Margins after pkcross

[Dimi Arv]

Hi all,

I am trying to verify results from SAS regarding a 2X2 bio equivalence study. I ran pkcross in Stata 16 v.1 and got the same results for the model parameters which are sequence, subject, treatment and period. Then, I have tried to ran margins and got the following error message:

. margins treatment
variable c1 not found
r(111);

Is it possible to run margins after pkcross ?

Thank you very much in advance for your help.

Best regards,

Dimi

[Joseph Coveney]

Is it possible to run margins after pkcross ?

Not directly, no.

You'll notice that the command's help file doesn't mention any postestimation commands, such as margins. That's because pkcross creates a couple of temporary variables (I think for display purposes) and then calls anova using them. After the pkcross completes, these temporary variables vanish and make the usual postestimation commands impossible.

Nevertheless, you can fit the same ANOVA model manually, and then margins becomes accessible. See below.

.ÿ
.ÿversionÿ16.1

.ÿ
.ÿclearÿ*

.ÿ
.ÿwebuseÿpkdata
(FictionalÿDrugÿConcentrationÿData)

.ÿpkcollapseÿtimeÿconc1ÿconc2,ÿid(id)ÿkeep(seq)ÿstat(auc)
................................
.ÿpkshapeÿidÿseqÿauc*,ÿorder(RTÿTR)

.ÿpkcrossÿoutcome
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿsequenceÿvariableÿ=ÿsequence
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿperiodÿvariableÿ=ÿperiod
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿtreatmentÿvariableÿ=ÿtreat
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿcarryoverÿvariableÿ=ÿcarry
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿidÿvariableÿ=ÿid

ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿAnalysisÿofÿvarianceÿ(ANOVA)ÿforÿaÿ2x2ÿcrossoverÿstudy
ÿÿÿÿÿSourceÿofÿVariationÿÿ|ÿPartialÿSSÿÿÿdfÿÿÿÿÿÿÿÿMSÿÿÿÿÿÿÿÿÿÿFÿÿÿÿÿProbÿ>ÿFÿ
ÿÿÿÿ----------------------+--------------------------------------------------
ÿÿÿÿÿIntersubjectsÿÿÿÿÿÿÿÿ|
ÿÿÿÿÿÿÿÿÿÿSequenceÿeffectÿ|ÿÿÿÿÿ378.04ÿÿÿÿ1ÿÿÿÿÿÿ378.04ÿÿÿÿÿÿ0.29ÿÿÿÿÿÿ0.5961
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿResidualsÿ|ÿÿÿ17991.26ÿÿÿ14ÿÿÿÿÿ1285.09ÿÿÿÿÿÿ1.40ÿÿÿÿÿÿ0.2691
ÿÿÿÿ----------------------+--------------------------------------------------
ÿÿÿÿÿIntrasubjectsÿÿÿÿÿÿÿÿ|
ÿÿÿÿÿÿÿÿÿTreatmentÿeffectÿ|ÿÿÿÿÿ455.04ÿÿÿÿ1ÿÿÿÿÿÿ455.04ÿÿÿÿÿÿ0.50ÿÿÿÿÿÿ0.4931
ÿÿÿÿÿÿÿÿÿÿÿÿPeriodÿeffectÿ|ÿÿÿÿÿ419.47ÿÿÿÿ1ÿÿÿÿÿÿ419.47ÿÿÿÿÿÿ0.46ÿÿÿÿÿÿ0.5102
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿResidualsÿ|ÿÿÿ12860.78ÿÿÿ14ÿÿÿÿÿÿ918.63
ÿÿÿÿ----------------------+--------------------------------------------------
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿTotalÿ|ÿÿÿ32104.59ÿÿÿ31

ÿÿÿÿOmnibusÿmeasureÿofÿseparabilityÿofÿtreatmentÿandÿcarryoverÿ=ÿÿÿ29.2893%

.ÿ
.ÿdisplayÿinÿsmclÿasÿtextÿe(cmdline)
anovaÿoutcomeÿ__000001/id|__000001ÿtreatÿperiodÿifÿ__000003,ÿ

