Hello,
Hoping you can help. I am currently carrying out a mediation analysis using PROCESS macro for SPSS. I have 1 predictor, 1 outcome and 3 mediators. I have been using Andy Field's book as a guide .This suggests using Kappa squared for reporting effect size of the indirect effect. However, this function is no longer in version 3 of PROCESS as I understand Kappa squared is no longer recommended to be used due to a mathematical error found in it.
Looking at Hayes' book it seems the partially standardised effect and completeley standardised effect have something to do with effect sizes. But he doesn't offer a guide for interpreting this. Is it a case of interpreting the coefficient as you might a usual correlation coefficient (e.g. using Cohen, 1988, 1992) criteria? If not, how else might I interpret the effect size?
Many thanks!
Rebecca
Hi Rebecca. See if you can find help in
37. Hancock GR. Effect size, power, and sample size determination for
structured means modeling and MIMIC approaches to between-groups
hypothesis testing of means on a single latent construct. Psychometrika.
2001;66(3):373–88.
We used it in this article
Article Assessing understandings of substance use disorders among No...
to calculate effect size based on mean difference & variance in a Multigroup confirmatory factor analysis (undertaken with Mplus with a structural equation modeling procedure).
As I remember (based on statistical advice we got), it was not possible to calculate effect size based on coefficients (as in our MIMIC model in the same article). Can somebody update on that?
john-kåre
Rebecca,
in addition to John-Kåre's advice, this might help:
Article Effect Size Measures for Mediation Models: Quantitative Stra...
Thank you both. I have also since had advice that Wen and Fen (2009) advise against use of effect size in mediation.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS