If we are to take this seriously, it would be useful to know whether we are missing the target by a little or a lot: if the average number of portions consumed is 4.5 then that is a lot less worrying than if it is 1.5 - particularly as the recommended figure of 5 is surely a somewhat arbitrary round number. And this is not the only respect in which the survey falls short.

It appears that the poll (conducted by YouGov) is based on people’s reports of their own behaviour: “the highest levels of consumption were reported in the south of England, where 26% said they ate at least five portions.” However, it is well known that people are often inaccurate in reporting their own behaviour, particularly if that behaviour might reflect poorly on them.

But in this case there is a further reason to think that the reported figure may be inaccurate: it really isn’t clear what constitutes a portion.

The story on the BBC website carries the following information.

WHAT IS ‘FIVE A DAY’?

• It equals around 400g of fruit and/or veg
• One portion is 80g
• A portion equals two or more small fruits ie satsumas, one medium-sized fruit ie an apple or banana, or half a grapefruit or one large slice of melon
• A portion of veg would be two broccoli spears or three heaped tablespoons of cooked vegetables such as carrots, peas or sweetcorn
• Potatoes do not count
• But fresh, frozen, tinned and dried fruit and vegetables do
• Smoothies can count for up to two of your five a day
• Pulses and beans count as one portion – no matter how much you eat

No doubt this is intended to be helpful, but:

“It equals around 400g of fruit and/or veg”: does this apply to everyone, independent of age, sex, body weight, activity level, metabolic rate?

“A portion equals two or more small fruits”: wouldn’t more than two small fruits be more than one portion?

“One medium sized fruit ie an apple …”: what is meant by ”medium’ here? a quick Google search suggests that apples range from about 100g to over 200g. (And ‘ie’ should be ‘eg’.)

“A portion of veg would be two broccoli spears”: are broccoli spears always 80g each?

“Three heaped tablespoons of …”: are all tablespoons the same size, does everyone heap them to the same extent?.

“Smoothies can count for up to two …”: what do ‘can’ and ‘up to’ mean here?

This isn’t just nit-picking. Vague, non-standard units and broad-brush averages really are unhelpful. Without clear, unambiguous and specific information a campaign such ‘five a day’ cannot be expected to succeed.

And if our diets do need improving for the sake of our health then that is serious.

________________________________________________________________________________________________

Neil Sheldon has taught at The Manchester Grammar School for 40 years. He is a Chartered Statistician and Fellow of the Royal Statistical Society. He has been an RSS Guy Lecturer since 2007.  He is also course leader for the Certificate in Teaching Statistics offered by the RSS Centre for Statistical Education

On reading a press release from the Institute of Advancd Motorists (IAM) which referred to 57% of bike riders – nationally – admitting to jumping red lights, Peter Walker smelt a rat and instinctually started asking questions. 

Q. Where did the figure come from? A. A recent ’poll’ of 1600 people.  Q. What sort of poll was this? A. Responses to a survey monkey poll on the IAM’s own website. SO not exactly a random sample then….more of a self selecting group (and one, which although very unlikely, could have included respondents who were all members of the anti-cycling lobby).

Worse…..if the press release made it sound like 57% of cyclists jump lights regularly, when you actually broke  the figures down, they showed that 1.9% confessed to this. Another 11.8% did so “sometimes” while 24.6% did “rarely”, and 19.1% had done so once or twice.

In summary, it’s clear that most of us have a strong interest in knowing if and how many cyclists are jumping lights and it’s wrong that misleading headlines – based on poorly interpreted data – are allowed to stand in the way of good information.

N.B. We have since learned that once the mistake was brought to their attention, the IAM re-issued the press release with a more nuanced introduction “57 % of of cyclists say they have jumped a red light at least once, with 14 %  saying they do so regularly or sometimes, according to the IAM’s latest online poll of 1600 people”. It also referenced the ”32% of drivers who say they jump lights”.

However, the press release headline still maintains that ”More than half of cyclists jump red lights”. It will be interesting to see how this is reported tomorrow.

See the RSS-getstats ‘CIPR, MRS and RSS’ ‘Best practice guide for using statistics in communications’  for PR colleagues writing press releases.

Appearing before the European Parliament’s economic and monetary affiars committee – a report of the event here - Radermacher called for common standards for auditing public spending – a proposal likely to highly controversial in the UK, where public bodies have only recently switched to a standard not common elsewhere in Europe.

