The point of an earlier post about historical models has not gotten through, judging from how concerned people are about the latest campaign brou-ha-ha, worrying about the debates, and so on.
Whether the incumbent party will maintain control of the White House can be predicted months, sometimes years, in advance. Helmut Norpoth and Allan Lichtman, whose models were discussed, have been doing that for several decades, and their models work retrospectively over the past 100 years (Norpoth) to 150 years (Lichtman).
The verdict is in: the incumbent party will lose control, meaning Trump is going to win.
Now, these models do not forecast what the important topics will be during election season, nor which positions on those topics will be the most popular. They do not predict the Electoral College vote (other than to say that whoever wins the popular vote almost always wins the EC), nor do they predict which states will side with which party. They do not give any hint of what the zeitgeist will be like, according to contemporary observers or future historians. They don't even necessarily know who the candidates will be. They simply measure signs of stability vs. disruption.
The kinds of things that these models look at are macro-level conditions that apply over the past four to eight years. For example, there's first-term incumbent advantage -- but an incumbent penalty if the party is going for three or more in a row. Was there discontent with the incumbent Presidential party during the previous mid-term Congressional elections? Which party had the more evenly contested primary battle (weak candidate), and which one had the more lop-sided battle (strong candidate)? Are people happy with the direction the economy has been heading over the past four years? This may differ depending on which section of society is responding. Do people feel more protected or less protected from foreign threats? And so on and so forth.
These thoughts and feelings have been brewing throughout the past four years, largely unconsciously. By the time the election season kicks into high gear, it is too late to alter a person's gut-level intuition about whether they're going to vote for the same party or changing the guard.
This is why campaigns largely do not matter, at least once the primaries are over. People's minds are mostly made up before the general election season has even begun. They will only respond to the output of campaign season -- from the candidates themselves, from the media, from social buzz -- by accepting something if it is concordant with their already formed decision, or rejecting it if it is discordant.
Someone who already felt content with continuing the status quo will seize on Trump's latest problematic words about women from 20 years ago, and dismiss the latest in a long line of leaks proving how corrupt Crooked Hillary has been for her entire career. Someone who already felt fed up with the status quo will seize on the video of Hillary collapsing and being dragged lifelessly into the car, while they will dismiss leaked audio showing Trump to be a skirt-chaser in the past.
These are ad-hoc rationalizations of long-formed gut-level intuitions. Arguing over them will not alter the outcome of the election. People might as well just shut up and wait until Election Day to do what they were already planning to do -- but with the team-vs.-team spectacle that politics has become, voters cannot help themselves. The media cater to this demand for fighting it out over every micro-event in the campaigns.
Trump was onto this in the primaries when he said he could shoot someone in the middle of Fifth Avenue and still win. He was right because Republican primary voters were fed up with the status quo of the party, for a very long time in fact, and Trump was the only unorthodox and disruptive candidate. Voters who wanted change had only one choice. Of course, you have to be in a league of your own in the way that voters desire -- otherwise you're out immediately like Jim Webb during the Democrat primary.
Please, keep this in mind as we run the gauntlet of the final month of election season. Don't obsess over every nano-fart in the news cycle. And do not spazz out about WHAT TRUMP MUST DO to win. People are going into the voting booth thinking of the past four years, not the past four days.
I'm sure it is possible to botch a campaign so badly that in spite of historical trends, it can go the other way. However, that isn't the case here. At Wisconsin, audience members at the even Trump and Pence were supposed to go to were met with boos as Paul Ryan tried to throw Trump under the bus. Trump also received a #yuge reception from an group of people outside Trump Tower. This controversy really is no different than the Khans, Judge Curiel, etc. With the debate tonight, the media will start whining about that tomorrow. By mid-week, this controversy will be totally gone and the media will be hyping something else. Meanwhile Assange promises to drop one new set of leaks per week for the next ten weeks to celebrate the Wikileaks 10th anniversary. Will Hillary be able to have a controversy at the ready for the remaining four weeks? I doubt it.
ReplyDeleteThe current situation is definitely felt by the voters. The economy is still terrible and the jobs report reflects that. People really don't want a proxy war with Russia in Syria, despite years of being told that Putin is a bad man. People still want actual immigration reform and not open borders. People don't want their health care costs to spiral out of control. Nothing Hillary is proposing seems to be what the people want; it's a sign of an out of touch elite that is thundering on with neoliberalism and neoconservatism, ideologies that are being rejected by both the left and the right.
How much do voters care about the past four days? Trump's lead shrunk by just under 1 point in the USC poll.
ReplyDeletehttp://cesrusc.org/election/
And his numbers have been gently falling since the start of the month, and hers have been slowly rising. So the brou-ha-ha did not cause the latest 1-point narrowing -- it was under way, and is part of the cycle I've been pointing out for awhile now.
