Do the polls matter anymore?

[SN2903]

Thoughts? They were way off in 2016 on the state level. In the Michigan Primary most polls had Hillary up 8-10 in the Michigan primary and Sanders won. The national level they were not that off.

 In 2018 some of the polling was horrendous especially in Indiana and Missouri.

[Yank2133]

Biden's numbers are scaring you huh?

[Gracile]

National polling in 2016 and 2018 (especially toward the end of the campaign) was pretty accurate. The state level is a bit harder, which is why we see final state results that vary wildly from their polling.

[Roronoa D. Law]

Yes, it still matters just because Trump won and Clinton lost the MI primary does not mean everybody losing before election day will be the winner. Plus let's be honest the Democrats should have seen MO and IN coming after Bayh and Kander.     

[Alben Barkley]

Biden's numbers are scaring you huh?

[SN2903]

Biden's numbers are scaring you huh?

He won't get the nomination and Trump hasn't started criticizing him yet. Sleazy Uncle Joe cannot win.

[Ilhan Apologist]

Did you, like, forget everything that happened in 2018?

[TML]

On paper, polls in some states in 2016 were off by significant margins.

However, if you examined the vote shares of the major candidates in specific polls, you would have realized that the potential for a last-minute surge to change the result existed. For example, a poll in late October in WI was Clinton 46 - Trump 40, which, although it indicated Clinton was ahead, it also left room for Trump to overtake her with a last minute surge (which did indeed happen, according to exit polls).

In addition, 538 states that polling accuracy is measured more by the degree to which the actual result is off by rather than the winning party. For example, if a poll said D+1, it would have been more accurate if the actual result was R+1 than if it were D+10.

[MT Treasurer]

Only if they fit my narrative

[Senator Spark]

Nope. They haven't mattered since 2016.

[Alben Barkley]

Nope. They haven't mattered since 2016.

If polls don’t matter in your head, that’s the most convincing and powerful sign imaginable that they do indeed matter in reality.

[Xing]

The polls could be wrong, sure. But assuming that they will be when the numbers aren't to your liking doesn't tend to work out very well for the people who do it. Polling errors (and the severity of them) tend to change quite a bit from year to year. Well, I guess not in every state.

[Rookie Yinzer]

If Donald Trump weren't such an abnormal candidate people wouldn't be questioning polls the way they do today. 2016 polls were not that far off from reality.

[SN2903]

Nope. They haven't mattered since 2016.

[Amenhotep Bakari-Sellers]

MI/PA/WI CO and Va are the 272 Dem blue wall. With Trump approvals, NV, IA, AZ, OH and FL can give the Dems 300 electors

[morgieb]

Yes, though there's more grains of salt that might need to be applied and it seems to be a lot harder to poll states without large minority populations (and a significant college/non-college white differential in polling)

[SuperCow]

They matter, but not this far away from an election.

[AN63093]

When did the polls cease to matter?

One of the most annoying things about 2016 is this meme that started that polls are now not to be trusted.  Polls, particularly the national ones, were generally about as accurate in 2016 as they ever are.

Now what did cease to matter, and what never should've mattered, were those silly forecasts like the ones the NY Times had, which proclaimed ridiculous statements like "Clinton has 99% chance to win election", which was not what any of the polls were suggesting and was a significant misinterpretation of them.  These forecasts were often cloaked under the façade of "data analysis" (like with its seemingly objective, data driven sliding scale), when it was pretty transparent that these journalists weren't actually doing any analysis and just saying what they wanted to happen.

These projections were obviously pretty stupid, but had the effect of fooling a lot of the useful idiot crowd into thinking the election was all but over.  For those of us that were actually analyzing things objectively, what we saw was that Clinton had a slight edge in the national PV but it was a volatile and uncertain race in some of the swing states that Trump had a good chance of flipping, which lo and behold, is exactly what ended up happening.  

For those that actually looked at projections like the NY Times' and were foolish enough to think there was anything scientific about it whatsoever.. well, honestly, those people got exactly the disappointment they deserved.

[SingingAnalyst]

Of course they matter, but one should keep in mind that some respondents either refuse to answer or outright lie, and that such respondents are likely not evenly distributed around the ideological sphere. In particular, as long as Trump is widely regarded as fascist, he will poll low.

[Some of My Best Friends Are Gay]

Polls this far out aren't really accurate, but once we get closer to the election you can actually start putting some faith in them.

The idea that because polls were (slightly) off in 2016 they don't matter at all anymore is just pure delusion coming from Trump supporters who don't want to admit how poorly he's polling right now.

It's like saying; because I rammed my car into a tree once, I now ram my car into a tree almost every time I drive.

Also, polls were accurate in 2018, the results lined up perfectly with the GCB and with Trump's approvals at the time.

[elcorazon]

Thoughts? They were way off in 2016 on the state level. In the Michigan Primary most polls had Hillary up 8-10 in the Michigan primary and Sanders won. The national level they were not that off.

 In 2018 some of the polling was horrendous especially in Indiana and Missouri.

Did the polls ever matter? They are just a snapshot of the best information that can be gleaned. If you're asking are they worse and now useless? the answer is no. If you're asking is there any information that's relevant we can get from polls, the answer is yes.
Dumb thread.

[Amenhotep Bakari-Sellers]

Just like we see in all elections there are registered voters and likely voters. There is nothing that will change the 272 map and IA, Va and OH are states that will tie election, not FL, AZ or NC.

[redeagleofficial]

Use Polls plus Trends.  Its how I got every state right in the midterm senate races.  I knew WV would be closer than expected, as well as Indiana and Missouri would defy the polls and go red by a decent margin, as they did.  Its also explainable that Trump over performed the polls in the rust belt in 2016 while underperformed in places like Nevada and Colorado.

[Progressive Pessimist]

I still say that they do, but Democrats need to ensure that they are leading Trump and other Republicans by at least the mid single digits and be at or above 50% if they want to be sure that they can win. Anything less than that and it's too close for comfort. That's at least how I read polls now. Democrats should be cautious this time and not put that much stock in any favorable poll being an inevitable win as was possibly the case in 2016.

[Vern]

They will give us an idea of what is happening. It's like someone who wears glasses trying to see without them. It's burry, you can't see all the little details but you can see the overall picture.