Friday, April 16, 2010

please step away from the calculator

More nonsense statistics hoping to prove something useless. My problem with this comes in the bar graph... which displays what appears to be a shocking difference in IQ between those who are religious and those who are not.

http://www.psychologytoday.com/blog/the-scientific-fundamentalist/201004/why-atheists-are-more-intelligent-the-religious

... but then I think... why is the y-axis for this graph chosen at '96-104'?

... well, it captures the data I guess... and the claimed error bars...
but why not start at '97'? that would make the difference look even more magnified in the graph.

From a brief look on our internets... one standard deviation in IQ is approximated by about 16 IQ points.... so now we need to envision this data plotted on a y-axis of *at least* 80-120 to get any sense of the scope of differentiation the data is actually meant to convey. Now place normal distributions on top of each data point with a standard deviation of 16... and bang, any differentiation is completely washed out in any reasonably scientific data analysis.

"there are lies... damn lies... and then statistics..." as they say.


UPDATE:
Sorry... I just couldn't help it...

0 Comments:

Post a Comment

Subscribe to Post Comments [Atom]

<< Home