Whether you are examining simple mediation or if contextual factors moderate the effects for multiple mediators, Elkhart Group Ltd. can help you use your data to understand the mechanisms linking variables to each other.
The indirect effect of an independent variable on an outcome via a mediator can be analyzed using mediation analysis. Our statistical consultants can help you conduct simple mediation using modern product of coefficient methods and bootstrapping or the classic approach proposed in 1986 by Baron and Kenny. For categorical mediators or outcomes, in addition to the product of coefficients approach, our statistical consultants can also use causal mediation analysis to provide estimates of the causal effects.
There may not be an indirect effect through a mediator in all contexts or for all people. For example, participation in classroom activities may mediate the relationship between attending lectures and final grades, but only for students who do not study very much. Questions such as these can be tested using moderated mediation. Our statistical consultants can help you answer whether you have moderated mediation or mediated moderation in your data, and if you do, create graphs showing how the strength of the indirect effect varies across levels of the moderator. We can also tell you at what level or region of the moderator the indirect effect becomes statistically significant.
One mediator may explain all of the association between an independent variable an outcome, but in many contexts, there are multiple pathways. Multiple mediation extends simple mediation by examining whether a set of variables accounts for the effect of an independent variable on an outcome. Our statistical consultants can give you bootstrapped estimates of the individual indirect effect of the independent variable on the outcome through each mediator separately, and also test the total indirect effect through all the mediators simultaneously to test the overall model.
Complex models of how multiple variables fit together including multiple independent variables and multiple outcomes can be analyzed using path analysis or systems of simultaneous equations. Using path analysis, our statistical consultants can also help you test multi-step causal chains where x -> m1 -> m2 -> y. We can also test almost any combination of multiple mediators, multiple causal steps, moderation, and multiple outcomes using path analysis.
Indirect effects are known to not have a normal distribution. Our statistical consultants can help use parametric or nonparametric bootstrapping to robustly estimate the significance and confidence intervals. Additionally, we can use Bayesian estimation for most path analysis or mediation models and give point estimates and credible intervals from the posterior distribution.
Premier: Our most bespoke offering. Let our consultants create solutions and reports that turn your data into strategic insight.
Academic: For accredited, not-for-profit universities. From methods write up assistance to advanced model analysis, we are experienced researchers who enjoy scientific discovery.
© 2013 - 2017 by Elkhart Group LLC