How to plot multiple multcompare plots from anova

It doesn't work when I try to use tiledlayout to make a combined plot

% Loading data
y = [52.7 57.5 45.9 44.5 53.0 57.0 45.9 44.0]';
g1 = [1 2 1 2 1 2 1 2];
g2 = {'hi';'hi';'lo';'lo';'hi';'hi';'lo';'lo'};
g3 = {'may';'may';'may';'may';'june';'june';'june';'june'};

% For loop with subplot
for n = 1:4
    subplot(2,2,n)
    
    hold on
    [~,~,stats] = anovan(y,{g1 g2 g3},'model','interaction','varnames',{'g1','g2','g3'});
    results = multcompare(stats,'Dimension',[1 2]);
    hold off
    
end

And tiledlayout also doesn't work

tiledlayout('flow')

nexttile
[~,~,stats] = anovan(y,{g1 g2 g3},'model','interaction','varnames',{'g1','g2','g3'});
results = multcompare(stats,'Dimension',[1 2]);

nexttile
[~,~,stats] = anovan(y,{g1 g2 g3},'model','interaction','varnames',{'g1','g2','g3'});
results = multcompare(stats,'Dimension',[1 2]);

Answer 1

[Walter Roberson]

multcompare() calls meansgraph(), which always unconditionally does a clf() which erases the entire current figure.

Therefore to plot multiple such graphs on the same figure, you are going to need to provide a figure for multcompare() to scribble into, and then you are going to have to extract the axes from the figure and move it to the desired target location.

However, there is the hidden difficulty that the graphs are interactive, each line having a buttondownfcn callback. To support that, the axes is given UserData and also the figure is given UserData as well. If you had more than one kind of multcompare, such as mixing anova1 with kruskalwallis then you would have a conflict when you merged the axes into one figure, but you should be okay as long as they are all for the same kind of graph.

[Ebsa Eshete]

Oh, I remember what I did in the past to make it work, I had to delete that clf() line from meansgraph(). I think I will try and do that again, but if that doesn't work, I'll do the method you described.

In order to transfer the axes, would I do something like

figure(1)

[~,~,stats] = anovan(y,{g1 g2 g3},'model','interaction','varnames',{'g1','g2','g3'});
results = multcompare(stats,'Dimension',[1 2]);
ax1 = gca()

[~,~,stats] = anovan(y,{g1 g2 g3},'model','interaction','varnames',{'g1','g2','g3'});
results = multcompare(stats,'Dimension',[1 2]);
ax2 = gca()

tiledlayout('flow')

nexttile
ax1

nextile
ax1

[Walter Roberson]

realfig = figure();
tiledlayout(realfig, 'flow')
tempfig = figure();

figure(tempfig);     %make it current
[~,~,stats] = anovan(y,{g1 g2 g3},'model','interaction','varnames',{'g1','g2','g3'});
results = multcompare(stats,'Dimension',[1 2]);
tempax = tempfig.Children;
realax = nexttile;
set(tempax, 'Parent', realfig, 'Position', realax.Position);  %move it between figures
delete(realax);   %or maybe set it invisible

figure(tempfig);    %make it current
[~,~,stats] = anovan(y,{g1 g2 g3},'model','interaction','varnames',{'g1','g2','g3'});
results = multcompare(stats,'Dimension',[1 2]);
tempax = tempfig.Children;
realax = nexttile;
set(tempax, 'Parent', realfig, 'Position', realax.Position);  %move it between figures
delete(realax);   %or maybe set it invisible

[Ebsa Eshete]

Awesome, thanks for the help again!

[Barbab]

Hello, I am trying to do the same, unfortunately the code you provided above for the multcompare plots is not working in my case. It says "Error using matlab.graphics.axis. Axes/set Invalid or deleted object." Could you suggest another solution?

Is there any easy way to get the data directly from multcompare and to plot it directly how I would like it to be?