I've set up a Google spreadsheet that gets most of the score data to make this graph:

http://imgur.com/a/JvtZ7

My thoughts on the results:

• The last 50 ranks got a single low rating which most likely means that these are incomplete submission
• Only the first 100 ranks were completely safe of adjusting scoring due to low rating count; more voters would be nice or obligue a number or votes if you submit. That, or people need to up their marketing game
• High correlation between "Gameboy Feel" and "Graphics"; it's a likely scenario that people rate "Gameboy Feel" by mostly judging the graphics
• Most "Overall" rating was lost on sound

Interesting! The Raw and Overall scores were a bit confusing to me in how they worked.

I think the only games I skipped over were the ones I couldn't play - games for Mac/Linux exclusively or they were missing a data folder they needed, etc. Other than that, I did try to find more games that weren't getting as many ratings or comments, but yeah, since there are so many games, the ones you hear about the most are more likely to be looked at. GameBoy feel was the one I had trouble rating - I ended up rating a lot of games high for GameBoy feel, provided that they satisfied the Jam requirements and their controls could actually work on a GameBoy in theory. :P

The raw scores = sum of all votes (divided) by the number of votes.

If the votes are below the median, then the raw score gets decreased depending on how far it is away form the median votes. Resulting in your socre.

This is to prevent a game that gets only 1 voter with full 5-stars to win the competition.

To make sure that low vote count is less of a problem I would suggest that GBJAM also allows voters from outside of the competition to vote, just make it so that those outside-votes only count as 10% of a vote and that the total of outside-votes can only count as 25% of your final voter count (this is to prevent score boosting using outsiders).

This would allow for more voters to make the voting more equal (as in, less decrease due to low vote count) AND it would fetch more publicity for GBJAM.

Everybody votes in a different way, so overall it will work out, but still, the correlation between Gameboy Feel and graphics is still very much there. Perhaps the category can be explained better to help voters.