The sinister world of statistics
Two worlds collide: Planet Statistics, a strict environment governed by evidence and hard data, has once again smashed into Planet Westminster, an ecosystem run on much more subtle rules around public opinion and mood management.
Immigration Minister Phil Woolas' 'leaked' letter bemoaning the "naive" and potentially "sinister" decision of the Office for National Statistics to release migration figures early because they were "topical", is part of the wreckage from this culture clash.
As is statistics watchdog Sir Michael Scholar's accusation that statisticians are being "pilloried" for publishing independent and objective statistics.
What this is really about is the well-oiled wheels of the government communication machine being frustrated by the grit of tough new rules on the use of official data.
It was Gordon Brown who, as Chancellor, championed the introduction of a strict code on the use of government statistics. He was anxious at the lack of public trust in official figures and promised a "new politics" free of spin and manipulation.
But almost immediately, his own staff fell foul of the new rules in spectacular fashion. Last December, a special advisor in Gordon Brown's office demanded the Home Office publish helpful statistics on hospital admissions for stab wounds even though No 10 had been told by senior stats people that the figures were "potentially inaccurate and may give the wrong impression".
After Sir Michael complained to Downing Street that the affair had been "corrosive of public trust in official statistics", the head of the Civil Service, Sir Gus O'Donnell, revealed in a letter that all officials and advisors had been told of "the importance the Government and permanent secretaries attach to the observance of the Statistics Code".
Every spin doctor and press officer was reminded that they could not play fast and loose with official figures. The code was clear. They must:
• Publish statistical reports in an orderly manner.
• Publish statistical reports according to a published timetable.
• Present statistics impartially and objectively.
• Ensure that those producing statistical reports are protected from any political pressures that might influence the production or presentation of the statistics.
This was a massive cultural shock for those in government who had been used to "managing" the flow of such information. Power had flowed from the spin doctors to the number-crunchers and ministers were frustrated.
Frustration turned to anger, however, when last month the ONS brought forward the release of sensitive new figures on immigration because the National Statistician, Karen Dunnell, thought they were 'topical'.
The statistics, showing how foreign migrant jobs had increased while domestic employment had fallen, emerged in the midst of the wild cat strikes over 'British jobs for British workers'. But the numbers were not due to be released until last week as part of the regular migration statistics.
In science terms it made sense: ongoing public debate about nationality and employment would be enriched by having hard data made available.
In political terms it was highly controversial: ministers have accused the ONS of "playing politics" and naivety in releasing information that inflamed an already difficult situation. The opposition has accused the government of launching "hysterical rants against the independent statistics office".
Ministerial teeth-gnashing is all the greater because of a perceived unfairness. While government must abide by the rules, opposition politicians don't have to. Cabinet Minister Kevin Brennan recently suggested there might be an agreement that if Whitehall stuck to the code, so must the government's opponents. But he must realise that this would be almost impossible to enforce and might frustrate scrutiny of the executive.
The big idea behind the creation of a statistics code was to "build public trust and confidence in the statistical system as a whole". The rows and scandals surrounding its introduction are likely to have achieved quite the opposite.
I'm 
© MMIX~RS~q~RS~~RS~z~RS~27~RS~)
Comments
Sign in or register to comment.
Interesting that the Government has been hoisted by its own "petard" on a number of occasions since the "no spin from statistics" policy was introduced.
Since the early days of Alistair Campbell and Mandy spin has become so endemic in Labour life, that they don't even know when they are doing it these days.
Complain about this comment
What concerns me about statistics is how people are allowed to use them as "science"
Bayesian calculation, that of primary use of prevalence data, or proximity data ie "A is next to B therefore A cause B" seems to rule the roost.
In short, there are too many people in government using path-of-least-resistance thinking in their synapses, with little accreditation to Hume's Law.
Complain about this comment
I disagree. Binding opposition MPs to the Statistics Code would node frustrate scrutiny. It would force politicians on all sides to debate the facts - all of them - in an informed manner and discuss the subtleties of the evidence instead of relying on bullish sound bites for the gutter press. Now wouldn't that be a good thing?
