I have never dealt with data on Nevada trial
awards, but my bias with similar data is
to consider a generalised linear model
with log link to ensure positive predictions.
My guess is that awards are highly skewed,
so that a straight linear model would
be dangerous.
Nick
[email protected]
Hyun Yoon
> Hi Nick,
>
> Thank you very much for the response. I'll look at
> the Miller citation.
>
> Regarding your query, here's a brief context to my
> problem. I am trying to measure the effect of a legal
> rule allowing pre-trial offers. I look at a period of
> 1990-2000, where Nevada changed its law in 1995, and
> the neighboring states did not. Using a
> differences-in-differences approach,
>
> AWARDS = a + b(Nev) + c(post1995) + d(Nev)(post1995) +
> f(control variables) + e
>
> I get a small and statistically non-significant change
> in damage awards.
>
> I would like to know whether there was a statistically
> significant change in the variance in AWARDS. In
> other words, did the rule reduce the variance in
> payouts, even if it did not change the mean.
>
> Thus, I am trying to do a diff-in-diff where I look at
> variance, not mean, of AWARDS.
>
> Have you ever confronted this problem?
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
