Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Multivariate linear regression

  1. May 22, 2010 #1
    I have a model y= beta0 + beta1 x1 + beta2 x2 + eps, eps~N(0,1).

    How to test hypothesis beta1=0 ? Is the same test for beta2=0?
  2. jcsd
  3. May 22, 2010 #2
    you can use a t test just like you do in simple linear regression,
    except the degrees of freedom is n-3 not n-2,
    You can test each variable separately.

    you have to be careful about any conlcusions you make for example if x1 and x2 are highly correlated.
  4. May 31, 2010 #3
    Can someone help me to do this in SAS?
    Is this correct?

    PROC REG DATA = dataset;
    MODEL y = x1 x2;
    test x1=0 ;
    test x2=0;

    How can I save p value from each test to a new data?
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook