Supplementary memorandum from Mr Philip
Redfern
1. This is in three parts. I) Support for
the proposal of A R Thatcher (former Director of OPCS)[17]
that the 2011 census should ask for information only about persons
present on census night. That information would be the main element
in estimating the resident population of each area. II) The need
for a population register. III) Demographic checks on the census
results. These last two topics are crucial to the future of population
statistics, but the ONS Memorandum to the Committee has little
or nothing to say on either of them.
I. THE 2011 CENSUS
FORM
2. As Thatcher points out in his Supplementary
Memorandum, a form asking only about persons present on census
night is simpler for the householder to complete. He does not
have to consider who is usually resident at the address but is
absent, nor does he have to give details of them. That will nudge
up response rates. The fieldwork and processing tasks are reduced
in scale and complexity, with significant reductions in cost.
Tables of the usual residents in each area (down to postcode level)
can be compiled by transferring a visitor back to the postcode
of his usual residence.
3. Under the proposal the only residents
of this country beyond census reach are those who are abroad on
census night. This is an increasing number, probably about two
million by 2011. An estimate of their numbers could be obtained
from the International Passenger Survey by questioning residents
returning to the UK in the days and weeks following census night.
The present sample size would yield data on about 4,000 residents
absent on census night, and give broad brush estimates by sex,
age and area of residence. Whilst this may seem inadequate, it
must be compared with the alternative ONS approach in which about
one-third of the two millions absent abroad would not be picked
up by census questions on absent residents.
4. The proposal to transfer data on a visitor
to the postcode of his usual residence dispels the ONS criticism
(Ev 216, #5.2.1) that data on migration, resident population and
journey to work would be inaccurate. The casualty would be data
on household composition which would omit absent residents; but,
as noted above, about a third of absent residents would be missed
by the ONS approach.
5. The proposal will eliminate the problem
of distorted population geography inherent in ONS plans (see Ev
164, #21).
6. On balance, the simplicity of the Thatcher
proposal—for the public, for enumerators and at all stages
of processing—together with the cost saving more than counterbalance
a marginal reduction in the quality of the household composition
data.
7. A final point on the census form: the
Evidence shows the pressure to add more questions, and ONS refer
(Ev 217, 5.2.3) to their bid for funding for an extra page
of questions per person. They will achieve that aim at the expense
of poorer response rates and threats to the credibility of the
basic census counts of the kind that so damaged the 2001 census.
Fewer questions must be asked.
II. THE NEED
FOR A
POPULATION REGISTER
8. The Evidence shows widespread concern
about migration figures based on small scale voluntary sample
surveys and on disparate and disjointed administrative records,
as well as doubt whether "tinkering" with existing migration
sources will yield worthwhile results. There is also concern about
the reliability of the 2001 census results. There is increasing
recognition that the longer term solution is a population register
that would coordinate the separate registers held by public agencies—particularly
now that the administrative case for such a register has been
accepted by the Government. Other countries provide examples of
this new approach.
9. Ideas for a population register, leading
ultimately to an Administrative Record Census to replace the conventional
census, are outlined by the Statistics Commission (Ev 36, 20-22).
The case is developed more fully by David Coleman (Ev 185-188)
and myself (Ev 163, 8-11 and Ev 165, Section II).
10. In 2003 ONS published "Proposals
for an Integrated Population Statistics System" which wrote
(Section 5.5):
"From 2013 onwards, the population statistics
database would be updated regularly using information from administrative
sources, the address register, the population register and surveys,
creating a longitudinal database covering the whole population."
11. This ONS vision quickly evaporated.
In 2005 an ONS-led project team (the Citizen Information Project)
recommended that a population register should be created centred
on the issue of ID cards, with a completion date (and therefore
value to statistics) in the 2020s. This plan is now being implemented.
The weaknesses of the plan are outlined in my Memorandum (Ev 165,
26-27) and are well known to ONS, but they are not mentioned
in the thin ONS comment on this topic (Ev 220, Section 6.2). Moreover,
ONS are not now playing a direct part in the development of a
population register. If such a register is to provide proper information
on migration before the 2020s, then it needs to be pursued with
urgency and with a high-powered person in charge.
III. DEMOGRAPHIC
CHECKS ON
THE CENSUS
RESULTS
12. Demographic checks are standard practice
in validating census results in any advanced country and may,
as in the US 2000 census, lead to population estimates that depart
from the census results. But the topic receives scant coverage
in the Evidence. The Statistics Commission makes brief references
to the check provided by the "roll-forward" from the
preceding census (Ev 34, 4 and Ev 35, 13). Coleman (Ev
191-192) notes the low sex ratios in the 1991 and 2001 census
results and discusses the demographic checks that ONS made in
those years. But the lengthy ONS Memorandum is silent on this
important methodology, on which validation of the census results
depends.
13. Beginning with the paper I read to the
Royal Statistical Society in 2003, I have on four occasions formally
put to ONS a demographic check on the sex balance of the 2001
population estimates which finds a deficiency of half a million
males in relation to females (HC 326-i, Ev 15, 32). Coleman
devotes the final paragraph of his Section headed THE CENSUS (Ev
192) to a description of my check. However, ONS have not replied
to any of my approaches nor made any reference to the sex ratio
problem in their Memorandum. This is not an adequate response
from a professional organisation. The validity of ONS's measure
of the change in the sex ratio needs to tested and the underlying
reasons for any change identified.
January 2008
17 Ev 39 Back
| 
