The Polls are All Wrong
And why politicians can afford to ignore them
To be fair, this is not just about politicians, it’s all government officials and people with power in general. Polls, quite simply, can be ignored. Which is ironic, considering the amount of time we spend obsessively talking about them. To be clear, I’m not saying politicians unanimously ignore all polls, because they will often cite them whenever it is to their advantage to do so. In fact, they may even take poll results into account when making actual policy decisions. What I AM saying is that any politician, at any time, can easily rationalize why any particular poll can be disregarded, and instead do whatever they damn well please.
For a representative democracy, this is a problem.
It’s a problem that should not exist.
I’ll get to why it is easy for our representatives to ignore poll results in a moment, but first let’s establish why this is a monumental problem for democracy. Then we will dive into why this problem need not exist in the modern era.
Is this a problem? And Why?
For supposedly being a world class beacon for democracy, we really only have one direct mechanism for citizen representation: voting. Outside of this there are several indirect mechanisms- protest, polls, etc.- but the net impact of these is debatable. In fact, I plan to do a similar breakdown/critique of voting, protests, and others in the future, but for now let’s focus on the current state of polling.
Political scientist Sidney Verba describes polls in this way: “Surveys [polls] produce just what democracy is supposed to produce – equal representation of all citizens.” First, I think the “equal representation” part is nonsense, and I’ll get to why shortly. The real question is, even if you except this as an accurate statement, does it matter?
The answer is no.
Consider the following areas in which, according to polls, have had a rough consensus amongst the US population across party lines for a considerable amount of time.
Biden & Trump rematch in 2024: nobody wants these guys, you know this, I know this, and the polls show it.
Abortion: the overwhelming majority of nuanced polls show broad support for abortion within the first trimester.
Getting money out of Politics (elections, congress stock trading, the Supreme Court, etc.)
Numerous immigration policies (yet solutions are constantly sabotaged by one party or the other).
Maintaining/Strengthening Social Security (and other “safety net programs” like Medicare, Medicaid, Unemployment Insurance, SNAP, etc.)
Increasing taxes on corporations and the very wealthy
Police Reform (specifically on improved training and accountability measures)
And many more. This report from the University of Maryland does a good job hitting some of the specific policies where consensus is found.
Notice anything about all the things I’ve just listed? We have made very little national progress on all of them. In some cases, we are going backwards. If equal representation and a responsive government is the goal of democracy, and polls are supposedly an “equal representation”, then where is the disconnect?
That’s not what OUR polls show…
Before we dig in, let me emphasize again that this is not about whether polls should be ignored. While they are certainly flawed on a fundamental level, they are also the best insight into public sentiment that we have. The question is: why are those in power perfectly comfortable cherry picking which polls they pay attention to or disregard altogether?
As most are aware, a politician’s top concern is getting elected/re-elected. Thus, at a very basic level they are constantly making a risk assessment as to what will cause them to gain or lose votes. One would think a system that brings them constant updates on public sentiment would be highly valued. And it is. Everyone – literally everyone- is constantly citing statistics from polls, yet we live in a paradoxical world where information is highly valued and completely ignored at the same time. Biden and Trump are still our choices, nobody is getting money out of politics or even seriously talking about, our tax structure is still ridiculous, and the list goes on.
The problem here is perception. Not just the perception of politicians, but OUR perception as well. And, unfortunately for polls, sometimes the gap between perception and reality is not that wide.
Let me explain:
Perception of Inaccuracy: There is a general perception amongst the population that polls are of questionable accuracy (again, paradoxically we still rely on them constantly). This perception especially increased after the dismal predictive capabilities of the polls in the recent 2016 Presidential race. Pew Research Center partly blames this on technology: “Technology has disrupted polling in ways similar to its impact on journalism: by making it possible for anyone with a few thousand dollars to enter the field and conduct a national poll…Some newcomer polls might provide good data, but poll watchers should not take that on faith” (Pew Research Center). They’re probably not wrong, at least in part, as they will rightfully point to rookie mistakes such as the overrepresentation of college voters in the Clinton/Trump race polling. However, I would argue just like journalism, there is a net benefit to having more voices in the mix. We perhaps just need better BS meters.
Perception of Bias: Our perception and reality are not so different here. From the agendas of certain polling organizations, to the participant selection, to the selection of questions, to the specific wording of the questions, it is hard to deny that bias is baked in to most polls. Partially this is a structural issue in the sense that all of the aspects just mentioned are entirely controlled by the polling organization. However, this is also a transparency problem. Polls are often cited without giving background to where they came from, and even when you have this information, it is often difficult to discern exactly how the results were calculated.
Perception of Representation: Answer this: have you participated in a poll? Do you know anyone that has? How many, and how often? The reality here is that an astonishingly small number of people participate in polls. For instance, Pew Research Center has an American Trends Panel of over 12,000 adults that serve as the center’s “primary source of survey data for U.S. public opinion research.” Yes, 12k to represent 260+ million adults…plus, you know, a few minor correction factors…Look I’m not trying to cast aspersions on the science of survey statistics, but it does not take a stretch of the imagination to believe we can do better than this.
