@Wolverine wrote: @ballardw wrote: Which means that you want TM_Group in your Tables statement(s).   Defining macros in the middle of other procedure code is bad practice. if you really need that %val_labs macro then place the macro code before the data step or procedure that want to use it. So the Tables statement should be "TABLES DEM_AGE*TM_group*ADRD_group DEM_SEX*TM_group *ADRD_group / rschisq;" , and I should drop the BY statement altogether? I thought about it some more and I think it is correct to have TM_group as a BY variable. The significant difference I'm looking for is whether there is a difference between the ADRD and non-ADRD groups within the MA group (which is equivalent to TM_group = 0), or between ADRD/non-ADRD within the TM group (TM_group=1). The BY groups are TM_group = 0 and TM_group = 1, and I am NOT comparing them in this significance test. ... View more

PLEASE Test your data step before posting. The variable type is wrong for Table and Value_labels, your value labels includes the delimiter so you need to quote the values.   data have; infile datalines dsd dlm=',' truncover; input TM_group ADRD_group Table:$20. Value_Labels :$20. WgtFreq Percent; TM_Adrd_group = cats(tm_group,adrd_group); datalines; 0,0,Table DEM_AGE,"2:Age Group [65,75)",48301097,56.0249 0,0,Table DEM_AGE,3:Age Group >=75,37912485,43.9751 0,0,Table DEM_SEX,1:Male,36957033,42.8668 0,0,Table DEM_SEX,2:Female,49256549,57.1332 0,1,Table DEM_AGE,"2:Age Group [65,75)",983069,26.8767 0,1,Table DEM_AGE,3:Age Group >=75,2674626,73.1233 0,1,Table DEM_SEX,1:Male,1363336,37.2731 0,1,Table DEM_SEX,2:Female,2294360,62.7269 1,0,Table DEM_AGE,"2:Age Group [65,75)",82517247,60.1757 1,0,Table DEM_AGE,3:Age Group >=75,54609890,39.8243 1,0,Table DEM_SEX,1:Male,63773136,46.5066 1,0,Table DEM_SEX,2:Female,73354000,53.4934 1,1,Table DEM_AGE,"2:Age Group [65,75)",1290895,25.8295 1,1,Table DEM_AGE,3:Age Group >=75,3706852,74.1705 1,1,Table DEM_SEX,1:Male,2215172,44.3234 1,1,Table DEM_SEX,2:Female,2782576,55.6766 ; RUN; proc format library=work; value $tm_adrd '00'='no TM, no ADRD' '01'='no TM, ADRD' '10'='TM, no ADRD' '11'='TM, ADRD' ; run; proc report data=have; columns table value_labels tm_adrd_group,(WgtFreq Percent); define table/group; define value_labels/group; define tm_adrd_group /across format=$tm_adrd. order=internal; run;   Though  I would create the group variable, or others prior to summary then generate the summary with that new variable so there aren't any questions about combining stuff. If you are doing confidence limits this would be required.     You show no TM, no ADRD no TM, ADRD TM, no ADRD TM,  ADRD without a clear definition of where those come from.   I have to guess that means the combinations of the variables Tm_group and Adrd_group. One approach would be to create a composite variable in a data step such as   TM_Adrd_group = cats(tm_group,adrd_group); Which should create a character variable with values of '00', '01','10' and '11'.   create format to show such things as "no TM, no ADRD" for the '00' version. Then in one of the reporting procedures assign that format to the composite variable. Either Proc Report or Tabulate would do such nesting the WgtFreq and Percent under the Tm_adrd_group ... View more

If your data is from a complex sample, sort of implied with "replicate weights" then Procs Means, Univariate and Freq should not be used as they will not apply the weight according to the sample design.   My personal macros come in two basic flavors: calculate the numbers and display the numbers. That way any data set with the numbers can be displayed. The TableN macro combines those two pieces and from the little I looked at the code I am not going to attempt to "fix" something that the author spent years writing. Especially where you should be using the Procs Surveymeans or Surveyfreq which use quite different output data sets, having to use ODS OUTPUT and resultant sets are much more complex.   IF the TableN macro can be convinced not to attempt to create confidence intervals but use variables from a data set directly then you would be better off summarizing the data with Surveyfreq/Means, restructuring the data set and sending that to the macro. ... View more

