Bush leads Gore by nine points

B) 2000

November 6, 2000 Presidential Tracking Poll

On election day, Portrait of America (POA) finds Texas Governor George W. Bush leads Vice President Al Gore by nine points, 49% to 40%. Ralph Nader has 4%. The Rasmussen Research telephone survey of 3,750 Likely Voters was conducted for Portrait of America on Saturday, Sunday, and Monday, November 4, 5, and 6.

In the electoral college, Bush leads 224 to 168, with 146 electoral votes in the toss-up category. To see the Portrait of America Electoral College summary click here.

A review of our tracking data over the past seven months indicates that Campaign 2000 has not been a volatile whirlwind as reported in many media outlets. Partisan voters knew back in March how they would vote. Less partisan voters have been gradually making up their minds over the past seven months in a manner that is anything but volatile.

-- Uncle Bob (unclb0b@aol.com), November 07, 2000

Answers

In your dreams.

D-Dubya-I has proven himself to be a loser, and will always be a loser.

-- (best@man.wins), November 07, 2000.

In your dreams.

If you are so confident about your statements, why do you continue to HIDE like the coward you are behind the ever changing fake ID you use?

You are a lothsome slime slithering bottom feeder. Stand up and be counted!

-- Ain't Gonna Happen (Not Here Not@ever.com), November 07, 2000.

I thought I would bring a couple of these 2 day old threads back to the top, just to show what kind of foolishness was being peddled as fact on the days before the election.

As of now, nearly 100 million people have voted and been counted and the REAL results announced. Not only is Bush not "up by 9 points", he is trailing Gore, though ever so slightly, in the popular vote.

-- Brian McLaughlin (brianm@ims.com), November 08, 2000.

Come now, Brian. These were polls being peddled as polls. As I recall, they tended to change rather wildly from day to day, which was frankly explained as an artifact of polling technique and sample size rather than underlying changes in the electorate.

I'm willing to accept that all of these various (and very different) polls represented genuine efforts to describe and track what was essentially a flat out tie. Ties are extremely useful for calibrating polling methods. POA, to be that far off, surely had some serious methodological problem, not likely to be just very bad luck stumbling onto such a nonrepresentative sample.

But these efforts are neither foolish, nor sold as "fact". They are good faith efforts, sold as poll results with margins of error.

-- Flint (flintc@mindspring.com), November 08, 2000.

Flint:

Ties are extremely useful for calibrating polling methods. POA, to be that far off, surely had some serious methodological problem, not likely to be just very bad luck stumbling onto such a nonrepresentative sample.

I doubt that this one will be that useful for calibrating polling methods; at least in certain states. When one suddenly has a large new group added to the studied population and that new group doesn't fit the previous profile, errors will occur.

That was my concern [as I stated in two threads yesterday; I was actually spending much time doing grunt election work]. I would love to say that this was a product of my genius, but it wasn't.

I talked to a couple of people who study polling methods.

Best wishes,,,,

Z

-- Z1X4Y7 (Z1X4Y7@aol.com), November 08, 2000.

These polls purported to be up-to-date on election day. Their data was collected on Nov 4,5 and 6. So what happened? Were the polling techniques uniformly bad? Did many people change their minds in the last days? (if so, why?--the DUI thing?) Was the Dem party more successful than expected at getting out the vote? Was the Republican party less successful? Did the early TV predictions of a Gore victory in FL cause some Bush voters in the West to stay home because all was lost? Did people lie to the pollsters? (I hope so).

Upcoming analysis of exit polls may explain what happened.

-- Lars (larrs@indy.net), November 08, 2000.

Lars:

The polling statistics can be slippery. Take an honest coin and start flipping it. Flip it 1000 times, and you are very unlikely to have 500 heads and 500 tails. One will predominate by a little bit. Flip it a million times, and the absolute size of the difference between total heads and total tails will grow, but the percentage difference will shrink. With 1000 flips, you might find heads has a 2% lead. With a million flips, heads has only a 0.5% lead. Yet the underlying probabilities are identical. Start over, and tails might end up with the lead.

These are fairly small samples that were taken, because polling costs money and takes time. To minimize expense and increase speed while maximizing accuracy, you can do several things. You can have rolling averages from poll to poll. You can use various ways of selecting unbiased samples (actually, *correctly* biased samples). If your method is ideal, you'll be within 3% one way or another. Best you can do within the time and money available.

Can your sample show a 9% difference when the underlying population is tied? Yes, of course it can. You can calculate the probability of being that far off with a given sample size and assuming a random selection from a correctly selected population. That probably might be only 1 in 100. But if you take 1000 polls, it will happen 10 times. If you average 3 polls together, the odds are lower, but never zero. One very bad sample can throw off a rolling average.

Poll results can also be sensitive to the purpose of the poll (doh!) You take your poll differently if you want to predict the final result, than if you want to know if a certain ad campaign is working or if you won or lost last night's debate. In the first case, you're trying to describing something more or less static, and in the second you are seeking out trends. Different methods, different results, because you have different purposes.

Finally, if we decided to hold the election all over again next week and everyone actually voted again, results would be different, possibly VERY different. So which one is "real"? That's a matter of definition, and NOT measurement.

-- Flint (flintc@mindspring.com), November 08, 2000.

best@man.wins, says:

BD-Dubya-I has proven himself to be a loser, and will always be a loser.B

A loser as opposed to someone of high achivement like yourself? Hawk, you continue to confirm what a pathetic little weenie you are. In the game of grown-up life you will never be a player.

-- I (h@ve.spoken), November 08, 2000.

Polls are like bureaucracies,it's somebodies job that they want to be justified or meaningful,when in truth it just ain't worth a shit.

-- capnfun (capnfun1@excite.com), November 08, 2000.