Thursday, December 2, 2010

Lies, Truth, and the Graphs in between

I really like the Flowing Data website/blog.  The blog is a wonderful presentation of the myriad and often intriguing ways of presenting data.  Sometimes data I really could care less about, but sometimes data that gives me a chance to look at what's meaningful in the method of presentation.

Lisa's course (PubH 6172, IH Applications) included a lecture on data presentation.  It was the first time I had a coherent, planned discussion about the whole matter.  One can look at something and say "that's not very clear" ... but why?  how do I improve it?  Have I lost the whole point I'm trying to make?  Could it be presented with a different style of graph to make it better?  Is it a 2"x2" picture in the paper which, in fact, only has 3 data points amidst the colorful image? (data density)

This is not only applicable to the NYTimes or Strib.  I'm trying to figure out how to present the data for my masters' paper.  The question is not only what to present ... the questions to ask are:
Do the graphs have enough data?
Do they have too much data?
Can I show the graphs to someone without the text of the thesis and explain my research?

The books Lies, Damned Lies, and Statistics and How to Lie with Statistics are some of many books on the topic of being deceived by statistics (intentionally or not).  If you are actively trying to avoid being lied to - or - are trying to assure you don't inadvertently deceive someone ... learning all you can about the topic is critical.
.