You already saw and worked with the IPCC probabilities. Below are key excerpts from the 2019 report.

How to communicate the assessed probabilities of outcomes

A calibrated language scale is used to communicate assessed probabilities of outcomes, ranging from exceptionally unlikely (<1%), extremely unlikely (<5%), very unlikely (<10%), unlikely (<33%), about as likely as not (33–66%), likely (>66%), very likely (>90%), extremely likely (>95%) to virtually certain (>99%).

These terms are normally only applied to findings associated with high or very high confidence. Frequency of occurrence within a model ensemble does not correspond to actual assessed probability of outcome unless the ensemble is judged to capture and represent the full range of relevant uncertainties”. (IPCC Report, 2019)

Nonetheless, there are other calibrated language scales used to communicate probabilities. Below you can find a table with another connection between verbal and numerical uncertainties, with probabalities (first row) and corresponding relative frequencies (second row).

For example, “virtually certain” event means that 999 out of 1000 times that event occurs.

Probability of Failure

Virtually certain	Very likely	Likely	Neutral	Unlikely	Very Unlikely	Virtually Impossible
0,999	0,99	0,9	0,5	0,1	0,01	0,001
999/1000	99/100	9/10	5/10	1/10	1/100	1/1000

Does this one provide a more accurate quantification of your uncertainty?

Notice that in the above table, the frequencies are included along with the probabilities. These frequencies and probabilities have been actually used in expert elicitations. We have noticed that experts could relate more to frequencies (of failure, for example) rather than to probabilities.

How can experts quantify their uncertainty, depending on the different settings where their opinion is elicited?

The next video will answer this question.