Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: chow test multicollinearity


From   [email protected]
To   [email protected]
Subject   st: chow test multicollinearity
Date   Sat, 26 Feb 2011 09:49:05 +0100 (CET)

Hi,

I run a chow test by running a single regression with dummy variables.
The F statistic is significant; however, the interacted terms are strongly multi-collinear.


What should I do in these case? 


a) is the chow test run through interacted terms (i.e. in one single regression) affected by the presence of multicollinearity among interacted terms?


b) I think that, given multicoolinearity, I cannot trust significance of interacted coefficiencts, and hence cannot say that the predictors are different in the two subgroups; correct?


c) is there a way to solve the multicollinearity? 


d) I know the command orthog to solve multicollinearity; however, can I apply this command after a predictor is squared? That is,
 ---- first calculate the standardized values of X 
 ---- then get the square of this standardized value
 ---- calculate the product of each of these two variables with a moderator Q
 ---- and THEN using "orthog"  on the values already standardized and interacted 
      (i.e., orthog  z_X   z_Xsquare   Q   z_X*Q   z_Xsquare*Q  )  
I fear this procedur might bias results. What do you think?



e) since my problem is with multicollinearity affecting interactions, could I get the chow test from moderated regressions, and attributing this chow test to coefficients gotten in separate regressions?



Thanks



*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/


© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index