Conventional methods for mediation analysis generate biased results when the mediator–outcome relationship depends on the treatment condition. This article shows how the ratio-of-mediator-probability weighting (RMPW) method can be used to decompose total effects into natural direct and indirect effects in the presence of treatment-by-mediator interactions. The indirect effect can be further decomposed into a pure indirect effect and a natural treatment-by-mediator interaction effect. Similar to other techniques for causal mediation analysis, RMPW generates causally valid results when the sequential ignorability assumptions hold. Yet unlike the model-based alternatives, including path analysis, structural equation modeling, and their latest extensions, RMPW requires relatively few assumptions about the distribution of the outcome, the distribution of the mediator, and the functional form of the outcome model. Correct specification of the propensity score models for the mediator remains crucial when parametric RMPW is applied. This article gives an intuitive explanation of the RMPW rationale, a mathematical proof, and simulation results for the parametric and nonparametric RMPW procedures. We apply the technique to identifying whether employment mediated the relationship between an experimental welfare-to-work program and maternal depression. A detailed delineation of the analytic procedures is accompanied by online Stata code as well as a stand-alone RMPW software program to facilitate users’ analytic decision making.
AbadieA.ChingosM. M.WestM. R. (2013). Endogenous stratification in randomized experiments(Working Paper 19742). Cambridge, MA: National Bureau of Economic Research. Retrieved fromhttp://www.nber.org/papers/w19742
2.
AlwinD. F.HauserR. M. (1975). The decomposition of effects in path analysis. American Sociological Review, 40, 37–47.
3.
AustinP. C. (2011). Optimal caliper widths for propensity-score matching when estimating differences in means and differences in proportions in observational studies. Pharmaceutical Statistics, 10, 150–161.
4.
AvinC.ShpitserI.PearlJ. (2005). Identifiability of path-specific effects. Proceedings of the Nineteenth International Joint Conference on Artificial Intelligence (pp. 357–363). Edinburgh, Scotland, UK.
5.
BaronR. M.KennyD. A. (1986). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51, 1173–1182.
6.
BeinE.DeutschJ.PorterK.QinX.YangC.HongG. (2015). Technical report on two-step estimation in RMPW analysis. Oakland, CA: MDRC.
7.
BollenK. A. (1989). Structural equations with latent variables. Wiley Series in Probability and Mathematical Statistics. New York, NY: John Wiley.
8.
BullockJ. G.GreenD. P.HaS. E. (2010). Yes, but what’s the mechanism? (Don’t Expect an Easy Answer). Journal of Personality and Social Psychology, 98, 550–558.
9.
CameronA. C.TrivediP. K. (2005). Microeconometrics: Methods and applications. Cambridge, England: Cambridge University Press.
10.
ChengT. C. (2007). Impact of work requirements on the psychological well-being of TANF recipients. Health and Social Work, 32, 41–48.
11.
CoffmanD. L.ZhongW. (2012). Assessing mediation using marginal structural models in the presence of confounding and moderation. Psychological Methods, 17, 642–664.
12.
CollinsL.GrahamJ.FlahertyB. (1998). An alternative framework for defining mediation. Multivariate Behavioral Research, 33, 295–312.
13.
CookT.SteinerP. (2010). Case matching and the reduction of selection bias in quasi-experiments: The relative importance of pretest measures of outcome, of unreliable measurement, and of mode of data analysis. Psychological Methods, 15, 56–68.
14.
DrakeC. (1993). Effects of misspecification of the propensity score on estimators of treatment effects. Biometrics, 49, 1231–1236.
15.
DuncanO. D. (1966). Path analysis: Sociological examples. American Journal of Sociology, 72, 1–16.
16.
GroggerJ.KarolyL. A. (2005). Welfare reform: Effects of a decade of change. Cambridge, MA: Harvard University Press.
17.
HamiltonG.FreedmanS.GennetianL.MichalopoulosC.WalterJ.Adams-CiardulloD.…BrooksJ. (2001). How effective are different welfare-to-work approaches?: Five-year adult and child impacts for eleven programs. New York, NY: MDRC.
18.
HastieT.TibshiraniR.FriedmanJ. (2009). The elements of statistical learning: Data mining, inference, and prediction (2nd ed.). New York, NY: Springer.
19.
HawkinsD. M. (2004). The problem of overfitting. Journal of Chemical Information and Computer Sciences, 44, 1–12.
20.
HeckmanJ. J.RobbR. (1985). Alternative methods for evaluating the impact of an intervention. Journal of Econometrics, 30, 239–267.
21.
HollandP. W. (1986). Statistics and causal inference. Journal of the American Statistical Association, 81, 945–960.
HongG. (2010a). Ratio of mediator probability weighting for estimating natural direct and indirect effects. In JSM Proceedings (pp. 2401–2415), Biometrics Section. Alexandria, VA: American Statistical Association.
24.
