topic Re: creating table in SAS Programming

creating table

MEL20 — Mon, 30 Nov 2020 06:52:44 GMT

I have the data in following format:

Age	Age_group	Gender	Weight_group	Height_group
11	1	Male	1	2
12	1	Male	1	3
12	1	Male	1	2
21	2	Male	1	3
14	1	NA	2	1
13	1	Female	2	2
23	2	Female	2	2
21	2	Female	3	1
24	2	Female	3	4
23	2	Female	3	4
13	1	NA	3	3

I need help to create the following table.

Characteristic		Height group
		Group 1	Group 2	Group 3	Group 4
Age	(mean)	--- (SD)	--- (SD)	--- (SD)	--- (SD)
Age	Group 1	---(%)	---(%)	---(%)	---(%)
	Group 2	---(%)	---(%)	---(%)	---(%)
Gender	Male	---(%)	---(%)	---(%)	---(%)
	Female	---(%)	---(%)	---(%)	---(%)
	NA	---(%)	---(%)	---(%)	---(%)
Weight	Group 1	---(%)	---(%)	---(%)	---(%)
	Group 2	---(%)	---(%)	---(%)	---(%)
	Group 3	---(%)	---(%)	---(%)	---(%)

Thanks in advance.

Re: creating table

ballardw — Mon, 30 Nov 2020 07:37:46 GMT

What does "--- (SD)" mean? "--- (%)"

Percentages imply a division with a numerator and denominator, so what would the numerator and denominators be?

By table do you mean a data set (used for further processing -if so this is going to be extremely difficult to work with) or a report that people read?

Re: creating table

andreas_lds — Mon, 30 Nov 2020 07:57:51 GMT

Please post the data you have in usable form and the output including the numbers you expect to see.

Re: creating table

MEL20 — Mon, 30 Nov 2020 09:19:28 GMT

data have;
input Age Age_group $ Gender $ Weight_group $ Height_group $;
datalines;
11 1 Male 1 2
12 1 Male 1 3
12 1 Male 1 2
21 2 Male 1 3
14 1 NA 1 2
13 1 Female 2 2
23 2 Female 2 2
21 2 Female 3 1
24 2 Female 3 4
23 2 Female 3 4
13 1 NA 3 1
;
run;

I am trying to get a table in following format:

Characteristic		Height group
		Group 1	Group 2	Group 3	Group 4
Age	Mean (standard deviation)	17.00 (5.66)	14.60 (4.83)	16.50 (6.36)	23.50 (0.71)
Age	Group 1	1 (50%)	4 (80%)	1 (50%)	0 (0%)
	Group 2	1 (50%)	1 (20%)	1 (50%)	1 (100%)
Gender	Male	1(50%)	2(40%)	0 (0%)	2 (100%)
	Female	0 (0%)	2(40%)	2 (100%)	0 (0%)
	NA	1(50%)	1(20%)	0 (0%)	0 (0%)
Weight	Group 1	0 (0%)	3 (60%)	2(100%)	0 (0%)
	Group 2	0 (0%)	2 (40%)	0 (0%)	0 (0%)
	Group 3	2 (100%)	0 (0%)	0 (0%)	2(100%)

I can do this using two separate procedures:

proc means data=have maxdec=2;
var age;
class height_group;
run;

proc freq data = have;
table (age_group gender weight_group) * height_group;
run;

I am looking for a way to do this in a single step.

Re: creating table

Ksharp — Mon, 30 Nov 2020 11:33:50 GMT

Check this marco.

https://communities.sas.com/t5/SAS-Communities-Library/Demographic-Table-and-Subgroup-Summary-Macro-TABLEN/ta-p/634030

Re: creating table

ballardw — Mon, 30 Nov 2020 15:52:28 GMT

@MEL20 wrote:

data have;
input Age Age_group $ Gender $ Weight_group $ Height_group $;
datalines;
11 1 Male 1 2
12 1 Male 1 3
12 1 Male 1 2
21 2 Male 1 3
14 1 NA 1 2
13 1 Female 2 2
23 2 Female 2 2
21 2 Female 3 1
24 2 Female 3 4
23 2 Female 3 4
13 1 NA 3 1
;
run;

I am trying to get a table in following format:

Characteristic

Height group

Group 1

Group 2

Group 3

Group 4

Age

Mean

(standard deviation)

17.00

(5.66)

14.60

(4.83)

16.50

(6.36)

23.50 (0.71)

Age

Group 1

1 (50%)

4 (80%)

1 (50%)

0 (0%)

Group 2

1 (50%)

1 (20%)

1 (50%)

1 (100%)

Gender

Male

1(50%)

2(40%)

0 (0%)

2 (100%)

Female

0 (0%)

2(40%)

2 (100%)

0 (0%)

NA

1(50%)

1(20%)

0 (0%)

0 (0%)

Weight

Group 1

0 (0%)

3 (60%)

2(100%)

0 (0%)

Group 2

0 (0%)

2 (40%)

0 (0%)

0 (0%)

Group 3

2 (100%)

0 (0%)

0 (0%)

2(100%)

I can do this using two separate procedures:

proc means data=have maxdec=2;
var age;
class height_group;
run;

proc freq data = have;
table (age_group gender weight_group) * height_group;
run;

I am looking for a way to do this in a single step.

None of the SAS statistical or report procedures will place two statistics into a single cell. I can get values as shown but not the layout with something like this as a "single step".

proc tabulate data=have;
  class age_group gender weight_group height_group;
  var age;
  table age *(mean std)
        (age_group
        gender
        weight_group)*(n colpctn)
       , height_group
       /misstext='0'
  ;
run;

Up to you to provide labels and maybe a custom format to show "Group x" instead of the value provided.