Thursday, May 29, 2014

Statistical Literacy

Two news articles over the past few days have reminded me of the importance of this.  Both from sources who ought to know better (or whose ideas have been reported in a way that throws them open to interpretation).

First of all, the Stephen Hawking analysis of England’s World Cup chances relied on 45 matches played since 1966.  Other people have already picked holes in this, but I want to add something I haven’t seen mentioned elsewhere, which is that the data that has been used to predict the current England team’s response to matches presumably includes matches played in the early 1970s, before all of the current team were born and before Roy Hodgson was a manager.

Given the way football, fitness, analysis and tactics have changed over the past few years, presumably the early 1970s are as irrelevant to England’s current team and its chances as I am.

The other article, from the BBC, reports a Lancet study about rising levels of obesity.

The article states:
More than half of the world's 671 million obese people live in 10 countries, ranked in order:
  • US
  • China
  • India
  • Russia
  • Brazil
  • Mexico
  • Egypt
  • Germany
  • Pakistan
  • Indonesia
Source: The Lancet
Hold on a moment.  China and India have rather big populations.  As does Indonesia.  Let’s have a look at where these countries sit in a list of countries by population


Rank in obesity article
Population rank
Percentage of global population
USA
1
3
4.44%
China
2
1
19%
India
3
2
17.40%
Russia
4
9
2.04%
Brazil
5
5
2.83%
Mexico
6
11
1.67%
Egypt
7
15
1.21%
Germany
8
16
1.13%
Pakistan
9
6
2.60%
Indonesia
10
4
3.45%

Ah – the six most highly populated countries are included on this list.

Adding up the percentages give us just over 55% of the world’s population.  So the article is actually saying “more than half of the world’s obese people live in countries that contain more than half of the world’s people”.  In other words: big deal.

It would be rather more interesting to have some more context for the initial statement in the article:
“The number of people in the world who are obese or overweight has topped 2.1 billion, up from 875 million in 1980, the latest figures published in the Lancet show.”

Estimates (from other sources) suggest that the current population is about 7.17 billion while in 1980 it was 4.4 billion.  This means that the obese population has more than doubled while the general population hasn’t doubled.  Is there a real increase in obesity, or is it, for example, the case that younger people have higher rates of obesity and there are now more younger people than there were 34 years ago?

So: we certainly need more people to be aware of statistical gibberish. 

[The Guardian article does actually appear to understand some of the problems inherent in the “Scientists have found…” type of press release.  It is worth a read.]

No comments: