Why political success leads to failure

And further thoughts on polling

Soon I’d like to move on from politics and discuss my recent trip to Japan. But first a few (hopefully) final thoughts on the election. I’ll start with my subtitle, the fact that polls underestimated Trump for the third time in a row.

Part 1: Two views of probability

Some pundits have defended the polls, pointing out that the miss was within the margin of error. People like me responded by pointing to the fact that all the state polls missed in the same direction. This graph is from an FT article that defended the polls:

There are three ways to think about this picture. One possibility is that the final state polls occurred a few days before the election, and in the final days of the campaign nationwide opinion swung 3% toward Trump. That would explain why all the state misses were in one direction. In my view, that explanation is unlikely.

Another possibility is that the misses were due to sampling error. But there are two types of sampling error, excusable and inexcusable sampling error. The excusable type occurs when there’s a plus or minus 3% attached to polls because the 1200 people being surveyed just happen to include more Democrats or Republicans than you’d find in a comprehensive survey of the entire population. That sort of error is inevitable in polling. But importantly, those errors should not be correlated across states.

The inexcusable sampling error occurs when your sampling technique consistently fails to reflect the proportion of Harris and Trump supporters in the general population. Thus imagine a set of polls of a Michigan governor race that only surveyed Detroit residents—they would all show similar bias. That sort of sampling error is inexcusable. My claim is this is exactly what the polls did—three election cycles in a row.

It all comes down to the question of how you think about probability. Recall the example of the coin flip, which is a 50-50 event. Even after the coin is flipped, but quickly covered up by a hand, you might say there’s a 50-50 chance or heads or tails. But after the coin has already been flipped the result is completely determined, even if the hand briefly prevents you from seeing it. During this period, the probability merely reflects the state of our ignorance, not the fact that the coin could go either way.

Those who defend the state polls are treating the election like it was still “up in the air” at the time of polling, and the results were correlated because the national coin fell toward Trump at the last minute, pushing all 50 states more in that direction. I think the results were already basically determined when the final state polls were taken—the coin had already landed—they simply sampled the wrong people. I see no sign that anything fundamental changed in the final few days of the campaign.

Of course I might be wrong. Obviously my interpretation flatters me to some extent, as I predicted roughly the final results in my pre-election post. We’d all prefer to believe that we are smart, and not just lucky. Indeed I often criticize bubble-mongers for confirmation bias, after they accurately predict that a certain asset market is showing irrational exuberance and is about to fall. Perhaps I’ve become “that guy”.

Part 2: Post-mortem on Wisconsin

Since I’m in the mood to pat myself on the back, let me revisit one more post, before moving on to the main topic.

In my tipping point state post I said:

Wisconsin differs from many other states in having a strong tradition of clean government. When I was young, our voters were extremely allergic to corruption, and I suspect that’s still true to some extent. Wisconsin residents looked at the politics of states like Illinois with total disgust. In 2020, Wisconsin was slightly redder than Pennsylvania. If this reverses in 2024, it will probably be because all the Trump scandals hurt him slightly more in the Badger State than in Pennsylvania.

OK, I deserve zero credit for such a wishy-washy prediction. But the actual data is quite interesting. Once again, here’s the FT:

Obviously the vast majority of counties shifted to the red, by an average of roughly 6%. Hispanic areas like Miami and South Texas shifted especially strongly, as did big cities (which often have large Hispanic populations.) But a few counties shifted blue.

In Wisconsin, Door County shifted blue. It’s a vacation area (the “thumb”), similar to the Michigan vacation counties across the lake that also shifted blue. Perhaps they drew some work-from-home Democrats.

The other three blue shifts were in Milwaukee’s three collar counties. Milwaukee County (which includes the city of Milwaukee and some inner blue collar suburbs) shifted red. The three suburban counties have lots of well-educated professionals who traditionally vote Republican. My claim is that this is exactly the group that I would have expected to be turned off by things like January 6. (To be clear, the counties still went strongly GOP, it was a blue shift at the margin.)

People are Republican for all sorts of reasons. Someone might think “I strongly oppose abortion, so I’m a Republican.” Or “I don’t like minorities, so I’m a Republican.” Or “I hate paying taxes, so I’m a Republican.” Or “I support the principle of small government, so I’m a Republican.” My claim is that this last group is especially concentrated on Milwaukee’s collar counties. These are the well-educated and idealistic Republican voters that worried about Trump’s misdeeds.

[After I got home from Japan I watched a Milwaukee basketball game I had taped and saw a Harris commercial full of top Trump officials saying he was totally unfit for office. This commercial was clearly aimed at exactly the GOP voters I have in mind.]

You also see those blue arrows in some suburban countries outside Indianapolis, Columbus, and Atlanta. And lots of blue arrows in Utah, which has a tradition of clean government. But the suburban counties outside Detroit, Chicago and Philadelphia have far too many working class suburbs to exhibit the blue shift. It’s very specific groups that shifted away from Trump. Milwaukee’s blue-collar suburbs are mostly within Milwaukee County.

