Alexandra,
> -----Original Message-----
> From: [email protected]
> [mailto:[email protected]] On Behalf Of
> Alexandra Guisinger
> Sent: Tuesday, September 02, 2008 3:41 PM
> To: [email protected]
> Subject: st: RE: "V_b-V_B is not positive definite" / XTOVERID
>
> Dear Mark and others,
>
> I am using stata 9.2 updated July 2007 and so was able to use both the
> -sigmamore- and -sigmaless- options after xtreg. However,
> even when I ran "hausman FE_all RE_all, sigmamore" and as a
> check hausman FE_all RE_all, sigmaless", I still received the
> message "(V_b-V_B is not positive definite)" in both cases.
>
> Given the advantages of the artificial regression approach, I
> went ahead and used xtoverid instead. So at this point -
> unless this is signaling a problem with my data - I am just
> curious as to why I am getting the message when using
> sigmaless or sigmamore should guarantee a positive outcome.
I'm curious too. I just tried reproducing the problem in Stata 9.2 and
couldn't. The following gets the "not pd" error message without the
sigma options, but add one of these and it goes away:
sysuse abdata
xtreg n w yr*, fe
est store fe
xtreg n w yr*, re
est store re
hausman fe re
hausman fe re, sigmamore
hausman fe re, sigmaless
xtoverid
Could you post your output to the list? And also check that your
versions of -hausman- and -xtreg- are up to date. Mine are
. which hausman
C:\Stata9\ado\updates\h\hausman.ado
*! version 3.0.14 18jan2007
. which xtreg
C:\Stata9\ado\updates\x\xtreg.ado
*! version 1.5.2 19mar2007
--Mark
>
> Best,
>
> Alexandra
>
> *
> * 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/
>
--
Heriot-Watt University is a Scottish charity
registered under charity number SC000278.
*
* 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/
