TY - JOUR
T1 - Causal Mediation Analysis With a Binary Outcome and Multiple Continuous or Ordinal Mediators
T2 - Simulations and Application to an Alcohol Intervention
AU - Nguyen, Trang Quynh
AU - Webb-Vargas, Yenny
AU - Koning, Ina M.
AU - Stuart, Elizabeth A.
N1 - Funding Information:
Trang Quynh Nguyen was supported by National Institute on Drug Abuse grant T-32DA007292 (Principal Investigator C. D. Furr-Holden). Elizabeth A. Stuart was supported by National Institute of Mental Health grant R01MH099010. Yenny Webb-Vargas was supported by National Institute of Biomedical Imaging and Bioengineering grant 5R01EB016061-02 (Principal Investigator M. Linguist). The alcohol prevention study was funded by the Dutch Health Care Research Organization, Grant 6220.0021.
Publisher Copyright:
Copyright © Taylor & Francis Group, LLC.
PY - 2016/5/3
Y1 - 2016/5/3
N2 - We investigate a method to estimate the combined effect of multiple continuous/ordinal mediators on a binary outcome: (a) fit a structural equation model with probit link for the outcome and identity/probit link for continuous/ordinal mediators, (b) predict potential outcome probabilities, and (c) compute natural direct and indirect effects. Step 2 involves rescaling the latent continuous variable underlying the outcome to address residual mediator variance and covariance. We evaluate the estimation of risk-difference- and risk-ratio-based effects (RDs, RRs) using the maximum likelihood (ML), mean-and-variance-adjusted weighted least squares (WLSMV) and Bayes estimators in Mplus. Across most variations in path-coefficient and mediator-residual-correlation signs and strengths, and confounding situations investigated, the method performs well with all estimators, but favors ML/WLSMV for RDs with continuous mediators, and Bayes for RRs with ordinal mediators. Bayes outperforms ML/WLSMV regardless of mediator type when estimating RRs with small potential outcome probabilities and in two other special cases. An adolescent alcohol prevention study is used for illustration.
AB - We investigate a method to estimate the combined effect of multiple continuous/ordinal mediators on a binary outcome: (a) fit a structural equation model with probit link for the outcome and identity/probit link for continuous/ordinal mediators, (b) predict potential outcome probabilities, and (c) compute natural direct and indirect effects. Step 2 involves rescaling the latent continuous variable underlying the outcome to address residual mediator variance and covariance. We evaluate the estimation of risk-difference- and risk-ratio-based effects (RDs, RRs) using the maximum likelihood (ML), mean-and-variance-adjusted weighted least squares (WLSMV) and Bayes estimators in Mplus. Across most variations in path-coefficient and mediator-residual-correlation signs and strengths, and confounding situations investigated, the method performs well with all estimators, but favors ML/WLSMV for RDs with continuous mediators, and Bayes for RRs with ordinal mediators. Bayes outperforms ML/WLSMV regardless of mediator type when estimating RRs with small potential outcome probabilities and in two other special cases. An adolescent alcohol prevention study is used for illustration.
KW - binary outcome
KW - causal inference, continuous mediators
KW - causal mediation analysis
KW - multiple mediators
KW - ordinal mediators
KW - structural equation modeling
UR - http://www.scopus.com/inward/record.url?scp=84947218138&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84947218138&partnerID=8YFLogxK
U2 - 10.1080/10705511.2015.1062730
DO - 10.1080/10705511.2015.1062730
M3 - Article
C2 - 27158217
AN - SCOPUS:84947218138
VL - 23
SP - 368
EP - 383
JO - Structural Equation Modeling
JF - Structural Equation Modeling
SN - 1070-5511
IS - 3
ER -