Thursday, March 23, 2006

Junk Study 

I don't critique news articles often but this one deserves it, primarily because it points to an issue that affects PR. It is an example of a junk study, research that has little or no meaning but generated publicity.

The article starts off plausibly and continues so until near the end when the writer reveals that there was only a .27 correlation between being self-reliant in nursery school and being liberal as an adult. Wha? When I took statistics, if your correlation wasn't in the .8 to .9 neighborhood, you dare not claim anything and even then, teachers hammered that correlation DOES NOT equal causation.

I'll give credit to the writer for revealing that fact, but I won't give the writer credit for authoring the story. There never should have been an article about data so weak.

Yet, we do stuff like this all of the time in PR. One of the regular pieces of junk floated to the world is the internet survey without statistical controls. There is no validity to these instruments and yet, they become the basis of news releases time and again.

No wonder we get little respect. It's about the same level of respect that should be shown to the authors of the original study on nursery children and their political leanings.


Post a Comment

This page is powered by Blogger. Isn't yours?