lukestein’s Twitter Archive—№ 9,860

@delong @Andrew___Baker Assuming you *need* the identifying variance from state-switchers, how much do you want to basically lie to unsuspecting readers about the size of your sample?