Falling response rates in the decennial Census provide tangible evidence of loss of trust. It was initially proclaimed that the 2001 Census was ‘the most accurate census ever’. In spite of a legal obligation to return a Census form the response rate in the 2001 Census was only 94% − lower than in any previous census. The responses rate in inner city areas was much lower. There were unexplained discrepancies in the sex ratios in different age groups and it is estimated that that up to 800,000 young men were missing.

The adjective ‘best ever’ could be applied to access to detail of the results. Internet publication fully exploited the digital technology. But the detailed results give only an illusion of accuracy.

The underlying population figures are estimates based on uncertain assumptions about the population missed. The statistics relating to the answers made to individual census questions are even more dubious. Missing detail is interpolated based on population characteristics. A substantial but difficult-to-estimate proportion of the 2001 Census statistics published are invented. The proportion is highest for inner city areas.

The unreliability of the census statistics means that estimates of the total UK population for the first time are not based primarily on the census, but depend upon data from administrative sources such as the National Health Service Central Records.

The Statistics Council that is supposed to be the watchdog made only a muted response. The Council acknowledged failure in the Borough of Westminster and other areas where the response rate was particularly low but implying that the Census was otherwise of good quality (http://www.statscom.org.uk/media_pdfs/reports/Census%20Report%20Final%2022.pdf ). The Council performs a useful watchdog role in relation to the minutae of departures from statistical righteousness. But if the watchdog really barks, as it should have done in the case of the 2001 Census, it would reduce trust in the statistics and thereby reduce the support it could reasonably be expected to get from the Government.

The main architect of the design of the 2001 census was an academic − Professor Ian Diamond − at that time at Southampton University. Diamond tried to solve the anticipated low response problem with a statistical trick. Diamond conceptualised what one distinguished commentator described as a ‘black box’ would transform incomplete data from the census into a consistent dataset for the whole population. But nobody could understand or explain what was in the ‘black box’.

In the event Diamond’s approach proved to be impractical as well as statistically dubious.

But neither Diamond nor the methodological strategy of the Census was castigated when the failures became apparent. Instead someone in Government must have given Diamond a good reference. Soon after the Census Results were published Diamond was promoted to become Chief Executive at the Economic and Social Research Council.

An unsound and inappropriate statistical approach was not the only problem with the 2001 census. It seems clear that the ONS label, that is clearly identifiable as belonging to central government, reduced the response rate. Illegal immigrants, illegal workers and failed asylum seekers, for example, have a strong motivation not to respond to the census because it is organised by central Government.

Receipt of government benefits may also be an important factor. Sixty percent of households in the UK are in receipt of benefit payments from government. The payment of benefits creates situations where individuals have strong motives for being less than fully responsive to a Government conducted survey. Single mothers who would lose benefit if they were found to be cohabiting with a boy friend are in that kind of situation. An individual in danger of being classified as a co-habitee has a motive for disappearing at census time.

It may be that such disappearances played a significant role in the unexplained undercount of some hundreds of thousands of young men who were seem to be missing from the Census results. Another theory is that there has been steady migration abroad. The mystery of the missing men was sufficient to stimulate the BBC into making a radio programme on the topic − Where have all the men gone? (http://www.radstats.org.uk/news/censusradio4.htm)

Counting the population creates a governmental dilemma. The solution favoured by the Government and by the Statistics Commission is to make increased use of dependence on administrative sources. But such reliance fosters a big brother image that threatens the scale and reliability of responses made to sample surveys conducted by government − and future censuses.

Questions about the value of a central government label, such as that of the ONS, is also evident in the detailed results of sample surveys. The UK has an exceptionally rich and coherent set of large scale social surveys. But they share a major weakness in that the numbers responding to questions about receipt of state benefits involving tests of eligibility is low. As a result these surveys consistently underestimate the proportion of the population in receipt of some government benefits.

The number of claimants of Job Seekers Allowance according to the Labour Force Survey, for example, typically ranges around 20% below the level recorded by the administrative statistics for the number of recipients of JSA. This inaccuracy limits comparability between the JSA series and the officially recognised International Labour Office series measure by the Labour Force Survey. The value of both series is diminished.

The government could use independent agencies to conduct surveys. That would create another dilemma. Use of independent agencies might increase response rates but it would reduce Governmental control of the information. But the tradition that central government should be the fount of official statistics is deeply embedded in the political culture. There has not been clamouring for the repeal of Clause 4 as far as the production of statistics is concerned.

Immigration provides an extreme case of a conflict between governmental need for control and the desire for reliable information. The Census should be a prime source for estimates of immigration, but the central government label has ruled out the 2001 Census as a reliable source. Statistics for immigration are produced on the basis of the International Passenger Survey, but the IPS was designed to measure tourism not migration.

The Government has not shown inclination to be involved with reform of the IPS or setting up some kind of new system to measure immigration and emigration directly. It is noteworthy that John Reid’s promises of extensive reforms at the Home Office have not yet included the promise of better statistics. Getting better statistics would depend upon gaining the trust of individuals whose positions are vulnerable to the tightening up of rules that Reid is promising to introduce.

Discussion of the government of statistics in recent years has avoided the question of public trust. Public discussion has been largely concerned with consolidation of the relationship between the RSS and the ONS. This noble minded debate focuses on matters such as the quality and integrity of statistics and the professionalism of the statistics profession. But neither the RSS nor the ONS have show expertise, interest, or even strong motivation towards getting the trust of the public.

The lack of clarity and consistency in the Consultation Document may indicate that the Government has recognised the limited value of the ONS/RSS debates. In the face of the commitment to make new laws is it possible that the Government will go back to basics? Is it possible that the Government will go back to those first term promises to increase public trust in statistics?