Just discovered the Causal Quartet by @lucystats.bsky.social@travisgerke.bsky.social@malcolmbarrett.malco.io#rstats
Wait, how is the true effect of X on Y zero if it’s mediated? Is that the standard interpretation? (In which case I just have gotten that wrong since forever…)
Surprised that I’ve missed this. Really interesting, thanks for sharing. Like Anscombe's quartet. Beautiful.
Love seeing ggfx in the wild!
It seems off to say that in the mediator scenario the true effect is 0 - the direct effect is 0, but it would be wrong to conclude x has no effect on y.
Very interesting!
Are there counterpart visualizations to the causal quartet that show the causal effects w/ raw data? I like Anscombe's quartet because you can show the data tables and ask people to look for patterns, they scratch their heads for a few min, then you reveal the scatterplots and they see it instantly