How can I compare regression coefficients in a same multivariate regression mode

I think this question has been answered in bits and pieces here and there, but I am still a bit unsure about what the best approach for this is: how to compare two coefficients from a regression to see if the effect strengths are significantly different.

For ex, I am interested in measuring attitude (IV1) and behavioral control (IV2)on the medication adherence (DV). I found that standardized betas were 0.30 and 19 for attitude and behavioral control, respectively. Is it reasonable to say the attitude is the strongest predictor of medication adherence?
Thank you very much!
yeah you can say that (0.30 effect is larger than 0.19 effect) since betas are standardized.

Besides OP said:
I found that standardized betas were 0.30 and 19 for attitude and behavioral control,
Although 19 is lager than 0.30 that does not mean that it is more important.

As I understand it, it is an often discussed question and I don't think there is any good solution to it.

I would just say: compare if one unit of change in IV1 and IV2 result in a large effect in your own view in the context you are investigating.


No cake for spunky
I agree with victor in that the use of standardized coefficients in linear regression is an acceptable way to tell relative impact (it is the primary reason standardized coefficients were created). This is more complex in logistic regression because there is no agreement on how to standardize coefficients and many different ones exist that won't lead to the same result commonly.

This is from Michigan State, which has an excellent reputation for stats.

If two independent variables are measured in exactly
the same units, we can asses their relative importance
in their effect on y quite simply
– The larger the coefficient, the stronger the effect

• Often, however, our explanatory variables are not all
measured in the same units, making it difficult to
assess relative importance

• This problem can be overcome for quantitative
variables by using standardized variables

Note that this author, although not all authors, argues that it is not a good idea to standardize dummy variables (because speaking of a one unit change makes limited sense when the unit can only change one unit).
Last edited: