Dr Badmath or How I Learned to Stop Worrying and Reject the U of T OCI Post-Mortem Survey

This post was originally submitted to Obiter Dicta but rejected because I wouldn’t publish without a pseudonym.

Not too long ago I was pointed in the direction of a certain article published by a certain student newspaper at a certain law school. Not to be coy unnecessarily, but it was Ultra Vires,[1] U of T Law[2]’s student newspaper, and the article was understatedly titled “Bay Street Hiring a Bloodbath This Year.”

I gave this article the full weight that I would give any other Infographic[3] posted on the Interwebs[4]. In other words, I chuckled and moved along. After all, one more anecdote[5] about how those who study law are bad at math[6] is just one more opportunity to generalize that all lawyers are bad at math[7] based on a small number of data points.

Misinterpreting anecdote as data in order to generalize is fun.[8] For example, if I were to say that all U of T law students are pompous jack-asses with massive inferiority complexes[9] because I’ve only ever met a couple of U of T law students and they were both pompous jack-asses with massive inferiority complexes,[10] then that would be generalizing based on anecdote.[11]

This example is a bit facetious[12] because all law students are pompous jack-asses with massive inferiority complexes. If you aren’t infuriatingly morally superior with a huge inferiority complex then how did you get in? UofT likes to pretend it’s superior to Osgoode because their parents didn’t hug them enough or whatever, and Osgoode likes to pretend it’s superior to York because clearly nobody is getting enough hugs. Meanwhile, everyone else on the whole damn planet is just happy they don’t live in Toronto.[13]

I got off track, what was I saying? Lawyers are bad at math? Generalizing from anecdotes? Oh, right, “Bay Street Hiring a Bloodbath this year.” So I set aside the article as being mostly harmless[14] and got down to the task of getting my own OCI applications in order. Simultaneously, I also engaged in “not giving said article a said other thought” because I’m really good at multi-tasking. Recently, however, it has come to my attention[15] that there is a push amongst those who should know better[16] to expand this “study”[17] replete with “data analysis”[18] to all the law schools in “Ontario.”[19]

So what’s wrong with getting a bit of data so we can be more informed? Wouldn’t it just help us all hold a light up to the OCI black box and totes empower the law student or some whatever and things? Well, I can hold a fork;[20] let me take a stab at answering[21] that, voice in my head.[22]

First of all, let’s all agree that the title of the article is hysterically hyperbolic.[23] If 2013 was a bloodbath (7.3% drop in total hires over previous year) then so was 2012 (6% less than 2011) and 2011 (9.2% less than 2010). Look, there’s definitely a trend here, you guys, I’m not gonna lie: in the numbers available in the Bloodbath data, there’s a 3 year trend in declining 2L hiring for of a total of 20.9% over the 2010-2013 range that the data covered so that a a mean annual trend is around -7.5%/year. That’s not nothing[24] and it is statistically significant.[25]

Next, let’s all look at the subtitle of the article: “UTLaw, Queen’s, & Western hold steady,[26] but Osgoode and Ottawa take major hits”.

Ermagerd, panic!

Over the previous year, the deltas[27] are:

School UofT Queens Western Oz Ottawa
Normalized +5.6% +0% +2.3% -15.6% -23%
Raw Count +5 +0 +1 -14 -9

But there is far too much noise in single-year data for individual schools to draw anything resembling a conclusion, especially with low raw counts. But yes, Ottawa and Oz both saw a statistically significant drop in 2L summer hires in 2013, and that drop is in line with the overall decline in 2L hires from 2010, though everyone is down. Except Queens. Clearly Queen’s[28] is the superior school cranking out superior law candidates. Super-inferiority friends assemble! A table for some reason:

School UofT Queens Western Oz Ottawa
Normalized -9.6% +0% -20% -20% -31.8%
Raw Count -10 +0 -11 -24 -14

BUT! The median school has dropped 4 ± 1 hires to Bay Street each year since 2010 (mean is also 4 ± 1), and Oz here is in the upper end of the range, with a decline of 16 and 14 in each of 2011 and 2013 respectively. If the 2014 hiring is indicative of this trend we will expect to see the median school down 4 placements with an error of ±1.[29]

“Oh nos!” you scream, falling to your knees as you realize the horror of the situation, having spent the equivalent of a small mortgage to attend school, and having nothing but a liberal arts major to fall back on, “My student loans!” Your spectacle makes me forget to continue my line-by-line critique as I stare at you in shock. Like a train-wreck I am unable to look away.

