Bjørn Sætrevik

I just updated a #preprint from a couple of years ago (mostly edited for readability): https://psyarxiv.com/f9382

We wanted to see if medical students making #diagnoses would show a #PrimacyBias, #ConfirmationBias and #overconfidence. We did three #preregistered classroom experiments. Results were... mixed. (1/4)

#diagnostics #DecisionMaking #primacy #CongruenceBias #confidence #preregistration #replication #OpenScience

Nov 18, 2022 at 9:59amWeb
Show threadBjørn SætrevikNov 18, 2022

We found a small effect of confirmation bias, and increase in confidence despite not getting more conclusive information. But we did not see a primacy effect of the symptoms presented first being emphasized, or of confirmation bias being driven by confidence.

(2/4)

Show threadBjørn SætrevikNov 18, 2022
I originally expected #primacy, #ConfirmationBias and #overconfidence to clearly emerge in our procedure and in our sample. I guess #replication is hard. On the other hand, one may argue that the original effects are typically shown using more artificial tasks and novice #decisionmakers. It could be that experts in more realistic tasks within their domain think more carefully and thus mostly avoid the biases (i.e. #NaturalisticDecisionMaking). (3/4)
Show threadBjørn SætrevikNov 18, 2022

To explore this further we are planning a follow-up experiment on the same kind of sample. Here we will revise the experiment design to improve on some of the weaknesses of the preprint study.

However, I'm also considering redoing the preprint experiment with a larger online sample of non-experts, to see if the biases would then emerge.

Comments or suggestions would be very welcome! (4/4)