Léo Zabrocki and I published a working paper arguing that causal inference methods can produce inflated published estimates.
They intrinsically reduce statistical power and it creates a trade-off between confounding and exaggerating true effect sizes. 1/N
#EconTwitter
They intrinsically reduce statistical power and it creates a trade-off between confounding and exaggerating true effect sizes. 1/N
#EconTwitter