Hello, statisticians. I would like to explore which variable (for example A or B) has a stronger correlation with C. Since this is a panel dataset and I will use mixed model. My question is how could I compare the effects of these two IDV (A or B) on DV (C).
My idea is to
1. check the coefficients after standardization? But here A is beta distributed, I'm not sure whether it's applicable.
2. Model C~A and C~B separately using lmm and compare the R square? Does this method make sense?
And besides mixed model, could you please give any suggestions about other methods? Thanks a lot for your reply in advance.
Best regards.
The only good answer is to run correlation between each two variables. Regards.
Dear Colleague
I am not a statistician but a physicist, and as a physicist, I would produce the time series of Pearson's correlation function of A~C and B~C separately. After that, if my objective were to analyze the data, I would integrate the time series to get the average correlation to decide which is more important. If I were interested in future behavior, I would check the average derivative of the curves, or their tendencies, to extrapolate which one will be most important in the long-time behavior. I think, as a physicist, that this last approach is nicer because if the derivatives of both are asymptotically equivalent, it suggests that you may be searching for a correlation of a combination of both variables: D = aA + bB. A quick reminder that I always forget in my analisis. Correlations only work if A and C or B and C present themselves in a linear relationship, so do not forget to linearize your data.
I hope I could help somehow.
Best regards
- Badges
- Science topic
- Similar topics
- Organizational Psychology
- Training