February Updates: Hayes (2013) and Analyses

Hi there! These updates pertain to the progress that I have made on my thesis during the month of February:

February has been an incredibly busy month! I have spent much of the month combing through my database, for any potential missing values amid my subscale item scores and my global scores for analyses. Though this process can be rather exhausting and time-consuming, it has been a huge help in ensuring that I am completely confident about the subscale scores that I utilize in my analyses.

I have also diligently finished reading nearly all of Hayes (2013), as well as his additionally relevant materials, such as his 2015 publication on the Index of Moderated Mediation. And after reading his works front and back, I began thoroughly conducting my analyses. After reading, rereading, and analyzing Hayes’s 2015 publication, however, I have decided to pursue simple mediational analyses, rather than to conduct moderated mediation analyses.

In his article, Hayes (2015) writes that significant conditional direct and indirect effects found in moderated mediation analyses should not be examined unless the Index of Moderated-Mediation is significant. With a dichotomous moderator (e.g., gender) in the model, the Index of Mod-Med equates to the relative difference between two conditional indirect effects. Hayes (2015)notes that in exclusively examining conditional direct and indirect effects (rather than the Index of Mod-Med), researchers are merely assessing whether the simple slopes are statistically different from zero. However, moderated-mediation would actually necessitate determining if the simple slopes are statistically different from one another. In this way, the Index of Mod-Med is a novel quantity that more accurately represents the presence or absence of moderated mediation.

Though the Index of Mod-Med is still a very recent value in the realm of quantitative psychology, I wanted to ensure that my methods are as statistically accurate as possible. As such, with preliminary analyses revealing no significant Index of Mod-Med values, I opted to remove gender as a moderator from my analyses, as recommended by Hayes in his (2013) guide. Unfortunately, initial analyses also revealed that most (if not all) Time 2 effects were not emerging. So while I will likely still devote some discussion of these effects in my thesis, I will likely end up pursuing simple mediation models with my Time 1 variables for my final analyses.

In other news, I officially finish my doctoral program interviews this month! This has been a very exciting process, but I am so happy to be finally nearing an end. I am hoping that by the end of March, I will know where I am headed post-graduation. Fingers crossed!

Nine months down, three to go! Thank you, as always, for reading and I look forward to reporting more progress in the coming weeks!

Hayes’s (2013) book:

Hayes, A. F. (2013). Introduction to mediation, moderation, and conditional process analysis: A regression-based approach. New York, NY: Guilford Press.

Hayes’s (2015) article:

Hayes, A. F. (2015). An index and test of linear moderated mediation. Multivariate Behavioral Research50(1), 1-22. doi:10.1080/00273171.2014.962683