The BESTDw.p format doesn't seem to kick into scientific notation like BEST does (or at least not as quickly), so you might try playing around with that some, possibly with a smaller value for w than 32.

But if that doesn't work, the approach you proposed sounds reasonable to me.

Mike Rhoads RhoadsM1@Westat.com

-----Original Message----- From: SAS(r) Discussion [mailto:SAS-L@LISTSERV.UGA.EDU] On Behalf Of Lorne Klassen Sent: Monday, September 19, 2011 4:10 PM To: SAS-L@LISTSERV.UGA.EDU Subject: BEST32. format for small decimal numbers

I'm using the BEST32. format for displaying standard error values in a PROC TABULATE report.

proc format library=work; value SE (default=32) .='x' other=[best32.]; run;

It works well since it chooses the best display for a variety of values such as:

38476362448 475 5 0 0.2 0.015 0.00204 0.0008 0.0000721

But as soon as there's a value that starts in the 6th decimal place or beyond, it displays the value in scientific notation. I never want values to display in scientific notation.

E.g.: 0.00000204 ==> 2.04E-6

One important thing to note is that we first round our SE values to the 7th decimal place. So I don't have to worry about anything beyond 7 decimal places.

Is this the correct way or only way to solve this? Any pitfalls with this? One minor downside is that it will give a trailing zero for certain numbers but that's not a major problem.

proc format library=work; value SE (default=32) .='x' 0<-<0.00001=[9.7] other=[best32.]; run;

Thanks