Paperonium

Public health · Nat Med · 2023

Summary: 36.6% vaccinated when offered cash, 30.3% when not. Its own counts give 33.6% and 28.4%.

Offered cash33.6%Its own count: 355 of 1,05836.6%The summarytoo highNo cash28.4%Its own count439 of 1,54430.3%The summary25%303540
Offered cash33.6%Its own count355 of 1,05836.6%The summarytoo highNo cash28.4%Its own count439 of 1,54430.3%The summary25%303540

Out of every 100 people in each group, how many got a vaccine shot

The paper says

A trial in Ghana offered people cash to get a COVID-19 vaccine. The summary atop the paper says 36.6% of those offered cash got a shot (355 of 1,058), against 30.3% of a comparison group offered no cash (439 of 1,544): 6.3 more per 100.

Its own data show

Divide those same counts: 355 ÷ 1,058 is 33.6%, and 439 ÷ 1,544 is 28.4%. The paper’s results section, describing the same people, prints exactly those two figures and a gap of 5.2 people per 100. Rounding can’t explain 36.6%: that would need 387 of the 1,058 vaccinated, not 355.

A percentage is a count divided by its group’s size, so anyone can check one. This is the trial’s only outcome checked against health office records; the others are what people told researchers. The summary, the part reviews usually quote, puts cash’s boost at 6.3 extra people per 100 instead of 5.2. Cash still helped, by less.

“The summary” means the paper’s abstract: the short summary printed at its top, which is the part most readers see.

Read the paper →Full technical write-up →

Financial incentives for COVID-19 vaccines in a rural low-resource setting: a cluster-randomized trial · Nat Med (2023) · doi:10.1038/s41591-023-02670-4