Algorithm for highest rated items

[Lucrezia]

Most applications, to show the top rated items (regardless of the nature of the items) sort the items by rating average and then dismiss the ones where the voters are not enough (or the other way around, it doesn't really matter).

I was wondering, is there a better algorithm for that?
For example, I would consider an item with 500 voters and 4,6 rating better than one with 50 voters and a rating average of 4,8 but with the common algorithm the second would be considered better (if the voter cutoff limit is lower than 50 of course).

[Ancient Dragon]

For example, I would consider an item with 500 voters and 4,6 rating better than one with 50 voters and a rating average of 4,8 but with the common algorithm the second would be considered better (if the voter cutoff limit is lower than 50 of course).

Statistically one may be a lot better than the other. You could calculate the standard deviation and error percentage.

[Lucrezia]

Statistically one may be a lot better than the other. You could calculate the standard deviation and error percentage.

Thanks for your reply.
However, could you elaborate?
I can't see how the stardard deviation and error percentage would help in calculating a top rated items list... :$

[thoughtcoder]

Your problem is a case of http://en.wikipedia.org/wiki/Bayes'_theorem#Bayes.27_theorem_for_probability_densities

That might be an oversimplification.