Burka gets 538's prediction...less than right

I just posted on some problems with 538's prediction of Perry +7. It wasn't really that newsworthy when you look at the inputs.

It looks like Paul Burka didn't quite understand the methodology:

Silver analyzed every governor's race in Saturday's New York Times, projecting the likely outcome. In the Texas governor's race, the prediction is Perry 52, White 45. This margin is slightly lower than the results of recent Rasmussen polls, which are the most frequent source of information about the race. Rasmussen has had Perry up by 8 and 9 points in his last two polls.

This is why Fivethirtyeight is forecasting that the race will tighten.

Fivethirtyeight isn't really predicting that the race will tighten. Silver is saying that the race is currently at +7 right now because he doesn't like Rasmussen's results.

By the way, one other thing worth mentioning that I haven't seen mentioned anywhere: the main reason Silver gives White a 17% chance of winning is precisely because his model recognizes that there is a greater margin of error because there aren't many polls. If more polls (like the WRS) poll show a similar outcome, White's chances of winning would be reduced substantially.

Posted by Evan @ 09/06/10 06:49 PM


Previous Entry | Home | Next Entry


No comments yet

Add Comments

No flames or impolite behavior. HTML will be stripped. URLs will be transformed into hyperlinks.