Serious question: if people treat p values -- continuous variables from 0 to 1 -- dichotomosly, what would stop them from treating CIs, which literally dichotomize the parameter space into "inside" and "outside", dichotomously? Telling them not to?
Conversation
Serious answer: telling them not to would be a good first step, could do no harm, and would address the deeper underlying issue (the urge to dichotomize), so why not? I think this is what the Retire paper and ASA special issue are doing.
2
1
How long have people been telling others not to do this with p values? People immediately started dictotomizing BFs. CIs are not a solution to the problem expressed (people understand them as little as or less than p values). They are just a marginal improvement in reporting.
2
7
CIs are "only a marginal improvement": the first figure in the Nature comment is a proof that your affirmation is wrong.
Look, I ask for measures of uncertainty when I review. But let's not kid ourselves: an interval is a *very* impoverished way of representing data/inference. Every time I see a mean+CI, and nothing else, I groan. It is an improvement, but a marginal one.