Abstract
In social and behavioral sciences, the mediation test based on the indirect effect is an important topic. There are many methods to assess intervening variable effects. In this paper, we focus on the difference method and the product method in mediation models. Firstly, we analyze the regression functions in the simple mediation model, and provide an expectation-consistent condition. We further show that the difference estimator and the product estimator are numerically equivalent based on the least-squares regression regardless of the error distribution. Secondly, we generalize the equivalence result to the three-path model and the multiple mediators model, and prove a general equivalence result in a class of restricted linear mediation models. Thirdly, we investigate the empirical distributions of the indirect effect estimators in the simple mediation model by simulations, and show that the indirect effect estimators are normally distributed as long as one multiplicand of the product estimator is large. Finally, we introduce some popular R packages for mediation analysis and also provide some useful suggestions on how to correctly conduct mediation analysis.
Similar content being viewed by others
References
Alfons, A. (2020). probmed: (robust) mediation analysis. R package, Version 0.7.0.
Allen, D.G., & Griffeth, R.W. (2001). Test of a mediated performance-turnover relationship highlighting the moderating roles of visibility and reward contingency. Journal of Applied Psychology, 86(5), 1014–1021. https://doi.org/10.1037/0021-9010.86.5.1014.
Allison, P.D. (1995). The impact of random predictors on comparison of coefficients between models: Comment on Clogg, Petkova, and Haritou. American Journal of Sociology, 100(5), 1294–1305. https://doi.org/10.1086/230639.
Baron, R.M., & Kenny, D.A. (1986). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51 (6), 1173–1182. https://doi.org/10.1037/0022-3514.51.6.1173.
Batailler, C., Muller, D., Yzerbyt, V., Judd, C., Ho, A., Kteily, N., Chen, J., Dohle, S., & Siegrist, M. (2020). Jsmediation: Mediation analysis using joint significance. R package Version 0.1.
Cui, G., Yu, X., Iommelli, S., & Kong, L. (2016). Exact distribution for the product of two correlated Gaussian random variables. IEEE Signal Processing Letters, 23(11), 1662–1666. https://doi.org/10.1109/LSP.2016.2614539.
Daniel, R.M., De Stavla, B.L., Cousens, S.N., & Vansteelandt, S. (2015). Causal mediation analysis with multiple mediators. Biometrics, 71(1), 1–14. https://doi.org/10.1111/biom.12248.
Eshima, N., Tabata, M., & Geng, Z. (2001). Path analysis with logistic regression models: Effect analysis of fully recursive causal systems of categorical variables. Journal of the Japan Statistical Society, 31(1), 1–14. https://doi.org/10.14490/jjss1995.31.1.
Frölich, M., & Huber, M. (2017). Direct and indirect treatment effects-causal chains and mediation analysis with instrumental variables. Journal of the Royal Statistical Society Series B, 79(5), 1645–1666. https://doi.org/10.1111/rssb.12232.
Guo, J., & Geng, Z. (1995). Collapsibility of logistic regression coefficients. Journal of the Royal Statistical Society Series B, 57(1), 263–267. https://doi.org/10.1111/j.2517-6161.1995.tb02029.x.
Guo, J., Geng, Z., & Fung, W.K. (2001). Consecutive collapsibility of odds ratios over an ordinal background variable. Journal of Multivariate Analysis, 79(1), 89–98. https://doi.org/10.1006/jmva.2000.1957.
Hayes, A.F. (2013). Introduction to mediation, moderation, and conditional process analysis: A regression-based approach. Guilford Press.
Kenny, D.A., Kashy, D.A., & Bolger, N. (1998). Data analysis in social psychology. In D. Gilbert, S. Fiske, & G. Lindzey (Eds.) The handbook of social psychology, pp 233–265, McGraw-Hill.
Kim, J., & Cicchetti, D. (2010). Longitudinal pathways linking child maltreatment, emotion regulation, peer relations, and psychopathology. Journal of Child Psychology and Psychiatry, 51(6), 706–716. https://doi.org/10.1111/j.1469-7610.2009.02202.x.
Kisbu-Sakarya, Y., MacKinnon, D.P., & Miočević, M. (2014). The distribution of the product explains normal theory mediation confidnce interval estimation. Multivariate Behavioral Research, 49(3), 261–268. https://doi.org/10.1080/00273171.2014.903162.
Lachowicz, M.J., Preacher, K.J., & Kelley, K. (2018). A novel measure of effect size for mediation analysis. Psychological Methods, 23(2), 244–261. https://doi.org/10.1037/met0000165.
