multiple comparisons test figure object selection

I have a done a multiple comparisons test using the command multcompare. Say for example I have the following image:

load carsmall
[p,t,stats] = anova1(MPG,Origin);

[c,m,h,gnames] = multcompare(stats);

I want to be able to automatically select which group to highlight blue via code, as I am trying to automatically save my figure with exportfigure.

I suspect the command to highlight the group blue is something like

f=gcf

f.Children.SelectObject=2

I am pretty sure I have been able to do it before, but I can't for the life of me remember what command I used, or where I put the code I used before.

I would greatly appreciate any help on this.

Thank you,

Luck

Answer 1

[the cyclist]

The only way I could see to do this is to set the control type to "dunnett", and set the control group.

load carsmall
[p,t,stats] = anova1(MPG,Origin);

% Set the desired control group name, and the index will then be found programatically
controlGroup = "Japan";
controlIndex = find(stats.gnames==controlGroup);

[c,m,h,gnames] = multcompare(stats,"CriticalValueType","dunnett","ControlGroup",controlIndex);

I admit that I don't understand why this method relinquishes the ability to interact with the chart to change the control group, or why the error bars on the control group go away.

[Luck Haviland]

Thank you for your contribution, the cyclist. While this is interesting, I require the standard deviation bars to stay there for my final pictures, so I cannot accept this as an answer.

My apologies,

Luck

[the cyclist]

Most of the multcompare code is written in MATLAB (as opposed to being buried in internal C code or whatever), so it is conceivable to me that you could dig into where this is happening, and make your own version. Be careful not to overwrite the original.