Actually, I guess the reality is most MPs come from a legal background and don't understand the difference between scientific discussion and the binary ("for" and "against") philosphical debate. In science you discuss the evidence and come up with a decision, whereas in a philosophical debate, to take a position and look for evidence to support it.
Complain about this comment
Nu Labour reserve the right to manage release of statistics.
The people of this country are not capable of facing the truth and the reality of national issues such as immigration.
The ONS must not direct the debate. The elected Government are the arbiters of what is revealed to the public.
This is a Labour Government which has controlled the agenda for so long it cannot accept being challenged by genuinely independent bodies.
The head of ONS will not be enobled and will certainly be replaced by a more friendly placeman.
Shame, Shame, Shame.
Complain about this comment
The sinister of statistics is made more sinister by incomplete reporting by the media which makes it easier to put on whatever spin you want. The figures do seem rather much larger than many would expect, BUT How do they compare with previous times? How many were born to British citizens while abroad (service families, international companies, VSO etc)? I am sure readers could think of other questions. In the absence of decent reporting politicians can try to get away with telling us that they mean X, or that they mean Y leaving us to believe whom we want, but unable to decide rationally whom we ought to believe.
That is before someone makes the unfounded assumption that all those not born in Britain are taking jobs from Brits and living off the welfare state. My friend had a quad heart by pass done by an Italian and was looked after by a specialist Phillipino cardiac nurse.
Complain about this comment
I just don't understand the desire to peddle false or misleading statistics. The only conceivable reason is to mislead either to a degree or totally. No government can ever get away with misleading the people. In the end the truth will out. Even those Soviet tractor factories closed down in the end.
When I learned my statistics I quickly understood that they can be easily manipulated. Context, construction and consistency are everything.
Sometimes the numbers tell you things you don't want to know, but as every business-person knows these have to be understood and taken into account. There is no point in lying to yourself. Measure, review and change is the best method and keep on measuring.
The problem with Planet Westminster is that there seems to be a priority in looking good. This is foolish.
The broader issue is that government is now too big and too complicated to respond to the initiatives of politicians. Before they can move on the political class will have to accept that the state is too complex and needs to be simplified in order to work.
What the public needs is government that works and that can honestly measure how it is working.
Complain about this comment
Westminster's position and directive to ONS is simple, if the truth hurts, don't tell it. Are there any questons?
Complain about this comment
Isn't this exactly what the Home office did with knife crime statistics?
You may have read about it on a blog somewhere.
Complain about this comment
Surely strict code and Gordon Brown is an oxymoron Mark. the only strict code Gordon Brown has ever understood is the bullying one he has used to charge his way through life. Mercifully, for him and for us, those days are about to change. For good. The sooner he slopes off to spend more time with his statistics about endogenous growth the better.
Complain about this comment
Woolas' letter was hardly "leaked", the recipient posted it on the internet.
I also commend the follow-up piece on Liberal Conspiracy; it seems to tear big holes into Woolas' account of events (which is what you've based your piece on).
Complain about this comment
Our company produces data on population by country of birth.
We were not at all keen to launch these with PR to the national press, and our PR agency advised that it could harm our reputation to do so - because of the risk of the data being distorted to make a 'good story'.
A lesson there for ONS.
It is not only govt politicians who spin. The duty is not only to produce accurate statistics, it is to do so in a way that ensures accurate statistics reach the public. One should not abdicate responsibility for how they are spun after publication.
Complain about this comment
This government really doesn't like the public to know the truth about anything, does it?
How outrageous that it should try to suppress the release of statistics that it doesn't like.
The sooner we're ride of this bunch of sinister buffoons, the better.
Complain about this comment
Statistics are a joke at the best of times. Look at the stats on drinking. If you drink, you'll get cancer. If you don't drink you'll get heart disease. So we get two groups saying we must have a glass at night before bed or none at all. Personally I say I'd rather live a few years less and die having had a happy life than worry about whether that night in the pub at the weekend is going to give me some nasty disease.
