News and Events

Read all the latest news within the Division of Psychology and Language Sciences.


Read more about the PALS Swan Silver Award

Divisional Subject Pool

Access and sign up for the Divisional Subject Pool.

Green Issues

The Division has won the Silver Award for the UCL Green Impact scheme

CPD@PaLS Courses

The Division runs Continuing Professional Development courses.

Communicating Environmental Risks


Communicating environmental risks: Clarifying the severity effect in interpretations of verbal probability expressions.

Communicating the risks of climate change is a major challenge. The body in charge of reviewing the science of climate change – the IPCC (Intergovernmental Panel on Climate Change) – use verbal statements of likelihood, rather than numbers, to communicate risks to policy makers and the public. For example, instead of saying “There is a 90% chance that…” researchers are encouraged to state that “It is very likely that…” The IPCC have even provided guidelines for how numerical risks should be translated into words (below).

The translations from numbers to verbal phrases used by the IPCC
Numerical probability Probability phrase
>99% Virtually certain
>90% Very likely
>66% Likely
33% to 66% About as likely as not
<33% Unlikely
<10% Very unlikely
<1% Exceptionally unlikely

The problem is that numbers cannot simply be mapped onto words. In our study, we tested the possibility that probability phrases referring to events of differing ‘severity’ (i.e., really bad vs. less bad events) would be interpreted as denoting different levels of numerical probabilities. Past research has yielded mixed or weak results, as a result of the difficulty of separately the influence of event severity from the influence of the general chance of an event happening. Typically, more severe events (e.g., hurricanes) are rarer than less severe events (e.g., a blustery day). We already know that the general likelihood of an event will lead to higher interpretations of probability phrases – If I tell you that it is likely there will be a hurricane tomorrow, you will assign this a lower numerical probability than if I tell you that it is likely that tomorrow will be blustery.

We (myself and Adam Corner [Cardiff University]) therefore set out to investigate the effect of severity on interpretations of probability phrases whilst keeping the general chance of the event happening (its ‘base rate’) constant. To accomplish this, we developed a method in which participants always judged the same event (in one case this was a 3 foot increase in global sea levels), but we changed the consequences associated with that event. In the low severity condition, participants were instructed to imagine that they lived on a high island, ‘Sulasemi’, that is protected from the sea by 50 foot cliffs, and so if sea levels rise by 3 feet, life on the island will be unaffected. In the high severity condition, participants were informed that the island was very flat, and so if sea levels rise by 3 feet, the island will disappear into the ocean. Participants were informed that the IPCC had issued a warning that by 2015 it is unlikely, perhaps very unlikely, that global sea levels will have risen by 3 feet or more. Participants were subsequently asked to provide a number between 0 (impossible) and 100 (it will definitely happen) that they thought matched the likelihood of this event happening.

We found that in the scenario above, as well as two others relating to the transportation of nuclear materials and the implementation of nanotechnologies, participants interpreted the probability phrases as denoting higher numerical probabilities when they referred to the severe event (i.e., when the island was flat) than the more neutral event (the high island).

The next stage in this work is to understand the full scope and implications of this effect, as well as the reasons underlying it. In summary, though, my work, together with that of other research groups (notably David Budescu [Fordham University, US] & colleagues), suggests that the use of words to describe risk likelihoods must be undertaken with extreme caution. I would always recommend that words be accompanied by numbers, to facilitate the most effective communication of true risk levels.

Download the full document in pdf format

Across both high and low probability phrases, participants interpreted the phrases as denoting a higher numerical risk when their island would be wiped out (high severity) than unaffected (low severity).