Numbers don’t always mean what they seem to mean | David Spiegelhalter and Anthony Masters

Last Wednesday, the Evening Standard trumpeted “Covid deaths rocket to 207 in deadliest day in nearly 6 months”. In fact, deaths within 24 days of a positive test were slightly down over the previous week. So what did it get wrong?

It appears it fell for the common misunderstanding, even after 18 months of pandemic, that the daily death figure represents those occurring in the last 24 hours, whereas it is deaths that have been reported in that period and reports are fewer on weekends and holidays. The spike of 207 picked up the backlog from the bank holiday weekend – only 50 had been reported the previous day.

Not all statistics mean what they appear to mean. Some do not match common understanding, such as politicians counting a new wing or major refurbishment as a new “hospital”. Others are more subtle, like the BBC headline, “Covid: disabled people account for six in 10 deaths in England last year”, accompanied by a photograph of a person in a wheelchair. In this context, “disability” means something much wider than this may suggest, being defined as some long-term limitation to daily activities, even if related to old age. The Office for National Statistics (ONS) data shows almost as high a share (54%) of non-Covid-19 deaths also met this definition. That reflects the observation that people who die of Covid are similar to those dying from other causes – Covid picks on the weak and vulnerable, just as normal life does. Exceptions include those with an increased risk of catching it.

When analysing the excess risk for people with a severe learning disability, ONS reports “the largest effect was associated with living in a care home or other communal establishment”. Economist Tim Harford talks about the dangers of “premature enumeration” – focusing on the numbers before being clear what they refer to. We are not immune. In last week’s column, we wrote about ethnic groupings of currently or recently pregnant women, but used an inappropriate classification of ethnicity of full-term live births.

It’s best to pause and make sure the numbers really mean what you think they mean.

David Spiegelhalter is chair of the Winton Centre for Risk and Evidence Communication at Cambridge. Anthony Masters is statistical ambassador for the Royal Statistical Society



Source link