st: Polynomial/Quadritic trend in residuals

["Abdul Q Memon" <a.memon@ucl.ac.uk>]

Hi

I am running GLM and GEE with AR1 error structure in STATA. It seems the
residuals obtained from the preferred model have polynomial trend in it.
There is no sign for the presence of linear trend. When I am trying to add
(time)^2 variable into the preferred model the polynomial trend
diminishes. Time is a variable having 1 to 1830 (1 represents 1st January
2001 and 1830 represents 31st Dec 2005). I have this data for 24 panels
and each panel has 1830 observations.

Variables in preferred model

1.	Road class (5 categories)+ vehicle class (5 categories)+Day of week (7
categories) +Month (12 categories) + Holidays+ Christmas + New year+
Interaction of Road class and Vehicle class (24 categories)+ Time

Could anyone add few suggestions how to accommodate the polynomial trend
in the above mentioned model. My initial  thoughts are to modify the Time
variable in such way that it can handle polynomial trend. Any suggestion
and comments on this  are more than welcome.

Regards

Memon


*
*   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/