Read Time:7 Minute, 10 Second
G. Elliot Morris would not want me to walk away with the notion that polls are fsking useless after reading Strength in Numbers: How Polls Work and Why We Need Them. Quite the opposite, actually. Morris has written a passionate defense of the polling industry and why we need polls. Unfortunately, it flounders on the fact that, by his own admission, most polls are garbage. It is a shame he doesn’t prove his point, because the book itself is largely a fascinating, well written exploration of the history of polls, their potential place in democratic government and why they, well, fsking suck.
The first part of the book is a review of the history of the debate about the value of public opinion, stretching from Plato to the enlightenment. After walking through a very high-level discussion of whether or not rulers should care about what the unwashed masses such as you and I think (and usurpingly coming down on the side of the unwashed), then whether or not our representatives should do as we say or substitute their expertise for our judgment (a more complex conversation that is one of the highlights of the book), Morris discusses the history of polling.
This section is fascinating. The focus is entirely American but given the audience that is understandable if a bit limiting. The evolution of polls from simple and unreliable straw polls to more and more rigorous and scientific and slightly less unreliable polls is a compelling read. And yes, Dewey Defeats Truman makes an appearance, but the true fallout of that headline is much more fascinating than you may have previously understood, on many levels.
The history, well compelling in its own right, does highlight one of the reasons that Morris considers polls useful: their ability to show politicians what the public is concerned about. Morris demonstrates that starting in the back half of the twentieth century, pioneered particularly by Nixon, American politicians came to reply upon polls and polling to replace their guts, random encounters, and newspapers/television reporting to determine how the public felt about them and the issues of the day.
To Morris, this is a good thing. Scientific, accurate polling acts as a potential check on out-of-control politicians between elections. It gives the public a means of making their feelings known and potentially redirecting a politician’s attention to what it considers import matters. Or putting breaks on politicians before they go too far down a path that the public does not with them to proceed along. Polling can act as a kind of non-binding referendum, a sanity check on the body politic. A representative can still choose to ignore such opinion, but then it is a choice. And the public can know that they made a deliberate choice and reward or punish them appropriately. In addition, Morris points out that policy polling can help inform representatives about what their constituents truly feel. He cites one study that shows representatives think their voters are almost 20 percentage points more conservative than they actually are. Letting representatives in on that secret would be an undoubted good.
Morris doesn’t really defend pre-election polls very strongly or clearly. I suspect that is because he does not have a lot of faith in them. He is probably right to. As he himself points out, many of the most common types of polls have failed miserably over the least two or three election cycles and may be permanently broken due to societal and technical factors. He himself admits that the only kinds of polls that might do better tend to be more expensive and harder to run. And even here, he mentions to organizations (YouGov and Civiqs) that do not rate as the amongst highest accurate pollsters (a B+ and a B- respectively) on 538’s accuracy ratings. Now, 538 is not the be all end all, but I do find it odd that Morris puts so mush stalk in polls that are only slightly above average compared to their more traditional peers.
Morris, to his credit, does not shy away from any of the problems with polls. The many, many, many ways modern polling fails are covered in delightful detail. The story of how one African American voter sent to 2016 averages on a roller coaster ride is almost worth the price of the book alone. But that coverage does highlight two major issues with Morris’s approach to election polls. First, he never considers voter suppression as a cause for poll failings. At a time when one party is trying to prevent voters from exercising the franchise in swing states, not exploring that possibility is confusing, to say the least. And it points out the other major issue with Morris’s approach: he doesn’t seem to understand or value one major use of pre-election polls — detecting fraud.
One of the reasons to pay attention to pre-election polls is to help determine if an election has been conducted fairly. If the election deviates significantly from the polls, then, possibly, it was not. Perhaps voter suppression was at play, or outright fraud, or some other factor. But you have polls to help you ensure that you know when to look deeper. Morris never brings that possibility up and I have to wonder if it is because he knows, at some level, that pre-election polling may be irretrievably broken.
Morris suggests, polls have significantly wider margins of error than are traditionally understood, the biases in polls (due to unresponsiveness and other social and technical factors) skew in the same direction, making aggregation less effective, and the more effective modeling and polling techniques are unlikely to be adopted due to cost issues (though this is more implied than stated). If Morris is correct, then the pre-election polls have precious little ability to predict anything. And if they cannot be understood to be a reasonable collective marker for the performance of elections, then they really do not have any value. They provide no meaningful information — you cannot look at something so broken and suggest that it indicates voter suppression or other forms of active fraud. What good are they, then?
Issue polling, while currently more accurate than election polling, has the same potential problems. Now that polling has become a culture war issue, how long until the right wing convinces people to stop answering issue polls, so that they can “prove” the country isn’t really pro-choice or pro-LGBQT? It is only a matter of time, even without a concentrated push, before the specific unresponsiveness to election polls driven by right-wing culture war mania drifts into a general unresponsiveness to all polling and issue polls suffer the same problems as elections polls.
And Morris is not wrong that we will lose something important when that happens. As he points out, a demagogue like Trump is bad enough when his claims to represent the will of the paper can be shown to be nonsense. But while Morris does have some technical fixes in his closing arguments, one of his suggestions is simply living with the notion that polls are just not very accurate. Well, Morris seems to do that in his book and as a result, the notion of polls as a check on election fraud is not even contemplated, because it is unviable given that advice. Pretty soon, we may lose the ability to discuss how issue polls can be used to check out of control politicians. Morris doesn’t seem to want to fully fac the implications of what he is describing — that perhaps polls cannot be saved, perhaps we need other methods of achieving what polls have previously given us. Maybe that is too much to ask in a book dedicated to arguing for the importance of polls, but it certainly feels as it would have made for a more honest look at the situation, we find ourselves in.
Despite the fact that I believe Morris failed in his objective, I do recommend this book. As I mentioned earlier, it really is well-written. The technical sections are explained well, the history and philosophy overviews are interesting, and Morris does cover important needs in a democracy that polling can and has helped fill. He is just doesn’t seem to be able to admit that his profession may no longer be in a position to practically fulfill them any longer. Maybe that’s another book, by another author, but I would have liked to see someone as well-versed in the need and the profession grapple with that question more openly.
Happy
0
0 %
Sad
0
0 %
Excited
0
0 %
Sleepy
0
0 %
Angry
0
0 %
Surprise
0
0 %
0
0