Lockhart, G., MacKinnon, D.P., & Ohlrich, V. (2011). Mediation analysis in psychosomatic medicine research. Psychosomatic Medicine, 73(1), 29–43. https://doi.org/10.1097/PSY.0b013e318200a54b.
Loucks, E.B., Huang, Y.T., Agha, G., Chu, S., Eaton, C.B., Gilman, S.E., Buka, S.L., & Kelsey, K.T. (2016). Epigenetic mediators between childhood socioeconomic disadvantage and mid-life body mass index: The New England family study. Psychosomatic Medicine, 78(9), 1053–1065. https://doi.org/10.1097/PSY.0000000000000411.
MacKinnon, D.P. (2008). Introduction to statistical mediation analysis. Taylor & Francis Group.
MacKinnon, D.P., Fairchild, A.J., & Fritz, M.S. (2007a). Mediation analysis. Annual Review of Psychology, 58, 593–614. https://doi.org/10.1146/annurev.psych.58.110405.085542.
MacKinnon, D.P., Fritz, M.S., Williams, J., & Lockwood, C.M. (2007b). Distribution of the product confidence limits for the indirect effect: Program PRODCLIN. Behavior Research Methods, 39, 384–389. https://doi.org/10.3758/BF03193007.
MacKinnon, D.P., Lockwood, C., & Hoffman, J. (1998). A new method to test for mediation. In The annual meeting of the society for prevention research, Park City, USA.
MacKinnon, D.P., & Lockwood, C.M. (2001). Distribution of products tests for the mediated effect. Technical Report. USA: Arizona State University.
MacKinnon, D.P., Lockwood, C.M., Hoffman, J.M., West, S.G., & Sheets, V. (2002). A comparison of methods to test mediation and other intervening variable effects. Psychological Methods, 7(1), 83–104. https://doi.org/10.1037/1082-989x.7.1.83.
MacKinnon, D.P., Lockwood, C.M., & Williams, J. (2004). Confidence limits for the indirect effect: Distribution of the product and resampling methods. Multivariate Behavioral Research, 39(1), 99–128. https://doi.org/10.1207/s15327906mbr3901_4.
MacKinnon, D.P., Warsi, G., & Dwyer, J.H. (1995). A simulation study of mediated effect measures. Multivatiate Behavioral Research, 30(1), 41–62. https://doi.org/10.1207/s15327906mbr3001_3.
McGuigan, K., & Langholtz, B. (1988). A note on testing mediation paths using ordinary least-squares regression. Unpublished Note.
Meeker, W., Cornwell, L., & Aroian, L. (1981). Selected table in mathematical statistics (Volume VII). The product of two normally distributed random variables. American Mathematical Society.
Micceri, T. (1989). The unicorn, the normal curve, and other improbable creatures. Psychological Bulletin, 105(1), 156–166. https://doi.org/10.1037/0033-2909.105.1.156.
Nadarajah, S., & Pogány, T.K. (2016). On the distribution of the product of correlated normal random variables. Comptes Rendus de l’académie des Sciences, Series I, 354 (2), 201–204. https://doi.org/10.1016/j.crma.2015.10.019.
Newland, R.P., Crnic, K.A., Cox, M.J., Mills-Koonce, W.R., & Investigators, Family Life Project Key (2013). The family model stress and maternal psychological symptoms: Mediated pathways form economic hardship to parenting. Journal of Family Psychology, 27(1), 96–105. https://doi.org/10.1037/a0031112.
Nübold, A., Dörr, S. L., & Maier, G.W. (2015). Considering the orphan: Personal identification and its relations with transformational leadership, trust, and performance in a three-path mediation model. Leadership, 11(2), 230–254. https://doi.org/10.1177/1742715014522679.
Ogden, C.L., Carroll, M.D., Curtin, L.R., Lamb, M.M., & Flegal, K.M. (2010). Prevalence of high body mass index in U.S. children and adolescents, 2007-2008. Journal of the American Medical Association, 303(3), 242–249. https://doi.org/10.1001/jama.2009.2012.
Pearl, J. (2009). Causality: Models, reasoning and inference, 2nd. Cambridge: Cambridge University Press.
Preacher, K.J., & Hayes, A.F. (2004). SPSS and SAS procedures for estimating indirect effects in simple mediation models. Behavior Research Methods Instruments, & Computers, 36(4), 717–731. https://doi.org/10.3758/BF03206553.
Preacher, K.J., & Hayes, A.F. (2008). Asymptotic and resampling strategies for assessing and comparing indirect effects in multiple mediator models. Behavior Research Methods, 40(3), 879–891. https://doi.org/10.3758/BRM.40.3.879.
