Census, is one needed

David Hansen made this Freedom of Information request to Scottish Government

This request has been closed to new correspondence from the public body. Contact us if you think it ought be re-opened.

The request was successful.

Dear Sir or Madam,

The last census in Scotland was rushed. MSPs were bounced into approving a set of questions before they had any time to consider the issues around this expensive, old-fashioned and rather Socialist exercise.

During the last census there was government spin that it was likely to be the last census, as information would be gathered in other ways. That has been dropped, though if there is a census in 2011 perhaps the same spin will be spun again.

Given that any government organisation sees keeping itself in existence as its main task, I am enquiring about what consideration has been given to having no census and thus saving me (and all other taxpayers) a lot of money. Having no census would also mean not employing torturers to run the census, but that is not a consideration in this request.

Please provide minutes of meetings, discussion papers and the like where the question of whether to have a census or not was discussed and the reasons for deciding to have one.

Please also note that "replies" which involve attachments in
proprietary file formats are not acceptable. A reply which is not in plain text format will be deemed to be a refusal to answer. If you are unable to produce some items in plain text format then please contact me to explain the particular difficulty with some classes of information so that we can agree a suitable format.

Yours faithfully,

David Hansen

Scottish Government

Dear Mr Hansen

I acknowledge receipt of your request which was received in our office
yesterday. We will respond to you in due course.

Yours sincerely

G Drysdale
(On behalf of GROS FOI Officer)

General Register Office for Scotland
Ladywell House
Ladywell Road
Edinburgh
EH12 7TF

show quoted sections

Scottish Government

Dear Mr Hansen

We have noted your request to have information provided in plain text
format where possible. If during our investigations it turns out that a
large number of documents have to be provided to you, would you be
prepared to accept these in Adobe PDF format? Otherwise it would be
helpful to have an indication of what formats would be suitable.

Yours sincerely

G Drysdale
General Register Office for Scotland
Ladywell House
Ladywell Road
Edinburgh
EH12 7TF

show quoted sections

Dear Sir or Madam,

Acrobat is acceptable, but only if there is something which cannot be conveyed by plain text. Logos and other unnecessary decoration can simply be left out, they add no information. If a document is held in a word processor then it can be output as text very easily.

Yours sincerely,

David Hansen

Scottish Government

1 Attachment

Dear Mr Hansen,

Thank you for your request for information under the Freedom of
Information (Scotland) Act 2002 (FOISA).  
 
We have provided, as confirmed with you via e-mail correspondence on 3
March 2009, plain text electronic copies of all information held by the
General Register Office for Scotland (GROS) on the 2 headings detailed
below: 

 
1.  what consideration has been given to having no census [and thus
saving me (and all other taxpayers) a lot of money. Having no census would
also mean not employing torturers to run the census, but that is not a
consideration in this request.]

2.  minutes of meetings, discussion papers and the like where the
question of whether to have a census or not was discussed and the reasons
for deciding to have one.

There are also links to documents published on the GROS and Office for
National Statistics (ONS) websites. The first relates to a paper titled
'The 2011 Census: A proposed Design for England and Wales, Discussion
paper', published by ONS in October 2003 and which provided a starting
point for GROS considerations into options for the 2011 Census. This
paper is available on the ONS website at
[1]http://www.statistics.gov.uk/downloads/t...

The remainder of the information is summarised in the bulleted list below,
with the information itself following in the order shown.

A.  'Cost-Benefit Analysis for 2001 Census' paper, discussed at the
2011 Census Programme Board on 24 August 2004
B. extract from the minutes of 24 August 2004, detailing the
discussion on the paper
C. 'A Statistical Strategy for the 2011 Census in Scotland' paper,
discussed at the 2011 Census Programme Board on 12 October 2004

D. extract from the minutes of 12 October 2004, detailing the
discussion on the paper
E. ministerial submission made on 19 November 2004 to the Deputy
Minister for Finance, Public Service Reform and Parliamentary Business

F. Autumn 2004 Census Consultation' document, referred to in
ministerial submission of 19 November 2004, available online at
[2]http://www.gro-scotland.gov.uk/files/con....
Results of that consultation are available online at
[3]http://www.gro-scotland.gov.uk/files/aut...

