In my view both statisticians were wrong. My answer would be “slim people lost weight and heavy people gained weight, both independently of their gender”. The first statistician fell into the trap of only looking at average (raw and bunt steaks are on average cooked to perfection) ; the second statistician is just struggling with (a variant of) Simpson’s paradox.
Great work, Peter! Thanks for producing such an engaging and high quality presentation :)
What happens to this argument when you are measuring latent scores (or latent difference scores) free of measurement error?
Fantastic.
Did he just say difference-in-difference -estimation is rubbish :O ?
In my view both statisticians were wrong. My answer would be “slim people lost weight and heavy people gained weight, both independently of their gender”. The first statistician fell into the trap of only looking at average (raw and bunt steaks are on average cooked to perfection) ; the second statistician is just struggling with (a variant of) Simpson’s paradox.