The UK Statistics Authority has issued for consultation a draft code of practice for official statistics. It is quite hard to get people to actually read it word by word. I think that when you do, you find there is a lot in it. It has 10 principles – covering everything from integrity, accessibility, methodology, resources, and value for money. These principles contain a total of 60+ practices and 3 further protocols on release practices, consultation and use of administrative data in producing official statistics. This 3rd protocol is completely new and may need some more work in light of responses to the consultation process. If you have views on the draft Code do let us have them by 30 September.
We didn’t pluck it out of air. The Code has a lot of history but it is based quite closely on the existing National Statistics Code, the European Code and the UN Fundamental Principles. It is shorter, simpler and more imperative in style than the current National Statistics code; it is closer to European Code; and, no longer seeks to set out exemptions and exceptions.
Crucially, the Code is a set of principles and rules for organisations that produce official statistics. It is not so much for individuals. So it says things like:
Principle 1.2 “Ensure that those producing such reports are protected from any political pressures that might influence the presentation of the statistics.”
Principle 10.1 “Ensure that statistical services have the staff, financial and computing resources to produce and disseminate statistics to the standards of this Code.”
While the Code covers a great deal it does not cover pre-release access arrangements because the Statistics and Registration Service Act says it cannot, but we are required to treat the new statutory rules (the pre-release access to statistics order) as if they were part of the Code. The problem here is that the new rules on pre-release access are loose and only apply to ‘statistics in final form’. So there is a further document being produced by the Cabinet Office which guides bodies on how the statutory rules should be interpreted. And Scotland has drafted a rather different set of statutory rules. So this area is very problematic. The Authority’s view is that what matters is how the various statutory orders are used in practice and that we will be looking to see a reduction in pre-release access. We have also rejected the argument that pre-release access is needed in order to be able to put out Ministerial statements at the same time as the statistics – why should we support the issue of political documents over the top of the statistical ones?
Meanwhile, the Code has to deal with pre-release access to statistics not in their final form which is not covered in the statutory orders.
Back to the Code proper. What is it for? What are we going to do with it? One thing it is not going to be is some sort of guarantee of the quality or reliability of the figures themselves. As long as the statistics are the best that can reasonably be produced, are useful to users, and are well explained when published, the Code does not require them to be any better than they currently are.
And just to deal with a point that causes some confusion; we see the Code as applying to all official statistics but only being used as a basis for formal assessment in relation to those defined as National Statistics. I have to say, I do not see great value in the National/Official distinction but it is written into the legislation and we have to live with it.
The Code must rather be seen as a tool which can be actively used, in conjunction with the Authority’s Assessment function, to achieve the goals of the Authority. If the Authority thought that the statistical service was fine as it is, then the Code could be used to keep it as it is. And if the Authority thought that the statistical service needed to be improved, then the interpretation of the Code can be ‘calibrated’ to put emphasis where it is needed.
This is an important point and is best illustrated with examples.
The Code says at Principle 2.3 “Investigate and document the needs of users of official statistics, the use made of existing statistics and the types of decisions they inform, and the key areas of emerging demand and unmet need. Take these into account in the planning process and report the results”.
Let me ask you. What is your advice to the Statistics Authority? How do you want us to use that requirement? To change current practice, or to keep them where they are now? That is a decision we must take. My point is that the Code per se does not include this calibration. This will come out of the Assessment process and the dialogue that supports that process. The Code tells us to look at the documentation of user needs – it does not tell us at what point to be happy with what we find.
Another potentially tricky example is in the preamble to the Code (at paragraph 11 of the Preamble) “It is implicit in the Code that, within those bodies that produce official statistics, there will be sufficient managerial separation between officials responsible for National Statistics and other staff of the organisation to ensure clear lines of accountability for observance of the Code”.
We could have added this to the Code as an imperative; and we may yet do so. Do you think there is sufficient managerial separation at the moment? Where would you set the bar in this area? Again, we can calibrate “sufficient managerial separation as we think right”.
One more example. At Principle 2.1 “…adopt systematic statistical planning arrangements, including transparent priority setting, that reflect the obligation to serve the public good”.
The phrase about the “public good” stems from the legislation. What it means is supporting decision-making where the decisions benefit the public. So all government use of statistics is deemed to be covered but so, too, is a lot else. How will we know if the public good is being served by the planning arrangements?
My answer to that is that we need to ask the users whether our statistical planning is meeting their needs – this ties-up with the requirement to investigate and document the needs of users, as I have already mentioned.
In practice, quite a lot of guidance on the interpretation of the Code is going to be needed and while the Assessment team will influence this, we see the main role as being for the National Statistician. There is an important point here. We are not setting out to over-ride the judgement of the National Statistician or the Government Statistical Service. We will challenge it and, of course, the Authority itself will have the last word, but day-to-day guidance will come from ONS.
I now want to touch on a few possibly controversial points. Firstly, the legislation does not restrict the Code to the production of statistics. Our consultation document raises the possibility of the Code covering Ministerial statements issued at the same time as statistical release. So it may be that we might require them to quote statistics accurately, including full references to the relevant statistics they are using. It could even go a bit further, perhaps saying explicitly not to issue Ministerial statements until after the statistics; or not to give the impression that the Ministerial statement is the main statistical statement. Or not to distract attention from the statistics by looking like a statistical release? I would be glad to hear what you think.
Principle 6.2 says that producers of statistics should “Provide full and helpful commentary on the relevance and reliability of statistics in relation to the range of potential uses”.
How do you rate the commentary you read now? Does it say much about relevance and reliability of statistics in relation to the uses? Does it say anything at all about the uses? For example, take crime statistics. How many uses can we realistically name for crime statistics?
Finally, I want to turn to the new Protocol 3 on the use of administrative sources. At this point, the Code introduces the idea of an Administrative Sources Statement which requires the body responsible for the production of the statistics to set out all the administrative sources it uses to produce those statistics, and to define procedures which will ensure that any change in the system takes account of the implications for statistics. The Statement will also contain arrangements for providing access for statistical purposes and ensuring security.
The key thing in all of this is that the National Statistician must be consulted. This is what our draft Code says about that:
Public bodies who produce official statistics should also:
“Consult the National Statistician before finalisation of the Administrative Sources Statement and address any points raised so that the UK Statistics Authority may be assured that statistical work based on administrative sources is being conducted in ways that serve the public good.” (Protocol 3, paragraph 7)
Do please let us have your views on our draft Code and the principles and procedures that will govern our Assessment work. We look forward to hearing from you.