So How Off Were Drudge's "Exit Polls"?

[West Coast Democrat]

Posted in LBN yesterday, Drudge claimed this:

LATEST EXIT POLLS SHOW 59% VOTE 'YES' FOR RECALL, TOP CAMPAIGN AND MEDIA SOURCES TELL DRUDGE REPORT, 51% FOR SCHWARZENEGGER, 30% FOR BUSTAMANTE, 13% MCCLINTOCK... DEVELOPING...

*******

Recall passed with 55% (with 99% reporting)
Arnold won with 48% of the vote (with 99% reporting)

Drudge was off by quite a bit. Where did he get these numbers exactly? Sure, a lot of us at DU expected the recall would pass, but why were any of us paying attention to Drudge or the other phony exit poll numbers floating around?

[Southsideirish]

1. Weren't they posted way before the voting ended ? (N/T)

[West Coast Democrat]

3. Yes, and it was wrong of him to do so

because the results couldn't have been accurate.

[Rashind]

2. Exit polls are the suck.

Well, I take that back. Exit polls are perfectly fine, if used correctly. However, media outlets love to treat them as the end all be all. Last night MSNBC called the election for Ahnold a clear and certain proclaimation of his victory, and somewhere below in the text is a note that it is merely an MSNBC projection based on exit poll data.

Where my journalistic integrity at?

[newyawker99]

4. Hi Rashind!!

Welcome to DU :toast:

[HitmanLV]

8. Drudge reporting them at 4pm PST...

...is what sucks.

[HitmanLV]

5. Exit polls are fairly accurate in predicting the results...

...usually, but not the margin exactly.

Truth is, the exit poll was correct in predicting all the results, and did it with some accuracy. It correctly predicted the results, and also the order of finish, close to what happened.

The exit poll said the YES recall vote was at 59%, it in fact was 55.4%. The exit poll was off 3.6 points - in a race with that overwhelming a margin, an easy one to call.

The exit poll said Arnld with 51%, Cruz with 30%, and McC with 13%.
Arnold in fact won 48.7% (2.3 points off), Cruz won 31.7% (1.3 points off), and McC won 13.4% (.4 points off).

These results were fairly close and fairly accurate. The exit polls reported by Drudge were those the media commissioned to be done so they could call the race when the polls closed. They are just a rough indicator.

For example, if the exit poll had Recall losing at 50.2 to 49.8, for example, the network would know calling it one way or the other early could be embarassing, since there is a decent chance they could be wrong.

With margins of 10 points and 20 points, this wasn't close. an easy one to accurately call at 8:01 PST.

Best as I can tell, Drudge's leaked numbers were the only ones out there Tuesday afternoon, and they weren't phony - the networks paid a lot to have that survey done, and it was a real survey.

They got what they paid for - fairly accurate results.

[stickdog]

6. Who did the exit polls?

And who paid for them?

[HitmanLV]

7. the networks paid for them....

...and an outfit called Edison/Mitofsky Research conducted the exit polls, which ask California voters their opinion on a variety of relevant issues, determine how they voted, and ask a number of demographic questions to allow analysis of voting patterns by group.

They then take this data to be cooked by their computers and analysts, comparing them to past voting patterns and other info. In the recall race, it wasn't even close - it was easy to call with margins of 10% and 20%.

I don't find accountants, actuaries, and related eggheads scary. They are paid to be right, not to push any candidate above anyone else.

The same types crunch the numbers when we win, and they crunch them when we lose too. In either case, when the margin is narrow, they need to wait for more real data to come in, which they compare to their mathematical models, and can make a projection.

[cthrumatrix]

9. interesting....i don't remember anyone saying that they ran across a

"exit poll" all night.

I'm not saying it wasn't done. It's funny how quiet this was done and that people on DU are unaware of this action.

I'm sure we will be hearing more about this.

[HitmanLV]

10. Nothing funny about it....

...they only poll a few thousand people to get their numbers, and refine the numbers down based on mathematical models to get their projections.

As for the people on DU being unaware, that's hard to beleive since every election has detailed exit polls done. Why in the world would this be done differently?

The poll I mentioned was based on 4214 post-election Interviews (400 Absentee, 3814 Exit). That's fairly 'quiet' in a state the size of California. Haha, they polled fewer people than who voted for George B. Schwartzmann, and I bet none of us know anyone who did that, and certainly no one who would admit to it! ;-)

I have been voting since age 18 (I'm 35 now) and only once was I polled - a phone poll the afternoon of the election last year, after the Nevada election.

The truth is, nobody either of us knows was probably polled after the 1992 election, the 1996 election, the 2000 election, the 2002 election, or the 2003 election. I don't think any DUers mentioned they ran across a poll in 2000, 2002, or 2003, either. There were detailed exit polls done for all of these elctions - overwhelmingly, they were accurate.

There is nothing here. A day at the office, in the world of actuaries and numbercrunchers, and the people who love them.

[cthrumatrix]

11. I'm sure there are plenty of people here who have data they have

demonstrable differences between exit poll data and actual results....that's when there's a problem.

[HitmanLV]

12. you misunderstand the point of exit polling

...they are intended to predict the results. They are accurate enough to do so within a few points (see my posts above), though they only are there to give the networks a solid ground to call an election for one person or another.

They are not meant to predict to within a fraction of a point. They are ment to accurately predict results, not the margins (though they can be very good at that, too).

In fact, they did accurately predict results.

I don't know what you mean between datta DUers have that differs from the exit poll and the actual reported results. Sorry.

[stickdog]

15. So was it a true "exit poll" or was it based on 'actual' results?

Like E/M's Realvote system?

[HitmanLV]

19. the projections were based on....

...exit polls.

The results are the results.

[West Coast Democrat]

17. People Should Not Post Drudge Stories in LBN

Drudge makes up a lot of his stories. That's a fact.

[DemocratSinceBirth]

13. I Don't Like Matt Drudge....

but he was reporting what appears to be good information....


We need to deal with reality and not create our own....

[HitmanLV]

14. exactly right

Taking issue with the fact that at worst, the exit poll came to withing 3.6% of the actual vote is downright comical.

Reality is always a good place to start. Kudos!

[West Coast Democrat]

16. I think it's downright comical

that anyone here at DU would take issue with another DU member taking issue with Drudge's numbers. They were WAY OFF!

[HitmanLV]

18. I posted up above....

..that they were fairly accurate, and predicted all the results correctly.

Getting mad at the exit poll is unproductive, sorry.

[DemocratSinceBirth]

20. Drudge Is The Pits......

but some of his "newsbreaks" are good....

[West Coast Democrat]

21. Others are Angry as Well....I posted more here:

http://www.democraticunderground.com/discuss/duboard.php?az=show_topic&forum=104&topic_id=505590

-----------------------------------

http://story.news.yahoo.com/news?tmpl=story&u=/ap/20031008/ap_on_en_tv/recall_tv_13

Networks Call Election After Polls Close
By DAVID BAUDER, AP Television Writer

NEW YORK - For television journalists, Arnold Schwarzenegger's easy victory in the California recall revived questions about whether they have a responsibility not to characterize election results before all voters get the chance to weigh in.

......

But on Fox News Channel more than an hour before polls closed, Sean Hannity noted during an interview with a Democratic California assemblyman that exit polls pointed toward a Schwarzenegger victory.

.........

Reporting early exit poll results is also dangerous because the numbers can change as the day goes on, said Tom Hannon, CNN's political director.

Yet the networks also face the pull of competition and the natural instinct not to sit on information that they know. Many network executives were aware that the Internet's Drudge Report was declaring the race a landslide well before polls closed.

more.....