On Thu, 12 Aug 2004, Cordula Stolberg wrote:
> Dear Statalisters,
>
> I (again) have a question about rescaling. I have a panel data set in which
> two variables (the dependent variable & one independent variable) are
> expressed thousands of dollars and the other independent variables are all
> index numbers or percentages. As I'm taking logs but had several negative
> numbers, I rescaled the whole dataset by adding a value to all variables
> such that the biggest negative value equals 1. The problem there was that I
This is very unlikely to be the right thing to do. Even if you got a
result, the values of the estimated coeficients would depend upon the
amount of the shift factor. People have been know to search for the factor
which maximized R**2, but I don't suggest that. It seems like an unusual
factor that would have a large positive effect when the dependent variable
is large in absolute value, but not much affect when it was near zero.
You don't say what you are modeling, or why you are taking logs, but if
the reason is to reduce heteroskedasticity, there is probably an
alternative multiplicitive scaling that can be combined with a linear
regression, to accomplish what you need.
For example, if the dependent variable is profits, you might divide
through by book value and run a linear regression.
Daniel Feenberg
*
* 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/