I think there will be others more knowledgeable than I, but let me start.
At 12:33 PM 2/21/2014, cynicalflyer wrote:
>I am running a Least Squares Dummy Variable Regression (LSDV1) ...
First, *exactly* how are you doing this? With a built-in SPSS
procedure, an R module, or what? And what is the syntax you're using
to set up your model, for whatever tool you are using?
>[I have] data from 21 states observed 3 times (2007, 2008, 2009) and
>dropping one of
>the dummy value states.
What are you observing; what is (are) your dependent variable, or
variables? More important, how many dependent-variable values do you
have, per state per year? Is 'year' an independent category, and if
so, how do you get away without dropping one of the years, as well?
Or are you pooling all years for analysis, and you have three values
per state? What DO you have?
>I have 8 IVs that I am seeking to use (ENTER method).
It sounds like those are distinct from the 20 dummy variables for
states. That sounds like 20+8=28 degrees of freedom in your model,
which is a lot, especially if you have only 21*3=63 observations
total -- that's a little over 2 observations per degree of freedom,
far too little.
Note, however, that I get the counts of observations, and of degrees
of freedom, by making highly speculative guesses about what you are
doing. We need to have these guesses corrected, and more about your
study filled in, to answer usefully.
>When I regress I find two strange things:
>1) 7 of the 8 IVs are excluded
>2) 1 of my 20 states (21 - 1 dummy dropped)
OK. Again. *Exactly* what software, and what syntax for that
software, are you using; and *exactly* what are the error messages you get?
=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD
Locked
