Search This Blog

Monday, November 1, 2010

Causuistry, Randumbness, and Regression to the Moon

Charles Seife writes an interesting book called “Proofiness: The Dark Arts of Mathematical Deception.”

Reviewed here by the NY Times.

The review is very interesting and defines the terms I used in the title of this post.  "Proofiness" - a pun on "truthiness" which is defined as a "truth" that a person claims to know intuitively "from the gut" without regard to evidence, logic, intellectual examination, or facts.

"Proofiness" is sort of the mathematical sub component of "truthiness" where you use statistics and numbers to prove something is true even though it may not be.

My interest here, though, is in a further refinement of "proofiness" which Seife calls "causuistry".  Causuistry is what he calls "wrongly implied causation."

First a little definition of what causation really is.  Remarkably most people really don't understand this.  Causation is the relationship between a first event and a second event where the second event is a consequence of the first event.  Consequence meaning that without the first event there would not be a second event.

For example, "I threw a rock" where the rock flying through the air is caused by my throwing it.  The rock would have remained in my hand had I not thrown it.  Note that this is very different than "I threw the rock and it hit you in the head."  The reason is that the rock "hitting you in the head" could have been caused by one of two things - you standing there and me throwing the rock at you or me throwing the rock and you positioning yourself such that it would hit you.  In this second case your being struck by the rock could have one of two causes or even multiple causes if I through the rock at you and you position yourself to be hit.

So the point here is very simple.  Causation means that something is the direct cause of the result.  It does not mean that there were multiple intervening factors in addition, that there is an unknown, tenuous link between events, or that we don't really know so we just said it because it seemed right.

Small children learn to question this very early.  "Don't touch the stove or you'll get burned."  Usually little Junior figures out he can touch the stove an not get burned - if only the burner is on or the stove is off - because the admonition "you'll get burned" is overly vague.

Now let's see what Seife might mean with this concept in a practical sense.  Let's take this headline from What They Think: "Quad/Graphics gets $46 mill from WI to create 1,300 new jobs."

Now at first blush this seems pretty clear, or maybe not so much if you think about it.

So reading the first paragraph we see "expand their operations and create up to 1,300 new jobs and retain 5,500 existing jobs in Wisconsin."  Now, in my book there is a difference between giving you X dollars and giving you up to X dollars.  Or giving you up to X bricks versus giving you X bricks.  Most people as children learn this by example, usually from older siblings, cousins or friends.

"Hey, little Johnny, I'll give you up to 100 sticks of gum for your dollar."  Little Johnny, not understanding the real difference between "giving 100 sticks" and "giving up to 100 sticks", loses his dollar and gets nothing because getting zero sticks fits the definition of "up to 100".

So is giving the "$46 mill" going to really "create 1,300" jobs in this case.

I don't see a single word that implies a causation relationship between the two.  Certainly the article implies this.  Certainly it seems like this would be the case.  But, on closer inspection we see that there is no causality between the two events.

Later in the article we see the $46 million in tax credits is provided "to support the expansion of Quad/Graphics and the creation of 1,300 jobs in southeastern Wisconsin".

This is not the same as the headline.  All this now implies the money will be used to do three different things: "expand Quad," "retain 5,500 jobs" and "create 1,300 jobs".  By the time we expand Quad, save 5,500 other jobs, what's going to be left for creating the 1,300 jobs?  Perhaps not much.

But you, as the tax payer, will feel good about spending the $46 million because in your mind you'll retain "create 1,300 jobs" even if there is no causal relationship between the two  Certainly the headline would like me to believe that it will be used to "create 1,300 jobs" - but is that just a case of causuistry or causality?

So this is a very simple demonstration of causuistry.  It seems like the $46 million is an investment in 1,300 new jobs.  But there isn't anything that tells me that's what will happen.  You might try and argue that this is just journalistic license and that there isn't room to be specific - but that's nonsense.  I say either you create jobs or you don't.

So I don't buy what the article is selling.

Isn't the job of the news and of journalism to provide accurate information?  How can I decide, for example if its my tax dollars being used, if this $46 million is a good investment?  Certainly if this same sloppy model was used on my 401(k) prospectus or on my savings account, e.g., get up to 50% interested, I would be suspicious or call it a down right fraud.

Why is it that today we've lost our ability to question this sort of thing.  As kids we were always pushing the limit - touching the stove to test out the model of world we were being provided by others.

But today we've lost this.  We don't see past the headlines.  We believe what's put under our knows as if its the truth. 

We don't use simple reasoning to question the most dubious claims...

Where did we go off the tracks?

No comments:

Post a Comment