Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

AW: st: AW: dropped interaction terms from FE model


From   P K <statistics_2009@yahoo.de>
To   statalist@hsphsun2.harvard.edu
Subject   AW: st: AW: dropped interaction terms from FE model
Date   Thu, 18 Jun 2009 02:22:31 +0000 (GMT)

Thanks, Martin.

I checked this and there is enough variation in the dropped variables. 

Also, STATA removes them when I try to estimate a RE model (instead of a FE model).

Any other ideas on why they get dropped?

Thanks!




----- Ursprüngliche Mail ----
Von: Martin Weiss <martin.weiss1@gmx.de>
An: statalist@hsphsun2.harvard.edu
Gesendet: Mittwoch, den 17. Juni 2009, 17:49:30 Uhr
Betreff: st: AW: dropped interaction terms from FE model


<> 

Just look at -xtsum- for your covariates after the estimation to see whether there is any within variation in the dropped variables. That would be the most likely culprit for an FE model. Note the covariate "black" in this example...


*************
//     Setup
webuse nlswork, clear

gen age2 = age*age
gen ttl_exp2 = ttl_exp*ttl_exp
gen tenure2 = tenure*tenure
gen byte black = race==2

xtset idcode

xtreg ln_w grade age* ttl_exp* /* 
*/  tenure* black not_smsa south, fe

capt ssc inst indeplist
qui indeplist, dropped
xtsum `r(X)' if e(sample)
*************



HTH
Martin


-----Ursprüngliche Nachricht-----
Von: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu] Im Auftrag von P K
Gesendet: Mittwoch, 17. Juni 2009 23:44
An: statalist@hsphsun2.harvard.edu
Betreff: st: dropped interaction terms from FE model

Hi,

I am estimating  the following FE panel model:

xtreg AverChangeROEadj2 strategy_01 duration_change_t1 ceo_change_t1 ceo_int_t1 tmt_turnover_t1 sizelnempl age prior_slack_avail2 ROEadj_1 external_change_dummy lgcount_strategy performance_crisis_ROEadj strategy_01_change_t1 strategy_ceoint ceoint_change change_ceoint_strategy, fe

However, the interaction between strategy_01 (a dummy variable) and change duration (continuous variable) gets dropped from the model (strategy_01_change_t1).

Also, the interaction between strategy_01 (a dummy variable), change duration (continuous variable) and ceo change (a dummy variable) gets dropped from the model. (change_ceoint_strategy)

Any hints why this could be the case and what to do about it (there is no multicollinearity or missing value problem)? 
Thanks, Pat


Here's the STATA output:


Fixed-effects (within) regression Number of obs      =       430 
Group variable: number Number of groups   =        60 




R-sq:  within  = 0.0998 Obs per group: min =         1 
between = 0.0009 avg =       7.2 

overall = 0.0013 max =         9 






F(14,356)          =      2.82 

corr(u_i, Xb)  = -0.9883 Prob > F           =    0.0005 








AverCh~Eadj2       Coef.   Std. Err. t    P>t     [95% Conf. Interval] 




strategy_01    2.161211   2.012137 1.07   0.284    -1.795958     6.11838 
duration_c~1    -.475537    .877321 -0.54   0.588     -2.20092    1.249846 
ceo_cha~e_t1    .7214819   3.349671 0.22   0.830    -5.866149    7.309113 
ceo_int_t1   -2.241493   4.682335 -0.48   0.632    -11.45001    6.967021 
tmt_turno~t1   -2.060425   1.605712 -1.28   0.200    -5.218299    1.097448 
sizelnempl   -4.503579   1.727811 -2.61   0.010     -7.90158   -1.105579 
age    -.464394   .4892887 -0.95   0.343    -1.426654    .4978657 
prior_slac~2    .5187986   .3436538 1.51   0.132    -.1570482    1.194645 
ROEadj_1   -.0276773   .0467122 -0.59   0.554    -.1195439    .0641892 
external_c~y    1.628413   1.329209 1.23   0.221    -.9856769    4.242503 
lgcount_st~y    1.426753   2.201901 0.65   0.517    -2.903616    5.757123 
perform~Eadj    5.296512   1.790716 2.96   0.003       1.7748    8.818223 
strateg~e_t1   (dropped) 


strategy_~nt    15.62659   5.616043 2.78   0.006     4.581794    26.67138 
ceoint_cha~e   -6.139526    1.89593 -3.24   0.001    -9.868157   -2.410895 
c~int_stra~y   (dropped) 


_cons    69.46962   40.03339 1.74   0.084    -9.262056    148.2013 




sigma_u   33.386383 


sigma_e   12.392948 


rho   .87889857   (fraction of variance due to u_i) 





F test that all u_i=0:     F(59, 356) =     0.86             Prob > F = .7540 


      

*
*   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/


*
*   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/



      

*
*   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/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index