In sheer panic, you turn to me – where I am standing, slack jawed with embarrassment for you as your unbecoming display unfolds in the middle of Gowlings hall: “Please,” you scream pleadingly, mere inches from my face while you grab my lapels in desperation. Somehow I have become a life-preserver as you grasp for any small sliver of hope to keep yourself afloat in a sea of terror, “What are my odds?”

“Well, imaginary friend,” I say in my most calming of voices, trying to gather myself after being confronted with such a raw display of emotion, “Helpfully, UofT has provided you with those numbers in a wonderful act of altruism. Oh, no they haven’t, they gave you statistics.”

Stop treating statistics as odds you crazy people! You are just making yourselves crazy! Statistics are for populations! When you say “odds” you are talking about[30] mapping a lossy context-less population amalgam of data onto an individual in a specific context and expecting me to give you a probability of success as if you were at a casino. I have no damn idea; [31] the math only works in one direction.[32]

This is the problem with this exercise, and this is where it starts to shuffle from the useless infotainment[33] into the realm of dangerous: because when you start to apply this to your own personal situation, things are going to go sideways for your mental health and the “facts”[34] may not reflect reality.[35]

I’m not gonna sugar coat it for you,[36] it’s brutal out there. Not just on Bay Street, not just in Law. It’s brutal across the board. Since the Global Financial Cluster Fsck[37] in 2008,[38] things have gone sideways everywhere. The law had a nice run at bankruptcies for a while, but those are starting to clue up.[39] In the 1990s and 2000s, everyone[40] was sure that the economy was totes gonna tank because of globalization or free trade or whatevs. But what’s taking your job isn’t some “foreign worker”[41] it’s some robot.[42] In a recent Economist article, they[43] suggested that some 47(-ish)% of current jobs are expected to be automated within the next 20 years.[44] Go ahead and Google “Stanford AND Codex”[45] or “Legal Informatics” if you want to cry. Computers are making research easier than ever, and you just don’t need as many articling students and 2Ls to do that. Just be glad you aren’t a doctor from India driving a taxi, that job is gone girl to Google pretty soon.[46]

Look, it’s not some mysterious dark art solemnization performed by the illuminati or Dick Cheney or something. Bay Street doesn’t need us. They want us, but they want the right us. The OCI process is about Bay Street firms finding the right resources (that’s us) to bring value to their companies (the aforesaid firms) for business reasons (the practice of law is a business model). They are making an investment when they hire 2Ls and Articling Students, and the OCI process is a way of doing due diligence on a massive scale. It ain’t pretty, but we’re at fault as much as they are. We visit the firms and speak with the lawyers. We see first hand that the folks we speak to are (for the most part) high strung, miserable, putting a good face on it, and putting on “lawyer robot face” and we think “Yes. Yes, this is what I want my future to look like.” We are lied to, abused and mistreated, we have an explicit demonstration that ethics is a handicap on Bay Street and we say “Yes please, I would like that for me.” We still think we want to work there.

The El Suck[47] rules are supposed to protect students from predatory hiring practices.[48] Think about that: potential future employees need to be protected from the employers. And these employers are professional liars[49] trained to find the loop-holes and work around the letter of the law. Did you really expect that they would follow the spirit of all those rules? The ones who are more obvious about being duplicitous are the ones that are bizarrely trustworthy, since all the firms are playing games with you and at least these firms respect you enough to stab you in the front. It would be lovely if everything was balloons and flowers and rainbows, but if that were the case most of us would be in art school, running a craft brewery, or swashbuckling a zombie or whatever. Me? I’d be an astronaut. Or maybe a space cowboy.[50]

So how does it work? Firms are sorting through a pile of applications, a pile that gets sorted in 3 steps: applications, OCIs, and in-firm interviews. At each stage the firm makes decisions about who to keep and who to reject. Maybe 10% make it to an OCI, maybe 100 get first interviews and that gets halved from first to second. Are you not an idiot by the criteria with which the firm measuring idiocy? Do you look like you’d be a good fit with the culture, however that is defined? Did you show the firm that you are interested in them enough without acting like an ass? Do you look like someone who might stick around to provide a return on the investment they are about to make in you? Is your plan to pay off your student loans and bugger off to Kokomo[51] and make cocktails with Tom Cruise? And most importantly, is the business group they pigeon-holed you in based on limited knowledge both a) hiring and b) confident that you will stick with them?

