Is it meanful to obtain standardized coefficients of binary/categorical exogenous variables?

[chenb...@gmail.com]

Hi all,

I know that standardized coefficients are quite useful to compare the effect size between different exogenous variables on endogenous variables.

However, I am a bit confused if it is meanful to obtain the standardized coefficients  when exogenous variables are in binary or categorical type, and endogenous variables are in contenuous type?

Or do you have any alternative choice, if I want to compare the effect size?

Thanks!

[Terrence Jorgensen]

I know that standardized coefficients are quite useful to compare the effect size between different exogenous variables on endogenous variables.

Not really valid

http://scholar.harvard.edu/files/gking/files/mist.pdf

But yes, it is done.

However, I am a bit confused if it is meanful to obtain the standardized coefficients  when exogenous variables are in binary or categorical type, and endogenous variables are in contenuous type?

They are called point-biserial correlations, which are simply Pearson correlations between 1 binary and 1 continuous variable.  

Cohen's d could be obtained by standardizing only the endogenous variable, so that the 0-1 metric is retained in the exogenous dummy code.  If you prefer that, I think you would need to fit your model to standardized (endogenous) data yourself.

Terrence D. Jorgensen

Postdoctoral Researcher, Methods and Statistics

Research Institute for Child Development and Education, the University of Amsterdam

http://www.uva.nl/profile/t.d.jorgensen