I like models also because the more data, the better. One aspect that I honestly don't understand is whether the models (perhaps indirectly) take account of the fact that in the real world, voter suppression, corruption, and all kinds of other bad things occur. You can see why that matters more this time.
As a historical example, perhaps Stacey Abrams would also have lost to Brian Kemp in a perfect "Sim Cities" type clean election for governor. But it's almost impossible to argue that that election was like that, with attempts to shut down voting places where Dem voters were in the majority, wait times in the long hours in some precincts and in minutes in other precincts favoring Kemp, and major election-day decisions by Kemp early on that aided him and hurt her, given that until close to the end Kemp was (oddly) also the guy in charge of administering the Georgia elections.
Should an election model in some way take account of this ("per our numbers, candidate X would win if everybody who's registered is able to vote, but we've had to deduct 3 points from them due to voter suppression")? Or is that something outside of the model that observers have to guess about and add to the projections, which kind of undercuts the models' own value?
Thanks for breaking down the concept and practice of election models and highlighting some of the major ones. I'm interested to see how much of a difference there is between Nate Silver and ABC's 538 - what do you think are the differences between their approaches right now?
Signage means little in the Trump era, due to the cult-like nature of his movement and its followers. Those folks seem hellbent on letting the world know how easily duped and misguided they are.
So you are saying this is like making sausage! Throw a little of this and a little of that and don 't tell anyone your secret spices. It is up to us to see which tastes better!
Who votes more regularly, women or men? young women or young men? I keep reading that this is becoming a gender election. I still am surprised that anyone answers a phone call from someone they do not know. Who answers phone calls? any polling on that?
Chris did a piece on polling a few months ago that addressed this. Data is not solely collected by phone anymore. Polls are put out through social media and other online platforms and both my husband I have received texts asking us to take part in a poll. Having said that, I have not and would not respond to a poll delivered via one of these means, and I still have a hard time believing the results of such polls are an accurate representation of the electorate at large.
No one answers the phone from an unknown number, so polling is very suspect. Look for actual things people are doing, yard signs, going to rallies, ect. If people go out of their way to do these things they surely are gonna go vote! Yard signs in my swing state are dominant by Trump! What is your experience?
No one answers the phone from an unknown number? No one? Not ever? On what basis are you making this broad declaration?
And most voters are NOT going to rallies or even putting up yard signs. (I live in an apartment. No yard signs here.) The people doing so are likely to vote, but are not at all a representative sample of the voting population.
WTF does that even matter? Outside of telling me that you're biased to trump/vance, are unserious as a voter, and are immature enough to let someone else provide your insults?
I agree...never answer a number you don't know! But also do people answer questions honestly?? So as far as yard signs go, easy bet that those people will vote but honestly there is a greater majority that silently vote which I think makes that unreliable too.
I am in SLC, UT.... a blue spot in a deeply red state. i ordered a Harris sign... hasn't come, yet. I haven't seen a ton of either sign. The ones that are out are for Harris/Walz, but again, i live in the tiny liberal spot in UT. I teach down at UVU, in Orem... deeply red. Definitely more Trump signage in that area and lots of bumper stickers. Of course, UT will go R, but the LDS folk i know don't always love Trump, they are historically conservative and therefore, liberals are scary. My students, god love them,( and i do! ) are not fully invested in all of this. I do not know how this can not be a concern to them! A
Thank you for doing what you do, few things are more important than sharing wisdom with the next generation. Young people have so many things going on today politics is certainly low on the priority list. I agree, UT is definitely a red state going for President Trump. Here in Michigan, a very important swing state, the yard signage is decidedly in favor of President Trump.
Something to emphasize is that models use probablity rather than voter intentions, i.e., a coin flip is 50-50, heads would probably range between 47-53 in a hundred real flips. Silver reminds us that Harris, at a 55-45% probability, is essentially tied with Trump. In 2016, Silver's 538 last election model had the HRC 71-29; in 2020, 90-10 for Biden.
In the first case, Trump threaded the probability needle; in the second, Biden did what was predicted, albeit with much lower margins than forecast. Models are best at following trends, although no model caught Trump's late surge in 2016.
We are actually talking apples and oranges. I thought you were referring to the polls, but you were referring to the odds. I was referring to the polls. Yes, the press had pretty much been convinced Clinton was going to win, as did Clinton. The polls, while favoring Clinton, were more conservative.
I like models also because the more data, the better. One aspect that I honestly don't understand is whether the models (perhaps indirectly) take account of the fact that in the real world, voter suppression, corruption, and all kinds of other bad things occur. You can see why that matters more this time.
As a historical example, perhaps Stacey Abrams would also have lost to Brian Kemp in a perfect "Sim Cities" type clean election for governor. But it's almost impossible to argue that that election was like that, with attempts to shut down voting places where Dem voters were in the majority, wait times in the long hours in some precincts and in minutes in other precincts favoring Kemp, and major election-day decisions by Kemp early on that aided him and hurt her, given that until close to the end Kemp was (oddly) also the guy in charge of administering the Georgia elections.
