Public perceptions

Why Are People’s Risk Perceptions So Skewed?

Yesterday I noted that, 18 months after the start of the pandemic, a sizeable chunk of Americans still dramatically overestimate the risks of Covid. In a recent poll, more than one third said the risk of being hospitalised if you’re not vaccinated is at least 50%.

Of course, you’d expect some people to get the answer wrong just because we’re dealing with a small quantity, and there’s always going to be some degree of overestimation. But many people were off by a factor more than 10. What accounts for this?

Interestingly, Democrat voters’ guesses were much higher than Republican voters’ – about twice as many Democrats said the risk of being hospitalised if you’re not vaccinated is at least 50%. This suggests a role for ideology.

Throughout the pandemic, the ‘Democrat position’ has been to support restrictions and mandates, whereas the ‘Republican position’ has been to oppose such measures. This is clearly visible in a plot of U.S. states by average stringency index. Almost all the ‘red’ states are on the left-hand side, while almost all the ‘blue’ states are on the right.

Given that partisans (on all sides) like to avoid cognitive dissonance, they tend to adopt beliefs that are consistent with their party’s platform. Since Democrat politicians have been busy imposing all sorts of restrictions and mandates, Democrat voters have adopted beliefs that imply those measures were justified.

Most survey respondents don’t know numbers like ‘the risk of hospitalisation for people who aren’t vaccinated’ off the top of their head. Instead, they probably make a guess based on all the relevant information they can recall.

Democrat voters, who’ve spent the pandemic consuming media like MSNBC, CNN and NPR, will recall numerous incidents of pundits saying that Covid is extremely dangerous, and we have to do whatever we can to stop the spread.

They will also recall that they were locked down for months, that their kids’ schools were closed, and that they had to wear a mask whenever they went to the grocery store. 

Putting all this information together, they will tend to assume that the risk of being hospitalised from Covid is extremely high. ‘Why else,’ they might ask, ‘would there have been so many restrictions?’

Note: Republicans also overestimated the risk of being hospitalised from Covid, albeit to a lesser extent than Democrats. This indicates that people’s skewed risk perceptions cannot be blamed solely on the content of left-wing media (or the policies implemented in ‘blue’ states).  

The psychological quirk that may account for people’s skewed risk perceptions has a name in psychology: the availability heuristic. As Steven Pinker notes, “people estimate the probability of an event or the frequency of a kind of thing by the ease with which instances come to mind”.

Because plane crashes always make the news, people tend to overestimate the risks of air travel. And they may overestimate the risks of Covid for the same reason.

Since the start of the pandemic, we’ve been treated to morbid ‘daily death numbers’ – but for only one cause of death. Perhaps if these figures had been reported for all causes of death, people’s risk perceptions would be slightly less skewed. (Or perhaps they’d just be terrified of everything…)

During a pandemic, we obviously do want people to take precautions; we don’t want them nonchalantly walking into a care home when they have a high fever and a nasty cough. Yet – contrary to what some in government seem to believe – we don’t want people to be utterly terrified either.

There’s been so much attention on people claiming Covid is “just the flu” that the media has largely ignored the other end of the spectrum: people who believe Covid is the bubonic plague!

We can agree it’s bad if people underestimate the risks. But it’s also bad if they overestimate the risks. We want them to have the right risk perceptions. That way, they can make informed decisions.

Democrats Still Dramatically Overestimate the Risks of COVID-19

In a previous post, I noted that people tend to overestimate the risks of Covid, especially the risks to young people – which are vanishingly small.

In a Gallup poll last year, 41% of Democrat voters in the U.S. said that the risk of hospitalisation is at least 50%! (And Republicans didn’t do much better). However, that poll was taken in December. Has people’s understanding improved since then?

According to a new poll, the answer is ‘not at all’. Gallup posed a similar question as before, only this time they asked about vaccinated and unvaccinated people separately.

Note: the questions were not identical. In last year’s poll, they asked, “What percentage of people who have been infected by the coronavirus needed to be hospitalised?” In the recent poll, they asked, “What percentage of people have been hospitalised due to the coronavirus?”

