help please :) adding confidence intervals to SGplot after repeated measures Proc Glm?

[GabbyJ]

Hi,

Could you kindly help me?

I would like to add confidence intervals to a graph following proc GLM with data from multiple timepoints.

I calculated the means in each group, but not CIs. Could you teach me how to calculate and graph the CIs?

Or would it be easier to use a mixed model?

Data, code, and sample graph are below. 

Thank you so much 🙂

proc glm data=exercise;
  class exertype;
  model time1 time2 time3 = exertype;
  repeated time 3 ;
  lsmeans exertype / out=means;
run;
quit;
proc print data=means;
run;

proc sgplot data=means;
	series x=_name_ y=lsmean / group=exertype;
run;
quit;

sample data:

data exercise;
  input id exertype diet time1 time2 time3;
cards;
1     1         1       85       85       88
2     1         1       90       92       93
3     1         1       97       97       94
4     1         1       80       82       83
5     1         1       91       92       91
6     1         2       83       83       84
7     1         2       87       88       90
8     1         2       92       94       95
9     1         2       97       99       96
10    1         2      100       97      100
11    2         1       86       86       84
12    2         1       93      103      104
13    2         1       90       92       93
14    2         1       95       96      100
15    2         1       89       96       95
16    2         2       84       86       89
17    2         2      103      109       90
18    2         2       92       96      101
19    2         2       97       98      100
20    2         2      102      104      103
21    3         1       93       98      110
22    3         1       98      104      112
23    3         1       98      105       99
24    3         1       87      132      120
25    3         1       94      110      116
26    3         2       95      126      143
27    3         2      100      126      140
28    3         2      103      124      140
29    3         2       94      135      130
30    3         2       99      111      150
;
run;

[Reeza]

Start by adding the CL option to the LSMEANS statement and that will give you the confidence limits to plot. 

https://documentation.sas.com/?docsetId=statug&docsetTarget=statug_glm_syntax10.htm&docsetVersion=15...

proc glm data=exercise;
  class exertype;
  model time1 time2 time3 = exertype;
  repeated time 3 ;
  lsmeans exertype / out=means CL;
run;

Then you can add them with a BAND statement. I don't know what the exact variables are that come out of the LSMEANS statement so you'll need to modify the code to take the correct names.

proc sgplot data=means;
	series x=_name_ y=lsmean / group=exertype;
        band x = _name_ upper=upper_limit lower=lower_limit  / fillattrs=(color=graydd);
run;
quit;

[GabbyJ]

Dear Reeza,

Thank you so much for your reply 🙂

The standard error comes out from the CL option in lsmeans, and I converted it to upper and lower CI's. However, when I graph the upper and lower CI's using band, something doesn't quite work. Could you kindly help? Thank you again! I very much appreciate your help 🙂 Code, data, and graph are below.

proc glm data=exercise;
  class exertype;
  model time1 time2 time3 = exertype;
  repeated time 3 ;
  lsmeans exertype / out=means CL pdiff;
run;
quit;

data means;
set means; 
upperci = lsmean + (1.96*STDERR);
lowerci = lsmean - (1.96*STDERR);
run;

proc print data=means;
run;

proc sgplot data=means;
	series x=_name_ y=lsmean / group=exertype;
    band x = _name_ upper=upperci lower=lowerci  / fillattrs=(color=graydd);
run;
quit;

Sample of "mean" data:

 Obs _NAME_ exertype LSMEAN STDERR upperci lowerci123456789

time1	1	90.2	1.84932	93.825	86.575
time1	2	93.1	1.84932	96.725	89.475
time1	3	96.1	1.84932	99.725	92.475
time2	1	90.9	2.95284	96.688	85.112
time2	2	96.6	2.95284	102.388	90.812
time2	3	117.1	2.95284	122.888	111.312
time3	1	91.4	3.47227	98.206	84.594
time3	2	95.9	3.47227	102.706	89.094
time3	3	126.0	3.47227	132.806	119.194

[GabbyJ]

Dear Reeza,

Thank you so much for your help and reply! The CI option output the standard error (which I converted to CI's) and tried to add them to the graph with the band function. The graph is not looking quite right. Could you kindly advise? Thank you again! 🙂

proc glm data=exercise;
  class exertype;
  model time1 time2 time3 = exertype;
  repeated time 3 ;
  lsmeans exertype / out=means CL;
run;
quit;

data means;
set means; 
upperci = lsmean + (1.96*STDERR);
lowerci = lsmean - (1.96*STDERR);
run;

proc print data=means;
run;

proc sgplot data=means;
	series x=_name_ y=lsmean / group=exertype;
    band x = _name_ upper=upperci lower=lowerci  / fillattrs=(color=graydd);
run;
quit;

 Obs _NAME_ exertype LSMEAN STDERR upperci lowerci123456789

time1	1	90.2	1.84932	93.825	86.575
time1	2	93.1	1.84932	96.725	89.475
time1	3	96.1	1.84932	99.725	92.475
time2	1	90.9	2.95284	96.688	85.112
time2	2	96.6	2.95284	102.388	90.812
time2	3	117.1	2.95284	122.888	111.312
time3	1	91.4	3.47227	98.206	84.594
time3	2	95.9	3.47227	102.706	89.094
time3	3	126.0	3.47227	132.806	119.194

---

[GabbyJ · Accepted solution]

I just figured it out, thank you Reeza for your kind help 🙂 I added the group function in the band command.

proc sgplot data=means;
	series x=_name_ y=lsmean / group=exertype ;
    band x = _name_ upper=upperci lower=lowerci  / fillattrs=(color=graydd) group=exertype transparency=.80;
run;
quit;