The UN has declared that today is World Statistics Day. For some strange reason they’re more interested in Official Statistics, like GPD, and rates of unemployment rather than the interesting stuff to do with estimating the variation in weights of squrgles in laboratory populations.
Actually, this is less about complicated data analysis than about getting the data, and summarising it: it’s about making sure we have good, reliable numbers. Here’s an official video celebrating today:
Independence of statistics is vital: during the 1980s, the Tory government in the UK changed how they calculated the unemployment statistics every few months – oddly, the number when down each time. This, of course, made it more difficult to chart the changes in unemployment. Nowadays the UK’s official statistics body, the Office for national Statistics (aka the ONS) is now independent of government, reporting directly to Parliament.
To me it’s obvious that we need good quality independent statistics. Governments need to plan where to spend resources in the future, so they have to know how much money they have (this is related to GDP, of course), how much they will bring in in taxes, and where they need to spend it. So knowing how many people there are over 60 means they can plan for how much to spend on pensions and other services for the elderly. Of course, we want to make cross-border comparisons: how much does government spending on science influence overall GDP, for example? Or how do different national approaches to AIDS prevention work? For all of these, we need statistics.
The nitty-gritty of work on official statistics involves not just the collection of statistics, but also their collation and reporting. All of this: carrying out efficient surveys, managing data bases (just checking you have the right numbers entered!), and then presenting them in a way that is intelligible (i.e. pretty graphs), and can be extracted and used by economists, epidemiologists and other professionals (i.e. boring tables and spreadsheets). Like most public services, a lot of this is mundane work, but vital for evidence-based policy.
If we don’t have good quality independent statistics, what are we to rely on? People generating their own numbers?
So hug a statistician today – see if they’re normal. And if they’re not, check the length of their tail.