Paperonium

Neurology · Brain · 2025

The summary’s test result needs 391 people. The study recruited 328; its table lists 146.

The study recruited 328 peopleThe summary’s score of 18.5needs 391 peopleimpossibleIts results’ score of 7.7uses 146 people0100200300400
The study recruited 328 peopleThe summary’s score of 18.5needs 391 peopleimpossibleIts results’ score of 7.7uses 146 people0100200300400

How many people each version of the test result needs

The paper says

Healthy volunteers and patients with front or back brain damage took a logic test. Testing whether the three groups differed, the summary prints an evidence score of 18.5 (bigger means a clearer difference) and a head-count number of 387. Front patients did worse than volunteers.

Its own data show

For the paper’s stated analysis, that number is the people tested minus 4; its other checkable results all fit. For this test its results print 7.7 and 142, and its table lists 146 people (48 healthy, 55 front, 43 back): 146 − 4 = 142. The summary’s 387 needs 391 people; only 328 were recruited.

The evidence score grows with both the gap and the number of people. Both versions show about the same gap, but the summary’s score is 2.4 times larger and implies 2.7 times the people. The three groups do differ: with no real difference, chance alone would produce gaps this big about 1 time in 1,500. The summary overstates that evidence.

“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 →

A right frontal network for analogical and deductive reasoning · Brain (2025) · doi:10.1093/brain/awaf062