Sunday, January 10, 2010

Lies, Damned Lies, and Statistics

This week two prime examples of botched statistics hit my RSS feed.

The first was an interesting post on how statistics can be mishandled and misrepresented.



In this particular case, the author discusses bit.ly's market share, and how it "dropped" significantly. In fact, it has not. It is still the leading URL shortener by a large majority, but it was never as big as statistics told us.

In older statistics, only the five leading services were taken into consideration, while in the new statistics, all URL shorteners were considered.

In the second example, the author examines how some things can be made to look bad. The point here is that taking the median of pay in the public sector as compared with the private sector, does not equate one for one the pay for different types of job. Thus, the claim that the public sector gets paid more is false.



Interesting.

No comments:

Post a Comment