Vice President of the United States Terrible Harris was the ex-president in the support poll published by the news agency Reuters on Tuesday Donald Trump’s ahead by a couple of percentage points.
However, this still doesn’t say much about how this duel would turn out, says the research doctor Oscar Winberg from the University of Turku.
According to a poll conducted on Monday and Tuesday, 44 percent of Americans would vote for Harris as president, and Trump would get 42 percent of the vote. However, the poll’s margin of error is three percentage points, which is more than the difference between the candidates.
There is no reason to draw too far-fetched conclusions based on one support survey, says Winberg.
– This kind of national opinion poll does tell about the feelings and political debate of the American people, but the election result is determined by the results of the individual states, Winberg reminds.
The election is still several months away, and according to Winberg, opinion polls are usually more accurate closer to the election. There are still many voters who are undecided. Harris is also not officially the Democratic presidential candidate.
At least these things are good to keep in mind when starting to interpret opinion polls conducted in the United States.
1. National and state level measurements
In the US presidential election, it is important to distinguish between the meanings of national and state-level measurements. The separation is related to the US electoral system.
National measurements reflect the candidate’s overall support, which does not necessarily predict the winner due to the electoral system in use in the United States.
The presidential election is won by the candidate who wins enough states to receive the support of more than 270 electoral votes per state. For example, in 2016, the Democratic candidate Hillary Clinton led the national polls and got the most votes, but Trump won the necessary states, i.e. got more electors and was thus elected president.
Monitoring state-level measurements is hindered by the fact that they are done less often than at the national level. It’s cheaper and easier to do a nationwide survey than more than 50 separate surveys, says Winberg.
2. The differences between the equatorial states are not necessarily visible
National measurements often hold their itch, but you can’t always interpret the winner of the election from them. If a single, smaller state swings for another candidate, it will not affect the overall support significantly, but it can change the result.
In practice, the result of the elections is decided in a few states, where the differences between the candidates can be very small. The differences can be several thousand votes. This means that it is practically impossible to accurately predict them in surveys.
Six states are particularly important in this election: Arizona, Georgia, Michigan, Nevada, Pennsylvania and Wisconsin. Many larger media houses also conduct support surveys especially in these states.
3. Many actors – many methods
There are hundreds of operators in the United States that conduct support measurements.
Actors also use significantly different ways of doing research, which affects how well the research done by different actors can be compared to each other.
– At the local or state level, measurements made by local operators can be more accurate than national measurements, says Winberg.
In the media, individual measurements sometimes get an unnecessarily large amount of attention, even though it is more important to look at the average of measurements and support trends, says Winberg.
It’s fun to measure different measurements and follow long-term trends FiveThirtyEight website. It also evaluates the reliability of different operators based on, for example, how the operators’ forecasts have been correct in the past.
4. Research does not necessarily reach everyone
Opinion polls conducted in the United States suffer from the same problem as, for example, surveys conducted in Finland. Surveys are no longer answered as quickly as before.
Reaching the younger population, especially by phone, is particularly difficult, as young people do not want to answer the phone or talk on the phone. Of course, research is also done in other ways.
In the United States, according to Winberg, it has also been observed that the highly educated respond to the non-educated more often in the measurements.
– This was not necessarily such a big problem twenty years ago, but after the rise of Trump, it has been noticed that the highly educated vote for Democrats more and more often, which means that an educational divide has formed between the parties, Winberg says.
It is therefore even more difficult to form a comprehensive picture of the feelings of different population groups, even though it is statistically possible to correct various biases.