The real issue as has been pointed out is that statistics need to be seen in a wider context. You cannot say "we did this, and now things have improved" unless you can prove that your change is what has caused the improvement (ie knife crime).
It also doesn't help when the govt gets caught red handed fiddling the figures by getting the police to arrest people for more serious offences just so they can then get them to start arresting for the real offences later in the year in order to say "we've brought down serious offences in the last year". No you didn't, you just rigged the stats to make it look like you did. Ever noticed that overall figures stay fairly close, it's only the class of crime that changes? And usually just before each election as well, hmmm....
Complain about this comment
11
Surely any stats about nationality will be used in a story? unless they say "only 1% of people are foreign", the Mail will pick it up
Raw facts are raw facts, (although statistics are often misleading by their very nature, let alone the spinning and cherry-picking of numbers)
The ONS should simply have a publishing mechanism to provide real stats to the public without the media's involvement - they shouldn't need to consider public opinion as they are an independent government office, in a very different position to yours as a company
Complain about this comment
'Figures never lie, but liars always figure.'
The idea of having an office that produces nothing but 'unbiased' statistics seems like a waste of time. I do not believe it is possible to produce unbiased statistics because the bias is built in to the choice of data that is counted.
To complain about manipulating statistics for political reasons is missing the point. The choice of which criteria to count and how to count them is always part of the manipulation, so statistics should always be viewed with a degree of skepticism.
Perhaps a more appropriate question would be how can people be taught that skepticism and taught to put statistics into perspective.
Complain about this comment
Re: 6 stanilic
Excellent post, indeed statistics ALWAYS need context.
And the press is often lax in that area.
I am alway irritated when I hear that xxx pounds was spent this year on yyy when no context is provided. This means that only very few specialists who understand all the ins and outs of yyy costing will appreciate the news report!
Complain about this comment
Re: 4 kcband8
Nicely put, I also get the feeling that nuLab is treating the British public like children.
And if they do nobble the head of the ONS then this matter should be exposed to all.
Be vigilant.
Complain about this comment
Good lord - the ONS is playing politics ? Or is it that the raw data which has not be pushed through the Westminster filters and spin doctors was just a tad embarrassing ? I;d be encouraged that the politicians are upset - its would seem the ONS is doing its job ..
Complain about this comment
I would prefer to consume my statistics raw, rather than wait for them to pass through the digestive system of a government minister.
Of course popular papers and pundits will pick a figure out of context and misuse it. The best course then is full publication of all the data and wide public discussion.
Complain about this comment
We all know that immigration is under control, our borders are safe and we have 'British jobs for British people'. Trouble is statistics can prove almost anything and because our dear leader is always right no one should EVER doubt what the government says.
Complain about this comment
The Labour party being caught out by rules that they themselves introduced, surely not?
Complain about this comment
In our society we are ashamed of an accent originating in Liverpool or Birmingham, but proud to be innumerate. We have never heard of Bayes and think correlation is a heavy metal band.
So long as our maths is awful, statistics will be abused. Possibly the only meaningful measure of the state of our society serves only to make dishonest newspaper headlines
Complain about this comment
Statistics are virtually meaningless. If you take five people, four of whom earn £25 per week and one who earns £400 per week, their average income is £100 per week. Although the figure is statistically correct, it does not in any way reflect the truth of the matter.
It isn't so much a lack of trust, it's just that if you spend more than five minutes looking beyond the headlines and apply some intelligence, it is abundantly clear that statistics can not be seen as factual and therefore are of little value in the real world.
Complain about this comment
Once more the Government has been caught out, it must be so frustrating being a Labour supporter, each time that the BBC trips over itself to regurgitate Labour 'facts', someone else pops up and proves how false and hollow the Labour claims actually were.
I note that the Government has been well and truly rumbled with its claims that Gordon Brown knew nothing about the "misleading crime figures".
