Hi:   WRDS is supported by the Wharton help desk. So you'll need to find out from them where you can upload your local data file. My guess is that SAS on the WRDS server cannot read your local file and so you'll need to upload your local file to a location on the server. You may need to get help from the Wharton help desk to find out how to do this.   Next, even if you do not have write access to sort FILE1 and write it back to the permanent location, my guess is that you can sort your FILE1 and use the OUT= option in PROC SORT to write the data to the WORK library. Then once both of your files are in the WORK library, you should be able to do a merge.   Conceptually, it would be like using SASHELP.CLASS and SASHELP.CLASSFIT. SASHELP is a read-only folder on the SAS OnDemand server, so although I can sort both of these files, I cannot write the sorted copy back to SASHELP because I do not have authorization to write to SASHELP. So I have to do something like this: ** when you do not have write access to a file; ** you can sort the file and write a sorted copy into WORK; ** and use the copy that's in work for any subsequent steps; proc sort data=sashelp.class out=work.class; by name age; run; ** instead of using SASHELP.CLASSFIT, you will need to upload; ** your local file to the server for sorting and merging; proc sort data=sashelp.classfit out=work.classfit; by name age; run; data inboth; merge work.class(in=inclass) work.classfit(in=infit); by name age; if inclass=1 and infit=1 then output inboth; run; proc print data=inboth; title 'After Merge'; run; title;   In this example, SASHELP.CLASS would be the FILE1 that is already on the WRDS server. You should be able to SORT the file as you need as long as you use the OUT= option and write the sorted copy of the data to a location like WORK where you have write access. Then in this example, SASHELP.CLASSFIT would be the FILE2 on your local machine. You'll probably need to upload your data file from your local machine to the WRDS server. If your File2 is a raw data file, then after you do the upload, you'll need to read File2 into SAS format prior to your sort. But then, you should be able to sort File2 as you need and then do the merge.   Hope this points you toward a solution. Cynthia ... View more

Hi:   Here's a solution example with PROC REPORT:   In this example, the character version of the main numeric variable is created in PROC REPORT as a COMPUTED item and so the numeric variable is still used for ordering, but the character version allows you to put your custom text on the break line. Cynthia ... View more

Hi:   What you currently show is what PROC REPORT will give you. I do not think there's a way to get PROC REPORT to do what you want to do.   I was pleasantly surprised when testing the Report Writing Interface that my program generated the correct results in both ODS RTF and ODS TAGSETS.RTF (yay!), as shown below:   I just used SASHELP.IRIS, and presummarized it to get the averages. There are 4 columns in the report plus the Species column, to mimic your columns and then using some headers that are similar to what you want, just as a proof of concept. Here's the code I used: %let mypath=c:\temp; ** presummarize IRIS data to get only averages; proc means data=sashelp.iris mean maxdec=2; class species; var sepallength sepalwidth petallength petalwidth; output out=work.allavg mean=; run; ods html path="&mypath." file="complex_column_span.html"; ods pdf file="&mypath.\complex_column_span.pdf" ; ods rtf file="&mypath.\odsr_column_span.rtf"; ods tagsets.rtf file="&mypath.\odstr_column_span.rtf"; title "Complex Column Spanning"; data _null_; set allavg(keep=_type_ species sepallength sepalwidth petallength petalwidth) end=last; by species; if _N_ = 1 then do; dcl odsout obj(); obj.table_start(); obj.head_start(); ** Header row 1; obj.row_start(type:"Header"); obj.format_cell(text: "Species", row_span:3,style_attr:"color=black backgroundcolor=white vjust=b fontweight=bold"); obj.format_cell(text: "N=111", column_span:2, style_attr:"color=black backgroundcolor=white"); obj.format_cell(text: "N=222", column_span:2, style_attr:"color=black backgroundcolor=white"); obj.row_end(); ** Header row 2; obj.row_start(type:"Header"); obj.format_cell(text: "Averages by Species", column_span:4, style_attr:"color=black backgroundcolor=white"); obj.row_end(); ** Header row 3; obj.row_start(type: "Header"); obj.format_cell(text: "SepalLength", style_attr:"color=black backgroundcolor=white fontweight=bold"); obj.format_cell(text: "SepalWidth", style_attr:"color=black backgroundcolor=white fontweight=bold"); obj.format_cell(text: "PetalLength", style_attr:"color=black backgroundcolor=white fontweight=bold"); obj.format_cell(text: "PetalWidth", style_attr:"color=black backgroundcolor=white fontweight=bold"); obj.row_end(); obj.head_end(); end; ** row for every obs; if _type_ = 1 then do; obj.row_start(); obj.format_cell(data: species, row_span:1); obj.format_cell(data: sepallength, row_span:1); obj.format_cell(data: sepalwidth, row_span:1); obj.format_cell(data: petallength, row_span:1); obj.format_cell(data: petalwidth, row_span:1); obj.row_end(); end; if last then do; obj.table_end(); end; run; ods html close; ods pdf close; ods rtf close; ods tagsets.rtf close; title; footnote;   I used the _TYPE_=1 rows from the PROC MEANS OUTPUT data set because I did not need the _TYPE_=0 row.   Depending on the structure of your data and what you need to do, the Report Writing Interface may be an option. Cynthia   ... View more