Let's get noisy about data in electioneering

Written by Web News Editor on . Posted in Features

In an  interview with the BBC on his new book ‘The Signal and the Noise: why so many predictions fail, but some don’t”  - Nate Silver suggests that there is a big gap between what we think we know in politics and what we actually do know.
 
It is easier, he says, to predict the outcome of baseball than a presidential election as there is a lot more data available on baseball (162 games are played every year) than there is for an event which only happens every 4 years. In elections, we are trying to make inferences about what will happen in the future based on a small amount of historical data.  Outcomes are inferred from various sources of data including factors such as the state of the economy and levels of unemployment at the time of the election (the assumption is usually that there is a direct relationship between these and election results – invariably, not the case in US elections).
 
Today, there is more data than ever to draw on to predict election outcomes but  “in some ways having more data can get you into more trouble if you don’t know what to do with it”.
 
There’s a wider point about our knowledge gap when it comes to politics. Baseball apart, there are few occasions when data is quite so abundant and the public’s interest in the information in data stronger than during an election. But how well do we grasp the information in data ? How well-equipped are we as voters to handle the ‘noise’ …the cacophony of data we experience ? Do we scrutinise the statistics in manifestos and in the content of pre-election ‘fact-flinging’ debates?.  How good are the media at reporting that data?
 
And politicians’ ‘knowledge gap?. how often are candidates really comparing like with like? how often do they exaggerate numbers, how much real data underpins the big promises they make? and then how evidence-informed is the subsequent media coverage?
 
Perhaps, we should then ask ourselves how much impact the ability of candidates to use statistics accurately and appropriately in the run-up to an election has on the final election results?. Few would disagree that evidence-informed debate and policy-making should be the norm in parliaments everywhere, let’s not overlook the need for evidence-informed electioneering too.
 

Mums know best but stats can help debate

Written by Web News Editor on . Posted in Features

To breastfeed or not to breastfeed?  With last week’s Unicef UK report ‘Preventing Disease and Saving Resources’, the case in favour of mothers breast feeding babies until they are at least 6 months old seems to be stacking up and there’s a lot of international, national and local data on the subject.
 
International data? UNICEF data shows that UK breastfeeding rates are increasing but remain among the lowest in Europe. 81% of British babies are breastfed when they are first born, but the figure for 1-week old babies drops considerably and by 6-8 weeks, just under 50% of babies are still breastfed and by 6 months the figure is less than 2 %  (although more are still receiving breast milk).
 
National data? the Department of Health (DoH)s’ Infant Feeding survey is run every 5 years.
 
Local data? There is a local case study in the UNICEF report. The DoH analyses local data too.  Since 2008/09, each primary care trust has been required to submit quarterly data on the prevalence of breastfeeding at 6-8 weeks.
 
And now, there are cited data projecting health benefits (for example that 10,000 fewer young children a year would require hospital treatment), and data projecting at least £ 31 m savings for the NHS (due to lower incidences of some cancers in women who have breastfed and in illness in babies breastfed for longer).
 
The rate of breastfeeding is a sensitive issue with complex societal, family, health (and now economic) aspects. But whilst statistical findings may build a body of evidence,  how to feed their children remains a mother’s choice.  The statistical findings of this report and other sources of data should not be viewed as advocacy or weapons with which to target mothers who do not breast feed but as knowledge to support well-informed debate.
 
An Independent article,  ’How women feed babies isn’t just about statistics’ puts this case clearly. However,  references in the same article, to “mind-boggling statistics” and to those statistics leaving mothers “feeling they have put their child’s health at risk” are less helpful. So, let’s not make statistics the bogey man in this debate.  Statistics are the information in numbers. More enlightening than ‘boggling’ and they serve an important purpose in public discourse and the development of policy.
 

