Data and Statistics

Data and Statistics

/

B: Humans are fallible, so we can’t always believe them.  Why is there so much quantifying of things, or finding recipes/formulas for them?  Can’t it be OK to like something and not know a reason why?  Our reliance on statistics shows that we have given up all hope on settling issues with appeals to moral principles.

Just as in math, many try to use functions to describe just about everything in the world: you put something in, you get some value (often a number) out.  We may think that for every situation, there is some prescribed method to deal with it.  This kind of thought may lead us to make some laws that try to blanket certain issues with a uniform solution of some kind (e.g., the Civil Rights Law).

We try to quantify things like:

  • “In such and such a situation, always do this to get the best result,” as if those situations are the domain of a certain all-explaining function
  • Music, movies, and the arts in general. For example, “three out of five stars.”  Also recall that scene from “Dead Poets Society” where the Robin Williams character talks about a silly poem-rating system
  • How to best raise children. “If your child is like this, always do that.  If he’s like that, always do this”
  • “Grading” dates by what happens at the end
  • Too many other things to mention here

Let me let you in on something: there isn’t an overarching, end-all, “all-explaining function” that takes care of every single possible situation.  If there was one, it would be this: “follow the Spirit.”

A: And just what the dickens is that?!

B (grinning): As I have now become fond of saying, we’ll talk about it later.  Trust me, we’ll get there.  But I got places to go, people to see.  And there’s a whole bunch o’ stuff I gotta talk about first.

A (sighing tiredly): Such as?

B: Well, there seems to have been a notion prevailing throughout contemporary society, for a while now, that if any statement is to be taken seriously, it must have data and analytical statistics to back it up.  One thing odd about this is that diametrically-opposing sides of the same issue seem to be able to use whatever data they can come up with to support their argument.  One would think that if only one of those sides is correct – which it must be in many cases – the data would support that one side only, leaving no doubt.  What giveth?

A: Oh, the people on the wrong side are too stubborn to admit they’re wrong, or they collected faulty data, or analyzed it improperly, or whatever.  It could be a bunch of different things, as simple as flat-out lying to the public, or more complicated, like the possibility that the data don’t actually contradict each other.

B: Indeed.  Any of those things you explicitly mentioned, or a whole bunch of others, are plausible explanations, undoubtedly.  From my own observation, it seems that you can use statistics to support just about any viewpoint you may have.

One of the most common misuses of statistics that I’ve noticed is the idea that correlation implies causation.  The offenders take a look at a couple of variables, notice that the statistical analysis yields the conclusion of a likely correlation, and decide, “ah!  The one variable influences the other, just like I thought it would,” conveniently forgetting the possibility of some other lurking variable(s) that might be the actual cause of the correlation.

To understand the meaning of some statistical studies, I believe it helps to know – among a bunch of other things – how and what data were collected, what analysis was done on the data, and what all that means with relation to the conclusion that is made.  For instance: some things may be very easy to measure, and may give us an idea about what’s going on, but they usually don’t tell the whole story, and far too often people make hasty and/or erroneous conclusions, prompting somebody to quote the classic line:

automotivator_inigo

If there is anything to remember with the interpretation of statistics, it might just be that line.

Post a Comment

Your email address will not be published. Required fields are marked *

*