Eye research · J Neurosci · 2018
“Significantly” fewer dying eye cells, says the paper. By its own rule, that could be chance.
How often chance alone gives this gap if the gene made no difference
The paper says
When the retina, the eye’s light-sensing layer, comes loose, its light-detecting cells die. The paper’s summary leads with this: a day later, mice lacking one gene had about half as many dying cells as normal mice. A note under a figure calls it “significantly less.”
Its own data show
The paper’s methods set one rule: a difference counts as “significant” only if its p-value is below 0.05. Beside “significantly less”, the figure note prints p = 0.067. That fails the rule, and the paper makes no exception. Of the 16 exact p-values the paper prints, this is the only one above 0.05.
A p-value is how often chance alone would produce a gap this big if the gene made no difference. The paper’s bar is 1 time in 20; this result is about 1 in 15. So by its own rule, the summary’s lead finding could be chance, though other experiments in the paper point the same way.