What a shame for the BBC that only one of its journalists has been following this story, had it been the Conservatives this story would be number one article on all news shows.
Impartiality is supposed to be the BBCs mantra, yet, after counting the amount of Labour claims that the BBC has had to correct, it is clear that the BBC takes everything that Labour says at face value, pity really.
Complain about this comment
Statistics are not meaningless in themselves, but our interpretation or deliberate abuse of them can make them so.
BrendanMaclean’s (comment 23) example of ‘average income’ is good one. The average of a set of numbers is exactly what it claims to be. It is not the most likely value, the most probable value, a typical value or even necessarily a possible value. Not even if the Chancellor says so!
No family has a fraction of a child. However, I would hope that the number of ‘families’ times the average number of children (per family) equals the total number of children. Again family needs a careful definition….. that a clever spin doctor can manipulate.
Complain about this comment
#25, you are quite wrong, the "average" may be the most likely/probable value - it's called the mode. It may also be a typical value, if by that you mean the value of which there are an equal number of greater and lesser values - called the median. The value that you allude to as not possible is presumably the arithmetic mean. However, these are all average values.
Stattos generally prefer the mean, as it is susceptible to further analysis such as standard deviation - but when was the last time you heard the words "standard deviation" in a news broadcast? No, that would imply a proper analysis of the figures!
The truth is that none of us have time to properly analyse every statistic that the news throws at us - so we need a sensible interpretation of results rather than meaningless figures.
Unfortunately, what we get, from politicians and journalists alike, is meaningless raw numbers spun to whatever end is desired. Sadly, that's not going to change anytime soon.
Complain about this comment
#26. I agree that strictly ‘average’ is a word that may be used for many different quantities (and even qualities). However, in most technical areas its default meaning is arithmetic mean. As such, I would suggest that an expert, who is interpreting statistical data for a wide audience, is being misleading if he gives any other meaning to average. And even for a technical audience, it should be defined if used for anything other than arithmetic mean.
We should not expect too much from the general community. Very few intelligent members of the public will have any idea about modes, medians or arithmetic means. Mode is a particularly unfortunate choice. Jargon may be precise and precision of expression and thought is essential in probability and statistics, but we shall never succeed in educating the general public if we insist on using that jargon.
I wonder if many people know or care what we are arguing about. Maybe our discussion serves to highlight the difficulties in trying to communicate these ideas. That does not mean we should abandon the effort. Mathematics, as someone said, provides a powerful means for accuracy and truth. My original point (#22) was that to make use of it, we need to provide a better mathematical education.
Complain about this comment
The missuse of statistical data is not only misused by MPs but also by some Local Government Community managers, which when challenged is defended as a working document.
After many e-mails including to the new Nottingham City Council Manager, Jon Collings who promotes transparency while in office, I still await an explanation of the difference between a working doc and a final draft when using gov statistics for a local gov document.
The plan in question and open to review I am assured, although on the public website is; NCC Clifton and Wilford, Neighbourhood Management Empowerment Plan Final Draft (2008-2011).
Although entitled final draft I am assured that it is a working document untill (2011).
confused, I am.
I agree with your endorcement of statistical ethics. statistics is a very useful tool when applied correctly.
Misplaced statical data being used, to benefit, target needs for further renumeration of services provision, amounts to fraudulent statistical use and should be treated as plagerism, to bring an abrupt end to statistical criminalty within Governmental documentation. If statistics are used to show change their should be a legal requirement to reference the said data to ensure proper placement of mathamatical facts being used as foundations for ideals.
When will we build an Educational provider statistical database. these statistics are needed to balance the educational statiistics recieved from educational reciever ability. There are no provider statistics to measure equality of service provision within education. How do I apply for this to be built.
I feel that without a "provider statistiscal database" educational statisics for learners are of no factual foundation base, when looking at predispositioned eduacaional ability from a nurture verses nature view point. I can never conclude that children can be born learners nor can I conclude that better educators provide better learning no matter the birth placement of the child.
Complain about this comment
View these comments in RSS