Thursday, February 28, 2013

What the candidates really think

Now that the SU elections have officially kicked off, candidates are running around campus trying to secure votes. One of the earliest features available to potential voters is the Candidate Bios page on the SU website.

Instead of telling you what I think of the bios just yet, I'd like you to take a look at these pretty word clouds I made of the bios.

The first one is just for the presidential candidates:

(click to zoom in)
And the second one is for all candidates:

(click to zoom in)
So what do the candidates care about? STUDENTS. Wow do they ever. They also seem to like "campus", "university", and "year" a lot.

What do candidates seem to forget about? Voting. On the first cloud you can barely see it between the S and T of 'Students', and on the second one it's hiding on the far right. Seeing as the language used when soliciting votes can apparently significantly effect the outcome, this is maybe a bit of a missed opportunity for our candidates here.

Come back for more election stuff in the near future!

Tuesday, February 26, 2013

The Economics of Roll Up the Rim to Win

Roll up the Rim is a great Canadian tradition. Once a year Canadians flock to Tim Horton's to buy their coffees in the hopes of winning something wonderful.

According to this year's statistics, just by buying a cup of coffee at Tim Horton's you could win one of:
  • 40 2013 Toyota RAV4 Vehicles
  • 100 $5,000 MasterCard Prepaid Cards
  • 1,000 Napoleon Gourmet Grills
  • 25,000 $100 Tim Cards
  • 10,873,327 Donuts, or
  • 32,619,980 Coffees
This is out of a total of 260,959,840 printed RUTR cups, making the odds of you winning something just slightly higher than 1/6.

Looking at the numbers, you have about a 1/8 chance of winning a coffee on any given cup you buy (ignoring regional variation). With these odds, you would need to buy 6 cups of coffee before expecting a greater than 50% chance of winning one. Assuming, then, that you got the largest size coffee you could as your freebie, and the smallest sizes possible for each attempt, this whole adventure would cost you a net $2.08. Six cups of coffee would gain you an expected $4.52 in winnings (including the chance of donuts and Barbecues, etc.), but would cost $6.60. Fascinatingly, as the largest coffee costs $3.89*, you may actually end up ahead after trying this.

What if your goal was to win a donut? You would expect to buy 17 cups of coffee before having a 50% chance of getting one of those. During this adventure you'd expect to win at least 2 coffees, and your net cost to try this would be $8.04.

Let's ramp it up. How many coffees would you need to buy to get a $100 Tim's Card? 7235. Oof. Not impossible, but a team effort would be required - that's 107 cups of coffee per day of the contest. You'd expect to win about 900 coffees and 300 donuts, but including winnings this would cost a net $3,646.27.

I like Barbecues. I don't think I'd like them enough to buy 180,883 cups of coffee just for the chance to win one, though. Sure, I'd likely get 17 of the Tim's Cards, 22,600 free coffees, and 7,500 donuts, but at a cost of $91,597.23 I think I'd rather just buy one...

It would take 1,808,833 cups of coffee before you'd have more than a 50% chance of winning one of those elusive $5000 MasterCards. In that time you'd likely win 7 barbecues that you could sell for $5,000. I really don't know why you'd bother past that... Anyway, this would lose you $917,747.41.

And what if you really like cars? You'd have to buy at least 5,522,083 coffees to have a reasonable chance at one of those Toyotas. This process would get you 2 MasterCards, 21 Barbecues, 530 Tim's Cards, 690,260 free coffees, and 230,100 donuts, but would cost you $2,801,508.19. For that price, you could just buy 85 of the cars yourself.

More importantly, this would involve you purchasing about 1.3 million liters of coffee over two months. If you think that sounds like a lot, just remember that's only about as much as the amount of oil spilled in the Deepwater Horizon oil spill every three hours.

So is Roll up the Rim worth it? Sure. If you already like Timmy's coffee, then go for it. If you're in it for the prizes, though, you may want to re-think things...

*Values for coffee and donuts based on ranges given in Tim Horton's website, and may not reflect prices at your local store. If you try anything mentioned above and it's more/less expensive than I say, don't blame me*

Wednesday, February 20, 2013

Female Candidates (Part 1)

There has recently been a lot of discussion recently regarding females being under-represented in SU elections on campus.

Quite a lot. All over the place. Honestly. Well, ok, mostly at the Wanderer and all over Facebook.

Frankly, yes, this is an issue. While there are no established barriers to female candidates, we cannot deny that there simply aren't any female candidates this year, and I would further venture that we can't deny that this is a bad thing - this really ought to be addressed.

Unfortunately, everyone seems to have wildly differing opinions as to why this phenomenon is occurring. While y'all can go debate fancy things like sociology, I'm thinking of (over the next little while) taking a look at it from a statistical point of view. Who knows - perhaps this could be similar to the Berkeley Sex Bias case I wrote about earlier, and the problems are more fundamental and basic. Who knows indeed...

In the meantime, here's a fancy thing I learned how to make after looking around on the internet. Enjoy! The size of each block is the size of the faculty, and the colour represents the faculty fraction that is female (with green meaning overwhelmingly female, and red being overwhelmingly male).

EDIT: Source Google Visualization API Sample

Tuesday, February 19, 2013

Voting Behaviour

Hey bloglodytes,

I realize it's been a month since my last post, and I apologize profusely. Not too much, though, since I can check the stats and you have not, by-and-large, been impatiently checking for regular updates. Shame.

On a related note, привіт to the randomly high number of readers from Ukraine!

This is just a short post discussing some of the preliminary results of an analysis I've been interested in doing for a while now. When it comes to SU voting on campus, people often talk about the "Lister vote" or the "Greek vote" as though they're massive and important voting blocks that need to be either wrestled with or courted. (Let me be clear here - when I say "people" talk about it, I really mean about six people, including me, who have nothing better to do and go out for beers altogether too infrequently)

If it turns out that there is any sort of consistency within these voting groups, that would certainly be cool to know as a potential candidate. Unfortunately, the only way to find this out would to look at anonymized versions of full ballots, and compare voting trends within a ballot. For instance, if a significant number of voters clump Lister or Greek candidates together on a ballot, it would be very easy to identify a trend.

Unfortunately the CRO is reluctant to share this sort of information (cough cough), so the closest we can come to this sort of analysis is by attempting to recreate ballots based on the results read-out that's provided following the elections. This is annoying.

With all that being said, the initial (VERY PRELIMINARY) results from last year's election suggest two cool findings about trends within voting.

The first one isn't really that surprising: voters who vote None of the Above as their first choice have virtually random subsequent vote selection. Any given ballot that ranked NotA first and a candidate second was evenly split between all remaining candidates (within 4%). When comparing these subsequent votes to the distribution of first round votes, they aren't even close.

The second one truly only comes from one datapoint, so I'm hesitant to say anything firm on it quite yet. However, an analysis of the Presidential race shows that voters who ranked Farid (Greek) higher than both Colten (Greek) and Adi (not so Greek) were more likely to vote for Adi than voters choosing their first round picks.

Don't get me wrong, in both cases voters choosing only between Colten and Adi preferred Colten by a ratio of about 3:2, however instead of seeing lots of Farid voters subsequently go on to prefer a fellow Greek (The "Greek Hypothesis"), Colten actually got 7% less of the vote share than when looking at first-round votes.

So that's cool.