Should an election model in some way take account of this ("per our numbers, candidate X would win if everybody who's registered is able to vote, but we've had to deduct 3 points from them due to voter suppression")? Or is that something outside of the model that observers have to guess about and add to the projections, which kind of undercuts the models' own value?
Thanks for breaking down the concept and practice of election models and highlighting some of the major ones. I'm interested to see how much of a difference there is between Nate Silver and ABC's 538 - what do you think are the differences between their approaches right now?
What does it impky if all three types (Modeling, Alan L, and Polls) have one candidate favored over the other.
As a consensus, is that more reliable or predictive?
Signage means little in the Trump era, due to the cult-like nature of his movement and its followers. Those folks seem hellbent on letting the world know how easily duped and misguided they are.
I like the model with the little fox on it, or whatever it is.
https://projects.fivethirtyeight.com/2024-election-forecast/
So you are saying this is like making sausage! Throw a little of this and a little of that and don 't tell anyone your secret spices. It is up to us to see which tastes better!
Chris,
Who votes more regularly, women or men? young women or young men? I keep reading that this is becoming a gender election. I still am surprised that anyone answers a phone call from someone they do not know. Who answers phone calls? any polling on that?
thanks
Ang
Chris did a piece on polling a few months ago that addressed this. Data is not solely collected by phone anymore. Polls are put out through social media and other online platforms and both my husband I have received texts asking us to take part in a poll. Having said that, I have not and would not respond to a poll delivered via one of these means, and I still have a hard time believing the results of such polls are an accurate representation of the electorate at large.
No one answers the phone from an unknown number, so polling is very suspect. Look for actual things people are doing, yard signs, going to rallies, ect. If people go out of their way to do these things they surely are gonna go vote! Yard signs in my swing state are dominant by Trump! What is your experience?
No one answers the phone from an unknown number? No one? Not ever? On what basis are you making this broad declaration?
And most voters are NOT going to rallies or even putting up yard signs. (I live in an apartment. No yard signs here.) The people doing so are likely to vote, but are not at all a representative sample of the voting population.
Let me guess, do you own a cat? I say yes….
Yes. And now let me guess: you can't back up your claims.
WTF does that even matter? Outside of telling me that you're biased to trump/vance, are unserious as a voter, and are immature enough to let someone else provide your insults?
I agree...never answer a number you don't know! But also do people answer questions honestly?? So as far as yard signs go, easy bet that those people will vote but honestly there is a greater majority that silently vote which I think makes that unreliable too.
I am in SLC, UT.... a blue spot in a deeply red state. i ordered a Harris sign... hasn't come, yet. I haven't seen a ton of either sign. The ones that are out are for Harris/Walz, but again, i live in the tiny liberal spot in UT. I teach down at UVU, in Orem... deeply red. Definitely more Trump signage in that area and lots of bumper stickers. Of course, UT will go R, but the LDS folk i know don't always love Trump, they are historically conservative and therefore, liberals are scary. My students, god love them,( and i do! ) are not fully invested in all of this. I do not know how this can not be a concern to them! A
Thank you for doing what you do, few things are more important than sharing wisdom with the next generation. Young people have so many things going on today politics is certainly low on the priority list. I agree, UT is definitely a red state going for President Trump. Here in Michigan, a very important swing state, the yard signage is decidedly in favor of President Trump.
Something to emphasize is that models use probablity rather than voter intentions, i.e., a coin flip is 50-50, heads would probably range between 47-53 in a hundred real flips. Silver reminds us that Harris, at a 55-45% probability, is essentially tied with Trump. In 2016, Silver's 538 last election model had the HRC 71-29; in 2020, 90-10 for Biden.
In the first case, Trump threaded the probability needle; in the second, Biden did what was predicted, albeit with much lower margins than forecast. Models are best at following trends, although no model caught Trump's late surge in 2016.
https://www.huffpost.com/entry/polls-hillary-clinton-win_n_5821074ce4b0e80b02cc2a94
I still have PTSD flashbacks to HuffPo saying Hillary Clinton had a 98.2% chance of winning
Thank you for this educational piece, Chris - I had heard way more about polls than models. Now I know more. 🤓
So insightful
And HRC was favored by 90% on Election Day, we all know how that went…..
Clinton was favored, but not quite the 90% you state. Here’s what 538 had on Election Day 2016: Clinton 45.7%, Trump 41.8%.
https://projects.fivethirtyeight.com/2016-election-forecast/national-polls/
Reuters 11/07/20216 @ 4:07 PM
Clinton had a 90% chance of winning.
VP Harris running the same campaign, only anti President Trump all day everyday!
Edmund Burke famously said “those who don’t know history are doomed to repeat it” you can see the train wreck coming a mile away
We are actually talking apples and oranges. I thought you were referring to the polls, but you were referring to the odds. I was referring to the polls. Yes, the press had pretty much been convinced Clinton was going to win, as did Clinton. The polls, while favoring Clinton, were more conservative.
There's a train coming, but I don't think it is the train you think it is. 🤔
That was that guy who said he'd eat a bug if Trump won. And of course he did!