G. 'Improving Population Statistics In Scotland' position paper,
Annex C to the ministerial submission on 19 November 2004

H. extracts from the '2011 Census Business Case' from November
2006, relating to consideration of the census and alternative methods of
collecting the data gathered by the census; and

I. the '2011 Census Business Case' in pdf format (40 pages with
tables)

If you intend to print this email please select Landscape option.

The information supplied to you continues to be protected by the
Copyright, Designs and Patents Act 1988. You are free to use it for your
own purposes, including any non-commercial research you are doing and for
the purposes of news reporting. Any other re-use, for example commercial
publication, would require the permission of the copyright holder. Most
documents supplied by the General Register Office for Scotland will have
been produced by government officials and will be Crown Copyright. You
can find details on the arrangements for re-using Crown Copyright material
on the OPSI website ([4]http://www.opsi.gov.uk/click-use/index.htm).

Information you receive which is not subject to Crown Copyright continues
to be protected by the copyright of the person, or organisation, from
which the information originated. You must ensure that you gain their
permission before reproducing any third party (non Crown Copyright)
information.

If you are unhappy with this response to your request, you may ask us to
carry out an internal review, by writing to:

Mr Eddie Turnbull
Head of Corporate Services
General Register Office for Scotland
Ladywell House
Ladywell Road
Edinburgh
EH12 7TF

Your request should explain why you wish a review to be carried out, and
should be made within 40 working days of receipt of this letter, and we
will reply within 20 working days of receipt. If you are not satisfied
with the result of the review, you then have the right to make a formal
complaint to the Scottish Information Commissioner.

Yours sincerely,

Robert Brown
GROS Information Manager

A) Cost-Benefit Analysis for 2001 Census

Introduction

1. The 2001 Census was criticised on a number of grounds, including cost
(£32m). Therefore, we must be able to justify the cost of the Census.
This paper has a simple cost-benefit analysis. It concludes that the
Census was justified.

Does the board think the approach is correct and does it think further
work is necessary?

2. The justification is based on the difference between the mid years
population estimates by Health Board Areas (HBAs) - with and without the
2001 Census. The Census provides, at the most basic level, a count of the
population. Annual population estimates are produced in inter-Census
periods by modifying the Census count by ageing, adding in births,
subtracting deaths and adding or subtracting net migrants. However, these
estimates become less accurate over time.

3. This is also true for population projections which are based on
similar data and take into account life expectancy, number of women of
child-bearing age, family size, etc.

4. There are important implications for government expenditure since
budget distributions are always to some extent based on population
projection figures. Differences between the true population and that used
in budgetary allocation lead to a misallocation of resources (either over-
or under-provision).

5. This paper measures the potential misallocation in the NHS
Scotland and compares it to the cost of the Census. I made assumptions
about how the budget is distributed (described below) and then I compared
the population projections from 2002 to 2011, obtained from the:

· 2000 MYE (derived from the 1991 Census); and

1 2001 Census.

6. I used approximate figures for the 2004-5 NHS Scotland (NHSS)
budget. I used the NHSS budget figures for this exercise because:

* Demand for health services and so expenditure is closely linked to
basic demographics such as age and sex:
* The NHSS budget, at about one third, is one of the biggest in the
Scottish Executive budget.

Details

7. The total Health Department budget for the financial year 2004-05
is £6.656bn. I have used this figure instead of the larger Department
Expenditure Limit (£8bn). The difference between the two figures is due
to NI contributions. This goes straight back to the Treasury however it is
allocated.

8. Approximately 80% of the total NHSS budget is divided between
HBAs. This money is distributed using the Arbuthnott formula according to
4 key elements:

* The size of each HBA's population;
* The age and sex profile of each HBA's population;
* Needs arising from ill health (morbidity) and life circumstances
(such as deprivation, poverty and ethnicity) in each HBA's
population;
* Unavoidable excess costs of delivering healthcare in rural and
remote areas of Scotland.

