topic Re: Generating Indicator Variables using age ranges in SAS Data Management

Generating Indicator Variables using age ranges

bgosiker — Sat, 06 Jul 2019 16:13:39 GMT

Hi Everyone,

I have a long dataset (multiple rows per person), and each row has a "start age" and "end age" to indicate the 0.5 year age range that the row is meant to represent. Ie someone would have age_start=20 and age_end=20.5. My LOWEST age_start is 17.5 and my highest age_end is 93, so there is at least one person for each 0.5 year increment between 17.5 and 93.

I am wanting to generate time indicators based on these age ranges so that I can align everyone on the basis of age. Some people have gaps in visit (note, ID=2 is missing visit=3, so there is a gap!). Basically starting with age_start=17.5 I would want some indicator variable (let's say age_X) to be age_X=0 and then increase by 1 for each progressive 0.5 years (ie for age_start=18 age_0=0 and age_1=1). So each row should only have ONE age_x variable that is 1 and all the others would be 0. There should be 151 indicator variables generated from this for each 0.5 year range between 17.5 and 93.

I think an array would be the best approach here, but was unsure how to get it to be based specifically on the age_start variable since each person can start at a different age and there can be gaps. Any help is appreciated!

An example of the data I have is below:

Data have;
 input ID  visit age_start age_end ;
datalines;1  1  17.5  18
1  2  18  18.5
1  3  18.5  19
2  1  25  25.5
2  2  25.5 26
2  4  26.5  27
3  1  51.5  52
3  2  52  52.5
3  3  52.5  53;
run;

Data I WANT:

id visit age_start age_end age0 age1 age2..... age15 age16 age17 age18......age68 age69 ag70.......

1 1 17.5 18 1 0 0 0 0 0 0 0 0 0

1 2 18 18.5 0 1 0 0 0 0 0 0 0 0

1 3 18.5 19 0 0 1 0 0 0 0 0 0 0

2 1 25 25.5 0 0 0 1 0 0 0 0 0 0

2 2 25.5 26 0 0 0 0 1 0 0 0 0 0

2 4 26.5 27 0 0 0 0 0 0 1 0 0 0

3 1 51.5 52 0 0 0 0 0 0 0 1 0 0

3 2 52 52.5 0 0 0 0 0 0 0 0 1 0

3 3 52.5 53 0 0 0 0 0 0 0 0 0 1

Re: Generating Indicator Variables using age ranges

arthurcavila — Sat, 06 Jul 2019 17:10:02 GMT

My first question would be why do you need those dummy variables? Procedures with a CLASS statement will generate the dummies you need internally.

Data have;
 input ID  visit age_start age_end ;
 array age{151} age1-age151;
 do i = 1 to 151;
 	if i = (age_start-17)*2 then age{i} = 1;
 	else age{i} = 0;
 end;
 drop i;
datalines;
1  1  17.5  18
1  2  18  18.5
1  3  18.5  19
2  1  25  25.5
2  2  25.5 26
2  4  26.5  27
3  1  51.5  52
3  2  52  52.5
3  3  52.5  53
;
run;

Re: Generating Indicator Variables using age ranges

bgosiker — Sun, 07 Jul 2019 21:01:18 GMT

I was needing to export my data to run analyses in Stata so thought I needed to generate the indicator variables for it. I tried running the code you gave and ran into an issue where the data step wasn't finishing and eventually had to break it.

Re: Generating Indicator Variables using age ranges

SASKiwi — Sun, 07 Jul 2019 21:18:02 GMT

@bgosiker - the program as posted by @arthurcavila works fine for me. You have obviously changed it so post the code you are trying to run.

Re: Generating Indicator Variables using age ranges

PaigeMiller — Sun, 07 Jul 2019 21:29:59 GMT

Doesn't STATA generate indicator variables when you try to fit a model? Doesn't SAS give the ability to run the same models?

Re: Generating Indicator Variables using age ranges

bgosiker — Sun, 07 Jul 2019 22:18:19 GMT

Dumb mistake - left off the run when I copied over!

Re: Generating Indicator Variables using age ranges

bgosiker — Sun, 07 Jul 2019 22:19:17 GMT

I wasn't sure if stata did and I'll be switching from a PC (with SAS) to a mac (with stata) in the coming days and wanted to make sure I was prepared if it didn't. If it does then that is great!