Troubled families and statistical significance

Written by Web News Editor on . Posted in Features

How many ‘troubled families’ are there? The government is ultra precise. It says, without hesitation, 120,000. Professor Ruth Levitas of the University of Bristol is sceptical. Do we have an exact definition of ‘troubled’? The Department for Communities and Local Government ‘makes a discursive move from families that have troubles, through families that are ‘troubled’, to families that are or cause trouble.
 
 The government figure comes from work done on 2004 data showing 2 per cent of families showed with a certain number of predefined indicators of disadvantage or problem. But coming from a survey, any figure is subject to sampling error and that means, the number of ‘troubled families’ could be as low as 60,000 or as high as 300,000.
 
 Levitas goes on to show how trying to allocate these families to particular areas is difficult, given the nature of the data.
 
 She concludes after inspecting the data base for the policy the government has not made rigorous use of the research ’In the term ‘troubled families’ it deliberately conflates families experiencing multiple disadvantage and families that cause trouble. The attributed costings are obscure and certainly open to question.’
 

Data say crime really does not pay

Written by Web News Editor on . Posted in Features

Today’s headlines - the Telegraph’s  ‘Crime really doesn’t pay – a third of bank robbers make nothing at all’ and the Scotsman’s ‘Crime does not pay very well: robbers on £ 12k a job’ - share new data-driven insight which can be used in fighting crime.
 
The articles pick up on a press release on an article which will be coming out in Significance, the magazine of the Royal Statistical Society and the American Statistical Association, later this month.
 
With unique acccess to British Bankers Association data, researchers at the Universities of Surrey and Sussex developed an economic model of bank heists, balancing  robbers’ efforts against their gains or losses. The model shows that contrary to common perceptions, in the UK, takings per successful raid are equivalent to less than six months’ average wage (and takings in an average US raid are considerably less). Factor in multiple raids to boost sub-average income and an increasing probability of capture and loss of earnings and this ‘career path’ makes less and less economic sense.
 

Grasping the immigration numbers

Written by Web News Editor on . Posted in Features

Immigration as defined by the public isn’t necessarily the same as defined by the official statistics, according to an Ipsos Mori survey on behalf of the Migration Observatory at Oxford University.
 
Thinking Behind the Numbers says ’members of the public and the government may be thinking about different things even when both are talking about “immigration”. Categories such as temporary immigrants and students loom large in official statistics, but less than a third of the public has in mind either of these categories when thinking about immigrants.’
 
People feel most strongly about low-skilled migrants and asylum seekers but the former often come from within the European Union, exercising a legal right, and the latter – a small proportion of the total flow – have a right under international agreements to have their claim for asylum assessed.
 
One of the largest categories of immigrants, students, seem to cause the public the least concern.
 

Cross your teas

Written by Web News Editor on . Posted in Features

NHS Choices looked at the health conditions and life styles of men in Scotland over three decades. It’s a fascinating piece of work but more suggestive than definitive. Researchers found that men reporting drinking a large amount of tea each day (more than eight cups) were also found in the NHS records as having contracted prostate cancer. The study does not show how serious their cancer (many men with prostate cancer die of other causes) nor can it check whether the men have been big tea drinkers all their lives; the study did try to screen out other lifestyle factors such as drinking alcohol.
 
Media reports said bluntly heavy tea drinkers are 50 per cent more likely to develop prostate cancer. But relatively few men get prostate cancer, whether they drink tea or don’t. The study found that overall 6.4 per cent of those who drank the most tea developed prostate cancer during the study period, compared with 4.6 per cent of those who drank the least. Those drinking a moderate level of four to six cups of tea a day were not at any increased risk compared with those who drank the least.
 
We all know that you can’t get a sexy headline out of caveats and qualifications but isn’t it time the media – especially in the age of Leveson – made a bit more of an effort to avoid empty and misleading sensationalism?
 

Happy Deathday!

