Questions tagged [mediation]
A mediator is a variable that plays an intermediate role in a causal chain, i.e., the 1st variable in the chain causes the mediator, which in turn causes the 3rd variable. Partial mediation is when the 1st variable causes the 3rd both through the mediator & directly.
577 questions
3
votes
0
answers
61
views
Comparing approaches to estimation/identification of controlled direct effects (DiD vs. G-estimation)
I am interested in estimating a controlled direct effect in a RCT with a longitudinal design featuring multiple measurement points (each wave includes the outcome, mediator and confounders). I am ...
- 41
4
votes
1
answer
55
views
Assumption check for every hypothesis?
It is unclear to when and where to check the assumptions. I'm doing a multilevel analysis on level 1 only with random intercept. I have 1 predictor, 2 mediators, 1 moderator and 3 outcomes. I have 19 ...
- 71
0
votes
0
answers
17
views
Inconsistent mediation with a small effect size for the indirect pathway
I'm conducting a simple mediation analysis with 3 variables: an independent variable X, mediator variable M, and outcome variable Y. I'm using Bayesian multilevel modeling for within-subject repeated ...
- 169
3
votes
1
answer
143
views
Frequentist alternatives to brms for mediation analysis with ordinal outcomes
I'm interested in conducting a mediation analysis with ordinal outcomes, and have not been able to find too many resources. I apologize if I've missed any.
My model contains:
X --> M --> Y
where ...
- 169
2
votes
0
answers
51
views
How to handle a nonlinear mediator-outcome relationship with zeros in a (moderated) mediation analysis?
I’m running mediation (Process model 4) and moderated mediation (Process model 7) analyses, but the relationship between my mediator and dependent variable looks non-linear in the scatterplot (log-...
- 21
1
vote
1
answer
46
views
How should I interpret a negative indirect effect (ab < 0) in mediation analysis when all treatment outcomes improved
I’m analyzing a pre–post clinical dataset using PROCESS Model 4 (SPSS) to explore whether ocular blood flow changes mediate the effect of hematologic recovery on retinal structural improvement after ...
- 81
1
vote
1
answer
82
views
Lavaan serial mediation model - covariance between mediators and adding covariates
I'm trying to set up a serial mediation in lavaan with two mediators, and have a couple of questions about setting up the model that I haven't been able to find the answer to so far.
I had initially ...
- 11
0
votes
0
answers
30
views
Mediation analysis with multiple correlated predictors
I have measurements from a clinical scale, a mediator and an outcome. I have performed a mediation analysis using the scale total using the paths: scale -> mediator -> outcome and scale -> ...
2
votes
0
answers
54
views
Multilevel modeling with lagged predictor in daily diary study (in SPSS)?
I don’t have much experience with data analysis and would really appreciate your help with a question regarding the analysis of a diary study (5 daily measurement points).
I’m including the following ...
- 21
0
votes
0
answers
63
views
Unconstrained model fits better but non-significant Wald tests?
I fitted two SEM models and the chi-squared difference test indicated that the unconstrained model had a significantly better fit than the constrained model (where intercepts and regression paths were ...
0
votes
0
answers
102
views
Reporting original or bootstrapped estimates and p-value for SEM analyses?
I ran a SEM analysis where I'm assessing the direct and indirect effects of X on Y through 2 mediators (M1, M2). The direct and indirect effects were evaluated using bias-corrected bootstrap method ...
4
votes
2
answers
82
views
SEM Model 2 vs. Model 3: Can Theory + AIC/BIC Outweigh a Just-Identified Model?
I’m comparing three structural-equation models:
Model 1 (full mediation) – IV → M1 → M2 → DV
Model 2 (partial mediation) – same as Model 1 plus IV → M2 and IV → DV
Model 3 (saturated) – same as Model ...
- 41
1
vote
0
answers
27
views
Create mediation model with binary mediator
My final goal is to simulate a simple mediation model where all 3 variables (treatment, mediator, and outcome) are binary. The user should be able to specify the share mediated in advanced. To achieve ...
- 11
0
votes
0
answers
33
views
Indirect effect in mediation analysis with log transformed outcome
I am running a mediation analysis using a binary treatment (X), a binary mediator (M) and a log transformed outcome (lnY). I am using a linear-linear model. To report my results for the second ...
9
votes
1
answer
989
views
What is the 'real-world' meaning of a mediation effect?
Social psychologist here. I find fascinating the idea that we are able to measure something as a 'mechanism of transfer of effects' from one entity to another.
I can read values on a table, but I don'...
- 582