Good survey methodology urges the use of seven-point scales, known as Likert scales, such as Very improbable (rated as 1), Improbable, Somewhat improbable, Neither probable nor improbable (4), Somewhat probable, Probable, and Very Probable (7). Most respondents don’t do well with more elaborate scales, there needs to be a neutral middle point (4) , and two calculations favor the seven-point choice.
If a second question asks about the importance of an event, again on a seven-point Likert scale, you can multiply the probability score by the importance score. Many two-by-two quadrants follow this methodology: respondents rate something on two scales, multiply the scores, then distribute them to the appropriate quadrant.
Another neat calculation becomes easy, as explained in Ron Hale-Evans, Mind Performance Hacks: Tips & tools for overclocking your brain (O’Reilly 2006) at 171. If you double the ratings, the scores that come from multiplying one scale score by the other for something, you get a result ranging from 2 to 98, which closely approximates a percentile scale (0-100).
Likert scales find many uses, such as in client satisfaction assessments, employee morale instruments, evaluations of law firms, software selections and anywhere a survey adds value.