I do wish polls would somehow include the expected probability of how the poll will translate to the voting booth. I have no doubt that polls can accurately gauge the sentiment of the population at a moment in time. However, not everyone will actually show up to vote. The election results may differ simply because some of the people polled decide to do something else with their time. If the polls show that 50 people want Trump and 50 people want Harris, what matters in the end is what percentage of people actually cast their vote.
It’d be helpful if polls could say something like, although the polling shows a 50/50 split between the candidates, the supporters of candidate A seem to be 10% more likely to actually cast a vote than the supporters of candidate B. Therefore, the results of this poll would be expected to translate to an election result of approximately 53/47.
Riffing on engagement: Some commentators use small-dollar donations as a proxy for engagement. Donations are a different kind of tool from approval polling, but it makes intuitive sense that the August donation imbalance between the two campaigns tells us something about likely engagement on November 5th.
You were the one who earlier in this thread pointed out how the big difference then and now was the paucity of “undecided” voters in this cycle’s polling. I don’t think it’s the voters being undecided that much less. I think reflects the
That tells you that something is different this cycle and suspect @DigitalC is right.
Am I missing it? I’m looking to see a breakdown by donation size, to see how many are “small” donations on each side. Bigger big donations and even more of them are important but not that engagement proxy we are looking for.
Here’s some brief articles from about a month ago (Reuters) (Al Jazeera) (Financial Times). These sources, and much of the popular media, use and cite campaign finance data collected by the non-profit Open Secrets.
I am hoping that Open Secrets updates soon, since it’s been roughly a month and the August campaign funding data were made public late last week. For now, the August 20, 2024 data is the most current on their site:
Elevator pitch version: harris picked up 1.5 million new donors, mostly of small amounts, representing 40% of all the campaign’s donors. (That counts Biden donors.) The new donors are a lot younger and more female than Biden’s donors. They are slightly better educated, and slighter blacker than Biden’s donors. They have similar household incomes.
This article focused on small donors. "Full details on donations by large donors will not become clear until October, when the campaign’s associated fund-raising committees are required to file reports with the Federal Election Commission. "
I didn’t look for data on Trump donors. But the article mentions he got a wave of donations after his felony convictions and after the attempts on his life.
Tying all this back into this thread’s topic: Small-donations-as-enthusiasm is the kind of information that traditional political polling doesn’t capture. It’s not “dollars = votes” because the example of 2016 (Hillary’s large war chest losing) can always be invoked. It’s more nuanced than that.
EDIT: Comparing to Harris’ small donor figure of 42% … Barack Obama was at 45% in 2008 and nearly 43% in 2012. I will be interested to see Harris’ small donor percentage when Open Secrets next updates.
All of these swing-state polls are very unreliable. Of course, they’re small samples and the error margins are bigger than the margins between the candidates. The pollsters tell you their error margin is about plus and minus 3 percent. That’s pure statistical error. That’s the error you would get if you had a huge jar of green and red balls and you took a sample and estimated the percentage of green and red balls. But human beings are not red and green balls. Most of us don’t respond to pollsters. In 2016, the pollsters underestimated Republican voting strength. So, like generals fighting the last war, they tried to correct it. And recently, they’ve been underestimating Democratic voting strength. For example, in the marquee special election of 2024, to the seat previously held by the disgraced George Santos, the New York congressional seat, a poll taken right before the election had the Democrat up by one point. He won by eight points, exceeding the poll numbers by seven points.
I shared more of his thoughts in another thread, but this part bears repeating.
Yeah, I still don’t know quite what to make of it. On this day in 2020, for example, Biden’s RCP average was 50%. Harris today is 49.7%. So basically no change. But back then Trump was at 43% and now he’s at 47.2%.
Are the pollsters just doing a better job of reaching Trump voters (or being more in aggressive in their weighting to account for them)? Or are there simply more Trump voters in existence and those “unreachable Trump voters” still exist and are going to swamp Harris on election day?
We do know that pollsters have made changes - they admit as much in various articles and interviews. We also know that they were very accurate in 2022, but that was a midterm and doesn’t really provide much insight.
I guess I would say odds are the polls are a bit better than 2020, simply because historically they have rarely been worse and some regression to the mean is to be expected. But I’d be pretty surprised if they are spot-on or if they miss in the other direction. It’s possible, but there does seem to be something specific to Trump that makes gauging his support via poll difficult.
If I had to bet today I’d say the polls will probably over-predict Harris’ margin by 1-2% (roughly half of the miss in 2020). And she probably needs to win the popular vote by 3% or so to win in the EC. So I’d really want to see a 2-point swing in her direction before I’d feel even remotely confident.
At least one pollster (I can’t remember which one of the top of my head) said that a lot of people in 2020 basically just yelled “We are voting for Trump!” and hung up, and they didn’t count those but now they are. It seems very unscientific to me.
Very unscientific? Compared to physics, yes. Compared to a newspaper column, no.
It may be comparable to surgery. In all but the simplest operation, there will be many judgment calls. Studies do not exist proving what every move of the scalpel should be. But surgery is highly informed by scientifically tested hypothesis. I similarly say that polling, while not science itself, is informed by science.
There is a reason there are a lot more questions in polls than who you are voting for. This sounds more like just giving Trump a handicap because they were so off last time.
I suspect that most pollsters are over-correcting for the shy-Trump-voter thing. Which is why many Senate race polls are not in congruence with the Presidential race polls. These days there’s just not that many split-ticket voters. Like generals fighting the last war, pollsters are correcting for the last election.
Which brings up the other thing I think pollsters are missing. There’ve been recent reports of lots more new voter registrations, especially of younger women, and you know those young women virtually all going to vote for Harris. Yet this doesn’t seem to be reflected in the polls. I suspect they’re mostly going under the pollsters radar.
One factor might be that they don’t answer pollster’s calls. Of all the groups of voters, I would guess that young women are the group least likely to answer calls from numbers they don’t recognize.
Meaning there was a 3% error due to systematic issues (like over- or under-represented sub-populations) that is independent of the margin of error from sampling?
I’m curious how they test and tune their models given there are so few data points from elections. From past elections I had the sense that 538 built their model by weighing each poll; the weight of a poll was based on their assessment of the methodology.
Thanks for the correction. He models using polls and presumably other data.
I keep saying this and have been saying it ever since Roe was overturned. It just gets me accused of irrational optimism. But I still think this is what is going to make the difference.
I would expect a newly registered voter, one that registered in the lead-up to the election, to be more likely to vote than your typical registered voter. I’m sure they have a good reason for registering at that time.