The denominator for the first question is ‘people who have been infected’, while the denominator for the second is ‘everyone’. However, many respondents may have assumed that the second question was referring to ‘people who have been infected’. This should be kept in mind when interpreting the results.

The chart below shows results for the version of the second question that asked about unvaccinated people:

Once again, 41% of Democrats (and 22% of Republicans) said that the risk of hospitalisation for those who aren’t vaccinated is at least 50%. The correct answer is less than 5%, so these respondents were off by a factor of more than 10. Only 42% of Republicans – and just 18% of Democrats – were in the right ball-park.

Democrats did do substantially better when asked about the risk to vaccinated people, as the chart below indicates. In this case, the majority of both groups were in the right ball-park. However, more than one in five respondents still gave an answer of 10% or more.

As I mentioned last time, part of this overestimation may reflect a general psychological tendency to overestimate small quantities; though I should stress, only part. After all, Republicans were much less likely to answer “50%” when the question referred to unvaccinated people.

It’s staggering that 18 months after the start of the pandemic, almost one third of Americans say the risk of being hospitalised from Covid if you’re not vaccinated is at least 50%. Clearly there has been a failure of communication on the part of public health authorities.

This finding may help to explain bizarre phenomena like the fact that young, fully vaccinated Americans are still wearing face masks outdoors.

One Stanford student, Maxwell Meyer, spent an hour ‘bike-spotting’ on a popular campus thoroughfare. For each bike that went past, he recorded whether the rider was wearing a helmet, a face mask, or both. Of the 400 cyclists that he observed, 34% were wearing a mask but no helmet! (And 7% were wearing both.)

Aside from some people simply being clueless about the risks, Meyer notes that wearing a mask has become a form of social signalling (‘I’m the sort of person who cares about doing his part’). Though of course, wearing a mask under such circumstances does approximately nothing – other than raise the question of how on earth you got into Stanford.

Even after lockdowns ended, various types of ‘Covid theatre’ have dragged on for months. This isn’t so surprising when you consider people’s skewed perceptions of the risks.

Is this the True Level of Opposition to Restrictions? Over Half of 18-34 Year Olds Have Deleted the NHS Covid App or Never Had It

Opinion polls are sometimes poor at distinguishing between virtue-signalling and what people really think – respondents tell the pollsters what they think they are supposed to say, especially on issues that have become moralised, like anti-Covid measures.

But actions speak louder than words. So the news that 19% of people have deleted the NHS Covid app (which pings you to tell you to self-isolate if you are identified as someone who’s come into contact with a person who’s tested positive) and so joined the 32% of people who never had it, according to a new ComRes poll, perhaps gives a better indication of how many people are not so keen on Covid restrictions. Among 18-34 year olds, over a third – 34% – have deleted the app, which is as many as still have it, while 21% never downloaded it in the first place – despite 98% owning a smartphone.

To my mind, statistics like these are a much more realistic indicator of who actually supports restrictions, since if you’re not willing to self-isolate when potentially infected, how can you be in favour of less targeted measures? This would mean just 42% of people are genuinely in favour of restrictions continuing.

True, you have to allow for the 16% of adults who don’t have a smartphone. If we assume this group splits in their views in the same proportions as those who do have a smartphone then we get 44.5% against restrictions in practice versus 50.5% in favour. This is probably an upper bound for those in favour, as some may just be saying they have the app even though they don’t, and some may have downloaded it just for appearance’s sake. Furthermore, some may not be supportive of measures beyond isolation of contacts (though I assume that anyone who favours more restrictive measures must favour self-isolation of contacts as it seems the bare minimum of restrictions beyond isolation of the infected).

Among 18-34 year-olds, those opposed to restrictions (by this measure) outnumber those in favour by 55% to 34%.

Such figures sound much more likely to me than the alarming support for draconian restrictions that often appears in opinion polls. They suggest that if politicians think the public are solidly behind the continuation of restrictions then they are in for a nasty shock come polling day. Politicians should pay closer attention to what people do than what they say.

