« McDespicable | platopal's Blog | Guarding polls »

Overconfidence



A Reuters post says that Obama's lead has dropped to 5 points, with a margin of error of 2.9 points, which essentially gives him only a one or two point lead. Why all the discrepancy? 


4 Comments

| Leave a comment
user-pic

It's a Zogby Poll. See my earlier post on this. He's a fraud. Nate Silver at 538 says his latest poll has equal numbers of Dems and Reps. Hardly likely since even Rasmussen weights his at D40-R33.

Zogby likes subscribers and he knows that if John McCain mentions his poll, which he did this morning on Meet the Press, Zogby makes another $500,000. Next week he will revert to the mean and claim how accurate he always was. John Zogby is a fraud.

user-pic

If I remember correctly, Zogby was the most accurate at the end of the Dem primaries. I've been following him more closely than the rest because of my perception. Please tell me if I'm wrong and I'll stop worrying so much.

user-pic

I like that sort of news. Now we can panick and work harder and come out to vote.

user-pic

There are a couple of errors in you post. The margin of error is 2.9% and, thus, your conclusion "which essentially gives him only a one or two point lead."

As Kevin Drum notes in this very informing discussion of the concept of the "margin of error", the concept is widely misunderstood.

Drum indicates that what the results of the poll you cite mean, with a 5 point lead and a 2.9% margin or error, is that there is a 96% probability that the leader by 5 points is "genuinely" the leader.

If you're interested in the ins and outs of polling, and polling shortcomings, as Candide recommends, check in at FiveThirtyEight.com and Polster.com. Check out the FAQs at the FiveThirtyEight. com site.

Leave a comment

platopal

user-pic

Following:
Followers: 1

Posts
Comments & Recommends


Favorites

All Reader Posts
How to use myTPM

Advertise Liberally
Share
Close Social Web Email

"To" Email Address

Your Name

Your Email Address