I have finished tabulating the final results for Interference's late 2011-early 2012 Rate the Song series! If you are just interested in seeing the results for the songs ranked by their mean scores, scroll down; they are in the post below this one. However, there is much more detail available in the full results packet, available here:
http://dl.dropbox.com/u/39648082/u2ratethesong2012.pdf
The winner of this contest (by mean), not particularly surprisingly, is Where the Streets Have No Name. The loser, in 214th place, is Drunk Chicken/America. A total of nineteen songs had scores of nine or above, while only six songs had scores of five or below. The average song's mean rating was a 7.3024362828 (though keep in mind that these scores have no intrinsic meaning whatsoever).
In the results packet, you will find references to the "reliability rating" of songs; in fact, there is a list of songs sorted in order of their reliability ratings. This is something that I devised to try to mathematically gauge… well, how reliably a song is liked. The goal of this metric is for a song with two votes, a 7 and a 9, to be less well favored than a song with just two 8s. It rewards consistency. The next paragraph will describe how I calculated reliability. Feel free to ignore it.
To calculate reliability, I first squared the standard deviation of each song. I did this to help correct for outliers. By squaring the standard deviation, smaller changes in standard deviation were magnified less than larger changes. As the maximum standard deviation under this was 5, squaring that left me with a number somewhere between 0 and 25. I then divided that by 25, leaving me with a number between 0 and 1. And then I added 1, leaving me with a number between 1 and 2. This number is a monotonic transformation of the standard deviation function (i.e., if you rank songs by standard deviation, and then by this number, your rankings will be identical), scaled to be between 1 and 2, and modified so that a few people dissenting from a large majority of people voting in one way (statistical outliers) is punished less than people in general voting all over the place. I call this number the adjusted standard deviation. I calculated (or rather, the computer calculated) songs' reliability ratings by dividing their mean scores by their adjusted standard deviations, so a song can lose anywhere between none and half of its mean rating dependent on its standard deviation. I basically created this formula as an experiment. Don't pay too much attention to it or its results. However, I am curious what people think of the rankings produced by "reliability". The formula was more-or-less arbitrarily chosen by me, so I'm not sure it and its results deserve too much emphasis.
Okay, enough with reliability. I would really like to thank everyone for participating in this. I seriously had a lot of fun doing it, and I hope that it helped make Interference a little more interesting during this time of utterly nonexistent U2 news. This is something that I would not at all mind doing again at some point. However, I think that it would probably be best for someone new to run any new song rating or voting contest in the near future. I tended to have a fairly strong voice on running this contest, and it's good to have a bit of diversity (I don't want everyone getting sick of me!). Later this year, it may be fun for someone to run a song survivor series. But once again… thank you to everyone for making this a fun contest.
Enjoy. Again, the most important results (songs ranked by mean rating) are below.
http://dl.dropbox.com/u/39648082/u2ratethesong2012.pdf
The winner of this contest (by mean), not particularly surprisingly, is Where the Streets Have No Name. The loser, in 214th place, is Drunk Chicken/America. A total of nineteen songs had scores of nine or above, while only six songs had scores of five or below. The average song's mean rating was a 7.3024362828 (though keep in mind that these scores have no intrinsic meaning whatsoever).
In the results packet, you will find references to the "reliability rating" of songs; in fact, there is a list of songs sorted in order of their reliability ratings. This is something that I devised to try to mathematically gauge… well, how reliably a song is liked. The goal of this metric is for a song with two votes, a 7 and a 9, to be less well favored than a song with just two 8s. It rewards consistency. The next paragraph will describe how I calculated reliability. Feel free to ignore it.
To calculate reliability, I first squared the standard deviation of each song. I did this to help correct for outliers. By squaring the standard deviation, smaller changes in standard deviation were magnified less than larger changes. As the maximum standard deviation under this was 5, squaring that left me with a number somewhere between 0 and 25. I then divided that by 25, leaving me with a number between 0 and 1. And then I added 1, leaving me with a number between 1 and 2. This number is a monotonic transformation of the standard deviation function (i.e., if you rank songs by standard deviation, and then by this number, your rankings will be identical), scaled to be between 1 and 2, and modified so that a few people dissenting from a large majority of people voting in one way (statistical outliers) is punished less than people in general voting all over the place. I call this number the adjusted standard deviation. I calculated (or rather, the computer calculated) songs' reliability ratings by dividing their mean scores by their adjusted standard deviations, so a song can lose anywhere between none and half of its mean rating dependent on its standard deviation. I basically created this formula as an experiment. Don't pay too much attention to it or its results. However, I am curious what people think of the rankings produced by "reliability". The formula was more-or-less arbitrarily chosen by me, so I'm not sure it and its results deserve too much emphasis.
Okay, enough with reliability. I would really like to thank everyone for participating in this. I seriously had a lot of fun doing it, and I hope that it helped make Interference a little more interesting during this time of utterly nonexistent U2 news. This is something that I would not at all mind doing again at some point. However, I think that it would probably be best for someone new to run any new song rating or voting contest in the near future. I tended to have a fairly strong voice on running this contest, and it's good to have a bit of diversity (I don't want everyone getting sick of me!). Later this year, it may be fun for someone to run a song survivor series. But once again… thank you to everyone for making this a fun contest.
Enjoy. Again, the most important results (songs ranked by mean rating) are below.