Evidence Suggests People Overestimate the Risks of COVID-19

One of the most surprising things to emerge from the pandemic, at least from a lockdown sceptic’s point of view, is how overwhelmingly the British public has backed the lockdowns. For example, a YouGov poll taken in March of 2020 found that 93% of people supported the first lockdown. Another poll taken in January of 2021 found that 85% of people supported the third lockdown.

While some lockdown sceptics claim these polls can’t be trusted, I suspect they’re not too far off the mark. And even if they do overstate support for lockdowns (due to unrepresentative samples or social desirability bias) the true number is unlikely to be more than 10 or 20 percentage points lower.

The high level of public support for lockdowns may explain why they’ve lasted as long as they have. Politics is notoriously short-sighted, so why would the Conservatives ease up on a policy that’s kept them ahead in the polls for most of the last 14 months?

Aside from the public’s longstanding reverence for the NHS, an obvious reason why support for lockdown is so high is that millions of people have been paid 80% of their wages to stay at home. In the absence of the Government’s unprecedented furlough scheme, many of these people would be out of work, and presumably much less supportive of lockdowns.

However, there might be a more important reason why support for lockdown is so high: the public overestimates the risks of COVID-19, especially the risks to young people. Let’s review the evidence.

In July of 2020, the consultancy Kekst CNC ran a poll asking Britons what percentage of the population has died of COVID-19. The correct answer at the time was around 0.1%. However, the median answer among respondents was 1%, and of those who ventured a guess (rather than saying “don’t know”) one in five said at least 6% of the population had died.

Last year, Gallup ran a poll for Franklin Templeton in which they asked Americans what percentage of people who’ve been infected with COVID-19 need to be hospitalised. Less than 20% of respondents gave the correct answer of “1–5%”. And a staggering 35% said at least half of those infected need to be hospitalised. Interestingly, Democrats were much more likely than Republicans to overestimate the risk of hospitalisation, as this chart reveals:

It should be noted that the poll also revealed some underestimation of risks on the part of Republicans. For example, 41% incorrectly stated that flu causes more deaths than COVID-19. This shows that results can vary depending on exactly which question you ask. (Notice that Republicans did overestimate the risk of hospitalisation; just to a lesser extent than Democrats.)

Likewise, a survey carried out by Ipsos MORI for Kings College London asked Britons what are the chances of needing hospital treatment if you catch coronavirus. The median answer among respondents was 30%, and of those who ventured a guess, one in four said the chances are at least 50%.

In March and April of last year, the economist Arthur Attema and colleagues carried out two surveys of the French population: one two weeks after the first lockdown began, and the other two weeks before it ended. They asked respondents, “Out of 100 people who are infected with the Coronavirus, how many of them die from the disease?”

In both surveys, the average answer was 16 (whereas the correct answer for Western populations is less than 1). The fact that the average in the second survey was no lower than the average in the first indicates that people’s understanding of the risks did not improve over time, despite more evidence accumulating that the IFR is less than 1%.

Members of the public seem to have a particularly skewed perception of the risk COVID-19 poses to younger people. The aforementioned Gallup poll asked Americans what percentage of those who’ve died were aged 24 and under. The correct answer at the time was around 0.1%, yet the average answer among Republicans was 8%, while the average among Democrats was 9%.

Likewise, a poll taken by Ipsos MRBI for The Irish Times asked people what percentage of those who’ve died were under the age of 35. The correct answer was around 1%, yet the average among respondents was 12%.

In November of 2020, Savanta ComRes ran a poll on behalf of The Conservative Woman and asked Britons to guess the average age of people who’ve died after testing positive for COVID-19. The correct answer is around 82. However, the median answer among respondents was 65.

Incidentally, one problem with asking people to estimate very small quantities (like the percentage of people who’ve died from COVID-19) is that humans have a tendency to revise small percentages upwards when they’re not sure. This “uncertainty-based rescaling” probably accounts for some of the overestimation in the surveys mentioned above.

However, taking all the evidence together, people – particularly in Britain – do seem to overestimate the risks of COVID-19. And this may help to explain their high level of support for lockdowns.