Note: Within an hour of this post, the elections committee broke the model! Good job, EC! (more t-tests, next time, I promise)

I did a really simple logarithmic fit to the data the elections committee has been releasing.

I’m going to go ahead and predict turnout this year will be 3781. Only 58% of last years ~6500.

ballots v. time

Note: I’d do some confidence intervals, but seriously, I just finished a STAT 306 assignment, and I’m tired of t-tests.

What’s different this year from last? This editor notes one big difference is the lack of a student-wide email, highly visible poll-booths, and highly visible campaigns. Post your thoughts in the comments thread.


10 Comments so far

  1. Alex Lougheed on January 28, 2010 2:28 pm

    What have you? Why is turnout down this year?

  2. Alex Lougheed on January 28, 2010 2:48 pm

    Apparently, within an hour of me posting this, the EC met my predicted target. I guess I kind of ignored that votes come in clusters around noon, and the last datum I had was at 9:30am. Hilarity. Total model fail.

    GJ on the EC on releasing the data though!

  3. Mac Attack on January 28, 2010 3:31 pm

    This is adorable, so freakin’ adorable. p value FAIL.

  4. Rory Green on January 28, 2010 5:04 pm

    So, what is quorum? Has it/will it be reached?

  5. Mitch on January 28, 2010 8:05 pm

    There was a massive spike in voting in the last hours of last year’s voting period.

  6. Mitch on January 29, 2010 7:54 am


  7. Ricardo on January 29, 2010 1:26 pm

    Also, saved votes will be submitted at 4pm. Expect another spike.


  8. Alex Lougheed on January 29, 2010 2:28 pm

    True fact. I need to read more 306.

  9. Nicholas FitzGerald on January 29, 2010 3:59 pm

    Proof that any model is only as good as the assumptions on which it is built.

    Or as us CS-types are fond of saying: “Garbage in, Garbage out”…

  10. Alison on January 29, 2010 4:06 pm

    haha good try though

Name (required)

Email (required)


Speak your mind

Spam prevention powered by Akismet