Tuesday, October 21, 2008

three on numbers

We did a lot of riffing about numbers back in September: here, here, here and here.
More below:

Here's a cautionary tale on a little numbers-tweaking by a New York Times freelancer who stretched a few stats to mean more than they actually did. Moral of the story: Don't twist. Don't stretch. Don't assume. Read about it on Gawker.com

For all you numbers geeks, the San Francisco Chronicle on Saturday took a look at all the new polling aggregators, including fivethirtyeight.com, which have sprung to the fore this campaign season. The piece also looks at reasons why some political polls can sometimes go wrong.

Finally, a quick read of Howard Kurtz's column in yesterday's WaPo makes you wonder if journalists who rely too heavily on polls in their reporting can actually affect the outcome of an election. You also wonder if over-reliance on data/handicapping the races can sometimes blind reporters to the other cues out there, so that ultimately the story goes south. bk

3 comments:

Moana F. said...

It's scary to think that stories written even for america's most famous newspaper can have mistakes. everyone makes mistakes but this seems like an intentional action to better the freelancers story and the point he/she was trying to make.

jeremy said...

Google Jayson Blair. This is very minor in comparison.

Still, it's never a good thing when reporters do this, because it crushes the credibility of that person and the news organization he or she works for. At least it appears the Times is doing a better job of policing.

In terms of unintentional mistakes, the Times probably runs the most corrections of any news organization -- not because they make the most errors, but they are the best at correcting them.

Melissa said...

Underreported fact: Nate Silver, the genius behind fivethirtyeight.com, started his statistical analysis with, yup, baseball. Further proof that sports people aren't all meatheads!