So what are your odds? Damned if I know. I don’t know the hiring committees at all the firms you applied to,[52] and I don’t know you well enough. I don’t know if this is your dream job, if you half-assed your application, or if your personality is a good fit for the firm culture. I know people who applied to 22 firms, got 22 OCIs, and 2 in-firm interviews. I know people who applied to 22 firms, got 8 OCIs and 6 in-firms. What does that tell me? One of those folks wasted less of his and/or her time doing OCI interviews, and nothing else.[53] So your odds are therefore shoulder shrug percent ± infinity-eleven. Behold the power of statistics![54]

None of this nuance is reflected in the UofT survey, and I don’t think it can be without making some pretty massive changes to the methodology, like starting with a particular question you wanted to answer and crafting the survey to reflect that. And either performing true random sampling or making the survey mandatory. The statistics don’t and won’t answer “why” the numbers are the way they are. The self-selection nature does not answer the number of people who applied for Bay Street jobs, nor the number of people who half-assed their applications. They don’t reflect the folks who targeted applications to a firm, practice area or work-culture. Most of all, there is no way of determining long-range career satisfaction from these numbers.[55] Maybe Oz students are more interested in public services, or space cowboys than UofT in working on Bay Street? I don’t know. I don’t have that information.[56]

There is no upside to this survey. While the “85%!” sticker that is prominently emblazoned on the UV cover for their article seems awesomesauce (it’s an A! Or a LMNOP in UofT’s grating [burn!] system![57]), it says up-front that there can be an up to 15% error in the results.[58] Voluntary response surveys are statistically flawed. This is why that survey you take on entry to law school, much like the old useful version of the census, is mandatory.[59] A voluntary survey like this provides a) no useful information to the survey taker, b) is likely to be statistically flawed, and c) can use used by someone with an inferiority complex to try to “prove” they are the somehow superior, because latin. Something-something-Mark Twain: lies, damn lies, and statistics. Also, it’s really easy to lie on these surveys if you wanted to, say, troll people who are bad at math. Just FYI.

This is why UV is bad for your skin: it makes you feel better briefly, but increases your likelihood of dying from a painful disease. That metaphor worked better in my head. Let me try again: The Bloodbath article in UV gives you the illusion of knowing something (your odds), much like a sun tan gives you the illusion of healthy skin. But in reality, the article is giving you a false sense of knowledge, much like a sun tan is a reaction of your skin to being exposed to ionizing radiation that causes cancer. Ok, that was a weak metaphor. I was told to end strong, but too late.

So you still want my advice? That’s stupid, but fine: First, personal information is valuable; don’t give it away unless you get something of value in return. Second, if you’re looking for happiness then you shouldn’t apply to Bay Street. But since you’re in Law School, being happy is clearly not a life goal: have at it. Third, before you go to Bay Street I suggest you do a little research to find the dialog that comes immediately before “Is not this a lamentable thing, that the skin of an innocent lamb should be made parchment?” By the time you’re done you’ll get it.

[1] That’s latin. Everything that’s latin is pretentious, but not everything pretentious is in latin. Obiter, for eg, is also pretentious. Et cetera. [UV]

[2] U of T’s motto is also in latin. So is Osgoode’s. Coincidence?

[3] Infographic: (noun) – statistics as interpreted by liberal arts majors and rhetorically glossed to be far more compelling than what the weak correlations and questionably sourced data actually show.

[4] Like the Internet, but with more spiders.

[5] The opposite of statistics.

[6] Well, maybe with enough anecdotes you can get statistics.

[7] Citation needed.

[8] Everyone knows this.

[9]  Or is it “inferiorities complex”?

[10] Ibid.

[11] Also, prejudice?

[12] Fancy!

