Numbers May or May not Lie

This here is an attempt to reconcile the following statements:

1) Numbers never lie.
2) There are three types of lies: Lies, Damn Lies and Statistics.

Gee, that's a toughie isn't it? I mean, time and again, statistics always prove themselves plus or minus some value. Car accidents happen pretty much with regularity. The house always wins. But then again, we also hear of people 'overcoming the odds' failing 'despite the odds' and I don't know of anyone with 1.7 children either.

Actually the answer is pretty simple: it isn't arrogance or such to assume that odds don't apply to you. They don't. In fact, under most circumstances, you'll have to be extremely foolish to believe that the odds do. The odds don't apply to you because you're only a sample of one. Furthermore, you could neither be called a random sample nor an independent variable. For goodness' sake, you're an individual... not the freaking population. You see, assumptions underlying most statistics just aren't met when it comes to you. Numbers don't lie in context within the given assumptions; statistics lie because it is taken out of context when applied to the individual.

This is also why numbers should never be taken seriously... if anything, they are more a reflection of the system itself, but even so, not all systems fit the assumptions of all statistics. Plus, it is very easy to game numbers to work for you. Methods of counting and formulae can change to suit your needs. There are also cheap PR methods like truncating curves or messing with the scale on the x and y axes, but those are just plain amateurish.

Therefore, never trust any standard whose sole metric is statistics. Especially if they never disclose their methods. Statistics can often give the wrong impression. Often, it can depersonalise tragedy or mask real problems inherent in the system. I know it really DOES gets on my nerves when so-called objective observers run around brandishing 'statistics' and little else. Obviously, those people need to go back and take a class in statistics. They don't know what they're talking about (in more ways than one). Sure, I know lots of people have an infatuation with all things scientific and objective (or in the Singapore context, pragmatic), but seriously... they need to learn to be more scientific and objective when making claims, by doing what science ideally does: by acknowledging the limits of their metric and stating their underlying assumptions.

I previously wrote what could pass for a poem called: A Statistic... it seems to not have any meaning... Why? Because I never told you what it was about. Here it is anyway, in full. Explanation follows...

A Statistic

e minus lambda times lambda kay
Over kay factorial;
There's got to be a better way
When lambda equals one.

The first two lines of course, refer to the formula for the probability distribution function of the Poisson distribution. It is a discrete probability distribution, for a number of discrete independent events in a fixed continuous space. What the heck does that mean? Who cares... anyway?

Let's just go to the last line first... lambda equals one? Lambda is the average number of events in that fixed space... what could it be?

Third line is one of the only clues (albeit deliberately vague), the other being the title... as to what the poem is about.

Let's just say that an average of one suicide in Singapore (independent discrete event) occurs per day (fixed continuous space). Source: Samaritans Singapore

So the poem is about suicide. Or more specifically, about reducing suicide into a statistic. Depressing isn't it? I know. I hope the lesson I try to teach isn't all too subtle for most.