I’ve been checking one of the three sites (538, Silver, or Realclear) almost every day, and posting about them here on the Dope about every other (i.e., second) day. Plus I listen to almost every 538 podcast (lately they do one once or twice a week), and occasionally I’ll see what Nate Cohn has to say in the NY Times.
As others have noted, this is not good for my mental health, and I WILL cut back, starting as soon as I post this (I just won’t have time for the next several weeks – crazy busy).
As for reasons, the same ones as others have noted, plus the state polls have a practical effect of helping me decide where I should put some effort into getting out the vote – do I bug my friends in Pennsylvania, or in Wisconsin? Do I sign up to send postcards to voters in Georgia, or in Michigan? Etc. (A few months ago, this was also helpful in deciding which Senate races I should send money to).
I never check the polls but do hear about them on the news. When polls are close all they tell you is that the race is close and the polls won’t reveal the winner. If they aren’t close they don’t get reported as much because that interferes with the most important purpose of elections, which is pollsters and media companies making money by the boatload. Next time you hear about the enormous cost of running a political campaign keep in mind that the vast majority of that money is spent on television advertising.
I very rarely check polls but I’m sure I will come October. I check individual down ballot polls more often than presidential when I’m curious about a race that I haven’t heard about for awhile.
Healthy doses of skepticism about polling fully acknowledged, including the likelihood of some systemic polling error, possibly more in Trump’s favor than not, there is an aspect of polling that I am becoming more obsessed about (yes I should consider a med eval …): keeping on eye on the electoral advantage figure. I am using the difference between the running weighted average for national polling and that of PA as the most likely tipping point state. The smaller that number the less anxious I am. Right now on 538 it is running at 2, on 270towin 1.8, and on RealClear Politics 1.3. That is consistently smaller than what is historically expected.
No doubt this is exactly why all the polls - national and in battleground states - showed a huge swing to Harris the moment she entered the race.
BTW, no polls seem to be announced on weekends, so there is no point to checking them seven days a week. Take time off and read a thread about how awful the Republicans are, making sure you click on all the links to mainstream media reporting on those behaviors.
Since you ask: never. If they show up in any of the news channels I sometimes look at, I’ll give them a glance. But I certainly don’t seek them out. I think the general concensus is that they are pretty unreliable anyway?
But when the populace is so evenly divided, a normal, typical, inevitable polling error has magnified effects (54% and 49% are the same distance from each other as 77% and 82%, but have very different consequences), so a lot of people make the same mistake you did.
In 2016 I trusted my friend’s recommendation that 538 was accurate.
I was let down.
Some weeks ago I again bookmarked 538, due to a post on here. Until today I hadn’t really looked at it. I just glanced at it, due to this thread. I’ll probably look at it off and on between now and November 5th. But I don’t put much faith in polls.
Yeah - as I understand it (not being a mathematician), the polls are not inaccurate so much as they do not say what many people think they do. It is not like gambling, where if you have a 1-2% edge and play enough hands, in the long run you’ll end up ahead. In an election, you get one roll of the dice. How much would you gamble on a single wager where you had a 55% chance of winning?
Plus, polls have margins of error, unlike games of chance. How good is the sample, how honest are the respondents, how fairly were the questioned phrased, what might cause a respondent to change their mind when it comes time to cast their vote…
I’m comfortable that polls do convey trends, and offer more predictive value when taken in the aggregate.
They may have some predictive value.
But the original question was, how often do you check them?
And then, why would you do so? Will it have any effect on how you plan to vote?
Presumably if you care, you are going to vote for the candidate you like anyway.
Good God, no.
(@Dinsdale, thanks for the pithy and spot-on summary of polling. I am not a mathematician either, but it seems “forecast” is a more accurate term than “prediction” for this sort of thing.)
Yeah they are not “unreliable” but they are neither precise nor accurate enough to be reliable winner prediction machines in a close contest.
And of course if it wasn’t a close contest and I wasn’t anxious I wouldn’t be interested in following them!
They are a tool of limited utility, but in terms of making guesses about the current mood, how it has changed, and results in the future, they are only actual data we got.
The closer the contest the more anxious I am, and paradoxically the more I will check on the status of a tool that is less and less useful for telling me what I most want to know.
Right. And, as I said a few posts ago, they can have some practical value, such as: “in early October, should I send $200 to a Nevada Senate candidate, or to a Texas one?”; or, “I have time to call or email five friends in mid-October, to suggest they talk to any of their friends that are on the fence about voting at all. Should I contact my Pennsylvania friends, or my Florida ones?”
That sort of thing.
But for that, I’d only have to check, say, once a week. So, yeah, the fact that I check them close to daily means it’s also a kind of spectator sport in which I have deep emotional investment. Like being on an airplane and watching the altimeter plummet. Does my watching the altimeter matter, really? No, but it’s hard for me to take my eyes off the numerical indicators as we get closer to crashing – or to landing, as the case may (yet) be.
Why do people who are interested in baseball check the standings every day? They make no difference; they change slowly over the season; the early standings have very low correlation to playoffs; the teams that make the playoffs are decided at the end of the season.
Yet, one does not need to be an obsessive fan of one team to want to check the standings on a regular basis. Many casual fans do this. Reasons are plentiful. It is fun. It is informative. It shows trends. It can be used for betting. The numbers can be played with.
Hundreds of gradual, accumulative measures can be found in daily life. Some people are interested in some; some people are not; very few are necessary. One can talk intelligently about a presidential race without knowing the results of any polls. But it’s harder if one ignores them.
And yes, polls are extremely accurate. Nate Silver, IIRC, called all 50 states in 2016 without running a single poll of his own, but just conglomerating the polls with better historic accuracy. He got almost all the congressional races correct. He got the popular vote correct. He did get the electoral college wrong, because three states had results within the margins of error.
But that’s not what people want to hear. They want a simple yes/no, who will be president? Polling cannot give them that and is not even set up in a way that a simple yes/no can be produced in our system. Education is needed to inform the voters of that truth. Obviously, the system is failing.
Almost completely agreeing with your post but this bit depends on what you’d define as extremely accurate. Systemic errors apparently can commonly be three percent ish. And we don’t know which direction that error, if it occurs, will be, with some believing in Trump’s favor, because the two past elections with Trump on the ballot were, or the other way, or a coin flip. So the whole set of results may even probably be shifted several points one way or the other than what polling says, or not. Not sure I’d count that as extremely accurate.
But that error, if it occurs, is by definition systemic, and consistent for the cycle. So for tracking relative positions polls are still of value. I don’t have huge confidence that polling of a Harris lead of 3% nationally and 1.5% in the most likely tipping point state will be a win in the EC, but I do have confidence that it is a major improvement in the odds over the same 3% nationally and 1% down in the most likely tipping point state. Precise odds I won’t believe.
I am sorry if I inadvertantly introduced a hijack.
I don’t agree that my statement was erroneous, but I have no interest in arguing about it.
If anyone wants to start a new thread about accuracy of polls, have at it.
Meanwhile, have a nice day. I’m done with this.
I can agree. Before that, I didn’t watch polls. And now I’ve gone back to not watching polls. If someone posts the results, I’ll read them, but I don’t go and check polls myself. I added the 538 site to my bookmarks (I believe Nate Silver actually has another site), but I haven’t actually looked at it, except when I read something that mentions polls.
So my answer is, not very often, and only if someone mentions polls first.
My response (and others’) was to your erroneous statement/question: