Mark is right, pie charts have a lot of problems and generally are considered bad practice.
Perhaps the second part of Mark's comment above was meant to be sarcastic. People who are serious about plots and have done the research on what is more understandable to a human mind, wouldn't necessarily assume that the above was better than a pie chart. Though marginally, it might be, as it's easier cognitively to compare the relative size of objects that are square vs objects that are at least partially circular (e.g., the slices you'd get in a pie chart). However, there's a better solution.
The problem with both pie charts and the graph above is that they are designed in a way that suggests that the most important aspect of the graph is the total amount (whether it's expressed in percentages or as a sum). If that were the case, you wouldn't need the graph at all. Just a summary statistic like total %, a mean, or total sum.
The real intended purpose of both the pie chart and the graph above is to understand the relative differences between the categories, NOT the total. So a simple solution is to just do a bar graph. In the case of the stacked bar graph above, you just need to "unstack it". I think graphs like the one you point to above are a sad byproduct of people wanting to make their results look more visually appealing and cool rather than thinking about the best way to communicate the information. Bar graphs are fantastic for this sort of thing, it's just that they are so ubiquitous they've become "boring" to people.
Pie charts have been out of favor for quite awhile. See Tufte's The Visual Display of Quantitative Information (full pdf of book at http://www.colorado.edu/UCB/AcademicAffairs/ArtsSciences/geography/foote/maps/assign/reading/TufteCoversheet.pdf ) came out in 1983. Tufte would argue that no chart at all should be used for 5 data points, and I tend to agree. See also http://junkcharts.typepad.com/ for on-going examples of bad charts curated by a Tufteite.
Hi, at least the top graph is understandable whereas the NBC2 one is just fraudulent if you compare the percentages to the bar heights, I guess the responses weren't what the network expected. cheers another phil
Pie charts have fallen out of favor in recent years. This would be considered a good plot by some.
ReplyDeletehttps://www.geckoboard.com/blog/pie-charts/#.V4hbp7x-yrU
Mark is right, pie charts have a lot of problems and generally are considered bad practice.
ReplyDeletePerhaps the second part of Mark's comment above was meant to be sarcastic. People who are serious about plots and have done the research on what is more understandable to a human mind, wouldn't necessarily assume that the above was better than a pie chart. Though marginally, it might be, as it's easier cognitively to compare the relative size of objects that are square vs objects that are at least partially circular (e.g., the slices you'd get in a pie chart). However, there's a better solution.
The problem with both pie charts and the graph above is that they are designed in a way that suggests that the most important aspect of the graph is the total amount (whether it's expressed in percentages or as a sum). If that were the case, you wouldn't need the graph at all. Just a summary statistic like total %, a mean, or total sum.
The real intended purpose of both the pie chart and the graph above is to understand the relative differences between the categories, NOT the total. So a simple solution is to just do a bar graph. In the case of the stacked bar graph above, you just need to "unstack it". I think graphs like the one you point to above are a sad byproduct of people wanting to make their results look more visually appealing and cool rather than thinking about the best way to communicate the information. Bar graphs are fantastic for this sort of thing, it's just that they are so ubiquitous they've become "boring" to people.
Pie charts have been out of favor for quite awhile. See Tufte's The Visual Display of Quantitative Information (full pdf of book at http://www.colorado.edu/UCB/AcademicAffairs/ArtsSciences/geography/foote/maps/assign/reading/TufteCoversheet.pdf ) came out in 1983. Tufte would argue that no chart at all should be used for 5 data points, and I tend to agree. See also http://junkcharts.typepad.com/ for on-going examples of bad charts curated by a Tufteite.
ReplyDeleteHi,
ReplyDeleteat least the top graph is understandable whereas the NBC2 one is just fraudulent if you compare the percentages to the bar heights, I guess the responses weren't what the network expected.
cheers another phil