Bad Stats dot Org

Measurement, Definitions and Interpretation

admin — Sun, 05 Sep 2010 12:12:49 +0000

Hurricane Earl is now a memory for most, but last Thursday while the storm was a major threat to the East coast states of the US I noticed a frequent reference to Hurricane Bob from back in 1991. In Fact the Washington Post went so far as to cited Hurricane Earl “as the most threatening hurricane to the East Coast since Hurricane Bob in 1991.” Hurricane Bob’s claim to fame was brushing the NC coast and then hitting parts of New England as a category 2 storm in 1991.

Question 1. What is the “East Coast.” Does Florida Count? Seems to me I remember Hurricane Andrew. That one hit as a Category 5 storm south of Miami in 1992 a full year after Hurricane Bob.

Question 2. Moving further North, are NC, VA, DE, and NJ part of the “East Coast?” Hurricane Isabel came ashore in NC at a category 2 storm in 2003. So I guess it must have been less threatening for some reason. That despite the fact that at one point in it’s life is had sustained winds of 165 mph compared to Bob’s peak at 115 mph.

That leaves me wondering about things like data sources, interpretation, what measurements are used to determine “most-threatening,” and how the “East Coast” is defined in the eyes of the writer.

Tansparency

admin — Fri, 03 Sep 2010 19:55:49 +0000

The American Association for Public Opinion Research (AAPOR) has developed a set of disclosure standards as a part of their transparency initiative. Last month they provided and updated listing the supporters of this initiative. An organization only damages their own credibility by failing to meet these standards.

The Law Of Large Numbers – No!

admin — Tue, 31 Aug 2010 21:43:51 +0000

Just yesterday the Washington Post got it wrong. In their Sports section there was a piece on Iowa’s prospects for the coming season.

The claim: “The law of averages says Iowa will pay the price this season for escaping so many games last year with close wins.”

The Problem: The implication is that because Iowa won so many close games last year they will not be able to win so many close games this year.

The law of averages makes no such claims. It simply states that in the long run if winning a lot of close games is uncommon it won’t happen very often. It makes no claims about the relationship between what happen one year and what happens the following year.

What the paper did was equivalent to claiming that flipping a coin and getting heads will increase the chance of getting tails the next time, or that rolling a pair of dice and throwing an eleven makes it more likely that you will get a lower roll next time. In reality you are likely to get a lower roll the next time simply because the probability of getting an eleven or twelve is not very high on any roll of the dice.