Data looks like this; start               end 20100313      20141231 20101127      20160421 20091021      20160519   and you had tried  month= intck('month', start, end); So this can be assumed you need difference in months. And the above did not work so both the variable start and end are not numeric or one of them not numeric.   1. Please identify the non-numeric type data first and change it to numeric data type using format yymmdd8.     e.g.1. if start is charecter then do as following     start1=input(start,yymmdd8.);     e.g.1. if end is charecter then do as following     end1=input(end,yymmdd8.); 2. Then try the intck function for the difference.   Please note I have used a different variable while changing datatype in the example.   Please let us know if it helped you. ... View more

Thank you, this helps a lot! ... View more

@hovliza wrote: Hi! If I want to calculate the difference between months I use this syntax; *months; DATA want; SET have; months=intck("month", start, end); RUN; But you only get the months without decimals. Is there an easy way to get the months with for example 2 decimals? Please describe exactly how you plan on using that fractional month part of the value. Anything I have worked on that needed a different interval worked much better by dropping to an actual different interval, either weeks, biweekly period, days or even 5-day periods. That way at least handles 0.1month = one of .28, .29, .30 or .31. Note that the intck and intnx functions supports multiples of the interval such as DAY3 or WEEK2 so you can easily specify a consistent period. ... View more

@hovliza wrote: Hi!   I was wondering what I am doing wrong. I want to make a starting point of my follow-up (date_start1) and I have different dates of measurement (date1, date2, date3 etc.) and I want to pick the date which is nearest to the date of questionnaire completion (dataq) as my date_start1. Most of the time date_start1 is date1. But sometimes date2-4 are closer to the date of questionnaire completion.     When I do this: DATA want; SET have; date_start1=date1; IF abs((date1-dataq)/365.25) GT abs((date2-dataq)/365.25) THEN date_start1=date2; ELSE IF abs((date1-dataq)/365.25) OR abs((date2-dataq)/365.25) GT abs((date3-dataq)/365.25) THEN date_start1=date3; ELSE IF abs((date1-dataq)/365.25) OR abs((date2-dataq)/365.25) OR abs((date3-dataq)/365.25) GT abs((date4-dataq)/365.25) THEN date_start1=date4; FORMAT date_start1 DATE9.; RUN;   I don't get good results. SAS doesn't pick this:  date_start1=date1;    It seems SAS only picks: IF abs((date1-dataq)/365.25) GT abs((date2-dataq)/365.25) THEN date_start1=date2;   What should I do? Thanks in advance!  There is NO way for SAS to have a value of date_start1= date1. (unless date1 matches one of date2, 3 or 4) Your if/then/else forces it to have one of the values of date2, date3 or date4. Perhaps you meant to have another else that left date_start1 with a value of date1 or a different test at the start before the assignment of date2? ... View more

Just post the data (in all datasets) for persons 1, 2 and 3 (as mentioned in your original question). ... View more

And there’s a difference between the terms database and data set with respect to SAS. They can be one and the same but typically databases have multiple datasets, not a single one. ... View more