Just need someone to do a sanity for me.

In linear regression, one might want to transform (log, power, etc...) a continuous independent variable so that it's relationship with the continuous dependent variable be more correlated for statistical significance during data processing. One can check how effective the transformation is by calculating the correlation between the two variables before and after the transformation against the dependent variable or looking at significant p-values after running the regression.

What about in logistic regression? Would it make sense to apply this same processing step of a continuous independent variable if the dependent variable is binary? How would one measure that the transformed continuous variable has a meaningful impact on the dependent variable other than just running the logistic regression and checking for significance?

Perhaps I should only use categorical inputs for logistic regression?

Thank you!