January 6, 2015 7:26 am

Mediation using bootstrapping tipsheets and published examples

Here are the links to three mediation tip sheets:


The PROCESS macro is the most versatile and tests a host of mediated effects including multiple mediation and moderated mediation.

Link to the PROCESS Tipsheet

Papers using PROCESS

Moderated Multiple Mediation

Loxton, N.J., Bunker, R.J., Dingle, G. A. & Wong, V. (2015). Drinking not thinking: A prospective study of personality traits and drinking motives on alcohol consumption across the first year of university. Personality and Individual Differences, 79 134-139. Used model 58

Tobin, S. J., Loxton, N. J., & Neighbors, C. (2014). Coping with Causal Uncertainty through Alcohol Use. Addictive Behaviors, 39, 580-585. Used Model 8


The INDIRECT macro is less versatile than PROCESS but is good for testing multiple mediation with either continuous or dichotomous outcomes. This was the predecessor to PROCESS and so the following papers can be used as models for writing up similar models using PROCESS.

Link to the INDIRECT macro

Papers using INDIRECT

Mediation w continuous DV

Davis, C., & Loxton, N. J. (2013). Addictive Behaviours and Addiction-Prone Personality Traits: Associations with a Dopamine Multilocus Genetic Profile. Addictive Behaviors, 38, 2306 - 2312.

Multiple Mediation w Dichotomous DV

Davis, C., Loxton, N. J., Levitan, R. D., Kaplan, A. S., Carter, J. C., & Kennedy, J. L. (2013). ‘Food Addiction’ and its Association with a Dopaminergic Multilocus Genetic Profile. Physiology and Behavior, 118, 63-69.

Multiple mediation w continuous DV

Hennegan, J., Loxton, N. J., & Mattar, A. (2013). Great Expectations. Eating expectancies as mediators of reinforcement sensitivity and eating. Appetite, 71, 81-88.

Yilmaz, Z., Davis, C., Loxton, N.J., Kaplan, A.S. Levitan, R.D. Dharmaraj, P., Carter, J. & Kennedy, J.L. (in press). Association between MC4R rs17782313 Polymorphism and Overeating Behaviour. International Journal of Obesity.

SOBEL (superseded by INDIRECT and PROCESS but the tip sheet talks through the basics of mediation using bootstrapping). This is still a good tipsheet as an introduction to the mechanics of mediation using bootstrapping. Although this macro does produce the results from the sobel test, these test is outdated and so just report the bootstrapped intervals

The SOBEL macro is the simplest macro for testing straight forward mediation analyses.

Link to the SOBEL tipsheet

More complex models

For those wishing to test mediation models with categorical predictors you can use Hayes' excellent MEDIATE macro. My student David Clark and I have used in recently in the following papers:

Mediation model with a three category IV and continuous DV ^foot

Davis, C. & Loxton, N.J. (2014). A Psycho-genetic Study of Hedonic Responsiveness in Relation to ‘Food Addiction’. Nutrients, 6, 4338 - 4353.

Two IVs in a multiple mediation model

Clark, M.D.T., Loxton, N.J. & Tobin, S.J. (2015). Multiple mediators of reward and punishment sensitivity on loneliness, Personality and Individual Differences , 72, 101-10

There is no tipsheet as yet....stay tuned

Those wishing more information on this macro and/or mediation etc are advised to refer to the Hayes website

Please note

I am unable to provide assistance on specific projects or tell you if your interpretation is correct. I simply don't have the time I'm afraid (there are budding psychologists who need be taught undergraduate statistics and personality theory). Hopefully though, these tipsheets will give you a good start. You can always consult the Hayes website for further information. His FAQ section is very informative. Plus there is a Facebook group with many enthusiastic participants. See Hayes. He da Man

Also be warned that I am trained as a psychologist, not a statistician and so these tipsheets are based on my understanding of the technique. Feel free to let me know if I've made any errors.

^foot: You can also use PROCESS to test this model, or if you also have a dichotomous DV, but you need to first set up dummy coded variables in SPSS. This is how I ran the Food Addiction Dx analyses in the paper btw.