Hello Kyran,

check this out:

Imai, K., Keele, L., Tingley, D., & Yamamoto, T. (2010). Causal mediation analysis using R. In D. V. Hrishikesh (Ed.), Advances in social science research using R (pp. 129-154): Springer.

Imai, K., Keele, L., & Tingley, D. (2010). A general approach to causal mediation analysis. Psychological Methods, 15(4), 309-334.

Be aware however, that this is some discussion about whether estimating a point estimate of an indirect effect makes sense, when one of the effects is nonlinear. For instance, on SEMNET (a global SEM-focused mailing list), Mirko Rönkkö said:

""In the case of non-linear effects, you should think carefully whether it makes sense to calculate a single point estimate (i.e. a number) for the mediation effect. If you hare using a u-shape effect modeled with a second-rider polynomial, then the answer is probably not. A general strategy for interpreting these effects is to calculate adjusted predictions. In the case of mediation models, you would first predict m for different values of x for all cases and then use the predicted m to predict y. Then plot the average y as a function of x to show the estimated full mediation curve. "

In this regard, the following paper may be helpful:

Antonakis, J., House, R. J., & Simonton, D. K. (2017). Can super smart leaders suffer from too much of a good thing? The curvilinear effect of intelligence on perceived leadership behavior. Journal of Applied Psychology, 102(7), 1003-1021"

Personally, I would like to add that I would focus on the causal identification of the indirect effect (exclude the chance for endogeneity due to unobserved confounding) and the significance of the indirect effect.

HTH

Holger

Dear Kyran J.R. Kuijpers

I recommended you consult Article Incorporating nonlinearity into mediation analyses

Best Regards