Please do not send HTML to the list.
This is explained in the FAQ.
I do not know the Wiggins and Ruefli
reference. Even if I did there will be
others who do not. The importance of giving full references
is also emphasised in the FAQ.
Whether the Stata is right is immaterial
if the statistics is wrong. I see here
no adjustment for, or recognition of,
the consequences of
1. dependence in the original data
or of
2. applying multiple significance tests on the same data.
Perhaps the missing reference explains all this.
In any case, how we can tell if the Stata is
right without a specification of what it is
supposed to do?
You might be better advised to draw graphs and think
about them.
Nick
[email protected]
irodriguez
I folowing wiggins and ruefli paper. They use an iterative kolmogorov smirnov approach to distinguish over time entries that perform significantly better or worse than their counterparts
My data set is a panel data with firms, industry, a variable of performance for 1990-2004 period.
I need help to know if this command is ok
Prep (Do file 1)
use ba se_perform
encode firm, gen(cod_firma)
keep if cod_sector==12
keep if initial_year==1999
reshape long roa, i(firm) j(year)
destring roa, replace
Coco (Do file 2)
capture gen estrato=1
capture gen fa=1
replace estrato=1 if estrato==0
sum cod_firma
return list
local numfirmas=r(max)
forvalues k=1/`numfirmas' {
replace fa=.
replace fa=1 if estrato==1
capture replace fa=0 if cod_firma==`k' & estrato==1
capture ksmirnov roa, by(fa)
capture return list
capture replace estrato=0 if r(p)<0.05 & fa==0
}
tab estrato
Una (Do file 3)
replace estrato=. if estrato==0
do coco
tab estrato
Media (Do file 4)
Summarize if estrato==1
Tab cod_firma estrato
Rep (Do file 5)
drop if estrato==1
replace estrato=1
do coco
tab estrato
*
* 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/