Written by Web News Editor on . Posted in Features

Stevie Wonder sang it with gusto over the Jubilee holiday but it may not be a happy birthday if you are over 60. A study of Swiss death certificates published in Annals of Epidemiology has found a significant number of ‘extra’ deaths occur on the day people are celebrating their birth.
 
Researchers checked the data was sound – some people’s birthdays are not exactly known. They calculated the increased risk as broadly the same for men and women at about 13 per cent, which suggests it’s more than coincidence, as NHS Choices helpfully explains.
 
But why might birthdays be lethal? The research establishes a correlation, but for cause we have to speculate. People might drink more, leading to more accidents. Birthdays see more of what doctors coyly terms ‘vascular events’, such as stroke: blowing out candles on the cake can be stressful. Sadly, birthdays are occasions for sadness, which may produce more suicides.
 
 

Good workmen don't blame their tools (so don't blame the stats)

Written by Web News Editor on . Posted in Features

There are stats – data about the economy, child poverty levels etc  – and then there is Statistics, the techniques used by statisticians to arrive at those figures.
 
Statistics is a toolkit. It helps us to understand the world around us and the society we live in. But ‘statistics’ is often taken to mean ‘my facts’. And when my facts are controversial (sometimes wrong) the other side has a go at ‘statistics’.
 
In a typical debate,  there is usually a point where opponents start picking holes in the other side’s numbers.  As their tenuous grasp on the numbers slips, ‘statistics’ become the common enemy.  Pundits and presenters then talk about statistical ‘fog’, statistical ‘spin’, ‘magic’, and so on. They are implying that the stats are inherently misleading, divisive, made up and difficult to grasp.
 
Confidence around stats seems to be at the heart of the problem. …not distrust of the production of stats per se but mistrust of the handling and reporting of the stats and lack of confidence that users and communicators of stats really know their stuff!
 
If there were more stats users and communicators (and listeners,viewers and readers) with the understanding and confidence to make sense of stats so they have an independent sense of when numbers are being used well or badly……debate would be far less likely to turn into a verbal brawl…and if and when it did, stats would be less of a ’whipping boy’ and more ’the good guy’ who intervenes and refocuses the discussion on the evidence at hand.
 

Gang border data and new approaches to policing

Written by Web News Editor on . Posted in Features

Statistics is all about measuring and finding patterns in data which offer up information and solutions.  It’s a ‘toolbox’ – a range of tools such as distributions, equations and models – which can be used to work things out in different arenas. It’s a matter of picking the right tool.
 
The Metro today ran a short piece ‘Gangs really do act like animals’ on a just-published, University of California study has used a tool – Lotka-Volterra equations (also known as predator-prey equations) - used since the 1930s by ecologists to model the way bees, chimps and lions compete for resources- to get a grip on gang crime. UCLA researchers have found links between between the way hunting animals map out their territory and the behaviour of street gangs especially in terms of their “range”
 
By modelling police record data  and other data on the gang’s bases,  researchers have worked out that 99.8% of crimes could be expected to occur within one mile of the borders between rival gangs (nearly 60% of crimes within 1/5 of a mile, nearly 90% within 2/5 of a mile). 
 
What solutions do these new findings offer?  researchers have been able to identify the real boundaries between gangs….ones that don’t just follow streets but run through backyards and alleyways.  This new knowledge wil help the police  to focus their efforts, allocate resources and achieve better outcomes. And incidentally, the study also offers an evidence-informed warning …. gut instinct might suggest that wandering into the heart of known gang territory would be the most dangerous thing to do….but most crime, it would appear, actually happens on or near the borders between rival gangs.
 

Join the RSS

Join the RSS

Become part of an organisation which works to advance statistics and support statisticians

Copyright 2019 Royal Statistical Society. All Rights Reserved.
12 Errol Street, London, EC1Y 8LX. UK registered charity in England and Wales. No.306096

Twitter Facebook YouTube RSS feed RSS feed RSS newsletter

We use cookies to understand how you use our site and to improve your experience. By continuing to use our site, you accept our use of cookies and Terms of Use.