How many people got ratings

Posted by (twitter: @LiamLimeGames)
January 3rd, 2017 8:01 pm

Wait! Don’t do your percentages yet!

Before you do percentages, you need to know how many people you were even competing with. Now it’s easy to just think: Oh, well there were 1489 jam entries and 901 compo entries, so it should be easy, right?

No! Bad boy!

Not everyone competed on all ratings, and more importantly, not everyone got enough votes to get a rating everywhere. You can only compete against the entries with enough ratings. You simply cannot make the assumption that all the entries with insufficient ratings “were worse” than the ones with. Statistics simply don’t work that way.

So here’s the short of it. This is how many entries got a rating in each of the categories. If you want to determine your position, use these numbers and not the global ones. Yes, your scores will be worse, but at least they will be accurate.

JAM ENTRIES

Humor: 643 entries received a score

Theme: 897 entries received a score

Fun: 885 entries received a score

Audio: 552 entries received a score

Mood: 826 entries received a score

Graphics: 848 entries received a score

Overall: 917 entries received a score

Innovation: 848 entries received a score

COMPO ENTRIES

Humor: 424 entries received a score

Theme: 624 entries received a score

Fun: 621 entries received a score

Audio: 446 entries received a score

Mood: 536 entries received a score

Graphics: 598 entries received a score

Overall: 640 entries received a score

Innovation: 596 entries received a score

How did I get these numbers?

By scraping the whole site. You can find a table with every single entry and data about it here:

Ludum Dare 37 Data (Big, big table)

If you would like to do analysis on this data, I also provide it in JSON form, available here: Ludum Dare 37 data (JSON)

Data from previous jams is also available: LD36, LD35, LD34, LD33, LD32, LD31, LD30, etc. Just edit the number in the URL for previous jams, or change it from .html to .json to get the json version.

If you just so happen to run http://ldstats.info/ you’re welcome to use this data and make the site way, way more useful! 😀

😀 LOVE YOU! <3

~Lime


12 Responses to “How many people got ratings”

  1. yumaikas says:

    Thanks for providing this data!

  2. random-storykeeper says:

    Ah well I can still say that we placed in the top 2ish percent for Audio, then. 😛

    I’m curious about the games for the Jam that actually opted out of certain categories. Because all of them would have had the opportunity to be rated given that they opted in, but they simply didn’t get the ratings…so could we still count out of potential? 😛

  3. Liam :D says:

    In statistics, the only thing you can do to fill unknown data (in this case entries with too few votes to get a rating) is to assume that the same ratings distribution (average star values) applies to the unknown ratings too, since they all belong to the same population. Obviously there will be some error in doing this, but we ~are~ trying to say something about missing data, after all. That’s the only thing we can do, we can’t simply assume that everyone did worse than everyone who did get ratings.

    I didn’t log whether or not individual entries opted in or out of each rating category though (that information is no longer available). But even if I did, there would be no statistical difference between entries which opted out of categories and entries which opted in, but did not receive enough votes. There are some ‘intuitive truths’ you could bring in, like “entries which opted out were probably bad in those categories”, but there is no way to prove or measure these truths*, therefore you’d just be intentionally introducing errors into your calculations to make your placement seem better than it actually is.

    With this in mind, the percentage you get by doing this is likely the best number you’ll get (other than absolute score. Your ‘placing’ is irrelevant and meaningless). For example… I got a 3.00 in Graphics, putting me in position 344 out of 598 compo participants who got ratings in graphics. This means I’m at the 57.53% mark. There were 901 compo participants, meaning 303 did not get a rating. The only assumption I can make is that the same ratings distribution from the 598 rated entries also apply to the 303 unrated ones, meaning I would be at the 57.53% mark even with all 901 entries rated. So my percentage wouldn’t change, but my position would. Instead of being in position 344, I’d be in position 518 if all 901 entries got a rating. Obviously I’ve introduced errors and uncertainty by making assumptions about unrated entries, but this is the only assumption you can make. I mean I could also say that I’m in position 344 / 901, putting me to 38.17%. It’s a more impressive number than 57.53%, but it’s also a completely wrong number. :)

    * There are many reasons to opt out, not just being bad. Could also not be present (audio) or not to be rated (downloaded audio/graphics) or incompatible with the game (humor/mood).

  4. rburema says:

    Great post!

    (In response to ‘No! Bad Boy!’) … but apearantly girls can still get their percentages the ‘old’ way? :-p :-p

    • LiamLime says:

      Heh, that wasn’t what I was going for but okay 😀 I was going for the sort of “Bad boy!” that you might say when you scold a puppy or something. Granted it’s more there to grab attention than anything else. 😀

      • rburema says:

        Thought so, that’s why I added the ‘ 😛 😛 ‘

        … I have to admit that I always thought of the unrated ones as ‘worse’, but your’re entirely correct that this isn’t something you can assume. It does mean that my games are consistently (and deservedly, I guess) in the bottom half & beyond though … but improving! And like a cartoon villain, I’ll get them next time. Next time!

        • LiamLime says:

          Haha, this was my 9th entry to LD and it came in at the 95% mark overall 😀 Tell me about being at the very bottom 😀

          For me, I never go into LD with a consideration for these numeric ratings. I only care for the written feedback I receive 😀

  5. Gipzo says:

    Thanks, Liam :)
    I thought that ldstats isn’t working anymore as it was only showing data for LD32 and LD33.
    But apparently username is case-sensitive 😀

  6. Jezzamon says:

    Interesting, there seems to be a pretty sharp decline in the number of compo entries… I wonder why?

  7. rburema says:

    Yeah… past few days have tought me that I care too much about those little numbers :-)

    I’ve already said a few things in the comments on other posts that are maybe overly harsh (to those people complaining about their results or the rating system in general). I think I’m going to give myself a time-out concerning this site for a few days.

Leave a Reply

You must be logged in to post a comment.

[cache: storing page]