function plotBootstrapCoeffOfVariationTest(widthCell, figDir, threshold, rSquareThreshold, varargin)


if nargin > 4
    xLabels = varargin{1};
    nameSuffix = varargin{2};
else
    xLabels = {'X OFF', 'Y OFF', 'X ON', 'Y ON'};
    nameSuffix = '';
end



qualityMeasure = 'responseQuality';
[isSignificant, cvArray, ~, ...
 centralDecisionLineCV, ldlCV, udlCV] = bootstrapCoeffOfVariation(widthCell);
error = (udlCV - ldlCV) / 2;
colors = getONOFFXYColors;
if numel(widthCell) > 4
    colors = repmat(colors, ceil(numel(widthCell) / 4), 1);
end
% Fill space.
hFig = createPrintFig(5 * [1 0.7]);
dummyVec = ones(1, numel(widthCell) + 2);
hAx = gca;
[hLine, hPatch] = plotErrorPatch(hAx, 0: numel(widthCell) + 1, ...
                                 centralDecisionLineCV * dummyVec, ...
                                 error * dummyVec, ...
                                 [1 1 1] * 0.2, [1 1 1] * 0.85);
hLine.LineWidth = 2;
hPatch.FaceAlpha = 1;
for iGroup = 1: numel(widthCell)
    scatter(iGroup, cvArray(iGroup), 50, colors(iGroup, :), ...
            'MarkerFaceColor', colors(iGroup, :), ...
            'MarkerFaceAlpha', isSignificant(iGroup), ...
            'LineWidth', 2);
end

ylabel('$$CV=\frac{s}{\bar{x}}$$', 'Interpreter', 'latex', 'Rotation', 90)
hAx.XTick = 1: numel(widthCell);
hAx.XTickLabel = xLabels;
arrayfun(@prettifyAxes, [hAx]); 
arrayfun(@offsetAxes, hAx); 

setFontForThesis([hAx], hFig)

figFileName = ['fhwmCV-' nameSuffix qualityMeasure '-' num2str(threshold) ...
               '-rSquare-' num2str(rSquareThreshold)];
% set(gcf,'renderer','painters')
% set(gcf, 'PaperPositionMode', 'auto');
print(hFig, [figDir figFileName '.pdf'], '-dpdf')
print(hFig, [figDir figFileName '.png'], '-dpng', '-r300')

end



function [isSignificant, cvArray, varargout] = bootstrapCoeffOfVariation(widthCell)
%% Bootstrap analysis of CV.
cvArray = cellfun(@calcCoeffOfVariation, widthCell);
bootsCV = cellfun(@(x) bootstrp(5000, @calcCoeffOfVariation, x), widthCell, 'uni', 0);
bootsCVArray = cell2mat(bootsCV);
meanCVPerBoot = mean(bootsCVArray, 2);
centralDecisionLineCV =  mean(meanCVPerBoot);
standardErrorCV = sqrt(mean((meanCVPerBoot - centralDecisionLineCV) .^ 2));
alpha = 0.05;
lowerDecisionLineCV = centralDecisionLineCV - norminv(1 - alpha / 2) * standardErrorCV;
upperDecisionLineCV = centralDecisionLineCV + norminv(1 - alpha / 2) * standardErrorCV;
error = norminv(1 - alpha / 2) * standardErrorCV;
isSignificant = cvArray < lowerDecisionLineCV | cvArray > upperDecisionLineCV;
varargout{1} = bootsCVArray;
varargout{2} = centralDecisionLineCV;
varargout{3} = lowerDecisionLineCV;
varargout{4} = upperDecisionLineCV;

end