I have never heard of this case before this post. After approximately 30 minutes of reading the two sites you linked to as well as the Wikipedia article, my current estimates are:
Knox guilty: 0.35
Sollecito guilty: 0.35
Guede guilty: 0.80
I think your opinion will roughly coincide with mine about Guede, but could differ dramatically about the other two.
I was curious, so I checked: if the 3 questions were independent (clearly they’re not), your estimate for none of the 3 guilty should be .192
I assume from your similar .20 probabilities that you see Knox and Sollecito’s guilt as highly correlated. This would have the effect of raising your p(none) higher than .192.
But on the other hand, if Guede is guilty, then that should decrease the chance that the others are.
So, it seems you at least thought about what it means to give p(none)
I say just give all 2^3 probabilities (one of which is redundant) :)
I have never heard of this case before this post. After approximately 30 minutes of reading the two sites you linked to as well as the Wikipedia article, my current estimates are:
Knox guilty: 0.35
Sollecito guilty: 0.35
Guede guilty: 0.80
I think your opinion will roughly coincide with mine about Guede, but could differ dramatically about the other two.
Update: After reading the other comments here, I’m revising to
Knox guilty: 0.20
Sollecito guilty: 0.20
Guede guilty: 0.70
None of the three: 0.20
I was curious, so I checked: if the 3 questions were independent (clearly they’re not), your estimate for none of the 3 guilty should be .192
I assume from your similar .20 probabilities that you see Knox and Sollecito’s guilt as highly correlated. This would have the effect of raising your p(none) higher than .192.
But on the other hand, if Guede is guilty, then that should decrease the chance that the others are.
So, it seems you at least thought about what it means to give p(none)
I say just give all 2^3 probabilities (one of which is redundant) :)