Soundcast - Podcast Responsive Theme

2d06 – Let’s ban election polls!

0:14:40 Download (MP3) Debate with us!

Click here to debate with us

BREXIT and the US election are only two examples of many where election polls published earlier seem to have gotten it wrong. Some even argue that election polls may even distort elections at large and deflect from real issues. Today Sebastian and Dirk debate whether we should therefore go ahead and ban election polls all together.

Image Source: CC0, Pixabay,
Audio jingles voice over by @berlinabby


Episode Transcript

Sebastian – Hello everyone! Welcome to the latest edition of, our podcast of debates. I’m Sebastian and my co-host is Dirk. How are you today, Dirk?

Dirk – As usual, I’m ready, I’m prepared, I’m out of words, I’m in a good mood and I can’t wait. Are you more prepared than usual? So I hope you learned your numbers.

Sebastian – Well, we’ll see, we’ll see! We’ll debate today on the following question, on the motion which is: “elections… election polls should be banned”. Maybe not elections themselves! Although that’s another topic, considering the results of some elections but ok, Dirk, you have the pleasure of starting this debate today as we have flipped the coin towards that, defending the case against the motion, which is again election polls should be banned.

Voice – OK let’s do this. Dirk goes first and argues against the motion.

Dirk – And there we are. Another election, weeks and weeks of reporting on numbers, and we all love that, right? And then bam! A result that we all believe has been predicted wrongly and this wasn’t the first time.  Actually there were plenty. It looks like if you research it, it’s as if almost every single election is a surprise to the pollsters. So let’s ban polls, right? They are altogether totally useless and, well, they screw the process and they lead away from actual topics. Wrong! I would start start by killing a myth. Actually, the polls haven’t been wrong, not even in the US election. Actually, the polls predicted a very short race. And polls are statistics, so there’s a margin of error. The margin of error can go either side and if you have multi millions of people, yes then the potential margin of error are a couple of million votes in the end. So that’s not a surprise. On the other hand, having proper data is important for the electorate, so everyone should have every data point possible to know what is on the agenda, what is important right now, what are the issues being debated, which candidate is up front, so it allows you to actually make an informed and data-driven decision. Next thing, if there is data like polls, it’s the right of the voters to learn that. It’s a freedom of expression, freedom of speech, it’s close to actually closing down an open communication if you just ban polls and data around elections altogether, just because you don’t like the result of it.

Voice – And now on to Sebastian. let’s hear his argument for the motion.

Sebastian – The election polls, it turns out, they’re perceived by people and by the press as being wrong over the past year. There was the elections in the UK earlier in the year in 2016, there was the election of Trump, there was Brexit. In every single case, they were perceived as wrong, irrespective of margins of errors by the way. Additionally, there is bias. It is not only about being wrong or right, it actually introduces bias and if you read a book like “Thinking Fast and Slow”, you will see also a number of biases and among them, there is this mechanism of “anchoring”: so if you know that a candidate is at a specific level, you may actually think well, you might as well vote for him or not, or you might not even show up because anyway there will be people going out there to vote, and it will seem that it’s a foregone conclusion. So there’s a risk of influencing elections in a way that is detrimental to the democracy by just letting people in it not being biased by what may or may not happen. And finally, it’s already the case in many countries that these election polls are being banned, from a few hours to a few days, even a few weeks, before the elections over time. And we’re not talking about some random countries. It’s the case in France: it used to be two weeks before the elections, it is now two days. There was a recent survey and it looked at 78 countries and I believe about 39 of them, half of them, actually had polls being banned right before the elections, precisely to avoid this bias. So my case here is that they’re wrong, irrespective of margin of error, or at least they’re perceived wrong. There’s a strong bias which is detrimental to the elections and people may not actually turn out. They don’t even understand the concept of margins of errors. And finally, it’s already the case, we can actually do this, it is actually implemented in a number of countries.

Voice – Now it’s Dirk’s turn, let’s hear it his rebuttal.

Dirk – Actually, there was just one argument. The one argument being maybe the data points are wrong and because the data points are wrong and the voters are stupid, we have to protect them from their stupidities or let’s move the data out of their sight. And I think, first off, there is nothing wrong with the data itself. There is something wrong with how we report it. And there’s something wrong with who we trust to bring data. So some countries, for instance the UK actually debate right now to introduce a penalty if you have grossly misleading data points. But if you are in the margin of error and you properly stated it, then there’s nothing wrong with actually communicating that. And secondly, we need competent presentation in the media, so media outlets actually should educate the voter on what it is that they are presenting, what the potential outcome may be. Again, in the US elections, actually the polls basically stated it’s a close call, it’s a close race, and then what happened was that there was an interpretation floating around saying that the likelihood of actually winning is this or that, and that was where the problem happened: that is where the misleading interpretation occurs. In general, voting tactical, staying at home, going to the voting based on data is actually also democratic. It’s a decision made by the electorate, so if people are not going to vote and then are burned because they misread the polls, maybe that’s teaching them a lesson, maybe next time they’re going to vote and vote as they really believe, and vote based on issues. If it’s really not the majority of the voters that wanted to have that outcome, then maybe those voters that haven’t been heard properly will next time make sure that they actually show up to the vote instead of believing a poll result that tells them their vote is not necessary anymore. So it could even be better for democracy to have an open debate about surveys and so on. Last, not least: I reiterate that I believe in order to make an informed decision, you need as much data as you can get your hands on. Just by hiding things, you actually don’t remove the data of the equation. So there are plenty of countries where people do bets on moving results and they will base those bets on data points, no matter if you communicate them openly in the media or not, no matter if you ban them or not. It’s just a different form of polling that you introduce and that’s a much more inaccurate polling than you can have from official outlets.

