GEE hierarchical stepwise regression

[hale be]

Hi! I've been trying to run a stepwise hierarchical GEE analysis on a panel data. I've managed to successfully conduct the stepwise regression, but when I try doing it hierarchically I keep getting an error message. I was wondering why this might be? All answers are much appreciated. Thanks!

[Joseph Coveney]

Hi! I've been trying to run a stepwise hierarchical GEE analysis on a panel data. I've managed to successfully conduct the stepwise regression

How did you manage to do that?

. stepwise, pr(0.2): xtgee mpg c.(headroom trunk weight length turn displacement gear_ratio), i(rep78)
xtgee is not supported by stepwise
r(199);

Which is consistent with the documentation for -stepwise-. (If that isn't what you were doing, then you should probably look into this.)

when I try doing it hierarchically I keep getting an error message. . . .All answers are much appreciated. Thanks!

You can fit the most common GEE model using -glm-, and -glm- allows -stepwise-. See below.

.ÿ
.ÿversionÿ16.1

.ÿ
.ÿclearÿ*

.ÿ
.ÿsetÿseedÿ`=strreverse("1603424")'

.ÿ
.ÿquietlyÿsysuseÿauto

.ÿsummarizeÿrep78,ÿmeanonly

.ÿquietlyÿreplaceÿrep78ÿ=ÿruniformint(`r(min)',ÿ`r(max)')ÿifÿmi(rep78)

.ÿ
.ÿ*
.ÿ*ÿBeginÿhere
.ÿ*
.ÿtempnameÿGEEÿGLM

.ÿ
.ÿquietlyÿxtgeeÿmpgÿc.(headroomÿtrunkÿweightÿlengthÿturnÿdisplacementÿgear_ratio),ÿ///
>ÿÿÿÿÿÿÿÿÿi(rep78)ÿfamily(gaussian)ÿlink(identity)ÿcorr(independent)ÿvce(robust)ÿnolog

.ÿmatrixÿdefineÿ`GEE'ÿ=ÿe(b)

.ÿ
.ÿquietlyÿglmÿmpgÿc.(headroomÿtrunkÿweightÿlengthÿturnÿdisplacementÿgear_ratio),ÿ///
>ÿÿÿÿÿÿÿÿÿfamily(gaussian)ÿlink(identity)ÿvce(clusterÿrep78)ÿnolog

.ÿmatrixÿdefineÿ`GLM'ÿ=ÿe(b)

.ÿ
.ÿmata:ÿmreldif(st_matrix(st_local("GEE")),ÿst_matrix(st_local("GLM")))
ÿÿ5.66491e-14

.ÿ
.ÿ//ÿSo,ÿ.ÿ.ÿ.
.ÿstepwise,ÿpr(0.2)ÿhierarchical:ÿ///
>ÿÿÿÿÿÿÿÿÿglmÿmpgÿc.(headroomÿtrunkÿweightÿlengthÿturnÿdisplacementÿgear_ratio),ÿ///
>ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿfamily(gaussian)ÿlink(identity)ÿvce(clusterÿrep78)
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿbeginÿwithÿfullÿmodel
pÿ=ÿ0.7652ÿ>=ÿ0.2000ÿÿremovingÿgear_ratio
pÿ=ÿ0.2567ÿ>=ÿ0.2000ÿÿremovingÿdisplacement
pÿ=ÿ0.7642ÿ>=ÿ0.2000ÿÿremovingÿturn
pÿ=ÿ0.0667ÿ<ÿÿ0.2000ÿÿkeepingÿÿlength

GeneralizedÿlinearÿmodelsÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿNumberÿofÿobsÿÿÿ=ÿÿÿÿÿÿÿÿÿ74
Optimizationÿÿÿÿÿ:ÿMLÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿResidualÿdfÿÿÿÿÿ=ÿÿÿÿÿÿÿÿÿ70
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿScaleÿparameterÿ=ÿÿÿÿ11.9814
Devianceÿÿÿÿÿÿÿÿÿ=ÿÿ826.7167277ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ(1/df)ÿDevianceÿ=ÿÿÿ11.81024
Pearsonÿÿÿÿÿÿÿÿÿÿ=ÿÿ826.7167277ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ(1/df)ÿPearsonÿÿ=ÿÿÿ11.81024

Varianceÿfunction:ÿV(u)ÿ=ÿ1ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ[Gaussian]
Linkÿfunctionÿÿÿÿ:ÿg(u)ÿ=ÿuÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ[Identity]

ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿAICÿÿÿÿÿÿÿÿÿÿÿÿÿ=ÿÿÿ5.359382
Logÿpseudolikelihoodÿ=ÿ-194.2971409ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿBICÿÿÿÿÿÿÿÿÿÿÿÿÿ=ÿÿÿ525.4322

ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ(Std.ÿErr.ÿadjustedÿforÿ5ÿclustersÿinÿrep78)
------------------------------------------------------------------------------
ÿÿÿÿÿÿÿÿÿÿÿÿÿ|ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿRobust
ÿÿÿÿÿÿÿÿÿmpgÿ|ÿÿÿÿÿÿCoef.ÿÿÿStd.ÿErr.ÿÿÿÿÿÿzÿÿÿÿP>|z|ÿÿÿÿÿ[95%ÿConf.ÿInterval]
-------------+----------------------------------------------------------------
ÿÿÿÿheadroomÿ|ÿÿÿ.0161296ÿÿÿ.4948527ÿÿÿÿÿ0.03ÿÿÿ0.974ÿÿÿÿ-.9537639ÿÿÿÿ.9860231
ÿÿÿÿÿÿÿtrunkÿ|ÿÿ-.0342372ÿÿÿ.0847613ÿÿÿÿ-0.40ÿÿÿ0.686ÿÿÿÿ-.2003664ÿÿÿÿ.1318919
ÿÿÿÿÿÿweightÿ|ÿÿ-.0038778ÿÿÿ.0012589ÿÿÿÿ-3.08ÿÿÿ0.002ÿÿÿÿ-.0063453ÿÿÿ-.0014104
ÿÿÿÿÿÿlengthÿ|ÿÿ-.0742607ÿÿÿ.0405048ÿÿÿÿ-1.83ÿÿÿ0.067ÿÿÿÿ-.1536485ÿÿÿÿ.0051272
ÿÿÿÿÿÿÿ_consÿ|ÿÿÿ47.38497ÿÿÿ6.952426ÿÿÿÿÿ6.82ÿÿÿ0.000ÿÿÿÿÿ33.75847ÿÿÿÿ61.01147
------------------------------------------------------------------------------

.ÿ
.ÿexit

endÿofÿdo-file


.