Sunday, December 10, 2006

The Ideas Issue Questions Its Own Existance

The NYT's Magazine has an end-of-year ideas issue including this article presenting an idea that I think should be widely publicized: studies showing negative results do not get published. Thus, even well thought out studies that get an outlier result will be published and mislead scientists and the public as to the true causal relationship between two events. The same reasons negative results don't get published in social science journals apply to journalism in general.

Unfortunately, the rest of the issue appears dedicated to providing a platform for spurious correlations, see below.

1 comment:

Anonymous said...

it's definitely an issue. firstly: negative results can certainly be informative. secondly: the data can lend support to the wrong conclusion -- by design, roughly 5% of "positive results" are simply the result of chance, and the percent of incorrect "negative results" is likely much higher.


y-h