Paperonium

Child psychiatry · JAMA Psychiatry · 2015

Paper: a bigger brain region goes with 4% lower depression odds. Its own range says 91%.

The paper’s number4% lower: outside its rangeCentre of that same range0.087 (91% lower)0.96×0.01 to 0.75should be inside0.010.750.01×0.03×0.1×0.3×1×no changelower odds
The paper’snumber4% lower: outsideits rangeCentre of thatsame range0.087 (91% lower)0.96×0.01 to 0.75should be inside0.010.750.01×0.1×1×no changelower odds

What the odds of depression are multiplied by, per extra thousandth of the brain

The paper says

Children with a larger right anterior insula (a brain region) were less likely to be diagnosed with depression later. Each extra thousandth of the brain it fills multiplies the odds (chance of diagnosis against chance of none) by 0.96: 4% lower. Range: 0.01 to 0.75.

Its own data show

Its opening summary and results print 0.96, above its own range’s top, 0.75. The range is centred on 0.087, 91% lower odds: 0.087 ÷ 8.66 ≈ 0.01, 0.087 × 8.66 ≈ 0.75. Its p-value (how often luck alone would give a link this strong) of 0.03 fits 0.087; for 0.96 it would be 0.97.

A range for odds is made by dividing and multiplying the best estimate by the same number, so the estimate sits inside, at the centre in multiplying terms (not the simple halfway, 0.38). The paper calls this the first brain warning sign of repeat depression in children. Bigger still means less likely, but 0.96 makes the link look tiny.

“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 →

Anterior Insula Volume and Guilt · JAMA Psychiatry (2015) · doi:10.1001/jamapsychiatry.2014.1604