Mergedlegend statement not merging legends

[Gregg_G]

I am trying to merge two legends using the mergedlegend statement in GTL, but I keep getting the first set of plots merged (regression line and individual data points), but the second set is still split into two individual legend entries.  Does anyone have any idea why this is happening?  I have been over and over the code.

I can't give you the data set, but you can pick any set of x's and y's to input.

Thanks!

Gregg

Code:

proc template; define style styles.mydefaultstyle;

  parent = styles.journal3;

  *style GraphAxisLines / linethickness = 2px;

  style GraphWalls / LineThickness = 1px FrameBorder = off;

  *style GraphError / LineThickness = 2px;

  *style GraphDataDefault / LineThickness = 2px MarkerSize =14px;

  style graphdata1 / Markersymbol = "trianglefilled" LineStyle = 1

  Color = black ContrastColor = black;

  style graphdata2 / Markersymbol = "circle" LineStyle = 2

  Color = black ContrastColor = black;

  end;

run;

Proc Template; Define statgraph plasmaconc2;

  begingraph;

       entrytitle "";

       layout lattice / columns=1 rows=1;

       layout OVERLAY /

            XAxisOpts = (display = (line ticks tickvalues))    

            YAxisOpts = (display = (line ticks tickvalues))    

            walldisplay=(fill);

                 scatterplot x=logminute y=logresponse/ group=trt name = 'scatter';

                 regressionplot x=logminute y=logresponse/ group=trt degree=2 lineattrs = (thickness=1) name='regression';

                 mergedlegend 'regression' 'scatter'/

                      across=1 location=inside autoalign=(topright)

                      border = false

                      valueattrs = (size=12pt);

                 endlayout;

                 sidebar / align=left;

                      entry "Log plasma propofol concentration (ng ml"{sup "-1"}")" / rotate = 90;

                 endsidebar;

                 sidebar / align=bottom;

                      entry "Log time (minutes post-induction)";

                 endsidebar;

            endlayout;

       endgraph;

end; run;

quit;

ods graphics / reset antialiasmax=1300;

ods listing style = mydefaultstyle

  gpath = "Z:\Desktop\testgraphics"; goptions device=png300;

proc sgrender data = concd template = plasmaconc2; run;

[Reeza]

I think it works for me in SAS 9.3, what version of SAS are you on?

[data_null__]

Or better still you could include a data step to generate a set of any x's and y's.  Don't make the forum have to work so hard to help you.

[Jay54]

As 'data_null' says, it is easier to help if you include a full working program.  Your code, modified using class data set works:

Proc Template; Define statgraph plasmaconc2;

  begingraph;

       entrytitle "";

       layout lattice / columns=1 rows=1;

       layout OVERLAY /

            XAxisOpts = (display = (line ticks tickvalues))    

            YAxisOpts = (display = (line ticks tickvalues))    

            walldisplay=(fill);

                 scatterplot x=height y=weight/ group=sex name = 'scatter';

                 regressionplot x=height y=weight/ group=sex  degree=2 lineattrs = (thickness=1) name='regression';

                 mergedlegend 'regression' 'scatter'/

                      across=1 location=inside autoalign=(topright)

                      border = false

                      valueattrs = (size=12pt);

                 endlayout;

                 sidebar / align=left;

                      entry "Log plasma propofol concentration (ng ml"{sup "-1"}")" / rotate = 90;

                 endsidebar;

                 sidebar / align=bottom;

                      entry "Log time (minutes post-induction)";

                 endsidebar;

            endlayout;

       endgraph;

end; run;

quit;

proc sort data=sashelp.class out=class;
by sex;
run;
ods listing;
proc sgrender data = class template = plasmaconc2; run;

[Gregg_G]

I appreciate the help - and I agree that including the data set would be helpful, but as it is already in publication, I cannot disclose it, even out of context.  When I tried to create my own data, the code works fine.  But, when I use the actual dataset, I get this (note the legend):

and the original data looks no different other than the numbers....   Is there anything in the code that might give me his sort of result?

[Gregg_G]

The original dataset looks like this (but longer), if that helps - I am assuming that there is something in the  data, not the code?

Obs	cat	trt	wt	minute	response	logminute	logresponse
1	1	P28	7.7	2	15899.88	0.30103	4.20139
2	1	P28	7.7	5	11304.92	0.69897	4.05327
3	1	P28	7.7	10	6959.90	1.00000	3.84260
4	1	P28	7.7	20	4612.61	1.30103	3.66395
5	1	P28	7.7	30	4291.60	1.47712	3.63262
6	1	P28	7.7	60	1359.31	1.77815	3.13332
7	1	P28	7.7	120	233.30	2.07918	2.36791
8	1	P28	7.7	240	183.00	2.38021	2.26245
9	1	P28	7.7	360	146.00	2.55630	2.16435
10	1	P28	7.7	720	58.60	2.85733	1.76789
11	1	P28	7.7	1440	.	3.15836	.
12	2	P28	4.2	2	7893.48	0.30103	3.89727
13	2	P28	4.2	5	5095.32	0.69897	3.70717
14	2	P28	4.2	10	3149.43	1.00000	3.49823
15	2	P28	4.2	20	4905.62	1.30103	3.69069

[Jay54]

It likelymeans that the values for TRT are not matching.  "P28" is matching, so those are merged.  But there are other values in the TRT column that are not matching.  They may look like "P", but are somehow different.

[Gregg_G]

Thanks for all the help.  I will go back and check the data set to see what went wrong.  The code appears correct for what I want done.

I will include a data set in further posts as requested - thanks for the etiquette tip, I am new around here!

[SASSLICK001]

Sanjay

I have the following

      seriesplot  x=eval(atptn-13*&offx) y=mean1 /   lineattrs=graphdata1(color=blue pattern=shortdash thickness=2) name='a1' legendlabel="Drug A";

      scatterplot x=eval(atptn-13*&offx) y=mean1/     YErrorUpper=High1 YErrorLower=Low1

                  markerattrs=graphdata1(color=blue symbol=trianglefilled size=9) errorbarattrs=graphdata1(color=blue pattern=solid)

                  name='a2' legendlabel="Drug A";

      seriesplot  x=eval(atptn-11*&offx) y=mean2 /   lineattrs=graphdata2(color=red pattern=mediumdash  thickness=2) name='b1' legendlabel="Drug B";

      scatterplot x=eval(atptn-11*&offx) y=mean2 /    YErrorUpper=High2 YErrorLower=Low2

                  markerattrs=graphdata2(color=red symbol=starfilled size=9) errorbarattrs=graphdata2(color=red pattern=solid)

                  name='b2' legendlabel="Drug B";

How do I use Mergedlegend

Currentl Im using the following way

mergedlegend 'a1' 'a2' 'b1' 'b2 / across=1 location=inside halign=right valign=top;

But its not working !!