There are limitations on the storage of large numbers. See the discussion on numeric precision at SAS(R) 9.2 Language Reference: Concepts, Second Edition. If you aren't using the number for mathematical operations, why not leave it as text? If you are doing mathematical operations, then SAS floating point representation might be good enough? If you insist on storing the entire number, try the w.d informat. And don't use the BEST format. The downside is increased disk space. You could use the COMPRESSION=BINARY option to offset this. data test; format b c d 32.; x = '0010000000130000123'; y = '001000000013000012'; z = '00100000001300001'; len1 = length(x); len2 = length(y); len3 = length(z); b = input(x, 32.); c = input(y, 32.); d = input(z, 32.); run; ... View more

You can use one of the CAT functions to count how many responses were 1,2,3,4,5 or 8 in each row. Something like: Data sum_response ;      set survey2;      count1 = countc(catt(of q1-q10), '1');      count2 = countc(catt(of q1-q10), '2');      etc . . .      count8 = countc(catt(of q1-q10), '8'); run; ... View more

The text comparison should work in this case (number-like). There are no letters that would mess it up (for example: JAN, AUG). The ASCII and EBCDIC collating sequences will both evaluate the comparison correctly, I believe. ... View more

Then, Sanyam's suggestion should work. But first make sure that there is only 1 CUSIP for each PERMNO in NAME1; otherwise, you will still have a many-to-one match between NAME1 and PINS. ... View more

The text-based comparison should still work in this case, provided all of the dates have a consistent form (YYYY-MM-DD HH:MM:SS) and the where clause contains a text value that is in the same form. But, I agree, a date conversion is preferable. ... View more

Question: do you also want to merge the name (COMNAM)? ... View more

Ideally, you should convert the text values to actual date values. However, something like the following should work? %let mydate = 2013-07-04 21:06:19; Data want ;      set your-file-name ;      where date >= "&mydate"; run; ... View more

Are these numeric values (with a SAS format) or text? ... View more

@Sanyam I'm not sure if LanMin's objective was solely to add the CUSIP to the PINS data. If so, that does simplify the query. However, there remain unanswered questions about the data (like the presence of duplicates in PINS, missing PERMNO, etc) and how LanMin wants to address these will, in my view, determine the coding approach. I'm at home and don't have access to the data/SAS, but I also seem to recall there were multiple CUSIP for some PERMNO; hence the original need to use YEAR_BG and YEAR_END as a matching criteria. ... View more

Your names1 data definitely has overlapping year records for a given permno. For example: PERMNO    CUSIP          COMNAM                             YEAR_BG     YEAR_END 10001          36720410     GREAT FALLS GAS CO     1986               1993 10001          36720410     ENERGY WEST INC           1993               2008 So, YEAR_END on the first record overlaps the YEAR_BG of the second record. In fact, your names1 data has many, many, MANY cases where the dates overlap. Most commonly, this is because the starting year of the next record is the same as the ending year of the previous record. Other observations: * There are many PERMNO in NAMES1 that do not exist in PINS (e.g. permno 10012) * there is one PERMNO in PINS that does not exist in NAMES1 (permno 88394) * there are multiple year records in PINS for the same PERMNO (e.g. permno 10597 has 2 records for 2001). This is in addition to the "problem" of overlapping records in NAME1. * many of the name records in NAMES1 for a given PERMNO are identical (e.g., the name is the same on record after record) although the starting and ending dates might change ... View more