Semipartial correlation and logistic regression I am running a series of multiple linear regression models as well as logistic regression models. I have calculated the semipartial correlations (sr) as a measure of effect size for the linear regression models and the odds ratio (OR) as a measure of effect size for the logistic regression models.
My first question is can I compare the sr to the OR? And if so, how do I do so? Based on Cohen's criteria my sr is small (<.3) and my OR is small (< 1.5).
My second question is, is it possible to calculate a sr for my logistic regression model when my predictor is continuous?
Thanks in advance!
 A: Is the outcome in your logistic regression a dichotomized version of the outcome in linear regression?
There is no direct equivalent in logistic regression to the semipartial correlation.  There can hardly be, when one type of model assesses linear relationships and the other assesses relationships via an S-curve.
"Can I compare" -- I see that you have :-)  Just be aware that Cohen freely admitted (insisted, when necessary) that these guidelines of his were rough and were not equally applicable to all situations.
I think you would do best to seek substantive interpretations:  describe the results from each regression -- semipartial correlation, odds ratio and not least the coefficient itself -- and try to interpret (and perhaps indirectly or informally compare) their meaning and their level of importance specifically for your context.
You can also generate standardized regression coefficients (linear) or their rough equivalent (logistic).  You might look up Scott Menard's online logistic regression book for info on that.
