Select Committee on Treasury Minutes of Evidence

Examination of Witnesses (Questions 1-19)




  1. Welcome back to the Committee and welcome to our first inquiry of this Parliament which is into the Census. For the convenience of the shorthand writer could you identify yourself and your colleague.
  (Mr Cook) I am Len Cook, National Statistician. John Pullinger is Executive Director in ONS.

  2. May we begin with a general question about the 2001 Census? Were you generally happy with the way it was conducted? Did it go to plan?


  (Mr Cook) There are four really big issues in a census which you have to get right. The first one is being able to get an agreement on what should be in the Census and get that agreement with Parliament, with the public in terms of the acceptability, and Ministers. That was successful; we ended up with a pretty good census form. The second is to be able to organise a field force, to get the questionnaires out and get them back in again. That went very well but it was hindered by the Post Office. I can talk about that in terms of the variability of performance of getting forms back from the Post Office, because we had a mail-back census this time for the first time. The third is to find ways of having everyone know about the Census, or nearly everyone, before it actually happens, so there is a consistent measure of trust across all different communities and regions. By and large that was done very well. The Census communication campaign went very well and the success of that was witnessed by the news media coverage in the month or so around the Census, where by far the overwhelming majority of criticisms about the Census were "Where's my form?" not "Can you trust these people to run a census?" and a lot of other issues. The fourth issue is about organising a system to process the Census in a reasonable time. There are two other issues. One is the ability to respond to events. We had several very important and large events this Census. One was the response to the ethnic question in Wales and that required us essentially to think of ourselves as running a separate census in Wales in terms of the communication programme compared with England. The fact that in Wales the Census response rate was higher than England in the end is testament to the success of that. The second is foot-and-mouth disease. Basically we were able to carry on with the Census despite the issues of foot and mouth. The third issue was our Census helpline. We were literally swamped and within three days we were able to fix that. My final comment is that it is important to recognise that the Census in Britain on 29 April was carried out at a time when you could say we had a fair wind behind us. If you think of the last couple of years, it was really quite a positive and supportive time for a Census Office to carry out a census in terms of the support of all communities in Britain. I believe we captured the benefit of that fair wind behind us in carrying out the Census.

  3. Are you on track to produce the results next year?
  (Mr Cook) As targeted, which is August next year for summary results of the British population where the Census is essentially used to update population estimates, produced only in summary, age and sex for regions. Then at the end of the year we shall have more detailed results of the key questions and then in the first half of 2003, there will be the full set of Census information released for the first time and once only.

  4. Given you have computers and presumably these forms are machine readable, why does it take two years to get all this stuff out?
  (Mr Cook) Because it is such a large Census. Sometimes in Britain people tend to forget how big the country is. This is an extremely large and complex Census questionnaire. We could have done it faster if we were to invest more money up front. We followed the scanning process adopted in the United States Census by Lockheed Martin; we have adopted that methodology. We have also chosen, through the one number Census, to use the Census coverage survey to give us estimates of the undercoverage which occurs in all censuses so that when we publish the results, it will cover for those people who are not included in the Census. Being able to put all that together has caused the time frame. We could have produced some results earlier and then revised them. But our consultations two and three years ago left us convinced that that was what people actually wanted us to do.

Mr Laws

  5. I want to focus on one particular part of the very helpful briefing document you sent out to us beforehand. In the section about outputs you say, "The effort and expense of taking a Census are worthwhile only when the results meet needs, and are delivered effectively". I want to explore those two points briefly. What are the needs that the Census meets? Why go through this vast operation costing about £¼ billion?
  (Mr Cook) We actually produce a wealth of information from a whole range of statistical processes, statistical surveys, administrative records which come from a lot of public processes. The Census is the one data collection we have which embraces the whole population, which tells us about each community, whether we define it by region, by age, by ethnicity. It is the key benchmark around which we are able to create other communities such as families, people in localities. There is no other source of information which gives us the opportunity to derive accurately communities which we cannot measure directly. It is because of the value of the Census that we can see changes not only in the totality of the population, but in its structure. This means we can forecast its change by knowing the changes in the age composition, the family structure, the household formation of the population. We can then produce reliable forecasts of the British population for several decades ahead.

  6. Could you not do that in any other way that was cheaper? You could poll people, using various forms of other market research which might give you a more cost effective solution. Is there anything particular in the data you get out of this survey that you could not get out in a different way?
  (Mr Cook) Ultimately every sample survey we have, whether it is a labour force survey, opinion polls, rests on having something solid in terms of a measure of the population that it is able to benchmark back to. For example, a large number of the sample surveys we carry out in the UK are either benchmarked directly to the 1991 population Census or population estimates we have made since April 1991 that update those population measures. You do need an anchor at some stage. We are seeing two important sources of information becoming available to us now. One is a lot more administrative records available to us through the computerisation which is occurring in public agencies such as the Department of Work and Pensions, where there is a very rich amount of statistical analysis of the beneficiary population. That gives us almost 100 per cent of people in the youngest ages and it gives us a very good measure of the population in the older ages. It does miss out the population in the middle.