9. The last three elements modify the results of the first element.
The first element rests on the Census and subsequent births, deaths and
migration. The second element also derives from the Census but is modified
according to needs of different age/sex groups. The third element is a
mixture of factors but many of them (e.g. incidence of long-term, limiting
illness) derive from the Census. The fourth element is not derived from
the Census.

10. Although the same indicators have been re-calculated using the
2001 Census data, they have only been available since March 2004. It will
therefore probably be some years before 2001 data is used in practice,
although the Arbuthnott Review did recommend that mid-year population
estimates should be used as an annual measure of the population.

11. Population and age/sex distributions for health spending are
mainly based on MYEs. They are annually updated and, since 2002, have used
2001 Census data. Therefore, I used these figures in the cost-benefit
analysis of the Census.

12. We also need the weighting that is assigned to age/sex population
figures. For each HBA, the relative spending factors give the cost of
providing medical services in that HBA relative to the average for
Scotland due to each Arbuthnott criterion. An overall relative cost factor
for the HBA is then calculated simply by multiplying them together.
Therefore the age/sex distribution has a weighting of 1/3rd.

13. Approximately 8% of the HBA budget is spent on `General Medical
Services' (GP funding). For this budget, the Community Health Index is
used to give the population by GP patient lists. This population is used
since demands on this service are dependent on GP lists rather than HBA
populations (in 1996 27,000 people were registered with a GP located in a
different health board area to the one in which they lived). Cost
benefits, therefore, cannot be applied to this portion of the NHSS budget.

14. The money at risk of misallocation is therefore:

[Proportion of NHSS budget allocated to HBAs] x

[1 - Proportion of HBA budget allocated to GMS] x

[Weighting assigned to age/sex data (assuming 1/3^rd because there are 3
factors)] x

[Total NHSS budget]

= 0.8 x 0.92 x 1/3 x £6.656bn = £1.635bn each year

15. I used this figure to calculate the overall misallocation from
2002 to 2011. Annex A contains the tables used to calculate the
misallocation (Please note that, due to rounding, some columns may not sum
exactly to the given total).

Table A1 1991 based HBA populations.
Table A2 2001 based HBA populations.
Table A3 HBA allocations on a 1991 population base.
Table A4 HBA allocation on a 2001 population base.
Table A5 Difference in allocation on the two bases.

Conclusion

16. The misallocation sums to £0 but, if presented on an absolute
basis, sums to between £21M and £31M each year and £284m over the
10-year period, almost 9 times the cost of the Census.

17. This analysis has not taken into account other major uses of
Census data for budget allocation such as the Local Authority GAE. We
could do further analysis if further justification was required. However,
we feel this analysis is conservative and does, of itself, justify the
Census.

18. ONS has drafted a more thorough preliminary 115 page business case
(see ONS CPB 2011 (04) 04 on the intranet). It concludes that the[ir]
sensitivity analysis based on the current quantified benefits and costs
estimates derived from the 2001 census suggests that the project [2011
Census] is economically viable. ... and also that a cautious approach has
been taken to estimating those benefits included.

