st: Question about Modified Park Test

["Insun Choi" <ischoi@rci.rutgers.edu>]

Dear all,
To determine family distribution for GLM model, I used Modified Park
Test(MPT).
The results was lambda=4. But I could not find the distribution when the
lambda=4.
What is the family distribution of GLM model when the result of MPT is
lambda=4 ??


The procedures which I used were as bellowing;
****************************************************************
glm tat_time  med_cpoe_ord page2 page3 page4 pfemale age2 age3 age4 female
emergen pulmonary family surgery cardio gastro do md pa rnfa central_drug
gast_drug card_drug antinf_drug horm_drug autom_drug oth_drug, family(gamma)
link(log)
predict yhat,mu
gen lnyhat=ln(yhat)
gen r2=((tat_time-yhat)^2)
glm r2 lnyhat, family(gamma) link(log) robust nolog
test lnyhat==0
test lnyhat==1
test lnyhat==2
test lnyhat==3
test lnyhat==4

************ TEST RESULT ************

. test lnyhat==0

( 1)  [r2]lnyhat = 0

           chi2(  1) =  160.88
         Prob > chi2 =    0.0000

. test lnyhat==1

( 1)  [r2]lnyhat = 1

           chi2(  1) =   95.91
         Prob > chi2 =    0.0000

. test lnyhat==2

( 1)  [r2]lnyhat = 2

           chi2(  1) =   47.65
         Prob > chi2 =    0.0000

. test lnyhat==3

( 1)  [r2]lnyhat = 3

           chi2(  1) =   16.10
         Prob > chi2 =    0.0001

. test lnyhat==4

( 1)  [r2]lnyhat = 4

           chi2(  1) =    1.26
         Prob > chi2 =    0.2618

Insun Choi

Insun Choi, Ph.D.
Assistant Research Professor
Department of Pharmacy Practice and Administration
Erneat Mario School of Pharmacy
Rutgers, The State University of New Jersey
160 Frelinghuysen Road
Piscatway, New Jersey 08854-8020
732-445-7993
Email : ischoi@rci.rutgers.edu



*
*   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/