In this post, Mark Bowker (Lund University) discusses their recent article in Journal of Applied Philosophy on our responsibility to be careful with scientific generalisations.

In a pandemic it is extremely important for the public to know how they can keep themselves and others safe. This requires effective communication to circulate information about scientific developments. In a recent article, I argue that even the most basic statements can be misleading, so we must think very carefully about the words they use. You may have heard, for example, that children do not transmit coronavirus, but this statement is not as simple as it may seem.

Why Language Matters

‘Children do not transmit coronavirus’ is what we call a generic generalisation, or simply a generic. It is a generalisation that doesn’t include a word like ‘always’ or ‘never’ to make the generalisation precise. ‘Children never transmit coronavirus’ is very precise. It tells you that it is impossible for children to transmit the virus under any circumstances. But the meaning of a generic is not entirely precise. What was does it mean to say simply that children do not transmit the virus? Does it mean that it is impossible for any child to transmit the virus, or does it mean that children don’t generally transmit the virus, though there might be some exceptions? Compare ‘Squares don’t have three sides’, which means that it is impossible for a square to have three sides, and ‘Dogs don’t have two legs’, which means that dogs don’t generally have two legs but allows exceptions like dogs that have been in accidents.

Neither of these interpretations is obviously the best, so different people might interpret the generic in different ways. Some might take it to be very strong, like ‘Squares don’t have three sides’, and others might take it to be weaker, like ‘Dogs don’t have two legs’. This isn’t a problem in itself. For many purposes, it doesn’t matter whether people take the generic to be stronger or weaker. Swiss coronavirus spokesperson Daniel Koch, for example, used the generic when explaining why Switzerland decided to allow grandparents to hug their grandchildren. In this context, perhaps, it doesn’t matter exactly what the generic means. Whether it is impossible for children to transmit coronavirus or it’s only rare, the risk to the population might be acceptable.

Problems arise, however, when generics are taken to answer other questions. Koch’s statement was used to support the global reopening of schools, for example. Afterall, if children don’t transmit the virus, then what is the risk? This reasoning is very natural, but it relies on a strong interpretation of the generic. If it is impossible for children to transmit the virus, then perhaps schools should be reopened. But if it is only rare for children to transmit the virus, then millions of children around the world mingling in schools with friends and teachers might still lead to dangerous spread. As a generic spreads throughout the community, it might be used to answer all kinds of questions, and there is a risk that these answers could be misleading.

What Can We Do?

So what should we do about this problem? One option is for scientists to avoid using generics altogether, but that risks throwing the baby out with the bathwater. Generics are extremely useful because they are a simple way of expressing complex findings that most of us wouldn’t have any chance of understanding. Rather than eliminating generics entirely, I suggest scientists think more carefully about the generics they use and the questions that people might take them to answer. If Koch wanted to avoid appearing to support the reopening of schools, for example, then he could have said ‘Children with COVID-19 pose a low risk to their grandparents’. This is still generalisation, but it doesn’t have any obvious connection with reopening schools.

Journalists have a role to play as well.  Various outlets reported Koch as having said that ‘Children cannot pass on coronavirus’. This is a very strong interpretation that is far more likely to lead people to strong conclusions. When reporting scientific generalisations, journalists should think about the questions that the scientist intended to answer and exactly how strong the generalisation needs to be to answer them. Like scientists, they should also think carefully about the questions that a generic might be used to answer and to avoid repeating generalisations that might be misleading.

We, the public, can also help. Just like the journalist, we should be careful not to exaggerate scientific generalisations or to repeat generalisations that might be misleading. We are all part of the network through which information spreads, and it is good for everyone if that information spreads efficiently and accurately. If you find yourself using a generic generalisation, you might want to clarify exactly what you mean.