If I gave a naive observer that chart, they would not know that anything "major" had happened over the weekend. It has been a non-event.
http://i.4cdn.org/pol/1476055676791.jpg
ReplyDeleteOnly 8% of Ohio Trump voters think less of Trump for this scandal. Also that does not mean that those 8% won't be voting for him either.
While cuckservatives are ready to throw him overboard yet again, the average public does not care. I hope once Trump gets elected, we start purging the party of these losers and traitors. Starting with Paul Ryan. On November 9, he needs to tender his resignation as Speaker of the House.
Agnostic,
ReplyDeleteMy problem with both Norpoth and Lichtman’s models is that the nation they use as their model no longer exists. To the America that could swing in eight years from Johnson’s 61% to Nixon’s 60% has been added a new voting bloc which follows Lee Kwan Yew’s observation on voting according to ethnic and racial interests. New York, California and Illinois are locked in Democrat. Two of those large states flipped from D to R from ’64 to ’72. That kind of change is unheard of in the last 20 years. This is because the old electoral elasticity exists mainly in the white vote and that percentage of the vote has shrunk enough to minimize the impact of any cyclical dissatisfaction shift. Future American elections will probably resemble those of a Latin American nation, where the “natural majority” party only drops out of power for short periods after significant crisis before returning to their normal position.
I can see the shifts in polling data from states in the Northeast and Midwest, but it just doesn't seem to be enough to overcome the locked in voting blocs there.
You know nothing at all about the past 30-40 years of electoral or demographic history.
ReplyDelete2012 electorate was 72% non-Hispanic white. Babies don't vote -- old people do.
How many Amerindian immigrants do you think there are in the Great Lakes, New England, and the Pacific Northwest? And of them, how many vote? (Spoiler: zero)
The whole country is not California and the ACELA corridor.
You're just bitter that white Americans have resoundingly rejected the Conservative Movement (TM) of the first Bush presidency. We're never going back there, thank God. Elitism is dead, whether globalist elitism or nationalist elitism. So is cognitive elitism, something else you probably fetishize.
And in any case, the historical models do not rely on any assumptions about demographics, only signs of disruption vs. stability, candidate strength with voters, etc.
Agnostic,
ReplyDeleteI'm on your side! The first election I could vote in was 1992 where I voted for Buchanan! I bought my Trump stuff in September of last year. I actually think that the one good thing we will get for sure out of this election is the end of the Republican Party as we knew it and the unmasking of the country club/Chamber of Commerce conservatives as the anti-core American quislings that they are and have been for a generation.
I will defer to your (and Norpoth and Lichtman's) knowledge of demographics and electoral history, but the thing that is hard for me to get past is that how all the polls-aside from the USC/Dornsife tracking poll-can be getting it so wrong. To what do you attribute this? Oversampling Democrats? Using past assumptions on turnout? Malice? All three? Shouldn't it be in the interest of some of these companies to try and get things correct, if only to help their future reputations for accuracy? I understand that USC's poll uses the methodology of RAND's poll in 2012 which was extemely accurate. Nate Silver seems to think they initially oversampled Trump supporters, which is skewing their totals. Is that possible?
The polls are becoming more and more bogus mainly because they think it'll help block Trump, and that is their main goal. They stack their sample with 70% who voted for Obama in '12, or 50% has more than a bachelor's education, or something stupid like that. Exaggerating, but true.
ReplyDeleteThen there is the change in the willingness of Trump supporters to respond to pollsters when there's a slump in the campaign:
http://akinokure.blogspot.com/2016/10/poll-shifts-after-events-are-illusory.html
That's why the panel polls like USC have been more immune to the wild swings that the RCP average shows. Emotions may be swinging that wildly, but intention to vote, and who to vote for, is not swinging that much, this late in the race.
Polling companies do not care about their reputations this time. They'll take the hit, and their future customers (media who commission polls) will understand and not take that into account. "You were terrible predictors because you were trying to stump the Trump, we won't hold it against you."
None of the organizations that did polling last time looks honest this time -- some worse than others, but all of them have green-lighted bogus polls to stop Trump. Quinnipiac, PPP, CNN, Reuters -- they were good last time. Not anymore.
The two who show up on RCP that are honest are both new -- LA Times / USC and Emerson. Emerson did a bunch of polls during the primaries, and were 94% accurate, but they didn't do the 2012 season.
The USC poll uses the RAND 2012 methodology, but it is being used by a new group this time, and a major newspaper at that. So credit to the LA Times for not junking their superior methodology. They have been far better in general at investigative and data-based reporting than NYT, WP, and other defunct papers.
Nate Silver argues that the LA Times poll is so in favor of Trump compared to the others because it's original sample was probably already too highly in favor of Trump.
ReplyDeleteWhat's the counter-argument?
