thank you! (the problem doesn't arise when I just use complete case analysis, but I take your point about whether or not MI makes sense given amount of missing data). ... View more

many thanks. Attached is proc MI output. Note that there are two, related analytic models and the variables in those have been included in the imputation model. I understood that to be best practice. But I wonder if I should have two imputation models, for to be used with each analytic model.  ... View more

thank you very much for the reply. I have revised the imputation model (See below) but am not sure I am understanding the best way to remove predictors. My understanding is that the imputation model should be congenial (ie consistent with ) the analytic model, i.e. imputation model should include at minimum the same variables that are in the analytic model. When I run the below model, I get the same warnings as before. If you could help me understand better how to further revise the model , I would be very grateful. Thanks.   proc mi data=siya3 nimpute=20 out=mi_fcs; class share agyw drugs depression ss worksite mobility pimp homeless moneybin2 unsafe disclose vs ed civil alcohol violrec viol; fcs plots=trace(mean std); var share agyw drugs depression ss worksite mobility pimp homeless moneybin2 unsafe disclose estigmasw estigmahiv astigmasw astigmahiv pills303 pillsreceive303 pillsgive303 vs ed civil alcohol violrec viol ; fcs discrim (vs= agyw drugs depression ss worksite mobility pimp homeless moneybin2 unsafe disclose estigmasw estigmahiv astigmasw astigmahiv pills303 pillsreceive303 pillsgive303 share ed civil alcohol violrec viol/ classeffects=include ); fcs discrim (share= agyw drugs depression ss worksite mobility pimp homeless moneybin2 unsafe disclose estigmasw estigmahiv astigmasw astigmahiv pills303 pillsreceive303 pillsgive303 vs ed civil alcohol violrec viol/ classeffects=include); fcs discrim (viol= share agyw drugs depression ss worksite mobility pimp homeless moneybin2 unsafe disclose estigmasw estigmahiv astigmasw astigmahiv pills303 pillsreceive303 pillsgive303 vs ed civil alcohol violrec / classeffects=include ); fcs discrim (violrec= share agyw drugs depression ss worksite mobility pimp homeless moneybin2 unsafe disclose estigmasw estigmahiv astigmasw astigmahiv pills303 pillsreceive303 pillsgive303 vs ed civil alcohol viol / classeffects=include ); fcs discrim (agyw= share drugs depression ss worksite mobility pimp homeless moneybin2 unsafe disclose estigmasw estigmahiv astigmasw astigmahiv pills303 pillsreceive303 pillsgive303 vs ed civil alcohol violrec viol / classeffects=include); fcs discrim (drugs = share agyw depression ss worksite mobility pimp homeless moneybin2 unsafe disclose estigmasw estigmahiv astigmasw astigmahiv pills303 pillsreceive303 pillsgive303 vs ed civil alcohol violrec viol / classeffects=include); fcs discrim (depression = share agyw drugs ss worksite mobility pimp homeless moneybin2 unsafe disclose estigmasw estigmahiv astigmasw astigmahiv pills303 pillsreceive303 pillsgive303 vs ed civil alcohol violrec viol / classeffects=include); fcs discrim ( alcohol = share agyw drugs depression ss worksite mobility pimp homeless moneybin2 unsafe disclose estigmasw estigmahiv astigmasw astigmahiv pills303 pillsreceive303 pillsgive303 vs ed civil violrec viol / classeffects=include); fcs discrim (ss = share agyw drugs depression worksite mobility pimp homeless moneybin2 unsafe disclose estigmasw estigmahiv astigmasw astigmahiv pills303 pillsreceive303 pillsgive303 vs ed civil alcohol violrec viol / classeffects=include); fcs discrim (worksite = share agyw drugs depression ss mobility pimp homeless moneybin2 unsafe disclose estigmasw estigmahiv astigmasw astigmahiv pills303 pillsreceive303 pillsgive303 vs ed civil alcohol violrec viol / classeffects=include ); fcs discrim (mobility =share agyw drugs depression ss worksite pimp homeless moneybin2 unsafe disclose estigmasw estigmahiv astigmasw astigmahiv pills303 pillsreceive303 pillsgive303 vs ed civil alcohol violrec viol / classeffects=include); fcs discrim (pimp =share agyw drugs depression ss worksite mobility homeless moneybin2 unsafe disclose estigmasw estigmahiv astigmasw astigmahiv pills303 pillsreceive303 pillsgive303 vs ed civil alcohol violrec viol/ classeffects=include ); fcs discrim (homeless =share agyw drugs depression ss worksite mobility pimp moneybin2 unsafe disclose estigmasw estigmahiv astigmasw astigmahiv pills303 pillsreceive303 pillsgive303 vs ed civil alcohol violrec viol / classeffects=include); fcs discrim (moneybin2=share agyw drugs depression ss worksite mobility pimp homeless unsafe disclose estigmasw estigmahiv astigmasw astigmahiv pills303 pillsreceive303 pillsgive303 vs ed civil alcohol violrec viol / classeffects=include ); fcs regpmm (pills303 =share agyw drugs depression ss worksite mobility pimp homeless moneybin2 unsafe disclose estigmasw estigmahiv astigmasw astigmahiv pillsreceive303 pillsgive303 vs ed civil alcohol violrec viol ); fcs regpmm (pillsreceive303 =share agyw drugs depression ss worksite mobility pimp homeless moneybin2 unsafe disclose estigmasw estigmahiv astigmasw astigmahiv pills303 pillsgive303 vs ed civil alcohol violrec viol ); fcs regpmm (pillsgive303 =share agyw drugs depressionss worksite mobility pimp homeless moneybin2 unsafe disclose estigmasw estigmahiv astigmasw astigmahiv pills303 pillsreceive303 vs ed civil alcohol violrec viol ); fcs regpmm ( estigmasw = pillsgive303 share agyw drugs depression ss worksite mobility pimp homeless moneybin2 unsafe disclose estigmahiv astigmasw astigmahiv pills303 pillsreceive303 vs ed civil alcohol violrec viol ); fcs regpmm ( estigmahiv = pillsgive303 share agyw drugs depression ss worksite mobility pimp homeless moneybin2 unsafe disclose estigmasw astigmasw astigmahiv pills303 pillsreceive303 vs ed civil alcohol violrec viol ); fcs regpmm ( astigmasw = pillsgive303 share agyw drugs depression ss worksite mobility pimp homeless moneybin2 unsafe disclose estigmasw estigmahiv astigmahiv pills303 pillsreceive303 vs ed civil alcohol violrec viol ); fcs regpmm ( astigmahiv = pillsgive303 share agyw drugs depression ss worksite mobility pimp homeless moneybin2 unsafe disclose estigmasw estigmahiv astigmasw pills303 pillsreceive303 vs ed civil alcohol violrec viol ); fcs discrim (disclose =share agyw drugs depression ss worksite mobility pimp homeless moneybin2 unsafe estigmasw estigmahiv astigmasw astigmahiv pills303 pillsreceive303 pillsgive303 vs ed civil alcohol violrec viol / classeffects=include); fcs discrim ( unsafe=share agyw drugs depression disclose ss worksite mobility pimp homeless moneybin2 estigmasw estigmahiv astigmasw astigmahiv pills303 pillsreceive303 pillsgive303 vs ed civil alcohol violrec viol / classeffects=include); fcs discrim ( ed=share agyw unsafe drugs depression disclose ss worksite mobility pimp homeless moneybin2 estigmasw estigmahiv astigmasw astigmahiv pills303 pillsreceive303 pillsgive303 vs civil alcohol violrec viol / classeffects=include); fcs discrim (civil=share agyw unsafe drugs depression disclose ss worksite mobility pimp homeless moneybin2 estigmasw estigmahiv astigmasw astigmahiv pills303 pillsreceive303 pillsgive303 vs alcohol violrec viol ed / classeffects=include); run;   ... View more