Datakindness helps not-for-profits see a little data go a long way

Written by Web News Editor on . Posted in Features

We hear a lot about the data revolution and the benefits that large organisations are reaping from big data but it is not always apparent how useful data analysis can be to smaller outfits in the not-for-profit sector too. This is a sector where it is often the case that data are collected  but then…because of limited capacity including know-now and resources, next steps  - analysing, interpreting and using the data to good effect - can be a struggle.
 
It is great to hear that a New York-based charity – DataKind  - with a mission to use data to serve humanity – is sharing the benefits of the data revolution and data’s worth to charities.  It’s doing this in an immediate and practical way, helping them to work more productively and efficiently and achieve fresh and quantifiable insight into their organisations’ achievements and impact.
 
In this week’s Economist, ‘Data Huggers‘ reports on a recent 2-day DataKind ”hackathon” in London when data scientists did just this. One of their teams helped a charity which offers counselling to troubled schoolchildren. By working through their data they enabled the charity to evaluate its work and, in particular, to see which of their audiences responded most to which services and where they were having most impact. During an earlier hackathon in the US, DataKind analysed months of a the text message data of a not-for-profit group which enables rural patients in India to use their mobile phones to secure the advice of doctors based in cities hours away.
 

Policing data is everywhere, insight nowhere

Written by Web News Editor on . Posted in Features

That uncompromising judgement – delivered to our recent parliamentary seminar by Rory Geoghegan of Policy Exchange – sums up the state of play on policing and statistics. Policing abounds with numbers. Police forces collect masses of data. But making the information sing and rendering the numbers intelligible is a different story.
 
Will the Police and Crime Commissioners shortly to be elected insist the data is made meaningful so that policing is based more securely on ‘insight’? Our audience of MPs, peers and their staff heard and applauded the challenge, but our speakers reserved their judgement.
 
Professor Allan Brimicombe of the University of East London is a specialist in the data of policing. To find out about crime, or at least activity that tends to involve police officers, you need to look, say, at hospitals. A ‘heat map’ of policing activity of the kind he has been compiling glows red around accident & emergency units. What that means is that a local strategy for policing necessarily involves collaboration with clinical commissioners and local authority health and wellbeing board. (Professor Brimicombe’s and Rory Geoghegan’s slides are on our resources pages.)
 
Look at how much data police forces already have, Rory Geoghegan said. But it is not managed. There is little online reporting. The data map is patchy, Professor Brimicombe said. ‘The statistics a PCC will need to devise a plan will not come served up on a plate.
 
‘The PCC needs an analytical capability independent of a police force and service providers. Any analysis function inherited from the police authorities may not be adequate to the task.’
 
Our partner for the event was SAS UK and Ireland, the software company. SAS’s  principal consultant on public security, Pete Snelling said policing data could be analysed. In a collaboration with Sheffield Hallam University and police forces, the company was putting together an automated alert system, which after processing the data on gun crime would flag patterns. ‘Big data’ including statistics derived from social media use were increasingly relevant to policing, for example allowing forces to anticipate a gathering of people that might have the potential for harm or violence.
 

Stats skills for social scientists - A major initiative

Written by Web News Editor on . Posted in Features

Within a few years, all students on social science courses at British universities will possess basic statistical capacity, if the hopes of the promoters of a new initiative are borne out.
 
With financial support from the Nuffield Foundation – its director, Sharon Witherspoon, is pictured left – the Economic and Social Research Council and the Higher Education Funding Council for England, new centres of excellence in teaching social science undergraduates quantitative methods are to be established. The plan is that they will boost capacity, first among those undertaking postgraduate work then later among academic staff, ensuring future supplies of sociologists, social policy specialists, political scientists who can pass on their quantitative skills.
 
The British Academy issued a position paper welcoming the £15.5m endowment. It has recently established a high-level strategy group to promote stats and quantitative capacity outside the so-called STEM (science and technology) subjects. Writing in THE, the group’s chair Professor Ian Diamond the vice chancellor of the University of Aberdeen and a distinguished social statistician deploreda gap in national capacity, that could reduce the overall quality of research and teaching in such subjects as sociology.
 
