Best method for doing Mixed ANOVA in STATA?

[butler bj]

Hello everyone, I'm trying to analyze this 2x2 factorial design I have, where one factor is between-subjects and the other is within-subjects. In SPSS it's straight forward, just do a repeated measures anova and add your between subjects factor. But I'm trying to learn STATA, so here I am. Should I try to do this using the split-plot ANOVA commands, or should use a mixed model using the mixed command?

Thank you in advance

[Joseph Coveney]

Unless I misunderstand you, your study sounds like a conventional randomized complete blocks design.

anova response treatment block

anova does have a repeated() option, but it isn't necessary here, because you have only two levels of your within-subjects factor. You wouldn't need to resort to mixed unless you have an unbalanced dataset.

[Joseph Coveney]

On further thought, it seems that I misunderstood. You can set up a split-plot ANOVA in Stata easily enough (see below) as shown in the example in the help file for anova, and illustrated at length at the UCLA ATS site.

.ÿversionÿ14.1

.ÿ
.ÿclearÿ*

.ÿsetÿmoreÿoff

.ÿsetÿseedÿ`=date("2015-11-06",ÿ"YMD")'

.ÿquietlyÿsetÿobsÿ24

.ÿgenerateÿdoubleÿuÿ=ÿrnormal()

.ÿgenerateÿbyteÿsubjectÿ=ÿ_n

.ÿgenerateÿbyteÿtreatmentÿ=ÿmod(_n,ÿ2)

.ÿforvaluesÿtimeÿ=ÿ1/2ÿ{
ÿÿ2.ÿÿÿÿÿÿÿÿÿgenerateÿdoubleÿresponse`time'ÿ=ÿuÿ+ÿ`time'ÿ*ÿ`time'ÿ*ÿrnormal()
ÿÿ3.ÿ}

.ÿpreserve

.ÿquietlyÿreshapeÿlongÿresponse,ÿi(subject)ÿj(time)

.ÿ
.ÿ*
.ÿ*ÿSplit-plotÿANOVA
.ÿ*
.ÿanovaÿresponseÿtreatmentÿ/ÿsubject|treatmentÿtimeÿtreatment#time,ÿrepeated(time)

ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿNumberÿofÿobsÿ=ÿÿÿÿÿÿÿÿÿ48ÿÿÿÿR-squaredÿÿÿÿÿ=ÿÿ0.5841
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿRootÿMSEÿÿÿÿÿÿ=ÿÿÿÿ2.59036ÿÿÿÿAdjÿR-squaredÿ=ÿÿ0.1115

ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿSourceÿ|ÿPartialÿSSÿÿÿÿÿÿÿÿÿdfÿÿÿÿÿÿÿÿÿMSÿÿÿÿÿÿÿÿFÿÿÿÿProb>F
ÿÿÿÿÿÿÿ------------------+----------------------------------------------------
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿModelÿ|ÿÿÿ207.3088ÿÿÿÿÿÿÿÿÿ25ÿÿÿÿ8.292352ÿÿÿÿÿÿ1.24ÿÿ0.3098
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ|
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿtreatmentÿ|ÿÿ9.6980773ÿÿÿÿÿÿÿÿÿÿ1ÿÿÿ9.6980773ÿÿÿÿÿÿ1.12ÿÿ0.3005
ÿÿÿÿÿÿÿsubject|treatmentÿ|ÿÿÿ189.7406ÿÿÿÿÿÿÿÿÿ22ÿÿÿ8.6245729ÿÿ
ÿÿÿÿÿÿÿ------------------+----------------------------------------------------
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿtimeÿ|ÿÿ3.1235067ÿÿÿÿÿÿÿÿÿÿ1ÿÿÿ3.1235067ÿÿÿÿÿÿ0.47ÿÿ0.5022
ÿÿÿÿÿÿÿÿÿÿtreatment#timeÿ|ÿÿ4.7466128ÿÿÿÿÿÿÿÿÿÿ1ÿÿÿ4.7466128ÿÿÿÿÿÿ0.71ÿÿ0.4094
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ|
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿResidualÿ|ÿÿ147.61956ÿÿÿÿÿÿÿÿÿ22ÿÿÿ6.7099801ÿÿ
ÿÿÿÿÿÿÿ------------------+----------------------------------------------------
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿTotalÿ|ÿÿ354.92836ÿÿÿÿÿÿÿÿÿ47ÿÿÿ7.5516673ÿÿ


