and longitudinal mediation. It addresses the challenges and considerations in each of these extensions. Additionally, the article discusses the relationship between moderation and mediation, presenting models such as moderated mediation, mediated moderation, and mediated baseline by treatment moderation. These models help in understanding how moderators influence the mediated effects. Finally, the authors address the causal inference aspect of mediation, critiquing traditional regression-based methods and introducing new approaches based on principal stratifications and latent class models. They emphasize the importance of causal interpretation in mediation analysis and suggest future research directions to improve the understanding and application of mediation in psychological research.