Preacher, K.J., & Selig, J.P. (2012). Advantages of Monte Carlo confidence intervals for indirect effects. Communication Methods and Measures, 6(2), 77–98. https://doi.org/10.1080/19312458.2012.679848.
Qiu, W. (2020). powermediation: Power/sample size calculation for mediation analysis. R package, Version 0.3.2.
Richiardi, L., Bellocco, R., & Zugna, D. (2013). Mediation analysis in epidemiology: Methods interpretation and bias. International Journal of Epidemiology, 42(5), 1511–1519. https://doi.org/10.1093/ije/dyt127.
Rucker, D.D., Preacher, K.J., Tormala, Z.L., & Petty, R.E. (2011). Mediation analysis in social psychology: Current practices and new recommendations. Social and Personality Psychology Compass, 5(6), 359–371. https://doi.org/10.1111/j.1751-9004.2011.00355.x.
Shrout, P.E., & Bolger, N. (2002). Mediation in experimental and nonexperimental studies: New procedures and recommendations. Psychological Methods, 7(4), 422–445. https://doi.org/10.1037/1082-989X.7.4.422.
Sobel, M.E. (1982). Asymptotic confidence intervals for indirect effects in structural equation models. Sociological Methodology, 13, 290–312. https://doi.org/10.2307/270723.
Steen, J., Loeys, T., Moerkerke, B., Vansteelandt, S., Meys, J., Lange, T., Legewie, J., & Fink, P. (2020). medflex: Flexible mediation analysis using natural effect models. R package, Version 0.6-7.
Taguri, M., Featherstone, J., & Cheng, J. (2018). Causal mediation analysis with multiple causally non-ordered mediator. Statistical Methods in Medical Research, 27(1), 3–19. https://doi.org/10.1177/0962280215615899.
Taylor, A.B., MacKinnon, D.P., & Tein, J.Y. (2008). Tests of the three-path mediated effect. Organizational Research Methods, 11(2), 241–269. https://doi.org/10.1177/1094428107300344.
Tein, J.Y., Sandler, I.N., & Zautra, A.J. (2000). Stressful life events, psychological distress, coping, and parenting of divorced mothers: A longitudinal study. Journal of Family Psychology, 14(1), 27–41. https://doi.org/10.1037/0893-3200.14.1.27.
Tekleab, A.G., Bartol, K.M., & Liu, W. (2005). Is it pay level or pay raises that matter to fairness and turnover? Journal of Organizational Behavior, 26, 899–921. https://doi.org/10.1002/job.352.
Tingley, D., Yamamoto, T., Hirose, K., Keele, L., Imai, K., Trinh, M., & Wong, W. (2019). mediation: Causal mediation analysis. R package, Version 4.5.0.
Tofighi, D., & MacKinnon, D.P. (2011). Rmediation: An R package for mediation analysis confidence intervals. Behavior Research Methods, 43(3), 692–700. https://doi.org/10.3758/s13428-011-0076-x.
Tofighi, D., & MacKinnon, D.P. (2016). Rmediation: Mediation analysis confidence intervals. R package Version 1.1.4.
VanderWeele, T.J. (2015). Explanation in causal inference: Methods for mediation and interaction. Oxford: Oxford University Press.
VanderWeele, T.J., & Tchetgen Tchetgen, E.J. (2017). Mediation analysis with time varying exposures and mediators. Journal of the Royal Statistical Society: Series B, 79(3), 917–938. https://doi.org/10.1111/rssb.12194.
Wang, W.W., & Yu, P. (2020). Nonequivalence of two least-absolute-deviation estimators for mediation effect. Complete Manuscript.
Wang, W.W., Yu, P., Lin, L., & Tong, T. (2019). Robust estimation of derivatives using locally weighted least absolute deviation regression. Journal of Machine Learning Research, 20(60), 1–49.
White, H. (2000). A reality check for data snooping. Econometrica, 68(5), 1097–1126. https://doi.org/10.1111/1468-0262.00152.
Yuan, Y., & MacKinnon, D.P. (2014). Robust mediation analysis based on median regression. Psychological Methods, 19(1), 1–20. https://doi.org/10.1037/a0033820.
Zhao, Z., & Xiao, Z. (2014). Efficient regression via optimally combining quantile information. Econometric Theory, 30(6), 1272–1314. https://doi.org/10.1017/S0266466614000176.
Acknowledgements
Wang’s work is supported by National Natural Science Foundation of China (No.12071248), and National Statistical Science Research Project of China (No.2020LZ26).
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of Interests
The authors declare that they have no conflict of interest.