HongG. (2010b). Marginal mean weighting through stratification: Adjustment for selection bias in multilevel data. Journal of Educational and Behavioral Statistics, 35, 499–531.
25.
HongG. (2012). Marginal mean weighting through stratification: A generalized method for evaluating multi-valued and multiple treatments with non-experimental data. Psychological Methods, 17, 44–60.
26.
HongG. (2015). Causality in a social world: Moderation, mediation and spill-over. West Sussex, UK: John Wiley & Sons, Ltd.
27.
HongG.NomiT. (2012). Weighting methods for assessing policy effects mediated by peer change. Journal of Research on Educational Effectiveness (special issue on the statistical approaches to studying mediator effects in education research), 5, 261–289.
28.
HuberM. (2014). Identifying causal mechanisms (primarily) based on inverse probability weighting. Journal of Applied Econometrics, 29, 920–943.
29.
ImaiK.KeeleL.TingleyD. (2010a). A general approach to causal mediation analysis. Psychological Methods, 15, 309–334.
30.
ImaiK.KeeleL.YamamotoT. (2010b). Identification, inference and sensitivity analysis for causal mediation effects. Statistical Science, 25, 51–71.
31.
ImaiK.KeeleL.TingleyD.YamamotoT. (2011). Unpacking the black box of causality: Learning about causal mechanisms from experimental and observational studies. American Political Science Review, 105, 765–789.
32.
JagannathanR.CamassoM. J.SambamoorthiU. (2010). Experimental evidence of welfare reform impact on clinical anxiety and depression levels among poor women. Social Science & Medicine, 71, 152–160.
33.
JoB. (2008). Causal inference in randomized experiments with mediational processes. Psychological Methods, 13, 314–336.
34.
JöreskogK. G. (1970). A general method for analysis of covariance structures. Biometrika, 57, 239–251.
35.
JuddC. M.KennyD. A. (1981). Process analysis: Estimating mediation in treatment evaluation. Evaluation Review, 5, 602–619.
36.
KangJ. D.SchaferJ. L. (2007). Demystifying double robustness: A comparison of alternative strategies for estimating a population mean from incomplete data. Statistical science, 22, 523–539.
37.
KlingJ. R.LiebmanJ. B.KatzL. F. (2007). Experimental analysis of neighborhood effects. Econometrica, 75, 83–119.
38.
KraemerH. C.WilsonG. T.FairburnC. G.AgrasW. S. (2002). Mediators and moderators of treatment effects in randomized clinical trials. Archives of General Psychiatry, 59, 877–883.
39.
KraemerH. C.KiernanM.EssexM.KupferD. J. (2008). How and why criteria defining moderators and mediators differ between the Baron & Kenny and MacArthur approaches. Health Psychology, 27, S101–S108.
40.
LittleR. J. A.RubinD. B. (2002). Statistical analysis with missing data. New York, NY: John Wiley.
41.
MacKinnonD. P. (2008). Introduction to statistical mediation analysis. New York, NY: Erlbaum.
42.
MacKinnonD. P.KrullJ. L.LockwoodC. M. (2000). Equivalence of the mediation, confounding and suppression effect. Prevention Science, 1, 173–181.
43.
McCaffreyD. F.RidgewayG.MorralA. R. (2004). Propensity score estimation with boosted regression for evaluating causal effects in observational studies. Psychological Methods, 9, 403–425.
44.
MorrisP. A. (2008). Welfare program implementation and parents’ depression. Social Service Review, 82, 579–614.
45.
MullerD.JuddC. M.YzerbytV. Y. (2005). When moderation is mediated and mediation is moderated. Journal of Personality and Social Psychology, 89, 852–863.
46.
PearlJ. (2001). Direct and indirect effects. In Proceedings of the American Statistical Association Joint Statistical Meetings (pp. 1572–1581). Minneapolis, MN: MIRA Digital Publishing, August 2005.
47.
PearlJ. (2010). The mediation formula: A guide to the assessment of causal pathways in non-linear models(Technical report R-363, July 2010). Los Angeles, CA: University of California, Los Angeles.
48.
PeckL. R. (2003). Subgroup analysis in social experiments: Measuring program impacts based on posttreatment choice. American Journal of Evaluation, 24, 157–187.
49.
PeckL. R. (2007). What are the effects of welfare sanction policies? Or, using propensity scores as a subgroup indicator to learn more from social experiments. American Journal of Evaluation, 28, 256–274.
50.
PetersenM. L.SinisiS. E.van der LaanM. J. (2006). Estimation of direct causal effects. Epidemiology, 17, 276–284.
51.
PowersD. E.SwintonS. S. (1984). Effects of self-study for coachable test item types. Journal of Educational Psychology, 76, 266–278.
52.
PreacherK. J.HayesA. F. (2008). Asymptotic and resampling strategies for assessing and comparing indirect effects in multiple mediator models. Behavior Research Methods, 40, 879–891.
53.
