Hello all, I hope you can help me with these questions.
I am developing a mediation analysis using the mediation package in R.
I need to analyze the influence of perceived learning (Mediating Variable –MV-) in the generation of pro-environmental outcomes (Dependent Variables DV)
I have different Independent Variables (IV) to test. One is numeric and another one is categorical. I have already developed the mediating analysis with the numeric IV and I got the results.
However, the motivation IV I have is categorical and has 4 values (Learning, Adventure, Wild, Social).
Do you know if creating dummies variables for this IV is a correct procedure to then run, the mediation analysis?
If the above is not possible, do you think that ANOVAs could be used as a substitute for regression models for mediation analysis?
If yes, is it correct to do the following?
ANOVA to see the relation between IV and DV
ANOVA to see the relation between MV and DV
But… How to analyze the relation: IV+MV DV?
Thanks in advance for your advice.
Dear: It is possible to analyze using a one-way analysis of variance table as well as a two-way analysis of variance table
Hello Daniela,
Yes, dummy variates may be used for IVs that are categorical. Yes, you can use good old OLS regression or path analysis (in the SEM arena) to evaluate the plausibility of your models.
Good luck with your work.
Hello Daniela Cajiao. If you have not already done so, I would urge you to read this article before trying to interpret your results. I consider it essential reading for mediation analysts. ;-)
Article Unwarranted inferences from statistical mediation tests – An...
Thank you all for your responses and Bruce Weaver for the paper. For now, I will proceed with the creation of dummies to do the regression with categorical IV. From my understanding, once the IV has been transformed into dummies I could run the mediation analysis. To give you a little more context, my study is about tourism and the generation of pro-environmental outcomes. What I want to do is to examine how environmental knowledge and satisfaction (MVs) mediate the relationship between motivation (IV -categorical) and pro-environmental outcomes (DV). Does it make sense for you to try mediation analysis?
What you are referring to is a parallel mediation analysis of two variables, if you are going to use regression models I recommend that you use JAMOVI's free access program that presents standardized coefficients.
You might wish to check this link. I think it would be helpful:
https://data.library.virginia.edu/introduction-to-mediation-analysis/
I think that you will find this approach much easier than trying to write and interpret an equivalent Analysis of Variance approach.
Best, D. Booth
No, ANOVA is not a substitute for Regression Model for Mediation Analysis, given the research objectives. I agree ANOVA and Regression can be used for continuous DV, and one to all category or continuous IVs(predictors). Specifically, would like to quote Jochen Wilhelm, 'ANOVA is a tool to check how much the residual variance is reduced by predictors in (nested regression) models, whereas the regression analysis aims to quantify effect sizes in terms of "how much is the response expected to change when the predictor(s) change by a given amount?" ' [See pls, https://www.google.com/url?sa=t&source=web&rct=j&url=https://www.researchgate.net/post/Can_anyone_help_me_to_get_the_core_differences_between_regression_model_and_ANOVA_model2&ved=2ahUKEwj2gtyYntrrAhXBShUIHfFQAKMQFjACegQIBBAD&usg=AOvVaw3oQEGpT1dOhtTDHk2u7H1U]
Yes, you can go ahead with using dummies for the Motivation levels, in your Regression Analysis, including Perceived Learning as a predictor that explains the underlying mechanism of the relationship between "numeric IV1" , then Motivation (IV2) and in the generation of the pro-environmental outcomes (DV).
David Morse Dear David are you suggesting ANOVA can substitute OLS regression in such cases in general?
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling
- Analyses of Variance
- ANOVA