topic Re: Creating a new variable from multiple observation. in SAS Procedures

Creating a new variable from multiple observation.

ebonfil89 — Sat, 17 Feb 2018 21:31:10 GMT

Hello,

I have a data set that is composed of participants and sedentary time. I want to create a new varible that is the average sedentary time for each participant. Meaning that I only want one sedtime for each participant.

I need a new variable because it want to then regress the average sedentary time with other predictors variables.

I am running SAS on SAS studio and I am a beginnier. Any help would be welcome. Thanks

example data.

participant Sedtime

1 4

1 3

1 2

2 1

2 5

2 3

3 4

3 5

3 4

Re: Creating a new variable from multiple observation.

novinosrin — Sat, 17 Feb 2018 21:42:46 GMT

do you mean -

proc sql;
create table want as
select participant, mean(sedtime) as avg
from have
group by participant;
quit;

Re: Creating a new variable from multiple observation.

novinosrin — Sat, 17 Feb 2018 21:44:02 GMT

proc summary data=have nway;
class participant;
var sedtime;
output out=want(drop=_:) mean=/autoname;
run;

Re: Creating a new variable from multiple observation.

Reeza — Sat, 17 Feb 2018 21:45:06 GMT

You can use PROC MEANS as illustrated here:

https://github.com/statgeek/SAS-Tutorials/blob/master/proc_means_basic.sas

Or you can add it in to your overall data set as illustrated here:

https://github.com/statgeek/SAS-Tutorials/blob/master/add_average_value_to_dataset.sas

And there are video tutorials here if you prefer that:

http://video.sas.com/#category/videos/sas-analytics-u/0

Specifically:

http://video.sas.com/detail/videos/sas-analytics-u/video/3306906230001/summary-statistics-using-sas-studio?autoStart=true&page=1

Re: Creating a new variable from multiple observation.

ebonfil89 — Sat, 17 Feb 2018 22:16:42 GMT

Thank you for the information. In this procedure, what is the new variable that was created?

Re: Creating a new variable from multiple observation.

novinosrin — Sat, 17 Feb 2018 22:23:58 GMT

Please test the demo that we share so you will have a clear understanding:

data have;
input participant   Sedtime;
datalines;
1                     4
1                     3
1                     2
2                     1
2                      5
2                      3
3                      4
3                      5
3                      4
;

proc summary data=have nway;
class participant;
var sedtime;
output out=want(drop=_:) mean=/autoname;
run;

Results:

participant	Sedtime_Mean
1	3
2	3
3	4.333333333

Re: Creating a new variable from multiple observation.

ebonfil89 — Sun, 18 Feb 2018 19:35:40 GMT

Thank you so much, option two worked for me.

Proc means data = WORK.IMPORT noprint nway;
class PQ_Study_ID_Unique;
output out = avg_values mean (sed) = ave_sed;
run;

Proc sort data = work.import out = sort;
by PQ_Study_ID_Unique;
run;

data sort_sed;
merge sort avg_values;
run;

proc print data = sort_sed;
run;

Re: Creating a new variable from multiple observation.

Reeza — Sun, 18 Feb 2018 19:53:54 GMT

Your MERGE doesn’t have a BY statement. I would expect to see one if you’re doing group summaries. Are you sure you’re getting what you expect?

Re: Creating a new variable from multiple observation.

ebonfil89 — Mon, 19 Feb 2018 21:07:43 GMT

Hello, I have added the "by" statement. I think that the data was already sorted by ID, so it did give me the same answers. Thank you for your feedback. Very much appreciated it.

Re: Creating a new variable from multiple observation.

Reeza — Tue, 20 Feb 2018 16:39:50 GMT

@ebonfil89 wrote:
Hello, I have added the "by" statement. I think that the data was already sorted by ID, so it did give me the same answers. Thank you for your feedback. Very much appreciated it.

If that was the case it means you have a single ID for each record which means there's no point to doing a PROC MEANS. I would be very seriously double checking your results. Something isn't aligning between what you're stating and your code.