Sebastian – Okay my three minutes!

Voice – Sebastian, let’s hear it!

Sebastian – Nothing is wrong with the data, you say?! And yes I’m going to get passionate now, my turn! Nothing is wrong with the data, really? You do know how polls are being conducted? The sample size of a supposedly representative set of demographics. Of course they’re completely wrong! From the very beginning, there’s bias by the pollsters. That’s the problem. Also not even is it just one problem, but it is a problem. Yes, data is wrong and will always be wrong because you’re using a sample size and demographics are way more complex than we think they are: gender, races, age, whether you’re rural or living in urban areas, of course data is wrong. And the problem is people, I’m sorry to say, but most people do not know how to read statistics. I mean who loves maths, apart from me, maybe you, nobody! Everyone hates maps, let alone statistics and probabilities. Nobody has any understanding of what the margin of error is. Even engineers, I can guarantee you, you can ask around, even today engineers they would be confused as to how we calculate a margin of error, so people do not know how to read the data. Sure, you can feed them with data, it can trick them with it. Now you say you want to educate people and give them more data? Fine, let the media focus on the programs themselves. What does the data bring you to know whether a person may be ahead of the race? Absolutely nothing! And if you want data, here’s the data you can analyze: whether a program a political program is actually financially doable, let’s look into the data. Journalists, most of them, I’m sorry to say, they’re not really good at this. There’s no one’s going to say Trump’s or Clinton’s or Sarkozy’s or Hollande’s program or Merkel’s program is actually viable, is that going to lead to something which is dramatic for the country’s finances. Let’s use that data, that’s way more interesting, more meaningful. Now there’s one more thing: we talk about Trump, we talk about Clinton, but who knows there were actually other candidates at these elections? Same in France, same in Germany. We don’t hear about them! You’re talking about freedom of speech. Seriously, there’s no equality of speech in the first place, let alone talking about freedom. We don’t hear about the small candidates. Why? Because the media only tells you about a big one, the big ones according to what? Opinion polls, which are wrong, which are wrong because the data sample size is wrong, because they just give you a misrepresented picture of what’s happening, because nobody knows the truth. So yes, you want freedom of speech, yes, go for it, let’s talk to everyone, all the candidates with equality of speech, of time in a media, but this is way more interesting. And finally, actually we may agree on this alternative to banning, although I’m in favor of banning it, like two days, maybe two weeks before the election. It’s also to add the fine, absolutely. You are talking about freedom of speech and talking about fines, I think you just lost your cause here. Yes, you know, the wider you are away from a margin of error, let’s have a hefty fine so that the media will think twice before publishing anything, whether it’s about publishing nothing at all or actually publishing something and then risking being bankrupt. But I’m surprised you bring the freedom of speech argument in that case but fine, I’m for banning and adding the fine. Yes, let’s do it!

Voice – Final statements.

Dirk – I love how you really think that you can remove the polling from the equation. You know what’s going to happen? Instead of looking at polling data that includes hundreds, thousands, maybe a couple of hundred who knows, people and yes it’s statistically flawed maybe, instead of that you’re going to poll your friends and your family. You turn around and ask the guy next to you what he thinks and that’s where you base your opinion on, so it’s not changing anything. Do you want to ban that too? Do you make it illegal the day before the election to ask your spouse what she’s going to vote for? I don’t think so. So yes, I would add a fee, I would educate, but I think the voter has a right to know what others, in his country, in his electorate, think of the issues, think of the candidates and think about who they’re going to vote for. The politicians have a right to that too because it helps them focus.

Voice – Sebastian.

Sebastian – Sure you can ask your wife, you can absolutely ask your wife who she’s going to vote for but this is private information. You’re not influencing anyone else beyond your close circle of friends. That’s perfectly fine with me, I’m not telling you to censor what you’re doing. So we’re talking about public influence and this is a problem because the data is wrong, because the sampling size has too much bias, because there’s no absolute truth, this is not a science when you’re actually polling people to ask whether they’re going to vote, whether they’re actually going to turn up. So the problem is there’s no right to know, there’s a right to know something that is wrong: you’re actually misinforming the public with erroneous data, which people can’t even read properly, even if it were correct. You want more education. I say I want to  have more democracy. Let’s hear about all the candidates in a more equal way, instead of influencing things the debate or the elections towards one or two major parties like it is in most of our modern societies. And yes, it’s banned as it is the case today in at least 39 countries, including democracies. Ban election polls prior to the elections at least two days maybe even a week before the elections, and have a fine for a media which misinforms the public.

Dirk – Alright done – and it kills me how you keep saying it’s not science. It is freaking science, it’s statistics, it’s maths.

Sebastian – Of course it’s science! I guess it’s my turn. I get nervous every week with it.

Dirk – Yeah, me too. I was short of breath with you know that.

Sebastian – I can see, you get serious and everything. I see your face and I’m actually reacting but you actually don’t even see me reacting when we talk about things like freedom of speech and I can’t stop smiling or laughing because it was our last debate. So I have to resist answering you and I have to focus on giving my speech.

Dirk – And now I know I went over for like 10-15 seconds just because…

Sebastian – Like always, like always! It’s the bias that you are trying to instill on our listeners, that you actually try to win because you have those extra seconds, I know!

Never miss an episode