[Mick]

This seems like an unrealistically precise number.  37% versus 35% versus 40%.  I think in these cases, the uncertainty associated with any model makes these models little better than a subjective evaluation; thus I might suggest bucketing these into a few subjective categories: “very likely” (95+%), “likely” (80 - 95%), “somewhat likely” (60 - 80%), “as likely as not” (40 - 60%),“somwhat unlikely” (20 - 40%), “unlikely” (5 - 20%) and “very unlikely” (< 5%)

I love unrealistically precise numbers, particularly those in the professional services world.  The first client feedback program I implemented at Arthur Andersen generated "personal client service scores" on a 1-10 scale in 12 categories down to, I kid you not, the sixth decimal place.  I only had a handful of partners call BS on that one, the rest nodded gravely and accepted the goofiness.

[Mick]

This seems like an unrealistically precise number.  37% versus 35% versus 40%.  I think in these cases, the uncertainty associated with any model makes these models little better than a subjective evaluation; thus I might suggest bucketing these into a few subjective categories: “very likely” (95+%), “likely” (80 - 95%), “somewhat likely” (60 - 80%), “as likely as not” (40 - 60%),“somwhat unlikely” (20 - 40%), “unlikely” (5 - 20%) and “very unlikely” (< 5%)

I love unrealistically precise numbers, particularly those in the professional services world.  The first client feedback program I implemented at Arthur Andersen generated "personal client service scores" on a 1-10 scale in 12 categories down to, I kid you not, the sixth decimal place.  I only had a handful of partners call BS on that one, the rest nodded gravely and accepted the goofiness.