You might say that the blue shifts are tiny. But keep in mind that even within these blue shift counties, there are many voters that shifted toward Trump. So in the specific category of well-educated idealistic Republicans, Trump was hurt significantly. Indeed this group has been gradually shifting away from the GOP for decades, partly due to other issues like abortion. In Trump’s favor, he made big gains among Hispanics, blacks, Asians and working class whites. Those gains were far more important than the erosion of support among well-educated white Republicans.

Part 3: Why political success leads to failure

Let’s suppose that the GOP brings a bunch of lawsuits and successfully rolls back woke policies on college campuses. Let’s suppose they secure the border, and keep inflation at 2%. What then?

In the short run, voters may reward them with re-election. But in the long run, they are digging their own (political) graves.

Voters have a short attention span. If you solve their problems they may thank you, but their attention quickly moves on to other problems. Unfortunately, you are the party in power. So you will be blamed for those other issues where the voters have switched their focus.

If you hire a plumber to fix a leak, you don’t want the plumber hanging around when the job is done. If you hire a political party to solve a set of problems, you lose interest in the party once the problems are solved. Oddly, in some ways the GOP would be better off if they tried but failed to smash wokism on college campuses, just as for decades they were better off when they tried and failed to ban abortion.

Part 4: Are we becoming more or less Christian?

The previous argument may also apply to religion, which bears some similarities to politics. I’d argue that the 19th century represented both the belated success of Christianity and the end of Christendom.

Here’s I’ll use the term ‘Christendom’ to mean something more than “85% of the population is Christian”. Rather, it generally refers to the region where Christianity is the defining feature of the culture. By this definition, Christendom has been eroding in the West; first in Europe, and more recently in the US.

The 19th century was the beginning of the end for Christendom, with the French Revolution being an important turning point. But in the long run it was the advance of ideas such as Darwin’s Theory of Evolution that began to erode religious belief. By the late 19th century. Nietzsche had declared “God is dead”. Of course, this was only true for a minority, but it was a very important minority—including many intellectuals.

I hate it when social scientists talk about the effect of “religion”. What the heck do they mean by religion? A set of written ideas from 2000 years ago? A set of (totally different) current beliefs? A culture? A tribe? An institution? In any case, you can argue that Christendom began dying in the 19th century. But you can also argue that Christianity only began to succeed in the 19th century, that the 1800s is when the West finally became Christian. To make this claim, I have to define Christianity not in terms such as institutions like the Catholic Church, but rather in terms of ideas like the Sermon on the Mount.

It seems to me that for its first 1800 years, Christendom mostly ignored the teachings of Jesus. Europe had a basically aristocratic culture, where the elite were especially respected if they engaged in warfare, and the poor were treated like dirt. That doesn’t seem very Christian!

In the 19th century, the West began to have a greater empathy for those at the bottom. Slavery was abolished. Democracy began to spread. Socialist ideas like income redistribution were adopted. Remember “Turn the other cheek”? In 1804, Hamilton was killed in a duel. By the late 1800s, dueling had been banned. Gradually over time, warfare became viewed less favorably.

In an earlier post, I discussed the prose of Herman Melville. In several novels, he made passionate arguments that society wasn’t truly Christian, as it had not truly absorbed the teachings of Jesus (which he viewed as profoundly important.) In White-Jacket, Melville made a strong argument against corporal punishment on US Navy ships. In Pierre, he portrays a character that sacrifices so much to help a poor woman that he is treated as if he is insane and is expelled from polite society.

Today, Melville has won his battles. Corporal punishment has been banned. Pierre would no longer be kicked out of polite society for his extraordinary act of generosity. We have finally become a Christian culture. (Judeo-Christian is perhaps more accurate.)

But Christianity in an institutional sense seems to be declining in the West. It’s as if the public is saying “We thank the Church for preserving the teachings of Jesus for 1800 years, but we don’t need you any longer. We have made these ideas a part of our secular philosophy, our social science, our politics, our culture. You’ve done your job, now please go away.”

Of course people like G.K. Chesterton and Ross Douthat would say that this won’t work, and perhaps they are right. I’m totally unqualified to predict the future course of society (even in a world where AI was not about to shake things up.) I’m just reporting what I’ve seen happen so far. After almost 2000 years, Christianity finally succeeded, and Christendom began fading away.

PS. Why does everything remind me of Dylan lyrics?

“There’s no success like failure”.

PPS. One of those three Milwaukee suburban counties is right on Lake Michigan. It is far less Republican than the other two. That’s consistent with my view that beautiful places like Marin County, Vermont, etc., tend to be Democratic. Republicans are more materialistic, liking big houses and big cars, while Dems like beautiful scenery. The Dems are the party of people with passports, people who go to art museums. On the other hand, Republicans are less neurotic. Each side has advantages.

BTW, the Great Lakes are like mini-oceans. On this very day in 1975, a 729-foot iron ore carrier sank during a severe gale in Lake Superior, with a loss of the entire crew of 29 sailors. These are not the placid little lakes you see in other parts of America.

PPPS. I liked Don Boudreaux’s essay on why his three siblings voted for Trump. I don’t agree with his siblings, but it’s a very thoughtful article.