Please register to participate in our discussions with 2 million other members - it's free and quick! Some forums can only be seen by registered members. After you create your account, you'll be able to customize options and access all our 15,000 new posts/day with fewer ads.
The outcome predictions have been getting increasingly better with every election. In 2012, 2 were 100% accurate, and all the others were 90% or more accurate.
That can only happen when all polls in general are becoming more accurate but only in part.
Outcome predictions are all based on electoral college votes. When we vote, we tell the electoral college how they should vote for us, but each state has its own rules that apply. In some states, the winner-takes-all states, all the electors are obliged to vote for the clear winner of the state. In others, the electors are only obliged to vote as their voters did.
So, in some states, the winner can have a 51% majority, but can get the additional 49% of the voters who didn't vote for him. But using the electoral college is still the most accurate predictor, as those factors are easy to make. And the electoral votes come in after the public has voted.
All the polling is less accurate, as it depends on two things: asking how a voter intends to vote, or asking a voter how he voted. In both, voter lying is a given, and has to ben factored in, but doing so is a much harder process.
I believe that in the near future, polling agencies will begin using discreet electronic visual recognition equipment to know if a voter is telling a lie or not. This will increase the accuracy of all polling dramatically, because they will have reliable data for the first time.
As long as all the voters remain anonymous, no laws would be broken. Other, similar equipment could be used for telephone polling and other methods.
As it is now, the pollsters are getting increasingly better at asking psychological questions that can reveal a lie, how strong the intention to vote is, a voter's commitment to a candidate, and other stuff. Those questions can be very accurate, but they may also be difficult to quantify accurately.
For sure, the old methods, which relied on intuition, crunching the wrong data sets, ignoring confirmation bias, and all the other stuff humans are prone to, is long gone.
2012 was the most recent election that some polls were the farthest from the outcome. Confirmation bias seems to have played a particularly large part in that. But overall, 2012 was also a year when the accurate polls were the most accurate as well.
Computer power has changed all this radically. As the computers grew more powerful and faster, so did the polling. Nowadays, the questions that are asked in a poll are all pre-tested by computer, which can show how close the question is to the statistical results that come from the question. Statistics weed out all the poor questions, and refine the good questions.
That's why every poll these days has some odd questions in it, like how much a person watches sports or something. Those are psychological indicators of confirmation bias and other stuff that's vague and hard to pin down.
See how the dozens and dozens of brexit polls predicted that outcome.
See how the carter/reagan polls predicted that race, right up to the last minute.
How "all star" poll guy nate silver predicted trump to have basically no chance of winning the primary.
And throw in our agenda laden media, mostly lib/left leaning, creative poll wording to sway a certain outcome, etc. and I think the simple answer is.....trust none of it. There's too much money, power, and agenda's at stake here. And a very, very nervous establishment who can lose a bunch of power/$ here.
Brexit polls showed that it was going to be close. Result, it was close.
"Close" doesn't count. They were wrong. That's the takeaway. And just because someone says they'd vote a certain way doesn't mean they will show up to do so.
Quote:
Originally Posted by Ace Rothstein
Wow, you found one that is almost 4 decades ago. Completely ignoring 2012, 2008, 2004, etc...
It's a valid example to show polls can't always be trusted. And the brexit was a recent poll. A recent and far off example. Variety to make my point.
Quote:
Originally Posted by Ace Rothstein
His model didn't show that. Silver tried to get into punditry and was wrong. Trump actually ended up under-performing in most primary polls.
His model failed. And he got it wrong many times on trump.
Again, the polls/"experts" were wrong. Some many times over. I think you'll admit this is a very unique election, not on the typical clown politician establishment side of hil, but for trump. Many polling people have said they feel he's underrepresented in the polls. Silent majority types don't like to tell who they're voting for and some people are shy about saying they are for trump. Plus today throw in many sleazy/agenda driven polls in the media that exist today, wording questions in certain ways for a desired outcome, poll more of one registered voting group then the other, etc. and it's obvious an educated person would question the final results of any of it.
