Hank First off, in the example you gave, in a SAS table a numeric missing value is usually represented by a dot/period.  So if the data is already in a SAS table both 45 and . are of the same type. If your data is already in 64 SAS tables, each table will have consistent types for each column, and you can tell the type from proc contents of the data table.  So even if the value looks like a character value (eg 12MAR45) it will just be the formatted representation of the underlying numeric value. Second, are you importing these tables from CSV or tab delimited files (or other text)?  If so (and you are using SAS in Windows), there is a way of forcing the the data types on import. Do you want to force all variables to character or numeric? A macro solution using arrays may be possible but would need to be quite complex: detect which columns are of the wrong type detect which order they are in, in the program data vector for each table create new arrays of the required column types convert the data, dropping the 'old' columns renaming the new columns if required. Richard ... View more

Regrettably not. That is because the data has to be fed into the dataset before it is converted, and SAS cannot tolerate a mix of character and numeric types in a column with the same name. The situation will be worse if some your input datasets will have different types the next time you import them. The most satisfactory solution is to ensure all the 64 datasets have the correct type data when you import them.  You may need co-operation from your sources to ensure type consistency. Failing this, It will be tedious for 64 datasets but your best course is to correct the type in each one before concatenating. data _newtable1 ;      set _table1 (rename = (var1 = _var1)) ;      var1 = input(_var1, ddmmyy10.);      drop _var1 ; run ; and so on until all _newtable datasets have consistent types. Then you can concatenate these tables. Richard ... View more

Have you tried exporting from Excel as a TAB delimited (.txt) file and then importing the data into SAS that way? Richard ... View more

You can find all the information on Unix commands on the internet but sometimes you do not know what command you want.  I would recommend you get a Unix book like "Unix in a nutshell" which will give you a good idea on how Unix works. Richard ... View more

Also, note that the Mixed option is not available for csv files but can be specified for Excel libnames. Richard ... View more

Assuming your environment is SAS on Windows, here is what you do: Run Proc Import but ignore the output Open a new program window Press F4 to retrieve the generated SAS code Examine the Informat statement for the cusips Change the Informat statement for cusips to use informat $char32. Change the Format statement for cusips to use format $char32. Change the Input statment to add $ after cusips Replace proc import in your original program with the amended data step code Rerun Richard ... View more

Another option is to use the EXCEPT set operator in Proc SQL. See Base SAS(R) 9.2 Procedures Guide Either the columns must be in the same order in both tables or you should user the CORRESPONDING keyword You can get the unmatched records in table first using something like this (untested) Proc SQL ;      Create table firstonly as           Select * from first                Except corr           Select * from second      ; Quit ; Then get the records in the second table not matched in the first by interchanging 'first' and 'second' in the code. Richard Message was edited by: Richard Carson ... View more

Something like this (not tested) Proc sql ;      create table want as           select a.*           from have a                , (Select product                     ,  count (distinct order) as tot                     from have                  )     b           where a.product = b.product                and b.tot = 5      ; quit ; Richard ... View more

Your requirement seems to translate to the conditions where DOB > '01MAY1983'D and DOB <= '01MAY1993'D if my mental arithmetic is correct Richard Message was edited by: Richard Carson Did some date arithmetic to update the figures ... View more