I have some longitudinal data on student reading scores.  There is a
baseline reading score at the start of a program and then tests one and
two years later. I am interested in the effects of a program (dummy
variable) administered in the first year on subsequent reading scores. In
looking at related studies, some analysts use the difference between
scores as the dependent variable and others use the baseline as a control
and the second measurement as the dependent variable. My assumption is
that the best way to analyze the data is Y_t+1=\phi Y_t + Z + treatment,
since it is a more general model allowing \phi to vary, rather than
differencing the data and restricting \phi to 1. Obviously, one could test
the restriction, but is there a reason to prefer differencing over the
more general model?

Any references or advice is much appreciated.

Thanks, Greg

Greg McAvoy
Associate Professor
Department of Political Science
University of North Carolina at Greensboro

**********************************************************
             Political Methodology E-Mail List
   Editors: Melanie Goodrich, <[log in to unmask]>
                    Xun Pang, <[log in to unmask]>
**********************************************************
        Send messages to [log in to unmask]
  To join the list, cancel your subscription, or modify
           your subscription settings visit:

          http://polmeth.wustl.edu/polmeth.php

**********************************************************