Dear Reader,
Joe Biden is leading the polls as the next US president. But are the polls really the best indication of who will actually win?
According to Jim Rickards, the answer is no. In fact, science shows that Trump is the favourite. Read on to find out more…
What the Polls Really Say
Biden’s huge lead in the polls and his lead in the betting odds are impossible to ignore. Even allowing for the margin of error (about 3.5% to 4.0%, depending on the poll), Biden will win if these polls are an accurate guide.
But they’re not.
The polls are riddled with errors, including oversampling Democrats, polling ‘all voters’ instead of ‘likely voters’, and organising and framing questions in such a way to lead the subject to a pre-determined conclusion about Biden as their choice for president — by starting with questions that cast Trump in a bad light.
The polls are as biased as ever (maybe more so) and the betting markets suffer from similar deficiencies, mainly relying on the weight of money to set odds instead of the number of bettors regardless of money. Big money bets for Biden create short odds from bookies, but in the voting booth money doesn’t matter; it’s one vote per person.
Also, don’t underestimate the extent to which political operatives throw money at these betting websites just to manipulate the odds and affect perceptions. Still, a 24.1-point lead is hard to discount entirely.
REVEALED: What’s Next for Aussie Gold Stock Prices? Learn more.
Skewed or biased polls?
On the other hand, Trump is not nearly as far behind as the polls indicate.
People don’t always tell pollsters their real views or won’t participate in polls at all. In this case, voters know ‘Trump’ is an unpopular answer for many, so they will say ‘Biden’ in reply to a pollster question — even though they intend to vote for Trump.
When the polls are adjusted for factors such as skew in selecting polling subjects, self-selection by those who do not participate in polls, and the social pressure to deny you are voting for Trump even when he’s your first choice, it appears the election is much closer than the polls indicate. This does not mean Trump will win, but it does mean that counting Trump out of the race is a huge mistake.
A Trump victory?
That said, there is some highly persuasive scientific research on polling that says Trump will win.
Since the 1940s, pollsters have asked the question, ‘If the election were held today, who would you vote for?’ This is the so-called horse race question, and the answer gets all the headlines and publicity. But the pollsters also ask another question: ‘Regardless of who you plan to vote for, who do you think will win the upcoming election?’
The first question (‘Who would you vote for?’) is called the voter intention question. The second question (‘Who do you think will win?’) is called the voter expectation question.
Just as the voter intention question receives huge publicity, the voter expectation question gets very little attention. After all, if you know whom someone is voting for, who cares what that person thinks about how others will vote?
It turns out ignoring the expectation question is a huge mistake. Over hundreds of elections for many decades, the answer to the expectation question is far more accurate in predicting outcomes than the answer to the intention question.
The reason for this is a bit counterintuitive, but here’s why it’s true: When you answer the intention question, you are a sample size of one. If 1,300 people are asked the intention question, the sample size is 1,300 people. In normal polling, 1,300 is about how many respondents you need to get a fairly accurate poll. The margin of error for such a poll is about plus or minus 3%.
I’ve done polling for presidential campaigns and I know how expensive and difficult it is to get 1,300 people to answer a series of detailed questions. Many polls use smaller sample sizes, some as small as a few hundred respondents. A smaller sample size increases the margin of error.
How to leverage your sample size
Now, think about the answer to the voter expectation question. You’re not just asking the voter about her intentions; you’re asking about their expectation based on the likely actions of everyone they know. Suddenly, when you ask one person about their expectations, the sample size increases from one person to 20, 50, 100, or even more people, based on the size of their social network.
And here’s the key insight: The expectation question includes the intention of the individual answering the question. In other words, when you ask the expectation question, you’re getting an answer to the intention question and much more. In the language of statisticians, the answer to the expectation question is information rich.
With the expectation question, you’re effectively expanding the sample size from one person to, say, 50 people. If your base sample is 1,300 respondents and the average social network of each participant is 50 people, then the sample size on the expectation question is 65,000 people. This much larger sample size means a much smaller margin of error and a much more accurate forecast.
This analysis is not just a hypothesis. Its validity is borne out by decades of hard data.
78% chance of Trump winning?
Of course, in most cases the intention question and the expectation question will forecast the same winner, even if the expectation question has a better forecasting record overall. That makes sense. In a landslide election like Lyndon Johnson in 1964 or Ronald Reagan in 1984, the winner in the individual question and the winner in the expectation question will be the same.
Where it gets interesting is when one candidate is leading the polls in the intention question, but the opposition candidate is leading in the expectation question. In effect, a poll respondent is saying, ‘I’m voting for A, but I expect B to win.’ What happens then?
The data is unambiguous. Here’s what economists Rothschild and Wolfers said in their 2012 paper, ‘Forecasting Elections: Voter Intentions versus Expectations’:
‘In the 77 cases in which the intention and expectation question predict different candidates, the expectation question picks the winner 60 times, while the intention question only picked the winner 17 times. That is, 78% of the time that these two approaches disagree, the expectation data was correct.’
Guess what? That’s exactly where we are today.
The average response to the intention question shows Biden with a 49.3% to 40.7% lead over Trump. But the average response to the expectation question shows Trump with a 55% to 45% lead over Biden. The research shows the expectation question has the right forecast 78% of the time when the two polls disagree.
Putting this data together and using the best available science shows that Trump is the favourite to beat Biden, according to the latest polls.
Regards,
![]() |
Jim Rickards,
Strategist, The Daily Reckoning Australia
PS: Market expert Shae Russell predicts five knock-on effects of the recent market crash that could be even bigger threats to the average investor’s wealth than the crash itself. Learn More.
Comments