‘The whole year there were three possible election outcomes’

Polling companies were heavily criticised regarding the US presidential election. How do you view this criticism?

– There are a number of misconceptions about polls. Poll findings are not prognoses. They provide a snapshot during the collection period and nothing else. That is why it is totally wrong when comparisons are made between election results and, in some instances, week-old polls, or when the polling average, which includes old polls, is used as an object of comparison.

– Moreover, poll findings must always be carefully interpreted, and especially with regard to margins of error. A poll of 500 people in a US state has a margin of error of almost 4.5 percentage points; in other words, if the poll says that ‘Biden is leading by 8 percentage points, with 54 per cent of the vote to Trump’s 46 per cent’, the actual estimate is that Biden is within the interval of 49.5 per cent to 59.5 per cent and Trump 41.5 per cent to 50.5 per cent. A Trump lead is therefore not out of the question (for example, 50.5 per cent to 49.5 per cent). This means Biden may have had a huge lead but nevertheless not a totally secure one. Few commentators take this into consideration. There are more misconceptions, but these are the reasons for many erroneous descriptions.

How do polling companies manage to defend their integrity vis-à-vis the media, both in Sweden and the USA?

– In Sweden, it is rarely a problem nowadays, especially not when our capable political reporters are involved in dialogue with us, the polling companies. They know the subject and understand the challenges. Perhaps they need to fight a bit against the media logic of finding conflicts and looking for ‘winners and losers’, but, in general, it is really good. However, when the Swedish media reports on the US election, other journalists and editors are involved, and on social media, many self-appointed experts also join the discussion. For them, the polling challenges are perhaps not as present. Then it can go wrong. For quite a long time, Biden was explicitly and implicitly described as the clear winner in this election. For those who read the polls correctly, there were, as has been the case the whole year, actually three election outcomes that were wholly possible: a landslide Biden victory, a narrow Biden victory or a narrow Trump victory. There was really no other way to call it. We ended up with the middle alternative.

What lessons can we in Sweden learn from the discussion about the importance of the polling companies in the US election?

– As we see it, the analysis companies in America, on the whole, did a fantastically good job in the election. For a long time, they described the possible outcomes and also Trump closing the gap at the end. This is despite the fact that several factors make it harder to conduct polls in the USA than in Sweden. The voter turnout is lower; it is harder to weight the polls correctly because they are unable to really know which voter groups will vote in the end, and the response rates are lower than in Sweden. From that perspective, it is not possible to compare Sweden and the USA. However, one important lesson is, as I have already mentioned, dealing with the prevailing misconceptions about poll methodology. We as polling companies are more than happy to help out and provide more and better explanations. One lesson might be to listen, first and foremost, to the pollsters, and not to experts forced to give ‘definitive’ answers in a 15-second segment on CNN.

Jonathan Wennö was interviewed by Daniel Nordlund, Senior Consultant, and Johannes Hylander, Senior Consultant and Partner, at New Republic. ‘Three Quick Questions…’ is a series of interviews conducted by New Republic.