post hoc tests for aoctool

Hi, I am running an ANCOVA using AOCtool and am not understanding the syntax for post hoc testing

[h,atab,ctab,stats]=aoctool(table_zero_some.var1,table_zero_some.var2,table_zero_some.group,0.05,'var1','var2','group')

I get

atab =

  5×6 cell array

    {'Source'          }    {'d.f.'}    {'Sum Sq'}    {'Mean Sq'}    {'F'       }    {'Prob>F'  }
    {'group'           }    {[   1]}    {[0.0869]}    {[ 0.0869]}    {[  0.1908]}    {[  0.6674]}
    {'activation'      }    {[   1]}    {[4.1257]}    {[ 4.1257]}    {[  9.0583]}    {[  0.0075]}
    {'group*activation'}    {[   1]}    {[0.9107]}    {[ 0.9107]}    {[  1.9996]}    {[  0.1744]}
    {'Error'           }    {[  18]}    {[8.1983]}    {[ 0.4555]}    {0×0 double}    {0×0 double}

I would like to use multcompare as I would in a standard ANOVA; however the matlab documentation says

multcompare uses stats.gnames(idx) as the control group

I'm not sure what this means or what the syntax is supposed to look like

This is the stats output:

stats = 

  struct with fields:

        source: 'aoctool'
        gnames: {2×1 cell}
             n: [2×1 double]
            df: 18
             s: 0.6749
         model: 5
        slopes: [2×1 double]
      slopecov: [2×2 double]
    intercepts: [2×1 double]
      intercov: [2×2 double]
           pmm: [2×1 double]
        pmmcov: [2×2 double]

Thanks!

Answer 1

[the cyclist]

After running aoctool and getting the stats output, you can simply run

[results,~,~,gnames] = multcompare(stats)

If you do that, then by default the first group listed in gnames will be the reference group. Alternatively, you can run

[results,~,~,gnames] = multcompare(stats,'ControlGroup',3)

if you want the 3rd group to be the reference group. (Different sources of the stats variable can have different methods of specifying the control group, but that is how you do it for aoctool, according to this documentation.)