Box plots for Kruskal-Wallis

Hello,

I'm plotting Kruskal-Wallis results for my paper. I'm not sure how to interpret folded edges on the plot (shown with an arrow on the picture) and how to add mean lines to the plots. Thanks for help

Answer 1

[fred ssemwogerere]

Hello, when the end point of a notch is lower than the 25th quartile of your data, that folding effect will occur. This effect is common to small datasets or small group size (when between groups of data).

% To calculate the end points of the lower and upper notch
lowerNotch=median(data)-1.57*iqr(data)/sqrt(numel(data)); % iqr is interquartile range
upperNotch=median(data)+1.57*iqr(data)/sqrt(numel(data));
% For a small value of "n" the notch will tend to extend beyond either the 75th percentile or 25th percentile (the upper and lower ends of the box respectively.
% If you compute the 25th percentile of your data and compare it to the end point of the lower notch, you will
% find this to be true. However, that won't be the case with the upper notch based on your plot.

[fred ssemwogerere]

To add marker corresponding to the mean of your data, to a box plot, you can get the current or active axes of the plot, get x-tick labels, hold on, and then using scatter plot, add a marker for the mean of the data.

% for example
figure;boxplot(h); % "h" is a vector representing your data
handl=get(gca); % get current axes 
hold on;
scatter(handl.XTick,mean(h),[],'filled'); % plot the mean of your data to its corresponding x-label ("handl.XTick")
hold off

[Sylwia Kaduk]

Dear Fred,

thank you for the answers. They help a lot. In therms of box plot, I would like to compare means with Kruskal-Wallis test and wanted plots specifically for it. I'm afraid that if I just order boxplot it will use different statistcial test and hence have different results and error bars. Or is there a way to order boxplots with a paritcular statistical test results?

Many thanks

Sylwia

[fred ssemwogerere]

Hello, you do not need to call the boxplot function. You can add the mean values on the same boxplot generated by the statistical test. With the boxplot from your statistical test, just get the handle for the current axes.

handl=get(gca);
hold on;
% Compute the mean value for each data group.
% Am assuming your different data groups are arranged in different columns of one matrix called: "gpdata".
avg=mean(gpdata); % a row vector output showing means for different data groups
% get the "XTick" of the current axes, and plot the mean values to correspond to their respective XTickLabel
scatter(handl.XTick,avg,[],'filled');
hold off
% This should be able to show the mean values as circular filled blue (default) markers on the same box plot generated from the statistical test

[Sylwia Kaduk]

Wow, it worked. Thanks a lot, it's a big help