Dear Statalist,
Thank you very much for all your previous help regarding my project.
I have panel data regarding individuals' (n=230) performance over 4 rounds - it's an experiment, so 920 observations.
I have used
and obtained some nice results.
Now I wanted to use margins:
but have realised that the number of observations is 920, without any groups.
I have used the margins command right after xtlogit.
did not work even though the instructions (https://www.stata.com/manuals/rmargins.pdf) say that vce is an option.
I would be most grateful if anyone could comment on the soundness of my margins command (especially regarding the standard errors) and whether these results can be trusted.
Thank you very much in advance!
Kind regards,
Mary
I am using Stata 17 SE.
Thank you very much for all your previous help regarding my project.
I have panel data regarding individuals' (n=230) performance over 4 rounds - it's an experiment, so 920 observations.
I have used
Code:
xtlogit reb_sa i.reb_ak i.reb_ua i.reb_g i.round i.risk i.T_C i.reby_n c.PPD_sa##c.r_sa, vce(cluster CASE) or nolog
Calculating robust standard errors ...
Random-effects logistic regression Number of obs = 920
Group variable: CASE Number of groups = 230
Random effects u_i ~ Gaussian Obs per group:
min = 4
avg = 4.0
max = 4
Integration method: mvaghermite Integration pts. = 12
Wald chi2(14) = 156.01
Log pseudolikelihood = -331.51916 Prob > chi2 = 0.0000
(Std. err. adjusted for 230 clusters in CASE)
---------------------------------------------------------------------------------
| Robust
reb_sa | Odds ratio std. err. z P>|z| [95% conf. interval]
----------------+----------------------------------------------------------------
1.reb_ak | 8.402469 2.551766 7.01 0.000 4.633439 15.23738
1.reb_ua | 3.879268 1.141404 4.61 0.000 2.17921 6.905584
1.reb_g | 2.941542 .8846838 3.59 0.000 1.631442 5.303692
|
round |
2 | 2.701071 .8781532 3.06 0.002 1.428234 5.108256
3 | 1.126661 .3371901 0.40 0.690 .6266768 2.025551
4 | 2.132915 .6343999 2.55 0.011 1.190687 3.820755
|
risk |
2 | 1.173569 .3310536 0.57 0.570 .6751374 2.039977
3 | 4.309221 1.364784 4.61 0.000 2.316395 8.016504
4 | 4.707272 1.72959 4.22 0.000 2.290935 9.672211
|
T_C |
TG | .6641125 .1871551 -1.45 0.146 .3822632 1.153774
1.reby_n | 1.090605 .2859005 0.33 0.741 .6524181 1.823092
PPD_sa | 1.266439 .0691242 4.33 0.000 1.137954 1.409432
r_sa | 1.04e+09 2.23e+10 0.97 0.332 6.42e-10 1.69e+27
|
c.PPD_sa#c.r_sa | 1.32e-06 8.41e-06 -2.12 0.034 4.77e-12 .3629207
|
_cons | .1143815 .0573874 -4.32 0.000 .042785 .3057879
----------------+----------------------------------------------------------------
/lnsig2u | .1390501 .387292 -.6200282 .8981284
----------------+----------------------------------------------------------------
sigma_u | 1.071999 .2075883 .7334366 1.566845
rho | .2588801 .0743063 .1405323 .4273383
---------------------------------------------------------------------------------
Note: Estimates are transformed only in the first equation to odds ratios.
Note: _cons estimates baseline odds (conditional on zero random effects).
Now I wanted to use margins:
Code:
margins,dydx(*)
Average marginal effects Number of obs = 920
Model VCE: Robust
Expression: Pr(reb_sa=1), predict(pr)
dy/dx wrt: 1.reb_ak 1.reb_ua 1.reb_g 2.round 3.round 4.round 2.risk 3.risk 4.risk 1.T_C 1.reby_n PPD_sa r_sa
------------------------------------------------------------------------------
| Delta-method
| dy/dx std. err. z P>|z| [95% conf. interval]
-------------+----------------------------------------------------------------
1.reb_ak | .2518499 .0379563 6.64 0.000 .1774569 .326243
1.reb_ua | .1543498 .0368114 4.19 0.000 .0822008 .2264988
1.reb_g | .1169143 .03533 3.31 0.001 .0476688 .1861599
|
round |
2 | .0969979 .0304955 3.18 0.001 .0372278 .156768
3 | .0127513 .0319273 0.40 0.690 -.0498251 .0753278
4 | .0759196 .0297198 2.55 0.011 .0176698 .1341693
|
risk |
2 | .0184477 .0323549 0.57 0.569 -.0449667 .081862
3 | .1474136 .0288938 5.10 0.000 .0907828 .2040445
4 | .1547088 .0334769 4.62 0.000 .0890953 .2203223
|
T_C |
TG | -.0406796 .0279514 -1.46 0.146 -.0954633 .0141041
1.reby_n | .0085604 .0259302 0.33 0.741 -.0422617 .0593826
PPD_sa | .0179221 .0045255 3.96 0.000 .0090524 .0267919
r_sa | 3.732591 1.870042 2.00 0.046 .0673752 7.397806
------------------------------------------------------------------------------
Note: dy/dx for factor levels is the discrete change from the base level.
I have used the margins command right after xtlogit.
Code:
margins,dydx(*) vce(cluster CASE)
I would be most grateful if anyone could comment on the soundness of my margins command (especially regarding the standard errors) and whether these results can be trusted.
Thank you very much in advance!
Kind regards,
Mary
I am using Stata 17 SE.
Comment