{"id":3604,"date":"2014-02-01T22:20:26","date_gmt":"2014-02-02T03:20:26","guid":{"rendered":"http:\/\/labrigger.com\/blog\/?p=3604"},"modified":"2014-02-02T11:44:09","modified_gmt":"2014-02-02T16:44:09","slug":"box-plots-vs-bar-charts","status":"publish","type":"post","link":"http:\/\/labrigger.com\/blog\/2014\/02\/01\/box-plots-vs-bar-charts\/","title":{"rendered":"Box plots vs. bar charts"},"content":{"rendered":"

\"box\"<\/a><\/p>\n

Nature Methods has a special on box plots<\/a>, and in particular, the web app BoxPlotR<\/a>.<\/p>\n

Box plots are great. However, the conventions for box plots are not completely uniform (see below), and that can lead to confusion and make it take longer for a general audience to interpret the graphical representation of the data and understand the story it tells. And furthermore, it’s usually pretty simple to supplement a humble bar chart (mean +\/- standard error or standard deviation) with a plot of all of the data points, so the reader can see the distribution.<\/strong> In 1969, when Tukey<\/a> came up with the box plot, we didn’t have the fast and powerful graphing tools that we have today. Even in the cases of large sample sizes, where it’s not practical to plot every point, a histogram can still provide more visual information than a box plot. For example, if the distribution appears bimodal, this is immediately obvious in a histogram, but not so in a box plot (nor a bar chart, of course).<\/p>\n

What do you think?<\/p>\n

Some features of box plots are always the same, others aren’t<\/h4>\n

Center bar<\/em><\/strong>
\nAlways the median.<\/p>\n

Top and bottom of the box<\/em><\/strong>
\nAlways the first and third quartiles.<\/p>\n

Whiskers<\/em><\/strong>
\nSometimes<\/strong> the max and min values, sometimes<\/strong> some extreme percentiles (e.g., 9th and 91st, or 2nd and 98th percentiles) to exclude the influence of extreme outliers, sometimes<\/strong> the whiskers are based on standard deviation, and sometimes<\/strong> none of the above.<\/p>\n

Data markers<\/em><\/strong>
\nRelated to the whiskers, if there are data points outside of the whiskers, they are sometimes<\/strong> drawn in the box plot, but sometimes<\/strong> not. And sometimes<\/strong> they’re outliers, and sometimes<\/strong> they’re just the extreme tails of the sample.<\/p>\n

Width<\/em><\/strong>
\nSometimes<\/strong> the width of the box plot is used to indicate sample size, sometimes<\/strong> not.<\/p>\n

Notches<\/em><\/strong>
\nSometimes<\/strong> notches are used to give a visual cue as to the potential significance for the difference between two means, sometimes<\/strong> not. However, the actual math behind the notches, is not as trivial as looking for overlapping standard error bars, which serves a similar purpose. Neither is perfect of course, but it’s easy to see the appeal of the simpler solution.<\/p>\n","protected":false},"excerpt":{"rendered":"

<\/a><\/p>\n

Nature Methods has a special on box plots<\/a>, and in particular, the web app BoxPlotR<\/a>.<\/p>\n

Box plots are great. However, the conventions for box plots are not completely uniform (see below), and that…<\/p>\n

Read More<\/a><\/div><\/p>\n","protected":false},"author":1,"featured_media":3618,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[7],"tags":[25,24],"class_list":["post-3604","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-software","tag-analysis","tag-dissemination"],"_links":{"self":[{"href":"http:\/\/labrigger.com\/blog\/wp-json\/wp\/v2\/posts\/3604","targetHints":{"allow":["GET"]}}],"collection":[{"href":"http:\/\/labrigger.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"http:\/\/labrigger.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"http:\/\/labrigger.com\/blog\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"http:\/\/labrigger.com\/blog\/wp-json\/wp\/v2\/comments?post=3604"}],"version-history":[{"count":17,"href":"http:\/\/labrigger.com\/blog\/wp-json\/wp\/v2\/posts\/3604\/revisions"}],"predecessor-version":[{"id":3622,"href":"http:\/\/labrigger.com\/blog\/wp-json\/wp\/v2\/posts\/3604\/revisions\/3622"}],"wp:featuredmedia":[{"embeddable":true,"href":"http:\/\/labrigger.com\/blog\/wp-json\/wp\/v2\/media\/3618"}],"wp:attachment":[{"href":"http:\/\/labrigger.com\/blog\/wp-json\/wp\/v2\/media?parent=3604"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"http:\/\/labrigger.com\/blog\/wp-json\/wp\/v2\/categories?post=3604"},{"taxonomy":"post_tag","embeddable":true,"href":"http:\/\/labrigger.com\/blog\/wp-json\/wp\/v2\/tags?post=3604"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}