Thread: Comparing regression coefficients - dependent samples

1. Comparing regression coefficients - dependent samples

Hi,

I used two methods (A and B) to obtain measurements in a group of subjects. The methods (ways of measuring) are different, but they should measure the same property.

I performed linear regressions to analyze the relation between the measurements (m_A and m_B) and another property (p) of the subjects:

m_A = b_A * p + c_A
m_B = b_B * p + c_B

I am interested in the regression coefficients (b_A and b_B). I would like to know if they differ significantly.

Is there a way to test for significant differences between the regression coefficients? Please keep in mind that the same subjects were used for both measurements.

2. Re: Comparing regression coefficients - dependent samples

A Wald test or any other type of F-test would come in handy, i think. In stata you just run the regression and then write 'test b_A = b_B'.

3. Re: Comparing regression coefficients - dependent samples

Thanks for your reply, but can you (or someone else) maybe explain this in a bit more detail? I cannot find any information about a Wald test comparing two coefficients. I don't have Stata, but I do have SPSS and R.