Hello again,  When I run PROC MI I get a few warnings, and am wondering what, if anything, I need to do. Here are the code and the warnings:   proc mi data=siya3 nimpute=20 out=mi_fcs; class share agyw drugs depression ss worksite mobility  pimp homeless moneybin2 unsafe disclose vs ed civil alcohol violrec viol; fcs plots=trace(mean std); var share agyw drugs depression ss worksite mobility  pimp homeless moneybin2 unsafe disclose  estigmasw estigmahiv astigmasw astigmahiv  pills303 pillsreceive303 pillsgive303 vs ed civil alcohol violrec viol ; fcs discrim(share agyw drugs depression ss worksite mobility  pimp homeless moneybin2 unsafe disclose vs ed civil alcohol violrec viol / classeffects=include) nbiter=100; run; WARNING: The regression with observed observations results in a perfect fit for variable          pillsgive303. Imputed values are the predicted values from the model. WARNING: An effect for variable estigmasw is a linear combination of other effects. The coefficient          of the effect will be set to zero in the imputation. WARNING: An effect for variable estigmahiv is a linear combination of other effects. The coefficient          of the effect will be set to zero in the imputation. WARNING: An effect for variable astigmasw is a linear combination of other effects. The coefficient          of the effect will be set to zero in the imputation. WARNING: An effect for variable astigmahiv is a linear combination of other effects. The coefficient          of the effect will be set to zero in the imputation. WARNING: The regression with observed observations results in a perfect fit for variable pills303.          Imputed values are the predicted values from the model. WARNING: The regression with observed observations results in a perfect fit for variable          pillsreceive303. Imputed values are the predicted values from the model.   Note that  estigmasw estigmahiv astigmasw astigmahiv, which were included in the analysis model and treated as continuous, are sum scores created by adding other variables together. Pillsgive303, pillsreceive303, and pills303 are also variables included in the analytic model that were created by summing other variables and treated as continuous. Do the warnings mean that I need to include the all the variables used to create those continuous variables in the var statement? If so, should they also be included in the class and fcs discrim statments if they are categorical, even if they themselves were not included in the analytic model? many thanks. ... View more

that did it. Hopefully that is all from me. Many many thanks for the assistance! ... View more

