Hi Maria,
If you're using an older version of the multcompare script that doesn't provide p-values, you can still perform multiple comparison tests with Bonferroni adjustment by manually conducting pairwise comparisons between your groups and adjusting the resulting p-values. The typical approach is to use a two-sample t-test (ttest2) for each pair of groups if your data are approximately normally distributed; otherwise, you can use a nonparametric test like the Wilcoxon rank-sum test (ranksum). After calculating the p-values for all possible group pairs, you apply the Bonferroni correction by multiplying each p-value by the total number of comparisons. Here’s how you can do this in MATLAB:
groups = unique(G); % G: vector of group labels
k = numel(groups); % number of groups
m = k*(k-1)/2; % total number of pairwise comparisons
pvals = [];
pairs = [];
for i = 1:k-1
for j = i+1:k
x1 = X(G==groups(i)); % X: data vector
x2 = X(G==groups(j));
[~, p] = ttest2(x1, x2); % or use ranksum(x1, x2) if nonparametric
pvals(end+1,1) = min(1, p*m); % Bonferroni-adjusted p-value
pairs(end+1,:) = [groups(i) groups(j)];
end
end
result = table(pairs(:,1), pairs(:,2), pvals, 'VariableNames', {'Group1','Group2','Bonf_p'});
disp(result)
This code will output a table showing each pair of groups and their Bonferroni-adjusted p-values, allowing you to identify which group differences are statistically significant after correction.