Martin McNicoll
GROS
18^th August 2004

Annex A

Table A1: 1991-based MYEs projections 2001-2011

HBAs 2001 2002 2003 2004 2005 2006 2007 2008
2009 2010 2011
Argyll & Clyde 421,817 420,163 418,568 416,958 415,356 413,727 412,104
410,453 408,792 407,130 405,468
Ayrshire & Arran 372,261 371,056 369,950 368,845 367,698 366,527
365,326 364,114 362,885 361,633 360,368
Borders 107,038 107,095 107,105 107,096 107,069 107,019 106,947 106,862
106,770 106,659 106,544
Dumfries & Galloway 144,973 144,278 143,768 143,239 142,682 142,097
141,483 140,845 140,191 139,520 138,837
Fife 350,847 350,954 350,853 350,760 350,643 350,508 350,351 350,174
349,985 349,800 349,600
Forth Valley 278,629 279,156 279,709 280,260 280,793 281,318 281,826
282,327 282,837 283,333 283,831
Grampian 521,429 519,688 518,342 516,945 515,456 513,901 512,276
510,585 508,845 507,091 505,282
Greater Glasgow 902,377 900,156 898,121 896,337 894,747 893,334 892,076
890,997 890,043 889,223 888,520
Highland 208,610 208,480 208,275 208,037 207,744 207,407 207,035
206,633 206,188 205,732 205,250
Lanarkshire 561,900 561,666 561,540 561,386 561,190 560,941 560,640
560,274 559,879 559,421 558,939
Lothian 787,343 790,484 793,327 796,281 799,280 802,350 805,476 808,642
811,865 815,132 818,455
Orkney 19,391 19,290 19,186 19,083 18,972 18,855 18,727 18,598
18,475 18,348 18,210
Shetland 22,188 22,068 22,037 22,001 21,969 21,938 21,904
21,876 21,840 21,810 21,771
Tayside 383,021 380,651 378,647 376,607 374,529 372,406 370,255 368,070
365,866 363,650 361,423
Western Isles 26,846 26,528 26,266 26,002 25,733 25,456 25,182
24,901 24,605 24,316 24,021
Total 5,108,670 5,101,713 5,095,694 5,089,837
5,083,861 5,077,784 5,071,608 5,065,351
5,059,066 5,052,798 5,046,519
Table A2: 2001 based MYEs projections (2001-2011)

HBAs 2001 2002 2003 2004 2005 2006 2007 2008
2009 2010 2011
Argyll & Clyde 420,491 418,750 416,233 414,498 412,733 410,950 409,144
407,367 405,616 403,844 402,092
Ayrshire & Arran 368,149 367,060 365,372 363,961 362,495 361,004
359,467 357,892 356,322 354,736 353,139
Borders 106,764 107,400 107,548 107,794 108,006 108,201 108,363 108,515
108,649 108,775 108,895
Dumfries & Galloway 147,765 147,310 146,695 146,199 145,676 145,113
144,523 143,894 143,237 142,569 141,884
Fife 349,429 350,620 350,992 351,390 351,742 352,066 352,369 352,665
352,963 353,238 353,520
Forth Valley 279,480 279,370 279,947 280,675 281,374 282,052 282,715
283,354 283,996 284,622 285,251
Grampian 525,936 523,290 520,953 519,311 517,564 515,716 513,801
511,797 509,733 507,623 505,453
Greater Glasgow 867,150 866,080 863,919 859,926 856,197 852,718 849,489
846,550 843,844 841,436 839,275
Highland 208,914 208,140 207,820 207,541 207,211 206,828 206,409
205,952 205,463 204,954 204,405
Lanarkshire 552,819 552,910 551,942 551,422 550,813 550,127 549,400
548,622 547,827 547,006 546,160
Lothian 778,367 779,100 781,697 784,840 788,023 791,216 794,452 797,767
801,142 804,567 808,088
Orkney 19,245 19,210 19,066 18,972 18,872 18,772 18,655 18,536
18,417 18,299 18,165
Shetland 21,988 21,940 21,829 21,864 21,899 21,920 21,948
21,979 21,998 22,026 22,064
Tayside 389,012 387,420 385,363 383,800 382,197 380,568 378,887 377,186
375,475 373,776 372,062
Western Isles 26,502 26,200 25,890 25,631 25,368 25,113 24,858
24,598 24,333 24,065 23,795
Total 5,062,011 5,054,800 5,045,266 5,037,824
5,030,170 5,022,364 5,014,480 5,006,674
4,999,015 4,991,536 4,984,248
Table A3: HBA allocations on 1991-based MYEs projections (2001-2011)
(£1000s)

