Heavy Lifting  thoughts and web finds by an economist 
I also contribute to Division of Labour  Load HL's Front Page 
Monday, June 06, 2005
When is a pie chart worthless?
When the sample size is 16? Here's the chart in question, from the Fourth Quarter 2004 Commercial Space Transportation report. The pie chart suggests that 6% of the space shots in the fourth quarter of 2004 were failures. Granted, 1/16 is approximately 6%. But what is the pie chart really supposed to convey? To me, the chart implies that there would be 6 failures if there had been 100 space shots rather than just 16. If the kid with the excel spreadsheet had just put 1 and 15 (and left out the percentages) I would have less of a problem with the graph. However, the kid did include the percentages and therefore the graph is basically worthless. For this reason, I don't like the use of percentages when the sample size is considerably less than 100. If we look at historical data from the United States, from 19962002 the U.S. had 160 earth orbit space shots and had a total of 5 failures (1 in 1996 and 4 in 1999) for a failure rate less than 2%. Here's a pie chart of earth orbit space shots since 1996: Is the original pie chart intended to convey that there was a higher failure rate than historical norms? Perhaps, but there is an integer problem: one failure becomes 6% but there couldn't have been a 2% failure rate. In my mind, graphs such as these just take up space but don't convey much information. I know it isn't that important, but it still drives me nuts.
Comments:
Post a Comment

Purchase
Le Chai  galerie du vin
Support
Popularity
Posts that contain Craig Depken per day for the last 90 days.
Contacts
Heavy Lifting's Main Page Email Me Atom Feed
Great Links
Money I Found Today Heavy Lifting  Firehose style (56k warning)
Recent Posts
 When is a pie chart worthless?
Archives
home
Visitors
Credits
Modified maystar design powered by blogger 