Between-subjectsÿerrorÿterm:ÿÿsubject|treatment
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿLevels:ÿÿ24ÿÿÿÿÿÿÿÿ(22ÿdf)
ÿÿÿÿÿLowestÿb.s.e.ÿvariable:ÿÿsubject
ÿÿÿÿÿCovarianceÿpooledÿover:ÿÿtreatmentÿÿ(forÿrepeatedÿvariable)

Repeatedÿvariable:ÿtime
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿHuynh-Feldtÿepsilonÿÿÿÿÿÿÿÿ=ÿÿ1.0476
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ*Huynh-Feldtÿepsilonÿresetÿtoÿ1.0000
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿGreenhouse-Geisserÿepsilonÿ=ÿÿ1.0000
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿBox'sÿconservativeÿepsilonÿ=ÿÿ1.0000

ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ------------ÿProbÿ>ÿFÿ------------
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿSourceÿ|ÿÿÿÿÿdfÿÿÿÿÿÿFÿÿÿÿRegularÿÿÿÿH-FÿÿÿÿÿÿG-GÿÿÿÿÿÿBox
ÿÿÿÿÿÿÿ------------------+----------------------------------------------------
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿtimeÿ|ÿÿÿÿÿÿ1ÿÿÿÿÿ0.47ÿÿÿ0.5022ÿÿÿ0.5022ÿÿÿ0.5022ÿÿÿ0.5022
ÿÿÿÿÿÿÿÿÿÿtreatment#timeÿ|ÿÿÿÿÿÿ1ÿÿÿÿÿ0.71ÿÿÿ0.4094ÿÿÿ0.4094ÿÿÿ0.4094ÿÿÿ0.4094
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿResidualÿ|ÿÿÿÿÿ22
ÿÿÿÿÿÿÿ-----------------------------------------------------------------------

.ÿ
.ÿ*
.ÿ*ÿMANOVA
.ÿ*
.ÿrestore

.ÿmanovaÿresponse1ÿresponse2ÿ=ÿtreatment

ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿNumberÿofÿobsÿ=ÿÿÿÿÿÿÿÿÿ24

ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿWÿ=ÿWilks'ÿlambdaÿÿÿÿÿÿLÿ=ÿLawley-Hotellingÿtrace
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿPÿ=ÿPillai'sÿtraceÿÿÿÿÿRÿ=ÿRoy'sÿlargestÿroot

ÿÿÿÿÿÿÿÿÿÿÿÿÿÿSourceÿ|ÿStatisticÿÿÿÿÿÿÿÿdfÿÿÿÿF(df1,ÿÿÿÿÿdf2)ÿ=ÿÿÿFÿÿÿProb>F
ÿÿÿÿÿÿÿÿÿÿ-----------+-------------------------------------------------------
ÿÿÿÿÿÿÿÿÿÿÿtreatmentÿ|Wÿÿÿ0.9503ÿÿÿÿÿÿÿÿÿ1ÿÿÿÿÿÿ2.0ÿÿÿÿÿ21.0ÿÿÿÿÿ0.55ÿ0.5854ÿe
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ|Pÿÿÿ0.0497ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ2.0ÿÿÿÿÿ21.0ÿÿÿÿÿ0.55ÿ0.5854ÿe
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ|Lÿÿÿ0.0523ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ2.0ÿÿÿÿÿ21.0ÿÿÿÿÿ0.55ÿ0.5854ÿe
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ|Rÿÿÿ0.0523ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ2.0ÿÿÿÿÿ21.0ÿÿÿÿÿ0.55ÿ0.5854ÿe
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ|-------------------------------------------------------
ÿÿÿÿÿÿÿÿÿÿÿÿResidualÿ|ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ22
ÿÿÿÿÿÿÿÿÿÿ-----------+-------------------------------------------------------
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿTotalÿ|ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ23
ÿÿÿÿÿÿÿÿÿÿ-------------------------------------------------------------------
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿeÿ=ÿexact,ÿaÿ=ÿapproximate,ÿuÿ=ÿupperÿboundÿonÿF

