I'm pretty sure that, when you evaluate something inside brackets like in Ken's code, it returns a value of 1 for true and 0 for false. In this case, when the code evaluates an 'a' value of 0.04 for example (a<0.05), it 'resolves' to 1 if true. Then the sum function is simply summing the resulting 1s and 0s for each condition across your dataset.

It's probably a bit counter-intuitive to use a sum when you want to count, but they're the same thing when the only candidates for summing are 1 or 0.

Does that help?

TF

yoonsup@gmail.com wrote: > Thanks all for your contribution. > > While I understand Tree Frog's code, Ken's code is not quite straight > forward to me although it did what I wanted. > > In SAS's SQL procedure user guide, the aggregate function, "SUM" is > described as "sum of values". > > With this, I'm guessing that Ken's advice on putting the condition > directly to the aggregate function is supposed to > > sum up all the values that meet the condition in the parenthesis > rather than counting them. However, it counts the number of rows > > that meet the condition, meaning Ken's code is just right for my > question. What do I not understand here in "SUM" function? > > Thanks. > > Yoon