Perception of Power: Or to be more accurate, lack thereof. To put it simply, polls have no teeth. You know it, I know it, politicians know it. Yes, sometimes the polls are correct, and a politician sometimes pays the political price for ignoring them when their name is on the ballot. But voting is very infrequent and a very poor barometer for real-time sentiment of the citizenry. Even when a poll proves to be accurate over time, it is still relatively meaningless if our politicians are under no obligations to heed it in the moment. Put simply, polls are a safety blanket. They make us feel better about the fact that our views are being recorded, all while they increasingly provide very little tangible benefit.
To summarize all of this, it boils down to trust. In their current state, no one trusts polls to universally and consistently reflect our values and needs. Thus, casting doubts on their results is easy. They are a tool used to confirm our existing bias when needed, and very little else. It’s what allows our government to consistently ignore areas of consensus in our supposedly democratic society. Its’ what allows Biden to respond "Well, I don't believe that's my approval rating. That's not what our polls show." in a recent ABC interview. Doubt rules supreme, thus our primary tool for systematically providing “equal representation” commands no action.
Why this problem shouldn’t exist.
Functional representation of a democratic society is a foundational problem that must be solved. More importantly, in the modern era, this seems like one of the more solvable issues on our plate. I put my thoughts on fixes into two buckets: Technology and Policy.
Technology:
Actual Representation: In a world where 85-97% of the US population have smart phones, it is somewhat ridiculous to think that we cannot conceive of a way to solve the “actual representation” problem. While I have no doubt that certain correction factors may still need to be employed, accuracy of results can be greatly improved simply by massively increasing the number of participants (i.e. into the several millions rather than thousands). This can be facilitated in several ways:
Creating a central platform that everyone has access to (perhaps connected to voter registration?)
Streamlined/easy user interface that anyone can use.
Polling window open for long periods of time to allow citizens to respond on their schedule (and from anywhere).
Bias: I agree that technology is a disruption to the field of polling. I disagree that this disruption needs to be a hindrance, rather than an evolutionary asset. Our entire polling strategy can, and possibly should, change. With the advent of technology that can aggregate and analyze large amounts of text/data, we should be rethinking the simple Question/Answer format of polling. Regardless of whether a polling company is truly doing its best to remove bias from their questions, the perception of bias is still there. (Honestly, I question whether removing bias is even possible for humans). Let’s look at the Pol.is software that I’ve mentioned in a previous post as a possible alternative strategy:
Rather than an organization crafting questions, the citizens craft statements on a given topic.
Citizens can then like/dislike statements from other citizens (picture doing this in a swipe left/swipe right fashion for a dozen or so statements)
The software maps responses in real-time to identify emergent consensus areas.
This process cycles, and more nuanced statements emerge that garner larger and larger consensus behind them. (i.e. the process of creating agreement is “gamified.”)
In other words, this process reveals the “bias” of the people, rather than the bias of the corporation/organization.
Policy:
Transparency: If we are truly trying to fix the trust problem, then we need a means of enforcing transparency around polls. This can be accomplished through official regulations around public/national polls, or perhaps simply designing trusted third-party certifications for pollsters. However, if we want mass participation and trust in the data, then I feel we need to look at solutions that involve a more centralized approach rather than the hodge-podge of disparate organizations we have now. It’s not that these avenues couldn’t exist, they just wouldn’t be considered the “gold standard.” Ideally this system would have de-centralized control (i.e. not directly/entirely controlled by government or any specific corporation).
Power: Representing millions of voices rather than a few thousand will inherently improve the trust in any poll’s results. This in turn increases its power (i.e. it is much harder for a politician to rationalize dismissing the sentiments of actual millions, especially if these millions are part of their specific constituency). However, I’m not convinced this will be enough. If we can design a system that is trusted and robust enough (possible, but still a big if), then I think we should consider actual laws that require the government’s response to any public sentiment that is not being adequately addressed. The exact structure of that law is an interesting question.
Creating a Perception of Trust
For a democracy to truly function, communication and responsiveness between the government and the governed is essential. Thus, it is imperative that both the government and the population trust any means that facilitate that communication. This is about instilling the perception of trust in one of the few mechanisms we have for actual representation in our country. Whether justifiable or not, our inconsistent trust in polling information makes this form of communication almost entirely ignorable when it benefits our bias to do so.
Finally, consider that this is not just about those with power listening to the larger population. This is about our ability to listen to each other on a large scale as well. How many times have you seen a poll referenced by the media, politician, or some political commentator and questioned its accuracy? Now consider how many times you did NOT question a poll’s accuracy because it served your own bias. Or maybe you doubt we have consensus on some of the very issues I listed at the beginning of this post. Would you doubt it less if you knew it was based on the responses of millions of Americans from every region of the country, not just a few hundred or thousand? Do you think you would be more willing to listen to opposing views if you trusted they weren’t just the skewed, terminally online rantings of a fringe few, but rather the genuine views of a large majority?
Again, I find myself going back to the quote from political scientist Jane Mansbridge, “Our structures of democracy, which basically evolved in the 18th century, are not sufficient to carry the load of the government coercion that we now need.”
I think it’s time for an upgrade folks. What are your thoughts?