.ÿ
.ÿ//Treatmentÿ×ÿTimeÿInteraction
.ÿmatrixÿinputÿYÿ=ÿ(-1ÿ1)

.ÿmanovatestÿtreatment,ÿytransform(Y)

ÿTransformationÿofÿtheÿdependentÿvariables
ÿ(1)ÿÿÿÿ-ÿresponse1ÿ+ÿresponse2

ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿWÿ=ÿWilks'ÿlambdaÿÿÿÿÿÿLÿ=ÿLawley-Hotellingÿtrace
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿPÿ=ÿPillai'sÿtraceÿÿÿÿÿRÿ=ÿRoy'sÿlargestÿroot

ÿÿÿÿÿÿÿÿÿÿÿÿÿÿSourceÿ|ÿStatisticÿÿÿÿÿÿÿÿdfÿÿÿÿF(df1,ÿÿÿÿÿdf2)ÿ=ÿÿÿFÿÿÿProb>F
ÿÿÿÿÿÿÿÿÿÿ-----------+-------------------------------------------------------
ÿÿÿÿÿÿÿÿÿÿÿtreatmentÿ|Wÿÿÿ0.9688ÿÿÿÿÿÿÿÿÿ1ÿÿÿÿÿÿ1.0ÿÿÿÿÿ22.0ÿÿÿÿÿ0.71ÿ0.4094ÿe
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ|Pÿÿÿ0.0312ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ1.0ÿÿÿÿÿ22.0ÿÿÿÿÿ0.71ÿ0.4094ÿe
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ|Lÿÿÿ0.0322ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ1.0ÿÿÿÿÿ22.0ÿÿÿÿÿ0.71ÿ0.4094ÿe
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ|Rÿÿÿ0.0322ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ1.0ÿÿÿÿÿ22.0ÿÿÿÿÿ0.71ÿ0.4094ÿe
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ|-------------------------------------------------------
ÿÿÿÿÿÿÿÿÿÿÿÿResidualÿ|ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ22
ÿÿÿÿÿÿÿÿÿÿ-------------------------------------------------------------------
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿeÿ=ÿexact,ÿaÿ=ÿapproximate,ÿuÿ=ÿupperÿboundÿonÿF

.ÿ
.ÿ//ÿMainÿEffectsÿofÿTime
.ÿmatrixÿinputÿTÿ=ÿ(0.5ÿ0.5ÿ1)

.ÿmanovatestÿ,ÿtest(T)ÿytransform(Y)

ÿTransformationÿofÿtheÿdependentÿvariables
ÿ(1)ÿÿÿÿ-ÿresponse1ÿ+ÿresponse2

ÿTestÿconstraint
ÿ(1)ÿÿÿÿ.5*0.treatmentÿ+ÿ.5*1.treatmentÿ+ÿ_consÿ=ÿ0

ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿWÿ=ÿWilks'ÿlambdaÿÿÿÿÿÿLÿ=ÿLawley-Hotellingÿtrace
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿPÿ=ÿPillai'sÿtraceÿÿÿÿÿRÿ=ÿRoy'sÿlargestÿroot

