Friday, 4 September 2015

Legal Highs and Bad Statistics Reporting

“'Legal highs' referrals 'rise by 700%' in Jersey” says the BBC News, getting in on the JEP’s act for misleading statistics. As the BBC news tells in the story, in 2013, seven people were referred to Jersey's Alcohol and Drug Service compared to 57 in 2014.

Now that is a massive rise, as the JEP mentions, a tripling, but it is small in comparison with the total Island population. Saying – rise by 700% - skews and distorts what is actually happening. The BBC - by putting it as a percentage increase - gives a very misleading picture.

A high percentage increase can result from low initial numbers, so these statistics should be considered in context. As a percentage of the total population of Jersey, they are very small numbers. 57 is around 0.057%.

The JEP is actually better at placing the statistics in a wider picture. It notes that:

“Michael Gafoor, director of the Alcohol and Drug Service, said that the significant increase last year was not just as a result of the number of users of NPS going up but also because people were becoming ‘more aware’ of the dangers and asking for help.”

Now that is significant, because it suggests part of the reason is increased awareness by drug takers of needing to seek help. So part of the increase may simply be more people becoming aware than in previous years – it may reflect not increased numbers, but the effect of better education in needing to seek help. That shows again how unreliable it is to take numbers as they stand, without considering the wider context.

We also glean some interesting figures from past years and the current year:.

“In total 57 people were referred to the service last year, compared to six in 2010, fewer than five in 2011 and seven in 2012 and 2013. So far this year 15 Islanders have been referred to the service, including five teenagers.”

Now 15 is still higher than previous years, but as we are now just in September, assuming even numbers per month on average, it suggests that the total for this year will be no more than around 23. It shows the dangers of taking just one figure in isolation.

We should also note that a calendar year is actually an artificial unit in which to count statistics. Any mathematician will tell you that choosing the starting and ending points for data may result in different statistics.

Even if this year’s figures got as high as 28, that would be half of 2014. But should we have a headline “referrals half in 2015 compared to previous year?”

I think not – the important thing about fewer referrals this year is that it shows the trends is not increasing year on year, but widely variant between years. That’s what you would expect with small numbers, where a relatively small variance can lead to 700%!

It may well be that 2014 was something of a "catchup" on referrals as people who had been unaware of their needs came forward, whereas in 2015, we are settling back to a more regular pattern, but one which does reflect increased awareness for referrals.

And by the way, Guernsey's 2014 report says: "There has been a significant reduction in the number of drug related referrals to this service compared to the previous year (37%)." But again, the figures are very small compared to the total population, and I'm not sure the word "significant" is the best one to use here. The words "promising" might be more accurate, as it leaves open the possibility of complex reasons for the reduction.

Typically a news source will try to make the numbers sound as dramatic as it can. I'm afraid that's the case with the BBC article here. Yet it does serve one good function - it enables me to point out how to look behind the figures.

A good strategy in assessing what has been reported is suggested by mathematician Robert Korn who says: "One simple strategy is to temporarily ignore the statistic that someone has presented and ask ourselves what statistic we would actually want in order to make a judgment about the issue involved". 

In this case,  the "bare statistics" do not actually tell the whole story, and we need more data to see possible causes for changes. We need more information - have the numbers increased in people taking legal highs, or is it a better educational programme - "because people were becoming ‘more aware’ of the dangers and asking for help". One indicates a worrying problem, but for the other an increase in figures may actually indicate a success!

Misleading statistics also feed the fear about the migrant crisis. As Patrick Kingsley reported:

"The migrants at Calais account for as little as 1% of those who have arrived in Europe so far this year. Estimates suggest that between 2,000-5,000 migrants have reached Calais, which is between 1% and 2.5% of the more than 200,000 who have landed in Italy and Greece"

He also notes that "the number of migrants to have arrived so far this year (200,000) is so minuscule that it constitutes just 0.027% of Europe’s total population of 740 million"

Why are these figures not more widely reported? That's something which the reader may care to ponder.

1 comment:

Póló said...

Statistics never prove anything, they only serve to illustrate.

I agree with you about the misleading reporting which psychologically inflates the increase enormously. The absolute numbers, as you point out, are quite small and some of the increase may be due to greater awareness of available services.

I don't follow your reference to a tripling. It is surely an 8 or 7 fold increase depending on how you look at it.

Way back, the head of the Irish Economic and Social Research Institute, himself a man of international repute, got so pissed off with newly graduated social scientists and journalists misusing statistics that he wrote a pamphlet about it and set out their proper use.