Illustration of intra-subject variability.

(Top) For each stimulus level x the distribution of proportion correct is plotted as a heat map based on 1000 templates created with the population average model (bold black sigmoid) and the intra-subject variability models using and . (Middle) The dashed black distribution curves correspond to the parametric log–normal and beta intra-subject variability models. The histograms (black outline) as well as the dashed black lines for the means show the parameter distributions of the 1000 templates including intra-subject variability. The white triangles indicate the threshold and slope of the population average model (without noise). As a comparison, the inherent method variability (histogram with gray fill) for the same population average model without intra-subject variability is plotted. (Bottom) The percentage of estimates within a tolerance interval (percentage within bounds, PCTw/iB) around the parameters of the population average model is plotted against the size of the interval (gray: method variability, black: intra-subject variability), together with the absolute difference of percentage (|Δ|, dashed black line). For both method and intra-subject variability, the normalized area under the curve (nAUC) is calculated.