PreacherK. J.RuckerD. D.HayesA. F. (2007). Addressing moderated mediation hypotheses: Theory, methods, and prescriptions. Multivariate Behavioral Research, 42, 185–227.
54.
RadloffL. S. (1977). The CES-D scale: A self-report depression scale for research in the general population. Applied Psychological Measurement, 1, 385–401.
55.
RaudenbushS. W.ReardonS. F.NomiT. (2012). Statistical analysis for multisite trials using instrumental variables with random coefficients. Journal of Research on Educational Effectiveness Special Issue on the Statistical Approaches to Studying Mediator Effects in Education Research, 5, 303–332.
56.
RobinsJ. M. (1999). Marginal structural models versus structural nested models as tools for causal inference. In Elizabeth HalloranM.BerryD. (Eds.), Statistical models in epidemiology, the environment, and clinical trials (pp. 95–134). New York, NY: Springer.
57.
RobinsJ. M. (2003). Semantics of causal DAG models and the identification of direct and indirect effects. In GreenP. J.HjortN. L.RichardsonS. (Eds.), Highly structured stochastic systems (pp. 70–81). New York, NY: Oxford University Press.
58.
RobinsJ. M.GreenlandS. (1992). Identifiability and exchangeability for direct and indirect effects. Epidemiology, 3, 143–155.
59.
RosenbaumP. R. (1987). Model-based direct adjustment. Journal of the American Statistical Association, 82, 387–394.
60.
RosenbaumP. R.RubinD. B. (1984). Reducing bias in observational studies using subclassification on the propensity score. Journal of the American Statistical Association, 79, 516–524.
61.
RubinD. B. (1978), Bayesian inference for causal effects: The role of randomization. The Annals of Statistics, 6, 34–58.
62.
RubinD. B. (1986). Statistics and causal inference: Comments: Which ifs have causal answers. Journal of the American Statistical Association, 81, 961–962.
63.
SchaferJ. L.KangJ. (2008). Average causal effects from nonrandomized studies: A practical guide and simulated example. Psychological Methods, 13, 279–313.
64.
SchochetP. Z.BurghardtJ. (2007). Using propensity scoring to estimate program-related subgroup impacts in experimental program evaluations. Evaluation Review, 31, 95–120.
65.
SheetsV. L.BraverS. L. (1999). Organizational status and perceived sexual harassment: Detecting the mediators of a null effect. Personality and Social Psychology Bulletin, 25, 1159–1171.
66.
ShroutP. E.BolgerN. (2002). Mediation in experimental and nonexperimental studies: New procedures and recommendations. Psychological Methods, 4, 422–445.
67.
SobelM. E. (2008). Identification of causal parameters in randomized studies with mediating variables. Journal of Educational and Behavioral Statistics, 33, 230–251.
68.
SpencerS. J.ZannaM. P.FongG. T. (2005). Establishing a causal chain: Why experiments are often more effective than mediational analysis in examining psychological processes. Journal of Personality and Social Psychology, 89, 845–851.
69.
Tchetgen TchetgenE. J. (2013). Inverse odds ratio weighted estimation for causal mediation analysis. Statistics in Medicine, 32, 4567–4580.
70.
Tchetgen TchetgenE. J.ShpitserI. (2012). Semiparametric theory for causal mediation analysis: Efficiency bounds, multiple robustness and sensitivity analysis. The Annals of Statistics, 40, 1816–1845.
71.
ValeriL.VanderWeeleT. J. (2013). Mediation analysis allowing for exposure-mediator interactions and causal interpretation: Theoretical assumptions and implementation with SAS and SPSS macros. Psychological Methods, 18, 137–150.
72.
van der LaanM. J.PetersenM. L. (2008). Direct effect models. The International Journal of Biostatistics, 4, Article 23.
73.
VanderWeeleT. J. (2009). Marginal structural models for the estimation of direct and indirect effects. Epidemiology, 20, 18–26.
74.
VanderWeeleT. J. (2010). Bias formulas for sensitivity analysis for direct and indirect effects. Epidemiology, 21, 540–551.
75.
VanderWeeleT. J. (2013). A three-way decomposition of a total effect into direct, indirect, and interactive effects. Epidemiology, 24, 224–232.
76.
VanderWeeleT. J.VansteelandtS. (2009). Conceptual issues concerning mediation, interventions and composition. Statistics and Its Interface, 2, 457–468.
77.
VanderWeeleT. J.VansteelandtS. (2010). Odds ratios for mediation analysis for a dichotomous outcome. American Journal of Epidemiology, 172, 1339–1348.
78.
WaernbaumI. (2012). Model misspecification and robustness in causal inference: Comparing matching with doubly robust estimation. Statistics in Medicine, 31, 1572–1581.
79.
WooldridgeJ. M. (2012). Introductory econometrics: A modern approach (5th ed.). Mason, OH: Cengage Learning.
80.
WrightS. (1934). The method of path coefficients. Annals of Mathematical Statistics, 5, 161–215.