If the data set you have requires weights to analyze / report properly and a macro or other tool does not support weights then the basic approach would be to use one of the procedures to generate the POPULATION estimates to have the counts and rates with confidence limits. Then use the report tool with the new data.   IF that tool wants to calculate confidence intervals then hope there is a way to override that behavior as it may have problems with the new data set and calculating incorrect confidence limits.   @Wolverine wrote: As I detailed in this post I'm trying to make a "Table 1" summary table and another additional table of descriptive statistics for dozens of variables. Generating the proper results is relatively easy. But rearranging those results for dozens of variables and putting them into a well-organized summary table takes a lot of code. The TableN macro does this extremely well, but it doesn't have a provision to add weights. Likewise, the code I wrote for the secondary table uses Proc Ttest, and there is no "SURVEY" equivalent (like Proc Surveyfreq is the equivalent to Proc Freq). Proc Ttest can use a person weight, but not sampling weights.   I'm supposed to have a meeting this afternoon to review the results. At this point, we just need initial results to start developing a theoretical model. So what kind of error will these results have? As I understand it, the standard error will be underestimated if weights are not used. As a result, some findings may appear to be significant but really aren't. Is that correct, and are there any other issues I should be aware of ?   ... View more

@Wolverine wrote: It should -- any proc with "survey" in the name should be able to use sample weights Generating the proper results is relatively easy. But rearranging those results for dozens of variables and putting them into a well-organized summary table takes a lot of code. The TableN macro does this extremely well, but it doesn't have a provision to add weights. Likewise, the code I wrote for the secondary table uses Proc Ttest, and there is no "SURVEY" equivalent (like Proc Surveyfreq is the equivalent to Proc Freq). Proc Ttest can use a person weight, but not sampling weights. ... View more

Depending on how "tall" your output is you might consider the system option  Orientation=Landscape; instead of the default portrait. Set this before the ODS RTF statement. You can also use OPTIONS PAPERSIZE= before the ODS RTF output is generated to create a wider/taller logical page to display output. I recommend providing a unit such as IN or CM with the option so you know what to expect.   Options papersize=(10in, 8in); would set the page to 10 inches wide and 8 inches high for example. The option also takes some paper names but that varies by location and system.     ... View more

Looks like the macro also doesn't accept a data step view as suggested by Reeza. Below fully working sample code that demonstrates how you can prepare the data. data have; infile datalines dsd dlm=',' truncover; input DEM_AGE DEM_SEX cohort_flag TM_group; datalines; 3,1,1,0 2,1,1,1 3,2,1,1 3,2,1,1 3,2,1,0 2,2,1,1 2,1,1,1 3,1,1,1 2,1,1,1 3,2,1,0 2,1,1,0 2,2,1,1 3,2,1,0 2,2,0, 3,2,1,1 3,2,1,1 3,1,1,0 3,2,1,0 2,1,1,0 3,1,1,1 3,2,1,1 3,2,1,0 3,2,1,1 3,2,1,1 3,2,1,1 ; RUN; proc format library=temp; value age2grp 1='1:Age Group <65' 2='2:Age Group [65,75)' 3='3:Age Group >=75' .='Inapplicable/Missing' ; value sex 1='1:Male' 2='2:Female' .='Inapplicable/Missing' ; value yesfmt 1='1:Yes' 2='2:No' .='Inapplicable/Missing' ; RUN; filename tableN url 'https://communities.sas.com/kntur85557/attachments/kntur85557/library/4477/5/tablen_05102022.sas'; /*filename tableN url 'https://communities.sas.com/kntur85557/attachments/kntur85557/library/4477/5/tablen_032020_pharmasug.sas';*/ /*filename tableN url 'https://gist.githubusercontent.com/statgeek/b55d964c99975c2ba23fa771afe616bc/raw/b30d55caf6ec4028f5afed3b260ce86b50419ddb/tablen.sas';*/ %include tableN /source2; data cohort_1; set have; where cohort_flag=1; run; %tablen( data=cohort_1 ,by=TM_group ,var=DEM_AGE DEM_SEX ,type=2 2 ,outdoc=C:\temp\test.rtf ); ... View more

@Reeza wrote: Have you looked into the TableN macro? The author does post on here. https://communities.sas.com/t5/SAS-Communities-Library/Demographic-Table-and-Subgroup-Summary-Macro-TABLEN/ta-p/634030 That looks like the answer to my prayers! Although, I'm still having some trouble with it. I'll create a new topic... ... View more