But Radermacher, formerly head of the German national statistics office,  insisted costs could be saved if there countries abandoned ‘national peculiarities’ in how they measured say exports. Too much ‘double measurement’ went on, increasing the burden of statistics on business.

He called for ’a European statistical system, not co-ordination of national systems’, bemoaning ‘a still deeply anchored reluctance to share data at EU level’.

As more and more decisions were precipitated, sometimes automatically, by movements in data, it was more necessary than ever to assure the quality and reliability of statistics. That raised questions about whether nation states’ statistics agencies were well enough supported, in terms of budgets, but also made the completion of a ‘real European statistical system’ more urgent.

Since Eurostat had been given more powers in 2009, progress had been made. What had happened in Greece and Italy could not happen again, Radermacher said, referring to allegations that official figures on debt and spending had been obscured or minimised to allow those countries to join the Eurozone.

Now Eurostat figures on the Greek public sector deficit could be accepted without reservations - the agency could respond proactively to concerns about national data and conduct audits, looking at contingent liabilities.

Economic and monetary union depended on reliable and high quality official statistics, which in turn made Eurostat one of the critical institutions of the EU, provided its standards were mandatory across the continent.

The Health and Social Care Act now coming into force will have profound and – it’s argued – damaging consequences for fairness in healthcare. The basis for NHS data now becomes care commissioning groups and that switch may undermine the availability of the information that’s vital for monitoring who gets to access services, where and how.

An article in the British Medical Journal argues that area based structures for collecting data and monitoring populations are going. Most information responsibilities switch to commissioning groups, based on GP practices, the boundaries of which are fluid and often ‘non-geographical’.

This loss of a population basis and responsibilities for comprehensive data collection and monitoring will help ‘make it almost impossible to take the action needed to tackle inequalities in health and in access to healthcare’. Till now, the administrative structure of the NHS in England has been based on resident populations within defined areas, often co-terminous with councils, allowing matching of data.

In future the basis will be GP practices. This could impede collection of routine health data and make it hard to monitor the effect of new legislation on the health of the population locally or nationally.

The case is argued in Dismantling the signposts to public health? NHS data under the Health and Social Care Act 2012, the article also available on the City University site. Its authors are Professor Allyson Pollock of Queen Mary University of London, Professor Alison Macfarlane of City University and Sylvia Godden, honorary senior research fellow at QMUL.

 What will happen to the NHS architecture of public health observatories and collaborations, which produce detailed profiles of the health status of local areas, is not clear. Optimists may say that it is in the interests of the commissioners to have good knowledge of the people in their catchment. Much may depend on how much data is shared – and the electronic and security protocols that permit data sharing.

The caterpillar is the larva of the oak processionary moth, and the story continues with the information that “Each caterpillar is covered in 63,000 hairs which can trigger potentially lethal asthma attacks” – and a warning that the London Olympics might be under threat.

Of course, from a statistical point of view it is the figure of 63,000 that demands attention:

• Do all of these caterpillars have exactly 63,000 hairs? Of course not! You don’t have to be a lepidopterist to know that big ones will generally have more hairs than little ones.
• Is it an average across all caterpillars? Possibly, but in that case it would be nice to know how variable caterpillars are in hairiness.
• How is the figure of 63,000 arrived at? It seems unlikely that anyone has actually counted that many hairs. Perhaps it is an estimate; perhaps it is more like a guess.

Oak processionary caterpillars

Wikipedia gives the same information in somewhat more careful terms: “The backs of older caterpillars are covered with up to 63,000 pointed defensive bristles”. This statement suggests that the figure of 63,000 is an extreme case rather than some sort of average. This is better, but still not satisfactory. Suppose you knew nothing whatever about the people’s heights and someone told you that people can be up to 8’11” tall. That statement is correct, but it gives you no idea at all of the heights of typical people. In that respect it is very misleading.

A Google search with the question “how do you count the hairs on a caterpillar?” doesn’t turn up anything very promising, but it is likely that someone somewhere has actually done some counting. Perhaps the technique is to count the hairs on a few square millimetres of caterpillar skin, and then measure the total area of hairiness and scale the figure up. Nothing wrong with that, but it is obviously not going to give a very precise answer.

Good statistics are expressed cautiously and with due regard for accuracy, precision and variability. If the story had said “caterpillars of this type can have tens of thousands of hairs” it would have passed harmlessly under the goodstats radar – but an extreme figure of dubious validity quoted as if it is the norm sells papers.