This is a theory paper with an application to DNA methylation open data. Our paper does not involve research on human participants or animals. Thus, we depended on those who collected the primary data about DNA methylation (Loucks et al., 2016), for compliance with the 1964 Declaration of Helsinki and its later addenda. The datasets generated during and/or analysed during the current study are available in an open software “JT-Comp”, http://www.stat.sinica.edu.tw/ythuang/JT-Comp.zip.
The work has not been submitted elsewhere for publication, in whole or in part. All authors have seen the manuscript, approved to submit to Current Psychology, and consented to its review by Current Psychology.
Additional information
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendices
Appendix : Appendix A: Proof of Theorem 2
Proof
We first consider the simple case where β0 = β1 = β2 = 0. The least-squares estimators for the simplified models are
where X = (X1,…,Xn)T, M = (M1,…,Mn)T, and Y = (Y1,…,Yn)T. By the above least-squares estimators, the difference estimator is
and the product estimator is
This shows that \(\hat {c}-\hat {c}^{\prime }=\hat {a}\hat {b}\). That is, the difference estimator is equivalent to the product estimator for the linear regression models with zero intercept.
The proof can readily be generalized to the models with non-zero intercept by replacing X and M by their demeaned couterparts and so is omitted. □
Appendix : Appendix B: Proof of Theorems 3
Proof
By substituting Eqs. 13 and 14 into Eq. 15, it follows that
where 𝜖i = (b1 + b2d)𝜖1,i + b2𝜖2,i + 𝜖3,i. Since 𝜖j,i for j = 1, 2, 3 are zero-mean distributed, 𝜖i is also zero-mean distributed with E[𝜖i|Xi] = 0. Taking expectation of Eqs. 12 and 20, we have
This leads to β0 = β3 + b2β2 + (b1 + b2d)β1 and \(c-c^{\prime }=a_{1}b_{1}+a_{2}b_{2}+a_{1}d b_{2}\).
For the simplified models with β0 = β1 = β2 = β3 = 0, the least-squares estimators are
where A1 = XTX, A2 = XTM1, A3 = XTM2, \(A_{4}={M_{1}^{T}}M_{1}\), \(A_{5}={M_{1}^{T}}M_{2}\), \(A_{6}={M_{2}^{T}}M_{2}\), A7 = XTY, \(A_{8}={M_{1}^{T}}Y\), \(A_{9}={M_{2}^{T}}Y\), and \(A_{10}={M_{1}^{T}}M_{2}\). By the above least-squares estimators, it is easy to verify that \(\tilde {c}-\tilde {c}^{\prime }=\tilde {a}_{1}\tilde {b}_{1}+\tilde {a}_{2}\tilde {b}_{2}+\tilde {a}_{1}\tilde {d}\tilde {b}_{2}\). That is, the difference estimator is equivalent to the product estimator for the linear regression models with zero intercept.
The proof can readily be generalized to the models with non-zero intercept and so is omitted. □
Appendix : Appendix C: Proof of Theorem 6
Proof
Suppose there are k mediators:
where Rj contains the non-constant regressors in the equation for Mj, which may include X and/or other Mj’s, and RY contains Mj’s that appear in the equation for Y. After substituting the equations for Mj,j = 1,…,k, into the equation for Y, suppose we have
where (α0,α1) are functions of the coefficients in the equations for \(\{M_{j}\}_{j=1}^{k}\) and Y, i.e.,
α1 does not depend on β1,…,βk+ 1 because it measures the sensitivity of Y to X while β1,…,βk+ 1 does not contain such information, \(c^{\prime }\) is the coefficient of X in the equation for Y, and 𝜖0 is a linear combination of the error terms in these (k + 1) equations, so it satisfies E[𝜖0|X] = 0.
Because all the coefficients are estimated by least-squares regression, they employ the moment conditions
If these moment conditions imply
then our result follows since we just replace E[⋅] by \(1/n{\sum }_{i=1}^{n}\) in the least-squares estimation. However, this indeed holds because 𝜖0 is a linear function of \(\{\epsilon _{j}\}_{j=1}^{k+1}\) so that
implies
Here, note that X must be a regressor in the Mj equation, otherwise \(\mathrm {E}\left [\left (\begin {array}{l} 1 \\ X \end {array}\right )\epsilon _{j}\right ]=0\) cannot hold such that \(\mathrm {E}\left [\left (\begin {array}{l} 1 \\ X \end {array}\right )\epsilon _{0}\right ]=0\) cannot hold and the equivalence result fails. □
Rights and permissions
About this article
Cite this article
Wang, W., Yu, P., Zhou, Y. et al. Equivalence of two least-squares estimators for indirect effects. Curr Psychol 42, 7364–7375 (2023). https://doi.org/10.1007/s12144-021-02034-6
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s12144-021-02034-6