@Wolverine    Don't be shy.  Mark your explanation as the solution, so we all know the problem has been solved. ... View more

You need to change your data structure by PROC TRANSPOSE to get this kind of graph.   data have; call streaminit(123); do total_qual_care_score=0 to 8; do id=1 to 100; TM_non_ADRD_group=rand('bern',0.2); MA_non_ADRD_group=rand('bern',0.8); output; end; end; run; proc transpose data=have out=temp; by total_qual_care_score id; var TM_non_ADRD_group MA_non_ADRD_group; run; proc summary data=temp nway; class total_qual_care_score _NAME_; var col1; output out=temp2 mean=mean lclm=lclm uclm=uclm; run; data want; set temp2; by total_qual_care_score ; if first.total_qual_care_score then call missing(cum_mean,cum_lclm,cum_uclm); cum_mean+mean; _lclm=mean-lclm;cum_lclm=cum_mean-_lclm; _uclm=uclm-mean;cum_uclm=cum_mean+_uclm; run; PROC SGPLOT data=want; vbarparm category=total_qual_care_score response=mean/ group=_NAME_ name='x' ; scatter x=total_qual_care_score y=cum_mean/markerattrs=(size=0) yerrorlower=cum_lclm yerrorupper=cum_uclm; xaxis label='Total Quality of Care Score'; yaxis label='Proportion satisfied'; keylegend 'x'/title=''; RUN;     ... View more

More cumbersome than needed I'm sure here.   data have; do j=1 to 100; array value(10); do i=1 to 10; value(i)=rand('bernoulli', 0.4); _missing=rand('bernoulli', 0.05); if _missing=1 then value(i)=.; end; output; end; keep j value1-value10; run; ods output summary=part1; proc means data=have stackods N NMISS SUM; var value1-value10; run; data want (keep=variable N_0 percent_0 N_1 percent_1 NMISS percent_missing); length variable $32. N_0 percent_0 N_1 percent_1 NMISS percent_missing 8.; set part1; totalobs=NMISS+N; N_0=totalobs-sum-NMISS; N_1=sum; percent_0=N_0/totalobs; percent_1=N_1/totalobs; percent_missing=nmiss/totalobs; ; format percent: percent12.2; run; @Wolverine wrote: I'm trying to format a frequencies output dataset. The data looks like this: varname COUNT PERCENT   20 .200000000 0 50 .500000000 1 30 .300000000   I want it to look like this: _NAME_ 0 PERCENT_0 1 PERCENT_1 missing PERCENT_m varname 50 .50 30 .30 20 .20   Current code: PROC TRANSPOSE data=have out=want; VAR varname; BY COUNT PERCENT; RUN;   ... View more

@PaigeMiller wrote: It looks like you are trying to create a report. If so, you really really really really ought to do this in PROC REPORT; and you really really really really should not be using PROC TRANSPOSE for this. (and by using PROC REPORT you won't have to have silly column names that are computer variable names like American_Ind_pt; you can have real words as column names, like American Indian Patients or something like that) A fair point, but this information is getting combined with other output to create a single large table. I just needed to get this file into the correct layout so it can be combined. Once that's done, I can use other syntax I have that will correct the column names, add titles, etc. ... View more

@ballardw wrote: You didn't change the relative counts of 1 vs 0. Those stayed the same. If you had changed the -1 to 0 the ratios would change. I think I understand now... because the relative counts didn't change, the odds of having a value of 1 vs 0 didn't change either. Thanks! ... View more

An alternate:   ARRAY dx_codes DX1-DX15; PTB_birth_array=0; do index=1 to dim(dx_codes ); IF dx_codes[index] in: (&PTB_all_list.) THEN do; PTB_birth_array=1; leave; end; end; The Leave instruction will terminate the checking after finding the first match.  This might be useful as the Index value will be that of the Dx_codes array element that was first found in the list, potentially very useful if there is some order to the list. ... View more

Alternative solution using the VVALUE function   data want; set test; stdrd = round(std,.01); meanrd = round(mean,.01); format stdrd meanrd 8.2; final_output = compress(vvalue(meanrd)||"("||vvalue(stdrd)||")"); run; ... View more