Jump to content

User:PeterLFlomPhD/Dynamite plot sandbox

From Wikipedia, the free encyclopedia

A dynamite plot is a variation on a bar plot in which lines and whiskers are added to indicate either the standard deviation, the standard error, a confidence interval or some other measure of dispersion. Sometimes the lines go only upward from the bar, other times they go in both directions. Dynamite plots are not a standard statistical graphic - they are not mentioned in Cleveland's books [1], [2] nor in Everitt's Dictionary of Statistics. [3] but they are commonly used [4]

Uses

[edit]

Critiques

[edit]

Dynamite plots have been critiqued for

  1. Providing too little information, [4] scoring low on Edward Tufte's data-ink ratio.
  2. Not showing sample sizes [4]
  3. Not showing the data [4] not following Cleveland's rule to show the data. [1]: 29 

Alternatives

[edit]

Recommended alternatives depend on the sample size, both overall and in each bar.

When N is small

[edit]

When N is large

[edit]

Implementation

[edit]
  1. Implementation in R
  2. Implementation in SAS

References

[edit]

Short items

[edit]

Long items

[edit]

[5] [6] [7]

  1. ^ a b Cleveland, William (1994). The elements of graphing data. Murray Hill, N.J: AT & T Bell Laboratories. ISBN 0963488414.
  2. ^ Cleveland, William (1993). Visualizing data. Murray Hill, N.J. Summit, N.J: At & T Bell Laboratories Published by Hobart Press. ISBN 0963488406.
  3. ^ Everitt, Brian (1998). The Cambridge Dictionary of Statistics. Cambridge, UK New York: Cambridge University Press. ISBN 0521593468.
  4. ^ a b c d http://biostat.mc.vanderbilt.edu/wiki/pub/Main/TatsukiKoyama/Poster3.pdf
  5. ^ http://biostat.mc.vanderbilt.edu/wiki/Main/DynamitePlots
  6. ^ http://emdbolker.wikidot.com/blog:dynamite
  7. ^ http://biostat.mc.vanderbilt.edu/wiki/pub/Main/TatsukiKoyama/Poster3.pdf