ÿÿÿÿÿÿÿÿÿÿÿÿÿÿSourceÿ|ÿStatisticÿÿÿÿÿÿÿÿdfÿÿÿÿF(df1,ÿÿÿÿÿdf2)ÿ=ÿÿÿFÿÿÿProb>F
ÿÿÿÿÿÿÿÿÿÿ-----------+-------------------------------------------------------
ÿÿÿÿÿÿÿÿÿÿmanovatestÿ|Wÿÿÿ0.9793ÿÿÿÿÿÿÿÿÿ1ÿÿÿÿÿÿ1.0ÿÿÿÿÿ22.0ÿÿÿÿÿ0.47ÿ0.5022ÿe
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ|Pÿÿÿ0.0207ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ1.0ÿÿÿÿÿ22.0ÿÿÿÿÿ0.47ÿ0.5022ÿe
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ|Lÿÿÿ0.0212ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ1.0ÿÿÿÿÿ22.0ÿÿÿÿÿ0.47ÿ0.5022ÿe
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ|Rÿÿÿ0.0212ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ1.0ÿÿÿÿÿ22.0ÿÿÿÿÿ0.47ÿ0.5022ÿe
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ|-------------------------------------------------------
ÿÿÿÿÿÿÿÿÿÿÿÿResidualÿ|ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ22
ÿÿÿÿÿÿÿÿÿÿ-------------------------------------------------------------------
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿeÿ=ÿexact,ÿaÿ=ÿapproximate,ÿuÿ=ÿupperÿboundÿonÿF

.ÿ
.ÿ//ÿMainÿEffectsÿofÿTreatment
.ÿmatrixÿdefineÿYÿ=ÿJ(1,ÿ2,ÿ1)

.ÿmanovatestÿtreatment,ÿytransform(Y)

ÿTransformationÿofÿtheÿdependentÿvariables
ÿ(1)ÿÿÿÿresponse1ÿ+ÿresponse2

ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿWÿ=ÿWilks'ÿlambdaÿÿÿÿÿÿLÿ=ÿLawley-Hotellingÿtrace
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿPÿ=ÿPillai'sÿtraceÿÿÿÿÿRÿ=ÿRoy'sÿlargestÿroot

ÿÿÿÿÿÿÿÿÿÿÿÿÿÿSourceÿ|ÿStatisticÿÿÿÿÿÿÿÿdfÿÿÿÿF(df1,ÿÿÿÿÿdf2)ÿ=ÿÿÿFÿÿÿProb>F
ÿÿÿÿÿÿÿÿÿÿ-----------+-------------------------------------------------------
ÿÿÿÿÿÿÿÿÿÿÿtreatmentÿ|Wÿÿÿ0.9514ÿÿÿÿÿÿÿÿÿ1ÿÿÿÿÿÿ1.0ÿÿÿÿÿ22.0ÿÿÿÿÿ1.12ÿ0.3005ÿe
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ|Pÿÿÿ0.0486ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ1.0ÿÿÿÿÿ22.0ÿÿÿÿÿ1.12ÿ0.3005ÿe
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ|Lÿÿÿ0.0511ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ1.0ÿÿÿÿÿ22.0ÿÿÿÿÿ1.12ÿ0.3005ÿe
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ|Rÿÿÿ0.0511ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ1.0ÿÿÿÿÿ22.0ÿÿÿÿÿ1.12ÿ0.3005ÿe
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ|-------------------------------------------------------
ÿÿÿÿÿÿÿÿÿÿÿÿResidualÿ|ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿ22
ÿÿÿÿÿÿÿÿÿÿ-------------------------------------------------------------------
ÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿÿeÿ=ÿexact,ÿaÿ=ÿapproximate,ÿuÿ=ÿupperÿboundÿonÿF

.ÿ
.ÿexit

endÿofÿdo-file


.

[butler bj]

On further thought, it seems that I misunderstood. You can set up a split-plot ANOVA in Stata easily enough (see below) as shown in the example in the help file for anova, and illustrated at length

Awesome, thank you for your help!