| |
Oxidy
Account closed
Registered: Mar 2003
Posts: 80 |
Voting average bug
When checking Vote statistics for Beertime 3 there's 5 votes:
8 2 votes
7 1 vote
6 1 vote
4 1 vote
CSDB says the average is 5.9, but it should be 6.6.
|
|
| |
Perff
Administrator
Posts: 1694 |
Its not a bug! :)
Thats because we don't use the mathematical average, but an "weighted average" which we calculate in a bit different way. (based on the spread of the votes and the number of votes and stuff like that)
There have been many discussions about this in this forum. Check in the old posts. There should be reading for some hours. ;)
/CSDb Staff |
| |
Oxidy
Account closed
Registered: Mar 2003
Posts: 80 |
An average is an average... |
| |
Perff
Administrator
Posts: 1694 |
Yes but it dosn't say "average" anywhere.
It says "weighted average"
As mentioned a number of times before. If just enough ppl voted there would be no problem.
Also then the weigted average would get real close to the average. |
| |
Oxidy
Account closed
Registered: Mar 2003
Posts: 80 |
I've read back some earlier posts covering the issue and I understand the meaning. But I couldn't find what the actual "weighting" was.
If we take the Beertime 3 example above, is there any other reason behind the 0.7 point loss than the fact that there was few votes??
Since the main problem seem to be few votes, there must be a way to get more users to vote. Sceners like to see their names mentioned. Perhaps a top voter chart of some kind. It sounds quite stupid, but perhaps it could work.
|
| |
MorGorr
Account closed
Registered: May 2002
Posts: 47 |
Quote: Yes but it dosn't say "average" anywhere.
It says "weighted average"
As mentioned a number of times before. If just enough ppl voted there would be no problem.
Also then the weigted average would get real close to the average.
Why don't you tell us the formula so that we can discuss it?
If, for example, a standard deviation is included in a way that it always leads to lower "weighted averages", then the formula would actually foster the success of downvoters.
If each vote gets a lower weighting the farther it is from the "regular" average, then it's OK.
But what the . do you do with sample size err number of votes?! |
| |
MorGorr
Account closed
Registered: May 2002
Posts: 47 |
Sorry, what I was trying to say is that:
If high SD = low mean, then formula = BS. |
