Public health · JAMA Network Open · 2024
Paper: food stamps help people short of food more. By its own rule, that’s not shown.
Difference per 100 patients skipping pills, with food stamps versus without
The paper says
Among blood-pressure patients short of food, those on food stamps (government grocery money) skipped pills less: 13.6 fewer in every 100 than similar patients without. Among those with enough food: no clear effect. The paper calls this a real difference between the groups.
Its own data show
Its Table 2 gives each result a range where the true effect plausibly lies: short of food, 25.0 to 2.3 fewer in 100; enough food, 15.8 fewer to 25.7 more. The paper’s methods say two results count as truly different only if their ranges don’t overlap. These overlap, from 15.8 to 2.3 fewer.
Overlap means one effect, say 10 fewer in 100, fits both groups. Tested directly, a gap this big would turn up by chance alone about 1 time in 8 even with no real difference. The result for people short of food still stands; the paper’s advice to aim scarce aid at them rests on an extra benefit it hasn’t shown.