Sorry for the vague question but this chart appears in Biddle et al. 2009 and I've not encountered anything like it before. It's a bar chart with beveled edges, sometimes 'horns'. What do these mean? Does this type of chart have a name?

Per https://meta.stackexchange.com/questions/244083/site-for-asking-about-charts I thought Academia was the best place to ask.

enter image description here

  • $\begingroup$ You might gain further understanding of the different types of boxplots by using this link that allows you to create boxplots with a built in sample of data or by uploading your own data. $\endgroup$ – FTF Sep 9 '15 at 22:14
  • 2
    $\begingroup$ The top answer to that meta.SE question makes it abundantly clear that questions about charts do not belong on academia.SE. I don't know how you could read that post and conclude anything else. That the asker says he's seen people ask about charts there doesn't make it on topic there. The question asks where they belong and the top answer addresses that fairly well (though it misses that data viz. is on topic here) $\endgroup$ – Glen_b Sep 9 '15 at 23:06

These are notched boxplots. The standard boxplot summarises a distribution in terms of interquartile range (IQR) and median, with 'whiskers' extending to either the full range or some heuristic multiple of IQR after which "outliers" are plotted as points.

The notches show a non-parametric estimate of ~95% confidence intervals of the median. The reference most often given for these is "Variations of boxplots" by McGill, Tukey and Larsen (1978). A recent reference is Visualizing samples with box plots in Nature methods.

Most statistical packages will let you add these as an option when drawing boxplots, for example in R the boxplot command has the argument notch. Note that there are various subtly different methods of calculating the notches which may vary between implementations.

  • $\begingroup$ Some use "Notched Box Whisker plot", to add the whisker part. $\endgroup$ – Laurent Duval Sep 13 '15 at 13:55

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy