“On average, everyone on earth has one testicle.”
We
are inundated with statistics. Statistics tell us – or purport to
tell us – who we are (on average) and what is going on in the
world. Statistics purport to tell us how likely something is to
happen and to whom. Stats present results as causes and paint the
world as a dour and malicious casino with us as roulette balls.
Screw stats.
What's the Problem With Statistics?
Statistics
are not inherently bad. In their proper place they are useful and
instructive. Like everything else in life they are often abused,
intentionally or not, and to bad effect.
When
statistics are presented in a popular setting they may reliably be
suspected as the misuse of numbers (which we for some reason view as
authoritative) for deception, or at best attention.
Statistics are Lossy
Statistics
by their nature come at a loss of information. This is not “bad”
(who wants to pour over raw data anyway) but it does mean that the
numbers will almost inevitably be ported to other contexts, in
defiance of & without the warnings & parameters originally
included by responsible statisticians.
One
of the most common ways that stats “lose” information is by
ignoring boundaries in the source information. What are in reality
multiple distinct entities are perversely slurred together into a
single fiction that accurately describes the desires &
perspectives of the reporting agency.
Example
1: look at the first sentence of this article. The statement that
everyone on earth has on average one testicle follows easily from the
truth that there are as many testicles on earth as people. An
extraterrestrial unfamiliar with earth or its lying ways might fall
into that trap and never know it.
Statistics
are a conceit used to reduces large sets of data into small portable
takeaways. They serve to obscure just as easily as to enlighten.
Statistics Obscure Agency by Masquerading as Probabilities
Statistics
are often used as probabilities, as if events are the product of
random number generators and not the result of natural forces or
human action.
Example
2: divorce statistics. It is stated that marriages have an 'x'
percent chance of ending in divorce, ignoring two facts:
- A relatively small number of serial monogamists distort the statistics of the larger population.
- Marriage is not like a radioactive atom; it does not have a 'probability' of decaying. A marriage lives or dies based on the choices of the man and woman engaged in it.
Statistics
are not probabilities. They observe what happens but can at best
speculate as to why. Probabilities are restricted to the quantum
realm or casino-style ignorance.
Comments
Post a Comment