Strengthening Constitutional Self-Government

No Left Turns

A note on statistical errors, ecological fallacies and biased sampling. or, a meltdown

Jay Cost calls himself a "methods hound" and on that basis has a "methodological critique" of some who talk of a GOP "meltdown." Almost Greek to me, but not quite.

The Washington Post reports on its own poll showing "Republicans are losing the battle for independent voters, who now strongly favor Democrats on the major issues facing the country and overwhelmingly prefer to see them take over the House in November." And the New York Times asserts that Bush now has a new title: "optimist in chief:" "President Bush and his political strategists may be the most outwardly optimistic Republicans in Washington these days, and perhaps the only ones." And Tom Daschle predicts that the Democrats will pick up 25 House seats and 7 Senate seats. Mr. Daschle, a former senator, said he has not yet decided on whether or not to run for president, but it looks like Barak Obama has.

Discussions - 1 Comment

A couple of notes on the Cost article: Cost understands the ecological fallacy! He is correct that trends that are apparent at the national level may well not occur at the state level. This is a beautiful example of the ecological fallacy, and he understands it better than Wilkepedia (and Dain) do.

Three other notes: (1) Even statisticians tend to generate more alternative hypotheses when they don’t like the published outcomes, as opposed to when they do.(2) I wish that Cost understood that the word "data" is the plural form of "datum." So, for instance, the data ARE, and the data SHOW, would be correct, while the data IS, and the data SHOWS, would not. (3) Still, I can’t argue with any one of the alternative hypotheses that he generated. I think, and hope, though, that the most obvious interpretation is that Republicans will lose control of both houses.

Leave a Comment

* denotes a required field
 

No TrackBacks
TrackBack URL: http://nlt.ashbrook.org/movabletype/mt-tb.cgi/9224