Is Kamala Harris really in the lead?

Kamala Harris has got the vibes...

The change of Democratic presidential candidate from Joe Biden to Kamala Harris has shaken up the election campaign. In the national polls, it is no longer Donald Trump who is ahead, but the current vice president (title chart). She has certainly benefited from the fact that she has received special attention as a new candidate and that her nomination at the party convention has generated positive headlines in recent days. But how meaningful are these figures? We show what matters.

... but is this enough?

In fact, Harris' small poll lead of usually one to two percentage points says little. Many national polls give a margin of error of plus or minus three percentage points. This means that if Harris' share of the electorate measured in a poll is 52%, the “true” share is between 49% and 55% with a probability of 95%. This error results from the fact that not all potential voters were surveyed in the poll, but often only 1,000 people. In a poll in which participants can only choose between Harris and Trump, even a lead of four percentage points (52% minus 48% for Trump) would not be significant.

This means that Harris' national lead is currently within the margin of uncertainty, even if – as is sensible – we look at the average of various institutes rather than individual polls.

In addition, the true error of the surveys is likely to be greater than the three percentage points stated. This is because there are other sources of error, for example if the selection of participants is biased (e.g. only participants with a landline in a telephone survey), non-voters are asked about their preferences, certain groups of voters systematically do not state their opinion or voters – as studies have shown – prefer to name the candidate in the lead regardless of their true attitude. The institutes do try to correct their results for such effects. However, this may introduce new distortions. In any case, researchers in the US showed in 2018 that even polls conducted shortly before an election typically deviate from the actual result by seven percentage points.

And there are still more than two months to go until the election, during which a lot can change. At the same time four years ago, Biden was leading in the polls by 7.4 percentage points. The polls conducted immediately before the election then suggested a Biden lead over Trump of 7.2 percentage points. In fact, the actual spread was only 4.5 percentage points. Trump also did better in 2016 than the polls had indicated.

For full text see attached PDF-Version.