Fool My Thrice: Polls and Reality
Some politics and math (don't panic), followed by coffee and the Bermuda Triangle
Come for the quiet ranting over how commentators are already making the same mistakes with polls that they repeatedly have. Stay for a story about coffee, one of my favorite tumblers, and the inevitability of supernatural loss.
Down the Polling Rabbit Hole
I’m not one to quote Ronald Reagan. But there they go again.
They, in this case, are journalists discussing recent presidential election results, not Jimmy Carter. Some people and outlets who do serious work have already expounded on shifts in support for political parties among various groups. The topic is identity politics and which demographic groups voted for whom. Some examples:
“A larger percentage of every racial minority voted for Trump this year than in 2016. Among Blacks and Hispanics, this percentage grew among both men and women, although men were more likely to vote for Trump than women.”
“The Democrat calculation was that in gaining such donor largesse, it could survive losing some working-class support, especially since they would never lose a key piece of the actual working class in poor Black and Latino voters. They assumed that a combination of always-crappy bipartisan-approved economic policies, and the Republicans’ dependably vicious messaging on race and immigration, would guarantee those votes would stay in pocket forever. The calculation held for decades, until now.”
“The multi-racial working-class party is only half-alive: Some commentators saw the increased turnout among Hispanic voters and black men as an opportunity to declare the GOP a “multi-racial working-class party.” This may be premature. Democrats did dramatically increase their vote share with college-educated whites, particularly in the affluent suburbs around major metro areas. We see hints of the shift in the ever-declining Republican vote in wealthy suburban redoubts such as the lily-white Darien, Conn., and increases in the Republican share of depressed, multi-racial Waterbury. But Republicans are still winning among households with six-figure incomes. There are also reasons to suspect that Republican Hispanics have more educational attainment and income than their Democratic counterparts.”
In public policy and as well as politics, issues of age, race, ethnicity, religion, gender, and the frequently overlooked category of class play important roles. People following public discourse and action naturally want to better understand the implications.
But from a point of math, experience, and reality, I’ll offer a different modified quote: Fool me once, shame on you. Fool me twice, shame on me. Fool me thrice, what the hell is going on?
Plumbing the depths of exit polls on surface is questionable when, as American University professor of communications studies W. Joseph Campbell put it, we just experienced an “embarrassing failure for election pollsters.”
There was no predicted blue wave, an observation that POLITICO Playbook was easily able to make. Nor a red one, for the contrarily minded. Democrats haven’t yet won a majority in the Senate and very well may not. (A twisted view on that in the next issue of this newsletter.) Their lead was cut down in the House. Biden seems to have squeaked out a win, as states must still certify their results.
As Playbook said: “The polling industry is a wreck and should be blown up.”
A Harsh Call
There are reasons public opinion polling is inherently difficult. The practice depends on the following:
Statistical accuracy requires accurate random sampling. Pollsters can’t force people to answer and potential participants are increasingly wary of responding at all, let alone accurately. The more reluctant the responders, the increasing chance that you work with a self-selected group that is no longer representative of everyone.
Many people have dropped their landlines and moved exclusively to mobile phones, largely, but not completely, related to demographics. Then, as Pew Research explains, two factors come into play. One is that telephone poll response rates have plummeted, meaning that, again, you’ve got self-selecting groups. Another is that cellular carriers often show polling calls as spam. (This may be the one time virtually all of us would agree with them.)
Polls are expensive to administer. The more information you want to see by group, the more people you must talk to. In its political polling, Gallup talks to hundreds or thousands of people a day all year long.
Polls are an attempt to measure and quantify psychological choices. People can change their minds on a whim and do, when the breeze of current events shifts one way and then the other. It’s like sailing a small boat on the Charles River in Boston. If you’ve ever done it, you know what I mean. If you haven’t, let’s just say that it’s more fickle than your most infamous crazy ex.
People might just lie. They either don’t want to admit supporting a given person or choice or they are angry at the entire media complex and want to throw a monkey wrench into the works.
