1

I am trying to estimate a regression model where I am interested in the effect of a certain magnitude, probability and expected value (probability * magnitude ) on reaction times. I was told that it is necessary to demean probability and magnitude before estimating its product. Can anyone give me an intuition in why you would use ev=demean(prob)*demean(mag)?

Thanks!

Laurie

Laurie
  • 179
  • 6
  • 2
    It reduces correlation between main effects and interaction effects. – Michael M Aug 06 '20 at 16:31
  • 3
    Does this answer your question? [Mean centering interaction terms](https://stats.stackexchange.com/questions/417029/mean-centering-interaction-terms) You don't _have_ to, it just puts the single-predictor coefficients into more interpretable ranges. – EdM Aug 06 '20 at 16:38

0 Answers0