Biden stands for, but how reliable are American opinion polls?

Put on the American news stations these days and youll be buried in a tsunami of numbers. Whos winning? What do the latest polls say? Election experts conjure up one state after another on magic walls with corresponding gauge dates. Panels of commentators analyze the daily flood of polls to grieve.

It is teeming with research agencies questioning voters about their political preferences, but there are huge differences in the quality of these polls. How can voters get their way into this endless figure? And to what extent are opinion polls reliable? The prevailing view is that the polls in 2016 were quite wrong. Most of them pointed to an election victory for Hillary Clinton, but to everyones surprise Donald Trump won.

In reality, the 2016 image is a lot more nuanced. First of all, lets make a difference between two types of polls: you have national polls and polls in the individual states. National polls only give an indication of the so-called popular vote: who gets the most votes across the country? Because there are large differences between national polls, we at DeccEit always use the average of those polls. The two main sources for this are Real Clear Politics and the FiveThirtyEight.

Biden stands for

Both use different models for their calculations, but their trends almost always show the same direction. The national polls have proved miraculously stable this year. Despite a pandemic, a serious economic crisis and major social unrest, there have been hardly any major shifts this year. Joe Biden has been ahead for quite some time and his lead has been fluctuating between 6 and 10 percentage points for months.

The stable picture in 2020 is one of the biggest differences from 2016. Although Hillary Clinton was often ahead of the national polls, the differences between her and Trump were smaller and there were times when both candidates were neck-to-neck. The national polls were a lot more capricious in 2016.

The last snapshot was not far off in 2016. Clinton was 3.2 percent ahead of the average polls on the day before the elections. She would eventually win the popular vote with 2.1 percentage points difference. In polling terms, that was quite accurate.

However, elections in America do not win by getting the most votes. Clinton had almost three million more votes than Trump, but she still lost. Thats because of the American electoral system.

The results in the individual States are therefore decisive. The battle is decided in the swing states, states that are not clearly Republican or Democratic. In most swing states, Biden is ahead, although the differences here are smaller compared to his lead in the national polls. In 2016, the polls in a few key states went to the fog. Particularly in the so-called blue wall states of Pennsylvania, Michigan and Wisconsin there were significant differences between polls and final results. It was those three states that would give Trump, close, victory.

You can see how the American dialing system works in this video:

Then how can you best deal with the polls? First of all, its not predictions, but snapshots. Often polls are influenced by the news of the day, such as disclosure about the pussy tape in 2016, after which Trump plunged into the polls. Yet that said little about the election day itself.

FiveThirtyEight prefers to talk about probability calculations. The datasite uses a model that converts all the polling data into a percentage chance that a candidate has to win the presidency. One day before the elections in 2016, the datasite Clinton gave a 71 percent chance. But that didnt mean that Trump, with his 29 percent, had already been eliminated in advance. Compare it to the state lottery. How many tickets would you buy if you knew you had a 29 percent chance of winning the jackpot?

Pandemic elections

Trafalgar Group, one of the few opinion polls to correctly predict Trumps profits in Michigan and Pennsylvania in 2016, goes one step further. It states that most voter investigations repeat the mistake of 2016 and systematically underestimate the support for the President. According to Trafalgar Group, Trump voters, much more often than Democratic voters, do not give an honest answer to polling questions. If thats true, the picture would look very different. In the major swing states, Trump would then be ahead of Biden.

It is therefore wise to make a big blow to the arm. Pollers also base their methods and models on election results from the past, but thepandemic has created a unique situation. Voters on both sides of the political spectrum are highly motivated to vote, which means that turnout is likely to be much higher than usual. That alone makes these elections a lot more unpredictable.

Americans vote massively early

In addition, many states have adapted their electoral laws so that there are more opportunities for voting. Tens of millions of voters have now cast their votes early in numbers that have not been seen before. More than ever, Americans cast their votes by mail. That fact, too, makes the election results unpredictable. There are doubts as to whether all these postal votes arrive on time, because the national postal service is failing after a series of cuts.

All in all, this makes an exceptional election year with many variables. This is a difficult task for opinion pollers.