Weight Categorical Variables in Proc Tabulate

[benbuck]

Hi all,

I am running the below code, attempting to create a table using Proc Tabulate. I am using categorical variables and am attempting to weight observations using the variable "discwt". However, I can't get the weighting to work. Regardless of what I do, my code only produces an unweighted table. 

I am running this on SAS 9.4 TS 1M1.

Thanks!

proc tabulate data = work.redo_t1;
	title 'Table 1a';
	class cat_flag race ; 
	table race, (cat_flag)*(N colpctn);
	var discwt / weight = discwt;
	weight discwt;
run;

[Reeza]

Can you generate some data we can work with to replicate and test your issue? 

[benbuck]

Here is code to produce sample data:

data redo_t1;
	input ID discwt cat_flag race;
	datalines; 
		01 5 1 1
		02 5 1 2
		03 5 1 1
		04 5 1 2
		05 5 1 1
		06 5 3 2
		07 5 3 1
		08 5 3 2
		09 5 3 1
		10 5 3 2
	;
run;

proc tabulate data = work.redo_t1;
	title 'Table 1a';
	class cat_flag race ; 
	table race, (cat_flag)*(N colpctn);
	var discwt / weight = discwt;
	weight discwt;
run;

and i'd like my table to look something like this:

                   Cat 1       Cat 3

Region 1     15           10

Region 3     10           15

and not 

                   Cat 1       Cat 3

Region 1     3             2

Region 3     2             3

[art297 · Accepted solution]

You want to use the FREQ option .. NOT the WEIGHT option. e.g.:

data redo_t1;
	input ID discwt cat_flag race;
	datalines; 
		01 5 1 1
		02 5 1 2
		03 5 1 1
		04 5 1 2
		05 5 1 1
		06 5 3 2
		07 5 3 1
		08 5 3 2
		09 5 3 1
		10 5 3 2
	;
run;

proc tabulate data = work.redo_t1;
	title 'Table 1a';
	class cat_flag race ; 
	table race, (cat_flag)*(N colpctn);
	var discwt / weight = discwt;
	freq discwt;
run;

Art, CEO, AnalystFinder.com

[benbuck]

Thank you for your help! This is exactly what I was looking for. I appreciate your effort and time.

[Bedi]

Your code works fine when the weights are integers , however if the weight are non-integers ,it  doesn't give correct results

Below is the sample code

data redo_t1;
	input ID discwt 3.1 cat_flag race;
	datalines; 
		01 0.2 1 1
		02 0.4 1 2
		03 0.4 1 1
		04 0.4 1 2
		05 0.4 1 1
		06 0.1 3 2
		07 1.2 3 1
		08 0.4 3 2
		09 1.0 3 1
		10 0.4 3 2
	;
run;

proc tabulate data = work.redo_t1;
	title 'Table 1a';
	class cat_flag race ; 
	table race, (cat_flag)*(N colpctn);
	var discwt / weight = discwt;
	freq discwt;
run;

When I run the attached syntax only the integer part in my weights is used for calculation.

Can you please suggest how to use the solution with the decimal weights.

[Reeza]

Try the WEIGHT statement in that case instead of FREQ.

---

@Bedi wrote:

Your code works fine when the weights are integers , however if the weight are non-integers ,it  doesn't give correct results

Below is the sample code

data redo_t1;
	input ID discwt 3.1 cat_flag race;
	datalines; 
		01 0.2 1 1
		02 0.4 1 2
		03 0.4 1 1
		04 0.4 1 2
		05 0.4 1 1
		06 0.1 3 2
		07 1.2 3 1
		08 0.4 3 2
		09 1.0 3 1
		10 0.4 3 2
	;
run;

proc tabulate data = work.redo_t1;
	title 'Table 1a';
	class cat_flag race ; 
	table race, (cat_flag)*(N colpctn);
	var discwt / weight = discwt;
	freq discwt;
run;

When I run the attached syntax only the integer part in my weights is used for calculation.

Can you please suggest how to use the solution with the decimal weights.

---

[Bedi]

Hi..I tried it but than it is giving un-weighted response