The online version of this article can be read here.

PS The Daily Telegraph website has a headline “Lethal errors in 2 million prescriptions”. If true, that would imply that vast numbers of people are dying from prescription errors – because that is what the word “lethal” means. The story itself refers to “potentially life threatening errors”, but does not reveal a single instance in which a prescription error has killed a patient. Just another example of using an extreme figure of dubious validity to sell a paper.

________________________________________________________________________________________________

Neil Sheldon has taught at The Manchester Grammar School for 40 years. He is a Chartered Statistician and Fellow of the Royal Statistical Society. He has been an RSS Guy Lecturer since 2007.  He is also course leader for the Certificate in Teaching Statistics offered by the RSS Centre for Statistical Education

The monitors have been looking at both Whitehall and councils as they push out reams of data on activities and spending, and they are disappointed. Amyas Morse, head of the NAO and comptroller and auditor general, says opening up ‘has potential’ but ‘what the government is lacking is a firm grasp of whether that potential is being realised.’ It needs to know much more about what sites the public are accessing, whether the information is getting through and to what effect.

The NAO notes that since last year England’s local authorities have been pumping out detailed lists of what they spend on, the public has remained largely indifferent. By contrast, websites offering crime information, including maps of where offences are thought to have taken place, are much visited.

‘If transparency initiatives are to be more than aspirations, then government needs to measure and monitor both their costs and benefits. This is vital for tracking success and learning what works.’

But were the figures right? By coincidence the Audit Commission has just been looking at how hospitals ‘code’ the duration of patient stays and found them not wholly reliable.

One NHS information analyst is sceptical, noting a suspiciously wide variation between hospitals. Is this a case of information being put into predefined boxes? Hospitals work with a set of templates, some to do with the time when patients were admitted. Hospital stays are getting shorter but the measuring boxes come from the days when patients were in for several days if not longer.

The first rule in statistics is to ask where numbers come from. You can perform all sorts of manipulations and draw all sorts of conclusions but they are going to be warped if the data you started with is flawed or suspect. A round robin Freedom of Information request elicited lots of data and The Times wrote its story. But it would have been safer to quiz the figures and go back to those who produced them, to check they were recording something real.

The unveiling of a statue of Don Revie, Leeds United and England manager, on 5 May 2012, will be the latest addition to the 120 statues in situ dedicated to specific sportspeople in the UK. 

There were just a handful of these statues before 1980 and another 10 or so were erected by 1995. As we might expect, when it comes to statues, football looms large with 50+  players, managers or chairmen who have been, or are soon to be depicted by statues inside or adjacent to the grounds they once performed at. Yet, as a % of the total potential candidates, this number is still pretty low. 

So why has there been such a massive rise in the sculptural representation of our sporting  ’greats’? What does it take to be honoured in this way?  which sports are most likely to yield a statue and why? why are there no statues of sportswomen? and what, more generally, does this new aspect of sports iconograpy tell us about society today?: our relationship with the past? and perhaps (even) the role of organised religion?

‘From pitch to plinth’, an event run by the RSS Local Group on 24 April (1530 hrs) at Leeds City Museum, is a chance to hear Dr Chris Stride, University of Sheffield, one of the statisticians behind the ‘The Sporting Statues Project’ , talk on this subject.

Arts bodies ‘could track ticket purchasers, measure and map audiences, note accounts, compare web analytics, monitor social, share data, share donors’, he says. ‘Imagine the questions we could start to answer. What time of day do people buy tickets online? What age do they start to give? What areas of the country respond to what web layout? Imagine the lifelong picture of arts audiences we could start to build.’

The government has talked about ‘armchair auditors’ scrutinising council spending. But councillors themselves are confronted with reams of numbers, about service performance, spending, social and demographic data about the area, polling data showing residents’ views. How able are they to assimilate and judge such material? If they rely on council officers to do the donkey work in collecting and interpreting the figures, they themselves still have to know enough to make decisions.

There’s talk of ‘localism’, and driving decision making down from central government – but also from councils to more local levels, estates, wards or neighbourhoods. There the data may be more contained, more specific to housing, parks or the street scene, but much of it is still going to be quantitative in nature.

We’d welcome your views on how statistically literate those getting involved in community and local decision making need to be.