Got it, that worked! thanks. I am now trying to get multiply imputed results for a different genmod model (same imputation model, dataset etc) where I am producing contrast estimates for several variables, this time including several categorical ones, and am getting error messages. Thanks in advance for your help. Here is the code and the error message I get when I run proc mianalyze: proc genmod data=mi_fcs; class pid ss (ref="0") civil (ref="0") depression (ref="0") alcohol (ref="0") ed (ref="0") mobility (ref="0") money (ref="0") unsafe (ref="0") /param=ref; model share= ed civil alcohol drugs depression ss mobility pimp money unsafe viol estigmasw estigmahiv astigmasw/Dist=poisson Link=log; Repeated subject = pid / type=Ind; by _imputation_; estimate 'Incomplete secondary vs. any primary or none' ed 1 0 0 /exp; estimate 'Complete secondary vs. any primary or none' ed 0 1 0 /exp; estimate 'Incomplete or complete university or technical vs. any primary or none' ed 0 0 1/exp; estimate 'Partner, cohab vs single' civil 0 1/exp; estimate 'Partner, non-cohab vs single' civil 1 0/exp; estimate 'Alcohol freq =<monthly vs never' alcohol 1 0 0 0/exp; estimate 'Alcohol freq 2-4x/mo vs never' alcohol 0 1 0 0/exp; estimate 'Alcohol freq 2-3x/wk vs never' alcohol 0 0 1 0/exp; estimate 'Alcohol freq =>4x/wk vs never' alcohol 0 0 0 1/exp; estimate 'Drug use <30 days: yes vs. no' drugs 1/exp; estimate 'Depression: mild vs none/minimal' depression 1 0 0 0/exp; estimate 'Depression: moderate vs none/minimal' depression 0 1 0 0/exp; estimate 'Depression: moderately severe vs none/minimal' depression 0 0 1 0/exp; estimate 'Depression: severe vs none/minimal' depression 0 0 0 1/exp; estimate 'Social support: moderate vs low' ss 1 0/exp; estimate 'Social support: high vs low' ss 0 1/exp; estimate 'Mobility: nights spent outside Durban <6 mos.' mobility 1 /exp; estimate 'Share earnings with pimp or other: yes vs. no' pimp 1/exp; Estimate 'Enough money for needs: A little vs not at all' money 1 0 0 /exp; Estimate 'Enough money for needs: Moderately vs not at all' money 0 1 0/exp; Estimate 'Enough money for needs: Mostly or completely vs not at all' money 0 0 1/exp; Estimate 'safe in daily life: Moderately vs Very much/extremely' unsafe 1 0 0 /exp; Estimate 'safe in daily life: A little vs Very much/extremely' unsafe 0 1 0/exp; Estimate 'safe in daily life: Not at all vs Very much/extremely' unsafe 0 0 1/exp; estimate 'Physical and/or sexual violence ever: yes vs no' viol 1/exp; estimate 'Enacted SW stigma level' estigmasw 1/exp; estimate 'Enacted HIV stigma level' estigmaHIV 1/exp; estimate 'Anticipated SW stigma level' astigmasw 1/exp; ods output estimates=estimate_ds(where=(index(label,'Exp')=1)); title 'Sharing (Ever)correlates - robust poisson- MI'; run; proc sort data=estimate_ds ; by _imputation_; run; proc mianalyze data=estimate_ds THETA0=1 ; by label; modeleffects LBetaEstimate; stderr stderr; run;   ERROR: Data set WORK.ESTIMATE_DS is not sorted in ascending sequence. The current BY group has Label        = Exp(Incomplete secondary vs. any primary or none) and the next BY group has Label =        Exp(Complete secondary vs. any primary or none). ... View more

got it, thanks. I ran it, but the results don't seem right to me unless I am reading them wrong. I am assuming that the estimate under "lbetaestimate" is not the beta but the pooled prevalence ratio (contrast estimate). That estimate and CI seem fine but the p-value (<.0001) seems off-- see page 31 of attachment. For all 5 imputations, the p-value is around .2 (see pages 16-30). any thoughts on what could be going wrong here? Many thanks.     ... View more

thank you. is the 'data test' code necessary in all cases, or can I just use proc genmod on?  I don't understand what is happening with the 'data test' code.  ... View more