HBAs 2001 2002 2003 2004 2005 2006 2007 2008
2009 2010 2011
Argyll & Clyde 134,833 134,487
134,135 133,773 133,415
133,051 132,690 132,322
131,950 131,577 131,203
Ayrshire & Arran 118,992 118,769
118,555 118,336 118,107
117,872 117,629 117,383
117,132 116,873 116,609
Borders 34,214 34,279 34,323
34,360 34,391 34,416
34,435 34,450 34,463
34,470 34,476
Dumfries & Galloway 46,340 46,181
46,072 45,955 45,830
45,697 45,555 45,406
45,251 45,090 44,925
Fife 112,147 112,334 112,435
112,534 112,629 112,720
112,807 112,889 112,969
113,049 113,125
Forth Valley 89,063 89,353
89,636 89,916 90,193
90,469 90,743 91,017
91,294 91,568 91,843
Grampian 166,673 166,343
166,108 165,851 165,568
165,266 164,944 164,603
164,246 163,883 163,501
Greater Glasgow 288,442 288,125
287,813 287,572 287,399
287,289 287,233 287,240
287,289 287,381 287,511
Highland 66,682 66,731
66,744 66,744 66,729
66,700 66,662 66,614
66,554 66,489 66,416
Lanarkshire 179,610 179,780
179,952 180,109 180,258
180,394 180,517 180,622
180,718 180,795 180,864
Lothian 251,672 253,020 254,230
255,471 256,734 258,029
259,350 260,691 262,055
263,436 264,839
Orkney 6,198 6,174
6,148 6,122 6,094
6,064 6,030 5,996
5,963 5,930 5,892
Shetland 7,092 7,064
7,062 7,059 7,057
7,055 7,053 7,052
7,050 7,049 7,045
Tayside 122,432 121,840 121,342
120,827 120,301 119,763
119,216 118,659 118,095
117,525 116,951
Western Isles 8,581 8,491
8,417 8,342 8,266
8,186 8,108 8,028
7,942 7,858 7,773
Total 1,634,972 1,634,972 1,634,972
1,634,972 1,634,972 1,634,972
1,634,972 1,634,972 1,634,972
1,634,972 1,634,972
Table A4: HBA allocations on 2001-based MYEs projections (2001-2011)
(£1000s)

HBAs 2001 2002 2003 2004 2005 2006 2007 2008
2009 2010 2011
Argyll & Clyde 135,648 135,279
134,720 134,356 133,988
133,616 133,238 132,866
132,498 132,117 131,736
Ayrshire & Arran 118,762 118,580
118,258 117,975 117,679
117,377 117,061 116,730
116,396 116,051 115,698
Borders 34,441 34,696 34,809
34,941 35,063 35,180
35,289 35,393 35,491
35,586 35,677
Dumfries & Galloway 47,668 47,589
47,480 47,389 47,292
47,182 47,064 46,932
46,790 46,641 46,485
Fife 112,724 113,269 113,604
113,900 114,188 114,471
114,749 115,025 115,298
115,561 115,823
Forth Valley 90,158 90,251
90,609 90,979 91,344
91,706 92,066 92,418
92,770 93,114 93,456
Grampian 169,664 169,051
168,614 168,331 168,020
167,680 167,320 166,927
166,509 166,068 165,600
Greater Glasgow 279,737 279,790
279,620 278,738 277,952
277,253 276,637 276,110
275,649 275,274 274,969
Highland 67,394 67,240
67,264 67,273 67,268
67,248 67,217 67,173
67,116 67,050 66,968
Lanarkshire 178,336 178,620
178,644 178,739 178,813
178,868 178,913 178,938
178,952 178,952 178,936
Lothian 251,096 251,691 253,007
254,400 255,820 257,256
258,714 260,199 261,700
263,213 264,751
Orkney 6,208 6,206
6,171 6,150 6,127
6,104 6,075 6,046
6,016 5,986 5,951
Shetland 7,093 7,088
7,065 7,087 7,109
7,127 7,147 7,169
7,186 7,206 7,229
Tayside 125,493 125,157 124,728
124,406 124,075 123,738
123,385 123,023 122,652
122,280 121,897
Western Isles 8,549 8,464
8,380 8,308 8,235
8,165 8,095 8,023
7,949 7,873 7,796
Total 1,634,972 1,634,972 1,634,972
1,634,972 1,634,972 1,634,972
1,634,972 1,634,972 1,634,972
1,634,972 1,634,972

