with modelling reproduction and reimplementation is v important with respect to theory/validation etc
-
-
Replying to @o_guest @Darren_Rhodes
In principal I stand by "pre-reg everything", but if you're not collecting new data, it's a promise no-one can check
1 reply 0 retweets 0 likes -
Replying to @annemscheel @Darren_Rhodes
model is implementation of theory imho so I could preregister a theory, but the point is to discover the theory
2 replies 0 retweets 0 likes -
Replying to @o_guest @Darren_Rhodes
Sure but there are tons of details in a model that can be tweaked and will lead to cap. on chance if done post hoc.
1 reply 0 retweets 0 likes -
Replying to @annemscheel @Darren_Rhodes
hence why you publish the specificatiom
1 reply 0 retweets 0 likes -
speciation captures theory, so like a preregistration which can then be tested like usual for empirical data
2 replies 0 retweets 1 like -
does the specification include a clear description of all the parameters you tweaked that didn’t work?
1 reply 0 retweets 1 like -
Replying to @talyarkoni @o_guest and
if not, you’ll overfit your computational model in exactly the same way as an empirical analysis
1 reply 0 retweets 2 likes -
true, I actually have an appendix in my PhD with every parameter setting for a model in tables
1 reply 0 retweets 0 likes -
but is that just for the final model, or for every model you discarded along the way to get there?
1 reply 0 retweets 0 likes
surprisingly yes, it was a replication so every model attempt as part of replication is in the appen
-
-
Replying to @o_guest @talyarkoni and
for both of the different models I replicated/failed to replicate
0 replies 0 retweets 1 likeThanks. Twitter will use this to make your timeline better. UndoUndo
-
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.