One of many examples is:https://twitter.com/kymberlylouise_/status/1065317963945848832?s=19 …
-
-
Sorry, I was conscious of this even as I was typing 'exploratory', 'confirmatory' etc. - it's buying into an already-skewed narrative.
-
But what problem is this narrative constructed to solve? I mean for a clinical trial you have protocol, CRF, SAP, audit trial etc. and it's clear what problems that solves and why it's appropriate
-
But how are the problems for other analysis settings characterised, and what has been proposed to solve them? So social psychology has set about cleaning house, but that's a very specific set of conditions
-
Exactly. There are two two things to bear in mind that are often left out: 1) all houses (=fields) get dirty — entropy increases; 2) no reason to assume modelling's house is THAT dirty — it's clean (given a realistic definition of clean). For 2) that doesn't mean: stop cleaning.
-
For 2) also that doesn't mean we're perfect. Nobody is perfect. But we're actually pretty good, we do replications (in all senses, including re-implementations) as a matter of course.
-
This is not to pat us on the back too much, but if dirt is defined in the way social defines dirt, then we're on the right track already. We have — no surprises — OTHER issues.
-
So, just out of politeness, at the very least the conversation should be “What’s SOP for your field, and why?” before bringing the Dyson to the front door
-
I mean notice nobody talks about a file drawer or a publication bias for models... because it's a DIFFERENT kind of work.

*deep breaths*
End of conversation
New conversation -
-
-
I don't mean in this thread, just to be clear. I mean the wider debate that is mainly about social — I have no idea what or why they are doing with it. The framing might apply to their work, but to mine & to my understanding of the field & science generally, it's just all so off.
-
Relatedly, I don't find the distinction between exploratory and confirmatory very useful for modelling purposes.
-
Yep, because it's nearly all theory driven and thus more like confirmatory work.. but abductive work doesn't work in the realm of confirm/deny.
End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.