Split a continuous variable with proc tabulate

[Demographer]

Hi,

I have a variable x1 that range from 0 to 1 and a variable x2 that range from -10 to +10. I want to create a crosstable showing the average of x2 for different ranges of x, such as 

x1	average of x2
0-0.1	#
0.1-0.2	#
…	#
0.9-1	#

I know I could easily split the variable x1 in a data statement, but I wonder if it this possible to split in within the proc tabulate statement.

[Astounding]

You can, but you have to define a format first to identify the splits.  For example:

proc format;
   value tenths 
   0-0.1 = '0.0-0.1'
   0.1-0.2 = '0.1-0.2'
   0.2-0.3 = '0.2-0.3'
   0.3-0.4 = '0.3-0.4'
   0.4-0.5 = '0.4-0.5'
   0.5-0.6 = '0.5-0.6'
   0.6-0.7 = '0.6-0.7'
   0.7-0.8 = '0.7-0.8'
   0.8-0.9 = '0.8-0.9'
   0.9-1 = '0.9-1.0'
;
run;

Having done that, you can apply the format within PROC TABULATE:

proc tabulate data=have;
   class x1;
   var x2;
   tables x1, x2*mean;
   format x1 tenths.;
run;

[ballardw]

Being a mathematician, I cringe when I don't know which interval an end-point value occurs in, or even if it appears in any. So if X has a value of exactly 0.1 should it be considered in the 0.0-0.1 range or the 0.1 -0.2 range?

@Astounding's format will place the value of 0.1 in the 0.1 -0.2 range.

[Demographer]

My example was just theoritical, so it doesn't matter.

[ballardw]

---

@Demographer wrote:
My example was just theoritical, so it doesn't matter.

---

I will politely beg to differ. Assignment to categories for analysis may be critical, especially if you have lots of values exactly on the endpoints of intervals. Especially in a group working environment. If you do an analysis with 0.1 in the 0.1 -0.2 group and another team assigns it to the 0.0-0.1 you will have fun reconciling things like N's not matching in different places, or model results if the categorical assignment is used in a class type variable.

You may also have to meet an external client or even government agency's definition and failure to do so could put you or your organization in a number of binds.

[Demographer]

I mean, this was not a real example, so I don't mind if it's 0-0.1[ or 0-0.01]. I just wanted to know if there is a code to do it automatically in proc tabulate. Of course, when I'll code it, I will be cautious that categories fit.