Hi - any thoughts on how to get the pooled contrast estimates from proc mianalyze? thanks so much in advance, hopefully this is the last question! ... View more

thank you for pointing out the mistakes. I fixed those and it ran fine. But: the proc mianalyze results still don't give me the contrast estimate (prevalence ratio) that I need. I just get the beta (lbetaestimate). Is there a way to get that? ... View more

Thank you very much. I tried that code (below) and got the following error messages:   ERROR: Invalid reference value for ed. ERROR: No valid observations due to invalid or missing values in the response, explanatory, offset,        frequency, or weight variable. NOTE: The above message was for the following BY group:       pills303=0 Imputation Number=1 ERROR: Data set WORK.MI_FCS is not sorted in ascending sequence. The current BY group has pills303 =        3.3333333333 and the next BY group has pills303 = 0. NOTE: The SAS System stopped processing this step because of errors. WARNING: Output 'estimates' was not created.  Make sure that the output object name, label, or path          is spelled correctly.  Also, verify that the appropriate procedure options are used to          produce the requested output object.  For example, verify that the NOPRINT option is not          used.     I do not know what could be wrong with "ed" -- it has four categories, 0-3. "Pills303" is a continuous various (mean=0.63, range: 0-11.3)     Here is the code I ran per your guidance:  proc genmod data=mi_fcs; class pid civil (ref="0") depression (ref="0") alcohol (ref="0") ed (ref="0") mobility (ref="0") unsafe (ref="0") money (ref="0") worksite (ref="0") /param=ref; model vs= pills303 agyw ed civil alcohol drugs depression mobility worksite pimp violrec homeless money unsafe estigmasw estigmahiv astigmasw astigmahiv disclose/Dist=poisson Link=log; Repeated subject = pid / type=Ind ; estimate 'pills303' pills303 1 -1 0; by pills303 _imputation_; ods output estimates=estimate_ds; title 'VS on pills given and/or received with MI'; run;   Your help sorting this out would be greatly appreciated. ... View more

Thank you so much, that worked! But I am not getting the contrast estimate results I need when I run that PROC MIANALYZE code. do I include the estimate statement in it? Please advise, many thanks again. ... View more

Thanks, here is the code for the imputation process overall. It is when I run the code under /*3. pooling phase*/, highlighted in blue, that sas gets hung up, i.e. keeps running and never completes the procedure:   /*1. imputation phase*/ proc mi data=siya nimpute=20 out=mi_fcs; class share give receive agyw drugs depression ss worksite mobility pimp homeless money unsafe disclose vs ed civil alcohol violrec viol; fcs plots=trace(mean std); var share give receive agyw drugs depression ss worksite mobility pimp homeless money unsafe disclose estigmasw estigmahiv astigmasw astigmahiv pills303 pillsreceive303 pillsgive303 vs ed civil alcohol violrec viol ; fcs discrim(share give receive agyw drugs depression ss worksite mobility pimp homeless money unsafe disclose vs ed civil alcohol violrec viol / classeffects=include) nbiter=100; run; /*2. analysis and pooling*/ proc genmod data=mi_fcs; class pid civil (ref="0") depression (ref="0") alcohol (ref="0") ed (ref="0") mobility (ref="0") unsafe (ref="0") money (ref="0") worksite (ref="0") /param=ref; model vs= pills303 agyw ed civil alcohol drugs depression mobility worksite pimp violrec homeless money unsafe estigmasw estigmahiv astigmasw astigmahiv disclose/Dist=poisson Link=log; Repeated subject = pid / type=Ind PRINTMLE; estimate 'pills303' pills303 1; by _imputation_; ods output ParameterEstimates=gm_fcs; title 'VS on pills given and/or received with MI'; run;   /*3. pooling phase*/ TITLE " MULTIPLE IMPUTATION Linear REGRESSION - FCS"; PROC MIANALYZE parms(classvar=level)=gm_fcs; class share give receive agyw drugs depression ss worksite mobility pimp homeless money unsafe disclose vs ed civil alcohol violrec viol; MODELEFFECTS INTERCEPT share give receive agyw drugs depression ss worksite mobility pimp homeless money unsafe disclose estigmasw estigmahiv astigmasw astigmahiv pills303 pillsreceive303 pillsgive303 vs ed civil alcohol violrec viol;/*all variables from all models*/ RUN;   ... View more

thank you so much. That enabled me to complete stage 2 of multiple imputation, but now I cannot get through the third and final pooling stage because when I run the code (see "/*3. pooling phase*/" in original post), SAS gets hung up running and never competes the procedure. Any ideas of how to fix this? Thanks! There are just 890 observations in the dataset.  ... View more

having the same problem. did you figure it out? ... View more