Saturday, 4 February 2012

A Favour To Fools

I'm still having a little trouble adjusting to being a statistician.  Nominally, I'm half-statistician, half whatever-I-want-so-long-as-I-keep-publishing, but that was an arrangement worked out before it was known (or at least, known by me) that my first major project would require a degree of un-fucking so extensive it almost makes me sympathetic for whomever is running the Kormen Foundation's PR wing right now.

Still, as little as statistics grabs me as compared to probability theory, it's clearly a very important field.  After all, it's thanks to the power of statistics that I can see through high-level political analyses like this one
As Mitt Romney dominated the Florida Republican primary Tuesday night, he also captured the bulk of the votes from Latinos in the state, with 54% of their ballots... [T]he independents there who voted for [Obama] in 2008...will be the prize in the November election... Obama -- who starts with a 60% lead among all Latinos in state polls -- may end up battling Romney over the growing Latino vote.
Now, I know what you're thinking, with your pathetic, unscientific minds.  You're all like: ZOMG!  54% is close to 60%, oh noes!  I mean, this shit seems airtight, doesn't it?

Oh, how I pity you people, scrabbling around in the dirt, waiting for someone with Godlike powers of deduction - I'm not saying my skills are Godlike, that's for history to decide - to peel away the layers of reality and reveal the truth, nestled in some hyperdimensional pocket reality that you would have no hope of reaching were it not for the warp-drives of the statistical mind.

Let me lay down some truth on you.  60% of Hispanics say they'll vote for Obama.  54% of registered Hispanic Republicans voted for Mitt Romney.  The missing number that none of you will have thought of, because of how only someone as smart as me could possibly think of it?


Only 31% of registered Hispanic voters in Florida are Republican!

Are the deep complexities of this baffling situation slowly swimming into focus?  Perhaps you require further hand-holding.  No, don't apologise: I don't mind.  It's the only way you'll learn.  Obama current carries 60% of the Hispanic vote, which we'll assume, hardly unreasonably, means 60% of the registered Hispanic vote.  Mitt Romney carried 54% of the 31% of registered Hispanic voters who identify as Republicans.  That's less than 17%  of the total registered Hispanic population (Lopez could have worked this out from the numbers he included in his own colum.  Alas!  If only CNN had access to calculators!).

Clearly we can assume many if not all of those Republican Hispanics who chose not to vote for Romney when offered Gingrich, Paul or Santorum instead will choose Romney when their only other viable option is Barack Obama.  Indeed, if they all choose Romney then - let me just check my figures using the super-charged Mathematron that is my mind - he could carry 31% of the vote from Republicans alone! Against a mere 60% for Obama!  To reiterate: ZOMG!

Oh, but what about the independent voters, Lopez wonders?  They, after all, are the "real prize" [1]. And if 54% of Republicans like Romney, then how can we know how popular he is with the the constantly obsessed-over swing voter?

Well, maybe we can do some approximating using actual numbers that actual Lopez put into his actual column, like an undergrad who knows they have to include their sources even though they can't wrap their feeble minds around what their betters were saying in the first place.  38% of the registered Hispanics in the state are Democrats, leaving 31% of them unaffiliated. Let's assume every Democratically registered Hispanic will vote for Obama - that means that if the election were held today, 71% of unaffiliated voters would go for the 44th President.

That number again: 71%.

Of course, that number might be too high, but that would only be possible if registered Republicans were planning to vote for Obama as well, which ain't exactly bad news for the Democrats.

Tell me again how the Hispanic vote might be a hurdle for Obama, would you please?

You're welcome, the internet.  Man, but do I need a cigarette...

(h/t Balloon Juice)

[1] It's been said before, but it's a source of continual irritation to me that so much time and energy goes in to working out how best to court people too fucking stupid to have decided whether they want to vote for the guy who'll slash taxes on the rich, dismantle the safety net for the very poor, and announce his arrival in the Oval Office by bombing every brown person between Tunisia and Turkmenistan the instant he's sat down, or alternatively some or none of those things as soon as he's told not to by Donald Trump.  I realise there are people out there who really like all of those fucking bullshit alternatives of shittiness, but I can't understand how there are people who just aren't sure.


Tomsk said...

What does "60% lead" mean anyway? I can think of at least three different ways of interpreting that phrase. There should be a law against using anything other than percentage points.

SpaceSquid said...

I have to confess I don't know, but I can't think of any interpretation which didn't give Obama more than 60% of the vote going in.

Maybe it means an 80:20 split, and Lopez realised framing it that way would make the rest of the piece seem like the waste of time it was? We can only speculate...