tag:blogger.com,1999:blog-7310946608587805029.post6102408472601211937..comments2024-03-13T01:48:29.943-04:00Comments on Unbiased Research: Reading Boxplot used to deal with skewed DataTJ Murphyhttp://www.blogger.com/profile/17292359594683490598noreply@blogger.comBlogger1125tag:blogger.com,1999:blog-7310946608587805029.post-63696872030570152202016-05-03T07:45:19.205-04:002016-05-03T07:45:19.205-04:00Box plots used to always confused me as well. Than...Box plots used to always confused me as well. Thank you for the detailed breakdown. It's clearly for me to see why one might want to use a boxplot instead of mean when data is very skewed. The inclusion of the median and the visual representation of the concentration of data is helpful with boxplots. One thing - I'm a little confused now at the substantial number of outliers listed above the box plots in the first graph. While I have no information about the graph, and thus don't know sample sizes, it seems to me that a large number of outliers are concentrated in regions far above the other sets of data. This got me thinking about a few questions. How many outliers is appropriate to have on a graph? At what point does a group of outliers constitute its own separate, valid cluster of responses? Is there a limit to the number of outliers one can include in data representation, or can the number be (theoretically) infinite? Jenniehttps://www.blogger.com/profile/11636361642099976352noreply@blogger.com