Monday, 29 April 2013

Crime is down - maybe

I read that the crime statistics say that crime is down.

I don't know.

When I started doing serious econometrics, I discovered that before you can apply sophisticated statistical techniques to numbers, you have to get the numbers. And that getting them isn't as easy as I'd thought.

A simple example - if you look at the population of Germany in the 1960s, it suddenly jumps by a couple of million in (as I recall) 1963; apart from that year, there's a smooth progression.

When you look into this, you discover that suddenly they started including the population of Berlin, whereas before, they hadn't.

More generally, when you look at a time series of statistics (such as crime figures), before you can draw any conclusions, you have to look at how the numbers were gathered, and whether there's been changes in definitions, or in categories, or in how the numbers are reported, and so on.

I haven't yet seen an analysis of how UK crime figures are prepared, and of what changes might be affecting the year-to-year comparisons.

So what I say is ... maybe.

No comments:

Post a Comment