I often get asked how this visualisation works, and in particular what the blue and darker grey bars do. “They’re just a measure of spread” is the quick answer. A longer answer is that their calculations can be inferred from other clues in the UI – not 100% helpful!
Each of the scores – 2.7, 2.9, and 3.4 in this example – is an interquartile mean, sometimes IQM or midmean. To calculate one of these, we take all the relevant scores, sort them, discard the top and bottom quartiles (and along with them any outliers), and calculate the mean of the remaining data points. It is described as a robust statistic, one that is not easily influenced by errors.
Looking at prompt 4.1 in the picture, we can say informally that the “average of the middle half” of the scores given to this prompt is 2.7 (on a scale of 1 to 4). We might guess that the majority answers lie between 2 and 3, with more 3’s than 2’s. Not “nailing it” yet, but “getting there”.
The calculations for the bars are very similar, but here we do want to be influenced by the extremes. The left and right ends of the darker grey bars show the mean of the bottom and top quartiles respectively, the most extreme answers at the low and high ends of the scale. Notice that for prompts 4.2 and 4.3, these bars extend all the way to the right. At a glance, we know that at least a quarter of the scores here were 4’s (since their mean is 4, and there can be no scores higher than 4).
The blue bars are also influenced by extremes, but moderated by more typical scores. The left and right ends here show the mean of the bottom and top halves respectively. Looking at prompt 4.3, we can infer that nearly half the scores here were 4’s. Awesome!