  7. If we did not have the Census done, what would be the major negative effects for public policies? What information would we lack which would then create problems?
  (Mr Cook) We would still have measures that count the population in some way. In the United Kingdom we have very good births and deaths certificates. Although we are an island country, we do not operate as a police state at our borders and it is very difficult to measure migration flows. In the last four years some 70 per cent of the population change in Britain has come from migration not from births and deaths. Our ability to measure inter-censual population change accurately is limited by our knowledge of migration flows.

  8. Do any other advanced countries use any other mechanism for finding this information, or in your view do all of them use a census of this kind?
  (Mr Cook) Some of the Nordic countries have ceased having censuses because of their ability to integrate information from a variety of public registers which exist. The one limitation they find is that their ability to generate family and household information is constrained solely by the knowledge that the state has about families and households. They have very powerful registers through tax, family benefits and other such processes.

  9. May I explore the other part of this sentence on the effectiveness with which you deliver the Census? The first issue is one of cost. Am I right that this particular Census cost about £¼ billion?
  (Mr Cook) Yes.

  10. Do you have any comparable figures for what the cost is in other countries?
  (Mr Cook) Yes, we do.

  11. Do those demonstrate that we are particularly cheap or not?
  (Mr Cook) We are not particularly expensive and we are not particularly cheap. We are considerably cheaper than we were in 1991 and the per head cost in the United Kingdom is less than that in Australia, for example.
  (Mr Pullinger) I think it is less than in Australia and the USA.

  12. It is a bit unfair to ask you now, but would it be possible to get some information from you on that?
  (Mr Cook) Yes, we can provide that.
  (Mr Pullinger) The reason we hesitate is that where you have a situation as in the Nordic countries where you get the same information from another source the cost comparison is not straightforward. In the USA, they have a mix of a short form and a long form designed to achieve slightly different ends. We can certainly give you the costs. In a large number of countries they are still doing traditional censuses in the same way we that we do.

  13. Obviously this information is only useful and effective if it is fairly timely, otherwise things have moved on. You did explain to the Chairman that the reason for it taking so long to put together is because we are a large country. India, United States are large countries as well. I am sure you have very good reasons, but India completed their census results one month after the census had been collected in; the United States appears to be more like nine months. Compared with other countries do we not seem to be quite leisurely about the way in which we put together all of the information?
  (Mr Cook) It is useful to look at a couple of things. One is that we could have produced, as we have done in earlier Censuses, based on simple counts of the Census forms, estimates of the number of people in each area which would have been the de facto population of people actually counted in a particular area. What we are providing is the people who are normally resident, the actual de jure population in an area and we are also adjusting that for the measure of the undercount which we estimate from the Census coverage survey. That takes into account the differential undercount by age and sex for example.

  14. So we are a lot slower than the US and India.
  (Mr Cook) The Indian count was actually a head count. Where I come from in New Zealand the Census results were published not too much after the Census but they were the first count of the forms themselves in the areas. The full results come out about a year after that. When you make those comparisons you need to compare exactly what you have.

Dr Palmer

  15. I cannot think of any other scientific endeavour where a population of objects or people is explored by looking at every single one. I should really like to take you back to this question of whether we would not be able to get an equally accurate picture by looking in more detail perhaps—so perhaps one could have more questions—at a population of let us say 100,000, where you do not feel that even if that were not done every time it would be an option to replace every second census for instance.
  (Mr Cook) Some countries have had, as we had in the past, samples of households who are asked large parts of the Census questions. For example, 10 per cent of the population may be asked the full form. In other countries, France and Israel—and we are investigating it ourselves—there is a rolling census where about 10 per cent of the population are included in a rolling census every year and we come back to a basic population register. That allows us once a year to have an updated view of the population. What is important, what causes the quest for detail, is that when it comes to the end uses of the Census, a large part of it is about understanding the age and family structures of often quite small family groups: it could be an ethnic community which lives in a particular part of the country and regional groups. In local government for example there is a huge interest in understanding the population in quite small areas. If you are planning transportation policies you actually want to know how people in this or that pocket use this particular mode of transport. For many of the uses of the Census in local government it is the very richness and detail of the Census which makes it valuable.

