Table 1 Overview of research questions, hypotheses and findings by country.

Pilot

(1) Do we observe a severity effect (i.e. are the same likelihoods interpreted differently according to the severity of impacts to which they refer?

H₁ Higher warnings will be issued for the same likelihood category when it refers to a significant or severe impact versus a minimal or minor one

Main

(1) Do we observe a severity effect in a dynamic IBW paradigm? (Replication of pilot)

H₁ Higher warnings will be issued for the same likelihood category when it refers to a severe impact versus when it refers to a minor one

(2) Is the severity effect more pronounced in sequential IBW scenarios, rather than single warning IBW scenarios?

H₂ The magnitude of the severity effect will be more pronounced at the second stage of a multi-stage forecasting scenario

(3) Does a severity effect have a continued influence when making sequential weather warnings?

H₃ Likelihood warnings provided for identical events (in terms of likelihood and anticipated impact) 24 h from the event will be higher where the three-day forecast referred to a more severe impact

(4) Does a severity effect manifest itself when translating weather warnings to numerical likelihoods?

H₄ Higher numerical likelihoods will be assigned to VPEs when these refer to a severe impact compared to a minor impact

(5) Are forecasters aware of the severity effect and what are their intuitions regarding its optimality?

Due to the exploratory nature of RQ₅, we did not have specific predictions regarding this research question

Little evidence thus far