I'm just gonna make an open invitation here to anyone with over 100 posts who wants to do like Eric did and post how you would distribute 200 points over your top 20 albums. I've got a default number system, but I'd like to make it more accurate by basing it on some more people's real point distribution.
Posts: 4001 | Location: NE Indiana | Registered: 14 April 2005
Originally posted by RavingLunatic: I'm just gonna make an open invitation here to anyone with over 100 posts who wants to do like Eric did and post how you would distribute 200 points over your top 20 albums. I've got a default number system, but I'd like to make it more accurate by basing it on some more people's real point distribution.
I don't qualify, but I think this is a very good idea.
Posts: 121 | Location: Boston | Registered: 13 December 2005
If it's out of 200 points, giving something 1 point is less then 1%, which seems very insignificant to me. I'll give the last ones two points instead.
This is part of my problem(besides it being a pain to assign points). I didn't want to give any entries the same amount of points because there are none where I can say I like this album exactly as much as I like this album. So if we decide to go this way, I think we need more than 200 points. OR, if we keep 200, we should limit the list to the top 10. Because if not, you're almost forced to double up on point values unless you use decimals, which would be stupid, or you have a huge range between say your top three or five that have a high score, and the rest of the list that would have relatively the same values.
Posts: 3130 | Location: FoCo | Registered: 07 January 2005
We're not actually doing this method, thought it is the most accurate way to do a composite list. I'm just trying to get an accurate idea of how many points should go to each rank. How many points should each person's #5 ranked album get? I'm using these test runs to determine that.
Posts: 4001 | Location: NE Indiana | Registered: 14 April 2005
Originally posted by RavingLunatic: We're not actually doing this method, thought it is the most accurate way to do a composite list. I'm just trying to get an accurate idea of how many points should go to each rank. How many points should each person's #5 ranked album get? I'm using these test runs to determine that.
Oh. Please disregard my bitching then.
Posts: 3130 | Location: FoCo | Registered: 07 January 2005
Originally posted by The Furnace Is Fiery: why does the total point value have to equal 200 anyway?? With a simple 20-1 structure the point total is 210.
I think the idea is most peoples #1 is probably more than one point better than their #2, or even if not, at least more than five points better than their #5. The idea is to distinguish just how much better one album is to another.
Posts: 3130 | Location: FoCo | Registered: 07 January 2005
Originally posted by The Furnace Is Fiery: why does the total point value have to equal 200 anyway?? With a simple 20-1 structure the point total is 210.
It's just an arbitrary number really. It's nice to have everyone be around 200 for this test run that way no one's numbers count more than anyone else's. I suppose it doesn't really matter if it's not exactly 200, just so long as it's in the 190-210 range if that makes things easier.
Posts: 4001 | Location: NE Indiana | Registered: 14 April 2005
When this test is done, RL should be able to average the scores at each rank and come up with a fair scoring system. It doesn't need to add up to 200. So far, everyone's distributions seem to be pretty much the same: spread out a bit in the top ten, and little or no seperation in the bottom 10.
----- Stay gold, Ponyboy. Stay gold.
Posts: 5349 | Location: Michigan | Registered: 19 June 2005