Writing in the Guardian Sharon Witherspoon said that UK social science had many strengths, ‘not least in our impressive data infrastructure (with birth cohort studies, longitudinal studies and others), and in imaginative qualitative research. But over the years it has become increasingly clear that it is weaker than it should be in quantitative research skills.’
 
The  BA Position Statement – Society Counts concluded the deficit in quantitative skills ‘has serious implications for the future of the UK’s status as a world leader in research and higher education, for the employability of our graduates, and for the competitiveness of the UK’s economy’.
 
Professor Diamond’s group includes representatives from the Department of Business, Innovation and Skills, the learned societies and exam boards and RSS getstats.
 

College of policing must help cops to count

Written by Web News Editor on . Posted in Features

A new initiative to improve police skills takes off in December, amid hopes that it will improve the use of data and evidence by officers as they strive to improve performance and drive down costs.
 
The Home Office has just announced that Alex Marshall, chief constable of Hampshire (pictured left) will head the College of Policing. It will be more of a ‘virtual’ college than a campus-based institution and support the training and capacity efforts of police forces rather than itself provide courses. But the government hopes that it will improve police operations and set standards for specialist skills and training, in fields such as investigation, intelligence gathering and firearms.
 
The government hopes policing will become cleverer, allowing cuts in what it calls bureaucracy as well as ‘driving down crime’. Home Office minister Damian Green said: ‘the college will be the engine of police reform, enhancing professionalism and setting the highest standards of integrity. It will allow us to develop the change in culture crucial to British policing.’
 
Mr Marshall has been chief constable in Hants for four years and is credited with establishing a national police air service, cutting costs on the way. He talked of the college cutting ‘unnecessary policies in policing’, replacing them ‘with practical, common sense approaches based on the evidence of what works.’
 
But assessing what works depends on more police officers being able to read the data and count the numbers.  A test for Mr Marshall and the new college will be how far it picks up the work commenced by the Society of Evidence Based Policing and similar initiatives, pushing officers and police managers to study crime patterns and police deployments closely.
 
Police officers are going to be able to do the numbers and the College of Policing will need to insist on more stats capacity.
 

NHS choices on the facts of cancer screening

Written by Web News Editor on . Posted in Features

We are all being encouraged to make informed decisions and choices about our health. But to do this we need communication around the benefits and risks of different screening and treatment options available to us to be as clear as possible.
 
Whilst recent media coverage  has focused on the ‘harm’ attached to breast cancer screening…NHS Choices has unwrapped the facts.
 
In the main, the media story has been the undoubted anxiety attached to false positive (or ‘false alarm’) results. Also concerns about ‘overdiagnosis’ (when patients receive  cancer treatment even though their cancer was unlikely to affect their life expectancy.)
 
Helping to balance this, NHS Choices  straightforward reporting of the findings of the Independent UK Panel on Breast Cancer Screening chaired by statistician, Professor Sir Nigel Marmot, will have done a lot to build understanding of the benefits v. risks of breast cancer screening.
 
The panel undertook an up-to-date assessment of the quantitative benefits and harms associated with population breast screening. It estimated that for every 10,000 women invited to screening from the age of 50 for 20 years:
 
  • 681 breast cancers will be diagnosed
  • 129 of these diagnoses will be overdiagnosed
  • 43 deaths from breast cancer will be prevented
 
i.e. for every death prevented, there are estimated to be three cases of overdiagnosis. This means that of the around 307,000 women aged 50-52 who are invited to screening in the UK each year, about 3,700 women will be overdiagnosed and about 1,300 deaths from breast cancer will be prevented.
 
We are also advised that after initial screening, approximately 1 in 25 women are called back for further follow up and about 1 in 5 of those will have breast cancer (i.e. to further allay fears, 4 out of 5 do not have breast cancer).
 
It’s very good to see this information communicated so clearly. Also to see these figures as head counts rather than what can sometimes be more abstract notions such as percentages or odds.
 
For more on this research, see “The Lancet” and Cancer Research UK.
 

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.