https://communities.sas.com/t5/Statistical-Procedures/proc-freq-with-multiple-imputation/m-p/722596#M35013 <P>Yes.&nbsp; You would have to combine the actual Chi-Square statistics from each of the tables.&nbsp; Dr. Paul Allison has a macro on his website that will compute the combined Chi-Square statistics.</P> <P><A href="https://www.sas.upenn.edu/~allison/combchi.sas" target="_blank">https://www.sas.upenn.edu/~allison/combchi.sas</A>&nbsp;</P> Mon, 01 Mar 2021 15:07:21 GMT SAS_Rob 2021-03-01T15:07:21Z https://communities.sas.com/t5/Statistical-Procedures/proc-freq-with-multiple-imputation/m-p/671680#M32106 <P>Hi,</P><P>&nbsp;</P><P>I am working with a multiple imputed dataset and I want to make a frequency table of the responders. With the statement I use, I get frequency tables for every imputed dataset separately, but how do I combine these results in 1 final result using proc mi analyze?&nbsp;</P><P>&nbsp;</P><P>proc freq data = test;<BR />tables responders*intervention ;<BR />BY _Imputation_;</P><P>RUN;</P><P>&nbsp;</P><P>Thank you for the help.&nbsp;<BR /><BR /></P> Thu, 23 Jul 2020 06:44:36 GMT https://communities.sas.com/t5/Statistical-Procedures/proc-freq-with-multiple-imputation/m-p/671680#M32106 ablond 2020-07-23T06:44:36Z https://communities.sas.com/t5/Statistical-Procedures/proc-freq-with-multiple-imputation/m-p/671767#M32112 <P>Because Proc FREQ does not report a standard error for the frequency, you would not be able to combine the estimates in Proc MIANALYZE.&nbsp; Instead you must use Proc SURVEYFREQ.&nbsp; Below is an example.</P> <P>/* Generate Data */</P> <P>proc format;<BR />value ResponseCode 1 = 'Very Unsatisfied'<BR />2 = 'Unsatisfied'<BR />3 = 'Neutral'<BR />4 = 'Satisfied'<BR />5 = 'Very Satisfied';<BR />run;</P> <P>proc format;<BR />value UserCode 1 = 'New Customer'<BR />0 = 'Renewal Customer';<BR />run;</P> <P>proc format;<BR />value SchoolCode 1 = 'Middle School'<BR />2 = 'High School';<BR />run;</P> <P>proc format;<BR />value DeptCode 0 = 'Faculty'<BR />1 = 'Admin/Guidance';<BR />run;</P> <P>data SIS_Survey;<BR />format Response ResponseCode.;<BR />format NewUser UserCode.;<BR />format SchoolType SchoolCode.;<BR />format Department DeptCode.;<BR />do _imputation_=1 to 2;<BR />drop j;<BR />retain seed1 111;<BR />retain seed2 222;<BR />retain seed3 333;</P> <P>State = 'GA';</P> <P>NewUser = 1;<BR />do School=1 to 71;</P> <P>call rantbl( seed1, .45, .55, SchoolType );</P> <P>Department = 0;<BR />call rannor( seed3, x );<BR />SamplingWeight = 25 + x * 2;<BR />do j=1 to 2;<BR />if ( SchoolType = 1 ) then<BR />call rantbl( seed2, .16, .21, .30, .24, .09, Response);<BR />else<BR />call rantbl( seed2, .18, .23, .30, .22, .07, Response);<BR />output; end;<BR />output;</P> <P>Department = 1;<BR />call rannor( seed3, x );<BR />SamplingWeight = 15 + x * 1.5;<BR />do j=1 to 2;<BR />if ( SchoolType = 1 ) then<BR />call rantbl( seed2, .10, .15, .33, .28, .14, Response );<BR />else<BR />call rantbl( seed2, .13, .20, .30, .26, .11, Response);<BR />output; end;<BR />end;</P> <P>NewUser = 0;<BR />do School=72 to 134;</P> <P>call rantbl( seed1, .45, .55, SchoolType );</P> <P>Department = 0;<BR />call rannor( seed3, x );<BR />SamplingWeight = 25 + x * 2;<BR />do j=1 to 2;<BR />if ( SchoolType = 1 ) then<BR />call rantbl( seed2, .16, .21, .30, .24, .09, Response);<BR />else<BR />call rantbl( seed2, .18, .23, .30, .22, .07, Response);<BR />output; end;<BR />output;</P> <P>Department = 1;<BR />call rannor( seed3, x );<BR />SamplingWeight = 15 + x * 1.5;<BR />do j=1 to 2;<BR />if ( SchoolType = 1 ) then<BR />call rantbl( seed2, .10, .15, .33, .28, .14, Response );<BR />else<BR />call rantbl( seed2, .13, .20, .30, .26, .11, Response);<BR />output; end;<BR />end;</P> <P>State = 'NC';</P> <P>NewUser = 1;<BR />do School = 135 to 218;</P> <P>call rantbl( seed1, .45, .55, SchoolType );</P> <P>Department = 0;<BR />call rannor( seed3, x );<BR />SamplingWeight = 25 + x * 2;</P> <P>if ( SchoolType = 1 ) then<BR />call rantbl( seed2, .16, .21, .30, .24, .09, Response);<BR />else<BR />call rantbl( seed2, .18, .23, .30, .22, .07, Response);<BR />output; output;<BR />output;</P> <P>Department = 1;<BR />call rannor( seed3, x );<BR />SamplingWeight = 15 + x * 1.5;</P> <P>if ( SchoolType = 1 ) then<BR />call rantbl( seed2, .10, .15, .33, .28, .14, Response );<BR />else<BR />call rantbl( seed2, .13, .20, .30, .26, .11, Response);<BR />output; output;<BR />end;</P> <P>NewUser = 0;<BR />do School = 219 to 274;</P> <P>call rantbl( seed1, .45, .55, SchoolType );</P> <P>Department = 0;<BR />call rannor( seed3, x );<BR />SamplingWeight = 25 + x * 2;<BR />do j=1 to 2;<BR />if ( SchoolType = 1 ) then<BR />call rantbl( seed2, .16, .21, .30, .24, .09, Response);<BR />else<BR />call rantbl( seed2, .18, .23, .30, .22, .07, Response);<BR />output; end;<BR />output;</P> <P>Department = 1;<BR />call rannor( seed3, x );<BR />SamplingWeight = 15 + x * 1.5;</P> <P>if ( SchoolType = 1 ) then<BR />call rantbl( seed2, .10, .15, .33, .28, .14, Response );<BR />else<BR />call rantbl( seed2, .13, .20, .30, .26, .11, Response);<BR />output; output;<BR />end;</P> <P>State = 'SC';</P> <P>NewUser = 1;<BR />do School = 275 to 328;</P> <P>call rantbl( seed1, .45, .55, SchoolType );</P> <P>Department = 0;<BR />call rannor( seed3, x );<BR />SamplingWeight = 25 + x * 2;<BR />do j=1 to 2;<BR />if ( SchoolType = 1 ) then<BR />call rantbl( seed2, .16, .21, .30, .24, .09, Response);<BR />else<BR />call rantbl( seed2, .18, .23, .30, .22, .07, Response);<BR />output; end;<BR />output;</P> <P>Department = 1;<BR />call rannor( seed3, x );<BR />SamplingWeight = 15 + x * 1.5;</P> <P>if ( SchoolType = 1 ) then<BR />call rantbl( seed2, .10, .15, .33, .28, .14, Response );<BR />else<BR />call rantbl( seed2, .13, .20, .30, .26, .11, Response);<BR />output; output;<BR />end;</P> <P>NewUser = 0;<BR />do School = 329 to 370;</P> <P>call rantbl( seed1, .45, .55, SchoolType );</P> <P>Department = 0;<BR />call rannor( seed3, x );<BR />SamplingWeight = 25 + x * 2;<BR />do j=1 to 2;<BR />if ( SchoolType = 1 ) then<BR />call rantbl( seed2, .16, .21, .30, .24, .09, Response);<BR />else<BR />call rantbl( seed2, .18, .23, .30, .22, .07, Response);<BR />output; end;<BR />output;</P> <P>Department = 1;<BR />call rannor( seed3, x );<BR />SamplingWeight = 15 + x * 1.5;</P> <P>if ( SchoolType = 1 ) then<BR />call rantbl( seed2, .10, .15, .33, .28, .14, Response );<BR />else<BR />call rantbl( seed2, .13, .20, .30, .26, .11, Response);<BR />output; output;<BR />end;<BR />end;<BR />run;<BR />title 'School Information System Survey';<BR />/*Run SURVEYFREQ by _IMPUTATION_ assuming the MI step is already done*/<BR />proc surveyfreq data=SIS_Survey;<BR />by _imputation_;<BR />tables Response*schooltype/wtfreq;<BR />ods output CrossTabs=mi_ctab;</P> <P>run;</P> <P>proc print;<BR />run;</P> <P>/*Sort the data by the TABLES variables which is called RESPONSE here*/<BR />proc sort data=mi_ctab;<BR />by response schooltype _imputation_;<BR />run;</P> <P>/*Run MIANALYZE with STDERR option*/<BR />proc mianalyze data=mi_ctab;<BR />by response schooltype;*this would be the TABLES variable;<BR />modeleffects wgtfreq;<BR />stderr stdDev;<BR />title 'Results of for Weighted Frequency';<BR />run;</P> Thu, 23 Jul 2020 11:16:19 GMT https://communities.sas.com/t5/Statistical-Procedures/proc-freq-with-multiple-imputation/m-p/671767#M32112 SAS_Rob 2020-07-23T11:16:19Z https://communities.sas.com/t5/Statistical-Procedures/proc-freq-with-multiple-imputation/m-p/722593#M35012 <P>Thank you for this reply, it worked!</P><P>Now I had another question. Is it possible to perform a chi-square test and have a p-value for the imputed frequency data (comparing responders to non responders )?</P><P>&nbsp;</P><P>Best regards</P> Mon, 01 Mar 2021 14:56:42 GMT https://communities.sas.com/t5/Statistical-Procedures/proc-freq-with-multiple-imputation/m-p/722593#M35012 ablond 2021-03-01T14:56:42Z https://communities.sas.com/t5/Statistical-Procedures/proc-freq-with-multiple-imputation/m-p/722596#M35013 <P>Yes.&nbsp; You would have to combine the actual Chi-Square statistics from each of the tables.&nbsp; Dr. Paul Allison has a macro on his website that will compute the combined Chi-Square statistics.</P> <P><A href="https://www.sas.upenn.edu/~allison/combchi.sas" target="_blank">https://www.sas.upenn.edu/~allison/combchi.sas</A>&nbsp;</P> Mon, 01 Mar 2021 15:07:21 GMT https://communities.sas.com/t5/Statistical-Procedures/proc-freq-with-multiple-imputation/m-p/722596#M35013 SAS_Rob 2021-03-01T15:07:21Z