Harry Shannon
What is it about people who write or speak for a living? You often hear them proudly announce that they can’t do math. Which really means they are admitting they’re not properly educated.
It has consequences. If journalists are innumerate, they can’t do their job well. Anyone with an agenda can easily mislead them, and in turn they can mislead us. A recent example – in the Daily Mail (January 29, page 6) a headline read: “2 Pfizer doses ’99.96% effective’.” Very impressive, even Pfizer hadn’t made that bold a claim for its COVID-19 vaccine.
The short article, by Kate Pickles, Health Correspondent, reports data from Israel, and claims that “Almost all cases of Covid are prevented with two doses of the Pfizer vaccine.” It states that “data … found that just 0.04 per cent of those vaccinated with both doses went on to develop coronavirus.” (I’ll ignore the sloppiness in confusing the virus with the disease it causes. Either that or, as my younger son suggested, some serious bioengineering is involved.) The story reported that of 700,000 people vaccinated, 300 “contracted the virus.” So simple arithmetic: 300/700,000 = 0.04%. And 100 minus 0.04 is 99.96.
Now the article doesn’t mention the 99.96% effectiveness in the headline. (The figure was in quote marks, suggesting that the editor who wrote the headline made it up as a quote, or someone said it, but their quote was edited out.)
But there are two big problems. First, the report doesn’t take time into account. It doesn’t state how long they looked at people after the vaccinations. Suppose they’d waited 10 minutes. There’d be no new cases, so by the Mail’s logic (if you can call it that) the effectiveness would be 100%! If they waited twice as long as they did, you’d expect twice as many cases and you’d have to say that the effectiveness was going down – and would keep going down the longer you wait.
Second, you don’t know how many of the 700,000 would have been infected if they hadn’t received the vaccine. You need a control group to estimate effectiveness. So how do you do that?
The Pfizer vaccine study was reported in the New England Journal of Medicine. Participants were randomized to vaccine or control group (‘randomized’ means it’s essentially a toss of a coin). Controls were given a placebo injection. There were almost exactly the same number of participants in each group. The researchers waited for 7 days after the second vaccine dose was given, enough time for antibodies to the virus to develop.
They then started counting the cases of COVID-19 for (on average) 6.6 weeks. There were 162 in the control group, but only 8, 95% fewer, in the vaccine group. And the 95% is the number that you saw in the news. (Technically, this was ‘efficacy’ not ‘effectiveness,’ but the difference – explained here – doesn’t matter for my point.)
It’s more than disappointing that the Mail’s Health Correspondent and the editor can’t get this right. Numeracy should be a job requirement for journalists.
This not to say that the vaccines aren’t effective. The evidence shows they are. I’ll be getting mine when it’s my turn.