Trust in political polls since the Nov 2016 U.S. election?

[PlanetJosh]

So the last thread on this topic based on searching this forum for "poll" in titles was in late 2016 with the last post in Dec '16. I suppose this could just be merged into that thread but it's been almost a year and a half.

How's your trust these days for polls supporting an opinion in this forum? Is the perception of the accuracy of the most independent polling companies affected by how certain polls blew it in the Nov '16 election? I think some conservative pollsters got it wrong too for that election.

 

[dank69]

The polls were pretty accurate in 2016. I bet you are confusing the polls with the predictions, which isn't surprising given that you're an idiot.

 

[fskimospy]

So the last thread on this topic based on searching this forum for "poll" in titles was in late 2016 with the last post in Dec '16. I suppose this could just be merged into that thread but it's been almost a year and a half.

How's your trust these days for polls supporting an opinion in this forum? Is the perception of the accuracy of the most independent polling companies affected today by how certain polls blew it in the Nov '16 election? I think some conservative pollsters got it wrong too for that election.

The national polls were highly accurate in 2016 so trust in them should have increased, not decreased. They were more accurate in 2016 than in 2012. State polling was mostly accurate as well, with a few exceptions. That’s pretty standard too though.

So basically the answer should be that your trust in polling averages should be fairly high overall considering the good track record. Nothing’s perfect though.

 

[Commodus]

The 2016 election mainly helped underscore how certain polls can have biases. Not necessarily intentional, but just by virtue of their sample sizes and the demographics they have access to.

For example, Trump loves to cite Rasmussen, because it gives him the most flattering scores (never mind that it has still tended to lean toward disapproval for him). But what he doesn't realize, or perhaps chooses to ignore, is that Rasmussen has tended to skew conservative for at least the past several years due to its small size and inclusion of just enough conservative sample members to lean in their direction. Unless opposition to the Republicans is so overwheliming that even habitual voters change their minds, it'll disproportionately favor the GOP.

This isn't to say there aren't left-leaning polls, but my stance now is that you need to look at the broader polling trend to know what's really going on.

 

[WelshBloke]

The polls were pretty accurate in 2016. I bet you are confusing the polls with the predictions, which isn't surprising given that you're an idiot.

The first half of this is accurate, the second is harsh but may or may not be accurate.

 

[trenchfoot]

The 2016 election mainly helped underscore how certain polls can have biases. Not necessarily intentional, but just by virtue of their sample sizes and the demographics they have access to.

For example, Trump loves to cite Rasmussen, because it gives him the most flattering scores (never mind that it has still tended to lean toward disapproval for him). But what he doesn't realize, or perhaps chooses to ignore, is that Rasmussen has tended to skew conservative for at least the past several years due to its small size and inclusion of just enough conservative sample members to lean in their direction. Unless opposition to the Republicans is so overwheliming that even habitual voters change their minds, it'll disproportionately favor the GOP.

This isn't to say there aren't left-leaning polls, but my stance now is that you need to look at the broader polling trend to know what's really going on.

I think although we need to emphasize how accurate (trustworthy) on average certain polls are and as you mentioned how we need to take a wider view at them to glean what's meaningful from them, we also need to take into consideration how these polls are often whipped through spin machines and re-purposed into propaganda where certain media groups will skew the data into an analysis that in so many ways will actually defy what the data objectively shows.

In so many ways, polls have been considered untrustworthy not so much from the actual numbers they produce but from the taint that they get painted with.

 

[nickqt]

Polls are estimates of possible results. Some polls are more accurate than others.

The results are the results. Some results are better than others.

You can use polls to estimate the results. You can use results to estimate the accuracy of polls.

But they're two totally different things.

It should be clear by now that polls are not pre-results. They are just estimates.

I wouldn't trust any poll to be 100% accurate. Doing so is a logical mistake.

 

[TheVrolok]

People are fucking stupid. The polls were pretty damned accurate in 2016. Unfortunately, the aforementioned stupid fucking people don't understand polling. Sorry for the language, this topic has been beaten to death so many tim will by idiots it's infuriating.

 

[Thump553]

Generally speaking the polls gave Clinton a 90% chance of winning the presidency. This gives The Donald a ten percent change. Sometimes shit happens which is the real take away message here, not that the polls were "wrong."

 

[vi edit]

Trump won by less than 90,000 votes spread out between three different states. That's not splitting hairs. That's splitting a split end on a hair. I don't think people appropriately respect how thin that margin was. If something like 1 out of 100(I think that was the stat I saw) had changed their vote to Hillary instead of Trump in those states we'd have Madam President. Popular vote was within the error margin.

The challenge with polling right now is that it's *very* hard to account for "independent" voters in a hyper partisan environment. The political winds shift so hard on a day to day basis that a single issue or story can sway independents away. Demographics and historical voting patters are getting thrown out the window. It's very hard to accurately poll things right now given how on edge so many people are.

 

[dank69]

Generally speaking the polls gave Clinton a 90% chance of winning the presidency. This gives The Donald a ten percent change. Sometimes shit happens which is the real take away message here, not that the polls were "wrong."

Those weren't the polls.