The counter-argument is that this proves Trump has more support than is commonly acknowledged, or measured by the ad-hoc polls.
ReplyDeleteFirst, he has nothing to support his argument, other than his own wishful thinking. "Uh, well, uh, maybe they already had more Trump supporters?"
Second, he can read the methodology in detail or in summary at the poll's website.
"The Daybreak Poll is based on an internet probability panel survey. Daybreak Poll members are participants in the ongoing UAS internet probability panel of about 5,500 U.S. residents who were randomly selected from among all households in the United States. Members of recruited households that did not have internet access were provided with tablets and internet service."
Probability sample means they sampled at random to make it representative, rather than a convenience sample of whoever feels like participating.
And it was not chosen with the purpose of monitoring this election -- it's an already existing, ongoing panel that allows them to study all kinds of things, including voter behavior.
The fact that they had to provide internet access and/or tablets to some participants, means that the sample is more representative of America than a generic internet survey. It reaches those who are older, less tech-savvy, and not in urban areas.
We also know that people are more honest in internet polls than with push-button phone polls or live person interviews. More anonymous. Especially worth having when some people are nervous to admit voting for the leading candidate to a person.
"Each night, Daybreak Poll results are weighted to match demographic characteristics (such as race and gender) from the U.S. Census Current Population Survey, and aligned to the 2012 presidential election outcome using how respondents tell us how they voted in that election."
So they weight the results to resemble the population at large, though not necessarily the electorate of any election.
Then they make that American-looking population resemble the 2012 electorate by voting behavior -- not race, gender, age, etc., but just getting the percent who voted for Obama vs. Romney correct. (That's how I interpret it.)
That does not pick up the people who didn't vote last time, or the last two times, or whatever. But as I discussed in another post, these infrequent voters tend not to make up much of the electorate, and will not this time either.
ReplyDeletehttp://akinokure.blogspot.com/2016/09/the-minimal-role-of-infrequent-voters.html
It goes in cycles, with peaks every 4-5 elections -- 1972, 1992, and 2008. Will not peak again just two elections after the last one. But will go up -- somewhere between 5-10% of voters this year will not have voted last time. But that's completely within the range of what's normal.
They will definitely be even more pro-Trump than the returning voters. But they are not that numerous to make a wild swing -- only if it's an even race among the returning voters (which is what happened in 1992).
Reading Silver and this guy (Sam Wang) is pretty depressing in terms of their exuberant confidence in the number-crunching the aggregation of polls (he's now at 95% HRC):
ReplyDeletehttp://election.princeton.edu/2016/10/10/some-secrets-are-not-all-that-dirty/#more-17917
Not to mention the mainstream media's full-court press on the "Trump is toast" narrative. Here's hoping your contrarian macro view holds up.
The Establishment has been saying that Trump is toast from the beginning, when Cruz won the Iowa primary.
ReplyDelete"he's now at 95% HRC"
ReplyDeleteHe was adamant that Trump was not going to win the nomination. He called Trump's expectations of victory "delusional" just a few weeks before he finished off Cruz in Indiana.
I remember real money prediction/betting markets giving Brexit 10-20% odds of winning within 1-2 weeks of the vote. The plain fact is that the fog of war is intense on this one (both due to the unusual nature of the election, and the fact that the MSM is spinning as never before), so anybody claiming 95% certainty of the outcome is shilling, period.
Sam Wang doesn't add any value or insight beyond what the polls themselves say. He wasn't trained as a political scientist, and doesn't have any historical perspective.
ReplyDeleteI wasn't trained either -- but that's why I looked to the pros who have been accurately forecasting elections for decades, and whose models retrospectively match the outcomes back 100-150 years. Wang has no such record.
A sperg who obsesses over the trees while not seeing the forest, he believes the opposite of the title of this post -- the election is about the past four days, not the past four years.
What was that other internationally famous vote over the same issues as our Presidential election...? Seems like it was just barely over three months ago... Brexit? Brexit anyone? Here is this failure's final post *on the day of the vote*:
http://election.princeton.edu/2016/06/23/brexit-home-stretch-polling/
He says it's Remain +1.0, +/- 1.8. If that 1.8 means one standard deviation, then the lowest that his 95% CI goes is 1 - 2*1.8 = Leave +2.6. In reality, it was Leave +3.8 -- or 2.7 SDs away from his prediction.
He posted a weak update as the votes were coming in that Leave appeared to be gaining likelihood, but failed to update further and admit his model fucked up royally.
He's just dressing up the publicly available polls to make them seem more intellectual, to be consumed by TED Talk nerds and despair cucks.
Forget this guy.
Good to know. I realize Wang (heh) is a poli-sci amateur, but it seemed compelling on the surface to this layman. Thanks for providing a countervailing analysis and some hope against the unprecedented propaganda blitzkreig.
ReplyDelete