[13] What’s the difference between New York and Toronto? Toronto thinks it’s the center of the universe, New York knows that it is. – Why do the trees in BC lean east? Toronto sucks. – Who was Toronto’s mayor until October 27th? Rob Ford! – Two-drink minimum, folks. I’m here all week.

[14] Like the Earth in general.

[15] This is the best clichéd phrase for pivoting in an e-mail to preparing to scold people for being childish. I suggest you all take note for future reference since you will be using it a lot.

[16] I already made the Ibid joke, so Ditto, I guess?

[17] Nope. Well, yes. But nope. I’d give it an LP (everyone knows that’s a C, you guys).

[18] A truck load of nope. Nope on a truck. Dumped on your front step. Nope nope nope.

[19] Okay, you got me. These quotes are unnecessary.

[20] See what I did there?

[21] All data below is sourced to UofT’s “Bloodbath” charts. If two people are in an elevator, and one of them charts, then logically everyone knows who charted. [Bloodbath]

[22] If I could remember who I  creatively acquired this joke from, this is where they would get credited.

[23] Not alliterative enough, though. I would have given it mad points for alliteration if it elicited excellent alliteration effectively.

[24] By definition, that makes it something. Logic.

[25] Significant in the statistical sense doesn’t mean “important” it means “this is something that exists in the data.”

[26] What now?

[27] Comes after Charlie and before Echo, but also a Greek letter used in this context to mean Δx = x2 – x1.

[28] Regina’s?

[29] I’m not taxing the math in the statistical analysis here too seriously because a) I don’t want to bother to look it up, b) it’s barely relevant, and c) lazy.

[30] I assume: in general, people are absolutely terrible at understanding probability.

[31] Ask Nate Silver.

[32] Or this.

[33] Remember that the “tainment” in “infotainment” comes from “entertainment” and is 67% of the word. So by the same analysis used in studies like this, infotainment is 67% entertaining and only 33% informative. So you only have a 1 in 3 chance of being informed by infotainment. QED.

[34] A factoid is like a fact but not a fact. True fact.

[35] That’s the kicker.

[36] Can you actually get coats made of sugar? That’d be awesome, but probably wouldn’t do too well in the rain.

[37] To ensure that your Global Financial Clusters are operating correctly, run fsck regularly as a part of standard global maintenance.

[38] We aren’t allowed to call it a depression, because science, or regulations or something. It wasn’t made clear at the time.

[39] Wild speculation on my part. I figured if UV could have at it, so could I.

[40] I’m just gonna go ahead and blame Naomi Klein, because research is hard and “No Logo” was too damn long. Also, uhm, Ayn Rand because those books were awful. Just really badly written.

[41] Cue NDP campaign ad.

[42] In the corner a Roomba is laughing maniacally.

[43] Is the Economist plural? Would I just refer to the author. Meh, can’t be bothered to look it up. It’s probably in the McGill Guide.

[44] An actual source, just this once.

[45] That’s a Boolean operator, neat!

[46] Seriously, what’s wrong with this country where doctors are driving taxis? I can’t even find a damn GP.

[47] That’s how you pronounce “LSUC”, fyi. Just like how “Lawyer” is pronounced “liar”.

[48] Well, and also less powerful players from the race-to-the-bottom economics of just buying talent. Though that argument is sort of specious since it assumes all students are driven solely by money. There are a few ethical ones who slipped through the application process, I’m sure.

[49] Wait, that was supposed to be spelled “lawyer”.

[52] Who am I kidding? I don’t know the hiring committee at any of the firms you applied to.

[53] Well, I also know that if happiness was a goal, then we’ve all wasted our time.

[54] Math!

[55] And if my experience was any indication, very few folks at these firms are happy.

[56] The Bloodbath data conspicuously lacked the Space Cowboy question. Conspiracy.

[57] Guys, if you do a one-to-one relationship to replace A, B, C with HH, HP, Ketchup, or whatever other delicious condiment, you aren’t fooling anyone. We get it: You like BBQ. I like BBQ, too. We should have a BBQ!

[58] I’d math this up to confirm it, or at least get a real number, but as a law student that’s frowned upon.

[59] Ain’t no census like a useful census ‘cuz a useful census is mandatory!