Vanity: VNS' Exit Poll Model

November 6, 2002 | NL

[NittanyLion]

We heard last night that VNS decided its exit polling models were unreliable; hence, no races would be called until the raw vote provided substantial-enough evidence to warrant a prediction.

Based on the final result, is there any doubt at all that VNS looked at the predictions and decided it simply wasn't possible the GOP would pickup 3-5 Senate seats and 5-15 House seats? I submit the models worked just fine, but when they didn't play into the media's preconceived notions the evidence was rejected as unreliable.

Perhaps it's understable; let's take a look at the Florida gubernatorial contest. The ballot amendment regarding a limit to class size passed, yet the candidate advocating this action was routed convincingly. Parents and retired folks are rumored to have voted for Jeb Bush.

How about Minnesota, where precincts just north of Minneapolis supposedly voted 55-45 for Coleman. As I understand it, the numbers defied historical trends.

Bottom line: what does this say about exit polling as we go forward? If my theory is the correct one, do we allow humans preconceived notions to override the statistical data? Is said data any good in the first place (reference 2000...)?

[NittanyLion]

What do you think?

[Reelect President Dubya]

I think VNS saw EXACTLY what was happening and suspended reporting their exit poll data because the did not want to discourage Democrat voters from coming to the polls. Had the Democrats been ahead in the exit polls they would have been crowing it from the rooftops.

[NittanyLion]

I think VNS saw EXACTLY what was happening and suspended reporting their exit poll data because the did not want to discourage Democrat voters from coming to the polls.

Also a good possibility. I certainly wouldn't put it past the media to do as you say. Especially considering NBC was calling races for the Democrat that were within 5-7 percent, and refusing to call Republican races even if the lead was 15%.

[Uncle Miltie]

VNS Modus Operandi: If Dems are winning, call it early. If Reps are winning, shut down the service.

It is all about turnout. VNS has perfectly aligned with the Al-Quaidacrat interests. Proof positive was last night's non-performance.

[js1138]

bump. Expect rumors and leaks on this later this week, if not this afternoon.

[phelix]

It sure seems suspicious to me!

[Wright is right!]

I believe you are 100% correct. The VNS's highly-touted top-to-bottom revamping of its equipment, software, and procedures SHOULD have worked perfectly - and, apparently, it did. BUT it produced a result that the liberal media couldn't believe - so it was scrapped.

This morning, the VNS is undoubtedly telling its member networks that they were RIGHT - and probably laying blame on the nets for failing to report what was happening.

Actually, though, it turned out for the best. All of the races were voted and counted in isolation - without regard to the results of other states' races. This MAY have been one of the deciding factors in the GOP's astonishing success.

Michael

[willgetsome]

KILL VNS

I don't know about you guys, but I had a great time tracking the election on my own via this site and other websites (e.g. official state sites, news sites, etc.). To Hell with VNS! Let the West Coast and Hawaii feel like their elections matter and let me figure it all out rather than having Brokejaw, Blather, and Jerkings "spin" it.

[NittanyLion]

The VNS's highly-touted top-to-bottom revamping of its equipment, software, and procedures SHOULD have worked perfectly - and, apparently, it did. BUT it produced a result that the liberal media couldn't believe - so it was scrapped.

Agreed. It'll be interesting to see if we ever find out who made that decision.

Actually, though, it turned out for the best. All of the races were voted and counted in isolation - without regard to the results of other states' races. This MAY have been one of the deciding factors in the GOP's astonishing success.

Also agreed. I'd prefer no exit polling for any elections going forward. Let the results trickle in on their own after a state's polls have closed.

[usapatriot28]

I think that last night proves that VNS is unnecessary (and, in my opinion, undesireable). I actually enjoyed seeing the actual results coming in and not some propagandist's interpretation of them.

[Tacis]

In 2000, VNS declared PanderAl the victor in FLA long before the data supported the claim, knowing that Al would probably win and that no one would ever notice. We did notice and they called it a "mistake." It was intentional to keep independents in California at home, a counter to Nader's strength.

The lied again, yesterday. They knew very early on that there would be a Republican landslide and to prevent doing damge to the leftists in the West, especially, California, decided to abort. As media scum, their objective was to report only info that helped (but did not hurt) dems or hurt (but did not help) good guys. They saw quickily that what they would report would only help patriots and hurt the anti-American leftists and they shut down.

NO QUESTION!! The media got their money's worth because VNS did the data collection but not the forecasting.

[killroy]

The way to kill VNS is for voters to lie to it. The bad results will kill that type of polling.

[NittanyLion]

They lied again, yesterday. They knew very early on that there would be a Republican landslide and to prevent doing damge to the leftists in the West, especially, California, decided to abort. As media scum, their objective was to report only info that helped (but did not hurt) dems or hurt (but did not help) good guys.

I must admit, it's hard to interpret VNS' actions (based on the context provided by the results) any other way. Drudge provided some early exit poll numbers that appear to be pretty solid (with a couple exceptions in Colorado and NC). VNS must've been looking at the same models...

[NittanyLion]

I actually enjoyed seeing the actual results coming in and not some propagandist's interpretation of them.

As did I. It's fun trying to predict what precincts have reported in and how the remaining votes will lean. We also heard a lot more analysis on the respective parties' election strategies (which I enjoyed) as opposed to how sucker moms voted and why.

It seems strange, but in a way I thought the lack of exit polling info led to a more substantive discussion.

[Route66]

Watching the election results last night was SO much better with every network scrambling for numbers and reports and doing their own forcasting.

A central clearinghouse for collecting actual election data is fine, but VNS simply had too much power over election outcomes when all major networks were reporting projected winners and declaring the whole thing over before the polls even close in the midwest much less the west.

I hope they do not resurrect this dead horse.

[Lou L]

I think exit polling has a purpose, albeit after the elections are finished, and all winners are announced. I think there is some merit in knowing the demographics of who voted for whom, and where they stood on the issues. I don't think that exit polling should be used to project winners, as we saw what happened in 2000.

[NittanyLion]

I think there is some merit in knowing the demographics of who voted for whom, and where they stood on the issues. I don't think that exit polling should be used to project winners, as we saw what happened in 2000.

I agree 100%.

[aristeides]

Drudge reported several VNS errors: Dole down in NC; Allard down big in CO; KKT ahead in MD; Jeb at only 51% in FL. I don't think it's any coincidence that all these errors were in the DemoRAT direction.

[Fred Mertz]

KILL VNS

I'm with you. We don't need no steenkin' VNS!

[NittanyLion]

Drudge reported several VNS errors: Dole down in NC; Allard down big in CO; KKT ahead in MD; Jeb at only 51% in FL. I don't think it's any coincidence that all these errors were in the DemoRAT direction.

This is pure speculation, but suppose Drudge only got hold of the raw numbers. If the sample was skewed toward liberal precincts, CO, MD, NC and FL may look to be leaning Democrat.

Once VNS applies their models - which I assume correct that bias based on historical voting patterns - the races may have come out strongly in favor of Allard, Ehrlich, Dole and Bush. The VNS folks, for whatever reason, decided a sweep of that magnitude wasn't in the offing. Hence the reason they refused to release the polls.