  16. Could you imagine that it would be possible to delegate aspects of the Census to take account of these local concerns so that you have a core of questions which will be asked for a larger population and local authorities could commission additional questions for their area?
  (Mr Cook) That is possible. In some ways you could say we do that nationally now. As in health, the interests of a particular sector of public policy nationally tend to place some weight on whether a particular question should be included or excluded. If you were to do that regionally it would have an impact on how you managed the Census regionally. It would have an impact on how you manage the national publicity campaign. To answer your question in a slightly different way, the shift in access to administrative data in the UK is moving us more and more to a Nordic country type model where we can look at the balance of what we collect in the Census and what we get from other sources. That has been very much part of the ONS policy and the question is: do we have a 2011 Census?

Mr Bear

  17. What were the main lessons which came out of the conduct of the 1991 Census?
  (Mr Pullinger) We judged the 1991 Census overall to be successful and we built an awful lot on the successes which took place that time. One particular example which has already been mentioned today was the inclusion of an ethnic question for the first time. It had been very, very contentious during the period up to that; certainly when we were thinking about the 1981 Census. That came off and it was quite ground breaking and something we thought was very good and has proved to be very useful. In the areas where it was less successful, the most critical one was that the coverage was much poorer amongst some groups than amongst others. Key to our whole strategy—and maybe helping to answer some of the earlier questions—has been to try to minimise the amount of differential undercoverage, for example the fact that young men are much less likely to respond than some other groups, certain ethnic minority groups are less likely to respond and other groups may be slightly more surprising. Babies are difficult to catch; people forget to put them on. Elderly women too who may be reluctant to open the door. In 1991 we found quite significant different patterns. For 2001 the critical lessons we learned were first of all to target our effort to those communities which were most difficult to get at, especially the inner city areas where we knew we were going to have problems. So we focused our resources much more. Secondly, and possibly most critically, this whole concept of taking account of the people we had missed in producing the results. That has added to the time to produce them. In 1991 what was happening was that we produced a first set of results, area by area, but people found it very difficult to understand what sense they could make of those results because it was not until a long time afterwards that we were able to say that 20 per cent of young men in some inner cities had been missed. It is very difficult to judge without having made those adjustments. Central to our strategy this time has been to look at the primary requirement of people, which is how many people of what characteristics exist in the population, rather than how many we happen to have been able to count in the Census.

  18. Can you say now that you have overcome those problems in the 2001 Census?
  (Mr Pullinger) I would be foolish to say the Census was easy: it was very, very hard. The problems we expected to encounter we did and in some senses with more of a vengeance than we expected. Finding people in the inner cities, persuading them to fill in the form, was a huge challenge. It was not just taking account of issues which happened last time, but changes in society over the last ten years have really made things more difficult. More properties with entryphones or security systems, more people not in the house at any sensible time or any time at all in some cases, more people living alone. We targeted that. Our enumerator field force were calling at all times of the day or night, they were calling back on several occasions. We believe we have been successful in all the fundamentals, but I am sure there will be some small pockets where we have struggled in the actual collection. We are confident that the overall strategy, which allows for those areas where we will have difficulty, were successful in the end.

  19. What were the main changes in 2001 compared with the conduct of the Census in 1991.
  (Mr Pullinger) The main changes of the conduct. Firstly we planned it much more thoroughly from an earlier time and that was a key recommendation which came out from the National Audit Office report. There was a lot of testing of questions, whether they would be acceptable to people, whether they would be useful; an awful lot of testing. We had 150,000 in a complete rehearsal of the Census just to see whether the procedures worked. An awful lot of upfront work. The second thing which was a big change was involving local communities, particularly amongst ethnic minorities. We had a scheme which involved about 3,500 different groups in the population who could help us get into particular areas; religious groups especially were very supportive of this as they saw it as a means of getting their own communities recognised in the Census. A lot of effort went into that. The adoption of a post-back methodology for collecting the forms. In the past an enumerator has gone to the door to deliver the form and then gone back to collect it. The issues to do with the Post Office have already been mentioned but this time it was a huge leap forward for us to be able to save the cost of people going twice, but also for the vast majority of people who are quite happy to post a form back for them just to do it and not have to fiddle around with that afterwards. That was on the collection side. On the processing side, we have made huge changes in the technology we are using and in the way we are acquiring the technology, using the opportunities of engaging the private sector. The Lockheed Martin contract has been pivotal to that. If you went to our processing site in Widnes, it really is state-of-the-art technology which is out-of-sight different to what we had before. Two more things then I will stop. The first thing is the coverage survey adjusting for the results. The final one will be producing the results using the Internet which we would not have done last time.

previous page contents next page

House of Commons home page Parliament home page House of Lords home page search page enquiries index

© Parliamentary copyright 2001
Prepared 26 November 2001