Left: Difference between the fitted (α, β, γ) and the input value when using a usual log-likelihood minimisation code, in σ. The stacked histograms corresponds to three different input values defined in the legend on the right. The black line is the model the histogram should follow if the results were unbiased (both values and error), which is a Gaussian with a null mean and unitary standard deviation. Right: Same as the left plot, but for the total-χ² method.