Re: [Gretl-users] Easy - Printing down my critical-p
Sebastien Rodrigue-Prive wrote:
Here it's my programation with Gretl :
open mensuelmemoire-sanslesnulles-160310
# create big list of variables
list biglist = I*EX I*IM
# create a list of regresor
list reg1list = TREND PRE911 POST911 TX_CHG Jan Fev Mar Avr Mai Juin Juil
Aou Sep Oct Nov Dec DICHO_OCT
# full results matrix, start empty
matrix Res = {}
loop foreach i biglist
# create multiplicator
# diminuer leS VARIABLES ( / 1 000 000)
PIB_IND_C_ = PIB_IND_C / 1000000
PIBN_US_ = PIBN_US / 1000000
series $i_def = $i / (PIB_IND_C_ * PIBN_US_) / DEF_US
# run the regression
ols $i_def 0 reg1list
# grab the coefficient and stan-error
matrix res = $coeff' | $stderr'
# and add to the big results matrix
Res = Res | res
endloop
# open a fil to write down the results
outfile "results par un million 100504-1.txt" --write
printf "%#12.6g", Res
outfile --close
Everythings worked very well.
The problem it's when I want to get in my res matrix the value of my
coefficient and my the value of my critical-p... doesn't work.
I tried $pvalue, doesn't work. I really need to get out the critical-p
for my presentation.
P-values for the coefficients, being derived results, are not
available via a direct accessor; you need to use the pvalue()
function. To get the two-sided p-values that are printed in the
gretl output you'd do something like this:
open data9-7
ols 1 0 2 3 4
matrix pvals = 2 * pvalue(t, $df, abs($coeff ./ $stderr))
print pvals
Allin Cottrell