st: -mfx-, categorical variables or dummy?, interpretation of betas with reg and probit

["Nirina F" <fstata@gmail.com>]

Hello again,

1- what is the star next to some variables in mfx. I believe it's not
an indication of significance because if I checked  the pvalue, some
of the variables with stars are not significant?

2- What is the difference in having a dummy 0-1 or having a
categorical variable 1-2? I run the regressions and I only found a
difference in the value of constants but not in the coefficients of
variables or Rsq or anything else.
I wonder if using 0-1 is easier for interpretation. And which one is
the best way?


[sysuse auto
-	reg  price mpg rep78 headroom trunk weight foreign
-	I get a _cons  -5630.414

-	gen ff=foreign
                      recode ff (1=2)
                      recode ff (0=1)
-	reg  price mpg rep78 headroom trunk weight ff
I get a _cons |  -9021.797


3- Linear probability model (plain reg)  and probit should not have a
big difference, the results of coefficient  from LPM and the marginal
effect from probit are approximately the same.
Now if one gets a value of coefficient above 1, how does one interpret
that? For an equation pr(y)=ax+b and a>1 (a being the marginal effect
if probit), do we say that probability of y in case of an increase of
x by 1 are more than 100%?

Thank you very much in advance,

Nirina
*
*   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/