topic Re: Mean with SE bars over time: i am stuck.. Need help in ODS and Base Reporting

Mean with SE bars over time: i am stuck.. Need help

Dcicantab5 — Sat, 11 Jun 2016 14:12:40 GMT

I am on SAS studio,

trying to create a mean with SE bars over time ( hand hygiene compliance over time, 2 groups of wards ('0' composed of 6 wards with low bed occupancy rates and '1' composed of wards with high occupancy rates).

I referred to

Sample 42542: Plot means with standard error bars from calculated data with PROC SGPLOT

Supposed to get 2 pairs of dots & 2 lines with error bars. Didn't happen:

Where did it go wrong?

my codes:

proc means data=WORK.HANDHYGIENE3 mean stderr;

class grouphh month;

var findings;

output out=out1hh mean = mean stderr=stderr;

run;

data reshape(drop=stderr);

set out1hh;

lower=mean - stderr;

upper=mean + stderr;

run;

ods listing close;

ods graphics / reset width=600px height=400px imagefmt=gif;

proc sgplot data=reshape noautolegend;

scatter x=month y=mean / yerrorlower=lower

yerrorupper=upper

markerattrs=(color=blue symbol=CircleFilled);

series x=month y=mean / lineattrs=(color=blue pattern=2);

title1 'Plot Means with Standard Error Bars from Calculated Data';

run;

Re: Mean with SE bars over time: i am stuck.. Need help

Dcicantab5 — Sat, 11 Jun 2016 14:23:21 GMT

the data file

Re: Mean with SE bars over time: i am stuck.. Need help

jklaverstijn — Sat, 11 Jun 2016 15:40:27 GMT

Proc means will calculate stats for all permutations of the classification variables. Add the NWAY option to only keep the grouphh*month interactions. That will take care of the erratic lines in your plot.

proc means data=WORK.HANDHYGIENE3 mean stderr nway;

Hope this helps,

- Jan,

Re: Mean with SE bars over time: i am stuck.. Need help

Dcicantab5 — Sat, 11 Jun 2016 17:16:39 GMT

Re: Mean with SE bars over time: i am stuck.. Need help

jklaverstijn — Sat, 11 Jun 2016 18:09:57 GMT

Maybe you need to group the series by grouphh:

series x=month y=mean / lineattrs=(color=blue pattern=2) group=grouphh;

That worked for me.

Regards,

- Jan.