R Squared and dummy variables

#1
Hey,
If you changed your model from one variable to a number of dummy variables, would the R^2 increase considerably, supposing that the dummy variables helped to create a more accurate estimate of the dependent variable?

Thanks
 

hlsmith

Omega Contributor
#2
Your question is not clear. So you are conducting linear regression and have a continuous predictor that you may discretize? Now your question is how it will impact the R^2 value for the model. Is that your question?


We can talk about it, but I would recommend to first do it and report your findings. This will likely answer some of the basic question about what you are actually working with.
 

Karabiner

TS Contributor
#4
If you changed your model from one variable to a number of dummy variables, would the R^2 increase considerably, supposing that the dummy variables helped to create a more accurate estimate of the dependent variable?
I'd guess that this could happen in case of a nonlinear relationship between the predictor and the dependent variable.

With kind regards

K.


@evelyn13
Seemingly the OP didn't mean dichotomization, but multiple dummies.
 
#6
I'd guess that this could happen in case of a nonlinear relationship between the predictor and the dependent variable.

With kind regards

K.


@evelyn13
Seemingly the OP didn't mean dichotomization, but multiple dummies.
True sorry...my points still stand though (including the link I attached).