.ÿ
.ÿ*
.ÿ*ÿBeginÿhere
.ÿ*
.ÿanovaÿoutcomeÿsequenceÿ/ÿid|sequenceÿtreatÿperiod

ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿNumberÿofÿobsÿ=ÿÿÿÿÿÿÿÿÿ32ÿÿÿÿR-squaredÿÿÿÿÿ=ÿÿ0.5994
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿRootÿMSEÿÿÿÿÿÿ=ÿÿÿÿ30.3089ÿÿÿÿAdjÿR-squaredÿ=ÿÿ0.1130

ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿSourceÿ|ÿPartialÿSSÿÿÿÿÿÿÿÿÿdfÿÿÿÿÿÿÿÿÿMSÿÿÿÿÿÿÿÿFÿÿÿÿProb>F
ÿÿÿÿÿÿÿÿÿÿÿÿÿ------------+----------------------------------------------------
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿModelÿ|ÿÿ19243.812ÿÿÿÿÿÿÿÿÿ17ÿÿÿ1131.9889ÿÿÿÿÿÿ1.23ÿÿ0.3504
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ|
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿsequenceÿ|ÿÿ378.04323ÿÿÿÿÿÿÿÿÿÿ1ÿÿÿ378.04323ÿÿÿÿÿÿ0.29ÿÿ0.5961
ÿÿÿÿÿÿÿÿÿÿÿÿÿid|sequenceÿ|ÿÿ17991.264ÿÿÿÿÿÿÿÿÿ14ÿÿÿ1285.0903ÿÿ
ÿÿÿÿÿÿÿÿÿÿÿÿÿ------------+----------------------------------------------------
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿtreatÿ|ÿÿ455.03758ÿÿÿÿÿÿÿÿÿÿ1ÿÿÿ455.03758ÿÿÿÿÿÿ0.50ÿÿ0.4931
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿperiodÿ|ÿÿ419.46755ÿÿÿÿÿÿÿÿÿÿ1ÿÿÿ419.46755ÿÿÿÿÿÿ0.46ÿÿ0.5102
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ|
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿResidualÿ|ÿÿ12860.782ÿÿÿÿÿÿÿÿÿ14ÿÿÿ918.62725ÿÿ
ÿÿÿÿÿÿÿÿÿÿÿÿÿ------------+----------------------------------------------------
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿTotalÿ|ÿÿ32104.594ÿÿÿÿÿÿÿÿÿ31ÿÿÿÿ1035.632ÿÿ

.ÿmarginsÿtreat

PredictiveÿmarginsÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿNumberÿofÿobsÿÿÿÿÿ=ÿÿÿÿÿÿÿÿÿ32

Expressionÿÿÿ:ÿLinearÿprediction,ÿpredict()

------------------------------------------------------------------------------
ÿÿÿÿÿÿÿÿÿÿÿÿÿ|ÿÿÿÿÿÿÿÿÿÿÿÿDelta-method
ÿÿÿÿÿÿÿÿÿÿÿÿÿ|ÿÿÿÿÿMarginÿÿÿStd.ÿErr.ÿÿÿÿÿÿtÿÿÿÿP>|t|ÿÿÿÿÿ[95%ÿConf.ÿInterval]
-------------+----------------------------------------------------------------
ÿÿÿÿÿÿÿtreatÿ|
ÿÿÿÿÿÿÿÿÿÿRÿÿ|ÿÿÿ151.4779ÿÿÿ7.577216ÿÿÿÿ19.99ÿÿÿ0.000ÿÿÿÿÿ135.2264ÿÿÿÿ167.7294
ÿÿÿÿÿÿÿÿÿÿTÿÿ|ÿÿÿ159.0197ÿÿÿ7.577216ÿÿÿÿ20.99ÿÿÿ0.000ÿÿÿÿÿ142.7682ÿÿÿÿ175.2713
------------------------------------------------------------------------------

.ÿ
.ÿexit

endÿofÿdo-file


.

[Dimi Arv]

Hi Joseph,

Thank you very much for your prompt response. That was very helpful.
Have a nice weekend,

Dimi

[Dimi Arv]

Hi Joseph,

Please, one more clarification. By using the command:

. margins treat

without any other option such as asbalanced, the results I will get are the least squares means (lsmeans in SAS). I am right?

Thanks a lot in advance,

Dimi