Good polling needs both financial resources and expertise to pull off. Then come additional problems faced in political polling. News and specialty outlets frequently depend on the polling of others, particularly when looking at state numbers. But the organizations doing state polling, because of fiscal realities, frequently can’t afford the effort needed to get more accurate results.
And then, many of the people reporting on polls don’t understand how they work. There are multiple ways surveys can go wrong: bad sampling of the overall group, bad questions, bad arrangement of questions, poor presentation of the questions to the subjects, mistakes in the processing of the data, and misunderstanding on the part of those who write about them.
All of these can come into play and do. That’s why I had been telling people not to hang on poll results. They ultimately mean nothing, given the limitations, the subject of discussion, and human nature.
The Exit Poll Crash
Exit polls are among the least reliable in politics, as Nate Silver at FiveThirtyEight—hardly holding a sterling prediction history himself—said in 2008:
1. They overstate Democratic voting and gave Barack Obama a nonexistent additional 7 percentage point advantage in 2008.
2. They don’t evenly sample across all precincts, which means the results aren’t random and are biased.
3. Democrats are more likely to participate.
This year held an additional pandemic-provided turn. A massive number of Democrats voted early or by mail. That created an inherent bias toward Republicans at precincts. The only way to counter that would be reaching the early voters by phone, but that’s become an unreliable way method as we’ve already seen.
The chances are very high that normally undependable exit polls are even worse than usual. And yet, immediately after a disastrous election polling season, we’ve got members of the press—and, presumably, political experts and the like—back on a polls-as-truth kick.
For what it’s worth, in my estimate, no one knows how people and groups voted. And even if they had some clue—an indication about how far off the polls might be from consistency, let alone reality—there’s a mentality that pushes the conclusion that all people-in-whatever-category vote this way or that.
The idea that assuming all people who are Black or Hispanic or Jewish or Asian or female or college-educated or belonging to any other grouping will act in some united fashion is wrong. They don’t. If politicians and people in policy want to better address people—or even sell them on candidates—they should focus on individuals and neighborhoods and regions to understand what various parts of the public demand.
And then they have to learn how to actually address the underlying concerns by doing something rather than thinking a song and dance should last until the next election.
A Coffee Cup and the Bermuda Triangle
You don’t need to indulge in conspiracy theories to know that the essential concept of the Bermuda Triangle as a place where things disappear is true. Every home intersects with a corner of the Triangle. You know it because you’ve had things go missing for days or weeks with not a clue as to where they went, or how they returned (if they ever did).
I once ended up with a Manchester United T-shirt that I have to this day. I don’t follow sports, especially English football/soccer leagues, and never would have bought it. And yet, my wardrobe has long had an addition while someone somewhere still wonders where their beloved team-emblazoned clothing has gone.
It’s right here, whoever you are. Thanks.
But mostly more existing items disappear for varying periods of time than new ones arrive. The latest is a coffee tumbler that I had reviewed for my old Inc column (sample sent over gratis) that was delightful. I had forgotten it existed for some time (as I migrate circularly from one coffee system to another), uncovered it, and began to put it to use, only to mistakenly lay it down in our own Bermuda Triangle coordinates.
I’ve since purchased a new one—a Bobble Presse (hate the name, but what can I do?). They’re about $22 to $25, depending on where you find them. It’s essentially a stainless-steel French press built into a highly insulated tumbler. The outer shell has an airspace, then the press plunger creates a third one, meaning the surface remains cool to the touch. The contents stay hot and a silicone cap keeps your lips from burning on metal edges.
Take coffee (ground for drip or a even a little coarser), put it into the tumbler, and water (I use it at 185° F), place the press insert with cap on just into the tumbler, and wait five minutes. Press down slowly and you’re good to go. For a twist, add a green cardamom pod and some cinnamon to the grounds.
Now I’m waiting for the original unit to reappear and mock me. This will happen, but only when I stop expecting it.