Maybe there’s some fancy statistical methods that mitigate this, but from what I know about study design, if 9 women are happy because their period is more manageable/gone, and one woman feels very down because of the hormones, that might show up as ‘no effect on mood’, and then this gets communicated as ‘it definitely does not have an effect on mood, we did proper studies’ even if there is a sizable minority that gets reliably suicidal on it. Not sure if there is a name for this, but I always try to keep this in mind for any drug. Humans are sometimes just very different in how they metabolize things.
Maybe there’s some fancy statistical methods that mitigate this, but from what I know about study design, if 9 women are happy because their period is more manageable/gone, and one woman feels very down because of the hormones, that might show up as ‘no effect on mood’, and then this gets communicated as ‘it definitely does not have an effect on mood, we did proper studies’ even if there is a sizable minority that gets reliably suicidal on it. Not sure if there is a name for this, but I always try to keep this in mind for any drug. Humans are sometimes just very different in how they metabolize things.
The method here would be to just estimate the entire distribution instead of just the mean.