So for this election......one shouldn't get arrogant/over confident. Here's some history for you:
"Just two days before the October 28 debate, Carter was eight points ahead in the Gallup poll. A week after the debate, he lost to Reagan by nearly ten percentage points."
So for the 100% certain.....polls get it wrong. And in big ways sometimes, right up to the week before the election. See above example. The other 100% certainty? Who wins will be known on/soon after election day.
Very accurate. There are too many polls showing the same thing for ALL of them to be as far off as they would need to be for Trump to win.
Yup.
Very close is when the surprises happen. Trump is not close.
At least, not to a win, he isn't. The worst defeat since polling began was 39%. Right now, the prediction of Trump's loss is at 37%. But I think he may close up on that number to the 39% figure; that's as low as his base support has ever gotten. He may go as high as 40%, the floor where his numbers have hovered around ever since the loss of his brief convention bounce.
In comparison, at the time of the Republican convention, Clinton was only polling 1 1/2 points higher than Trump. Now, she is 7 points ahead in the latest polling aggregate; it was averaged 7 hours ago, up one point from the 6 point lead she's had for almost a month.
The bump is believed to have come from the Friday Night Video. One poll after the debate had her at 15 points above Trump as the debate winner, but I think that one is an outlier. I haven't seen an aggregate on that yet, but the few I've seen gave Hillary a 2-digit lead coming out as the debate winner.
Does a big lead in the debate mean very much? No. But a long aggregate lead does.
Considering that the 2012 election was decided by 4/5 of a point, her lead is still substantially better than Obama's was then.
You are absolutely right, Steve. Close doesn't count. But probability has proven to be seldom wrong, too, by your own recognition.
"Close" doesn't count. They were wrong. That's the takeaway. And just because someone says they'd vote a certain way doesn't mean they will show up to do so.
You don't understand what I'm saying when I say close. The polling aggregate before the vote was .5% in favor of staying, suggesting a very close race. 4 of the last 11 polls released were pro-Brexit. There was a split unlike the polling right now.
Quote:
Originally Posted by stevek64
It's a valid example to show polls can't always be trusted. And the brexit was a recent poll. A recent and far off example. Variety to make my point.
When you have to go back 4 decades and 9 elections to make a point, you're really stretching. You have numerous elections since then when the polls were spot on. There are more polls and better polls now as well.
Quote:
Originally Posted by stevek64
His model failed. And he got it wrong many times on trump.
Again, you are confusing his prognosticating with his polling data. Notice how all but 1 of the 7 examples are well before any vote was cast.
Quote:
Originally Posted by stevek64
Again, the polls/"experts" were wrong. Some many times over. I think you'll admit this is a very unique election, not on the typical clown politician establishment side of hil, but for trump. Many polling people have said they feel he's underrepresented in the polls. Silent majority types don't like to tell who they're voting for and some people are shy about saying they are for trump. Plus today throw in many sleazy/agenda driven polls in the media that exist today, wording questions in certain ways for a desired outcome, poll more of one registered voting group then the other, etc. and it's obvious an educated person would question the final results of any of it.
So for this election......one shouldn't get arrogant/over confident. Here's some history for you:
"Just two days before the October 28 debate, Carter was eight points ahead in the Gallup poll. A week after the debate, he lost to Reagan by nearly ten percentage points."
So for the 100% certain.....polls get it wrong. And in big ways sometimes, right up to the week before the election. See above example. The other 100% certainty? Who wins will be known on/soon after election day.
If 1980 and Brexit are all that you are going off of, you're going to be incredibly disappointed on election night. Hillary has a commanding lead right now and there is nothing Trump can do to turn it around.
Please register to post and access all features of our very popular forum. It is free and quick. Over $68,000 in prizes has already been given out to active posters on our forum. Additional giveaways are planned.
Detailed information about all U.S. cities, counties, and zip codes on our site: City-data.com.
Please register to participate in our discussions with 2 million other members - it's free and quick! Some forums can only be seen by registered members. After you create your account, you'll be able to customize options and access all our 15,000 new posts/day with fewer ads.
