--- bumbuminc <[email protected]> wrote:
> In the moment i am trying it with the standard "reg" command, like:
>
> *-------------------- begin example ---------------
> sysuse nlsw88, clear
> gen ttl_exp2 = ttl_exp^2
> gen wage_log = ln(wage)
> reg wage ttl_exp ttl_exp2 grade
> *--------------------- end example ----------------
> At first, I calculate the logarithm of the wage because i want to
> minimize anomalies.
What I did with -glm- is largely but not exactly similar to what you
are doing. I am modeling the mean wage which is linked to the predictor
variables through a log function, while you are modeling the mean
log(wage). The former is probably closer to what you really want to do.
Another way of thinking about the difference between these two models
is that you are modeling the geometric mean of wage, while I am
modelling the arithmatic mean of wage. See:
Roger Newson (2003) Stata Tip 1: The eform() option of regress. The
Stata Journal, 3(4): 445.
http://www.stata-journal.com/article.html?article=st0054
> A multivariate analysis via odd ratios is then a better resolution, i
> presume, or?
No. I said that there is a peculiarity in my sub-sub-discipline, which
makes it unnecesary for me to transform the factor change (exp[b]) into
percentage change (100*{1-exp[b]}). This peculiarity is that people in
my sub-sub-discpline are used to interpreting a very similar statistic,
the odds ratios.
-- Maarten
-----------------------------------------
Maarten L. Buis
Department of Social Research Methodology
Vrije Universiteit Amsterdam
Boelelaan 1081
1081 HV Amsterdam
The Netherlands
visiting address:
Buitenveldertselaan 3 (Metropolitan), room N515
+31 20 5986715
http://home.fsw.vu.nl/m.buis/
-----------------------------------------
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
