Didier Fourcot Mar 16, 2016:

Statistical wording In fact the study is on a number of people, this is not 1 SD of HDL cholesterol on 1 person that leads to 0.65 hair loss.

Thus:
- 1000 persons have x% risk or hair loss with HDL cholesterol average y SD s
- among these people the ones with HDL cholesterol value (y + s) have a risk of hair loss 0.65.x IN AVERAGE (not every single person has 0.65.x, but the average of this population with this HDL cholesterol value)

SuzyKeller (asker) Mar 16, 2016:

Indeed, thanks a lot :)

Tony M Mar 16, 2016:

@ Asker Right! So my guess was pretty close then after all ;-) Glad you got it sorted!

SuzyKeller (asker) Mar 16, 2016:

I actually found the original article online and it makes more sense: Every standard deviation increase in HDL cholesterol was associated with a 0.65 times lower risk of hair loss.

Tony M Mar 16, 2016:

@ Asker Yes, I was thinking the same thing! And if the rest of the document seems to bear it out, then that is probably the explanation. Frankly, you ought to go back to your customer and say it is impossible to translate this gibberish, for fear of making a significant mistake.
My GUESS would be that in that case they are trying to say something like "the median hair-loss risk was reduced by a factor of 0.65"

SuzyKeller (asker) Mar 16, 2016:

I was also very confused by the structure of the whole sentence... Seems like a non-native wrote it.

Tony M Mar 16, 2016:

@ Asker Statistics isn't my field, but 'reduced of median' makes no sense to me at all!

I'm wondering if it might be a typo for 'off-median', which might then have some sense at least?!