Is it normally distributed?

Posted in

#1 by barfboy
2017-04-21 at 03:24
< report >I'm finishing up a semester of statistics for science majors (a bit more rigorous than regular statistics but doesn't require calculus later stats courses would).

So I thought I'd have fun and check the stats on my votes. The Rsquared value of my votes frequency is .9375 and appears to be normally distributed and contains no outliers.

I can't do much more since I'm not at school but it's fun to check and make sure my voting is useful and not clustered or skewed.
#2 by bunny1ov3r
2017-04-21 at 17:11
< report >Your votes are not truly normally distributed. How can it be when the voting range is truncated at 0 and 10 and are voted in discrete values?

And what R-squared? Did you mean that you calculated what normal distribution would be like given your data's average and standard deviation and did a regression of your sorted votes on that? Cuz otherwise it does not make much sense.
#3 by barfboy
2017-04-21 at 22:32
< report >Yeah, I tried cheating because I don't have access to minitab. I can't look at the normal distribution of frequency data? I thought I could.
#4 by bunny1ov3r
2017-04-22 at 00:18
< report >I am just saying that at its best the data will be approximately normally distributed, not truly so.

Also, this "normal distribution" could be self-fulfilling in a way. If you really like the idea that your voting scores should be normally distributed, you could subconsciously be scoring your titles in a way that makes it look normal. Personally, I make somewhat of an active effort to make it look kind of "normal" since I score by comparison anyways.
#5 by loctar87
2017-04-22 at 04:46
< report >It been a while since I've taken any statistics, but from what I can recall there's no particular reason a single person's votes should have a normal distribution. It may be fun to look at and compare, but from a statistical viewpoint it's mostly nonsense. And unless you're trying to prove/disprove something about that user, like "if this user played all games they would have an average vote of 5", even if it was normally distributed it wouldn't be much use.

However, it could be reasonable to expect a VN's overall vote count to be normally distributed, so long as there's a good sample size. I think?Last modified on 2017-04-22 at 04:55
#6 by then00bavenger
2017-04-22 at 05:34
< report >In most cases if a game is lousy enough to get less than a 5 I probably wouldn't even bother completing it. I mean, I can't imagine the ending being so fucking awesome that it blows my mind and changes my 3 to a 7+.

In that sense I think it'd be pretty unusual to have evenly distributed votes unless you're a masochist or something. I also think naturally you'd gravitate more to games that you would expect to enjoy instead of games you won't.Last modified on 2017-04-22 at 05:38
#7 by formis
2017-04-22 at 13:03
< report >Gotta agree that it's not really meaningful whether a single user's scores are (aproximately) normally distributed. Everyone has likes and dislikes and you're likely to look out for and read VNs that are closer to your likes, naturally that results in a skewed rating. The only scenario for a single user where normal distribution might hold meaning is if you purposely read only randomly picked novels regardless of what you enjoy to read.Last modified on 2017-04-22 at 13:15
#8 by sakurakoi
2017-04-22 at 13:28
< report >I'm still waiting for the day when the awful score system gets replaced by a proper and simple (is the best) like and dislike/recommend and not recommended one, maybe even a proper review section instead of just having users start threads.

naturally that results in a skewed rating
yeah, how dare people to only read what they like and refuse to arbitrarily rate a work with an even more arbitrary number when they did not completely finish it, to 100%, every route and bad end, just because that's the only possible base line.

(and indeed and seriously, I am one of those who either rates 10, for completely finished something and finding no flaw, or does not rate at all. That's also why, albeit counting to one of my favorite series, Grisaia sure has its flaws so I can not measure it against others that I found near flawless, or near near flawless, or near near near... well, you get the drill)

by the by, having a normal distribution (whatever formula you used) is sure not normal... for since we, well, I am at least, are human (you all could also just be part of my imagination and if I cease to exist, so do you), ya know, humans are those animals which can and do reflect on their past experiences quite a lot, which vary greatly and have their own little personality and quirks, like cats... just are cats on average a bit more stupid.
#9 by formis
2017-04-22 at 13:59
< report >@8 I'm not entirely sure whether that was a jab at me, or the OP, I'm gonna assume at the OP given the rest of the comment. :P

On a side note (off-topic), I always try to finish a VN to 100% simply because it would annoy me to no end if I didn't, regardless of any rating, just completionist things.Last modified on 2017-04-22 at 14:04


You must be logged in to reply to this thread.