Table A5: Misallocations by 2001 and 1991-based projection differences
(2001-2011) (£1,000s)

HBAs 2001 2002 2003 2004 2005 2006 2007 2008
2009 2010 2011
Argyll & Clyde 815 792 585 584 573 565 548
544 548 540 533
Ayrshire & Arran 230 189 297 361 428 495
568 654 737 822 912
Borders 227 417 486 581 671 764 853 943
1,028 1,115 1,201
Dumfries & Galloway 1,328 1,408 1,408 1,434 1,461 1,485
1,509 1,527 1,539 1,551 1,560
Fife 576 935 1,169 1,366 1,559 1,750 1,942 2,135
2,330 2,512 2,698
Forth Valley 1,095 898 973 1,063 1,151 1,237 1,323
1,402 1,475 1,546 1,613
Grampian 2,990 2,707 2,505 2,479 2,452 2,414 2,376
2,324 2,263 2,185 2,099
Greater Glasgow 8,705 8,334 8,193 8,833 9,447 10,036 10,596
11,130 11,640 12,106 12,542
Highland 713 509 520 528 539 548 556
559 563 562 553
Lanarkshire 1,274 1,160 1,308 1,370 1,445 1,525 1,604
1,684 1,766 1,843 1,927
Lothian 576 1,329 1,223 1,071 914 773 635 492
355 223 88
Orkney 10 31 23 27 33 40 45 50 53
57 59
Shetland 1 24 3 28 53 72 95
116 136 157 184
Tayside 3,061 3,318 3,387 3,579 3,773 3,975 4,169 4,364
4,557 4,755 4,947
Western Isles 32 27 38 34 30 21 13 5
7 14 23
Total 21,633 22,079 22,117
23,340 24,529 25,701
26,833 27,928 28,996
29,988 30,937

B) Extract from minute of 2011 Census Programme Board held on 24 August
2004 which details the discussion on this matter.

16. Cost -Benefit Analysis for 2011 Census

16.1 The Chairman welcomed the paper and said that it delivered
what he had been looking for, justification of the Census based on
the difference between the mid year population estimates by HBA
with and without the 2001 Census. It was recognised that while
the method used was fairly primitive, it confirmed the
applicability to Scotland of the ONS approach.

16.2 In summary it was noted that the justification was reached
because the misallocation which had been estimated at between
£21m and £31m each year from 2002 to 2011 was 9 times the cost
of the Census. Peter Scrimgeour said that it was important to
note that funds misallocated were not automatically lost, only a
small percentage of that figure would be. It was agreed that
caveats be placed on the conclusion to reflect this.

Action: Ian Mate to amend conclusion to Cost Benefit Analysis
paper to ensure caveats are in place to reflect that only a small
percentage of misallocated funds are automatically lost.

C) A Statistical Strategy for the 2011 Census in Scotland

Introduction

1. This paper addresses the statistical framework for the 2011 Census in
Scotland.

2. The Censuses in England and Wales and Northern Ireland are the
responsibility of the Office for National Statistics and the Northern
Ireland Statistics and Research Agency respectively. However, there is a
need for consistent and comparable Census outputs across the UK.
Achieving this is a guiding principle of the GROS 2011 Census programme.

3. This document focuses on the key statistical aims and principles
and identifies the high priority areas of statistical research and
development over the coming years.

Context

4. The 2011 Census context will be different from previous censuses.
Technology and Society are changing rapidly. The design of the 2011 Census
must increasingly embrace flexibility and its rate of response to change
must become not only faster but more radical. It must also reflect the
lessons from 2001 and build on what went well then.

5. The response rate in the 1991 and 2001 UK Censuses fell compared
to 1981. Differential response, from which some areas and population
subgroups had response rates significantly lower than the national
average, increased. These trends also occur in surveys and election
turnouts.

6. We think that contacting and getting a response from some sectors
of society will be increasingly difficult. This is the key issue we face
when designing the 2011 Census.

7. People in Scotland are becoming more mobile and their living
patterns more complex. More people have more than one address. This
provides problems for both data collection and outputs. We must have a
clear definition of where people should be counted by the Census.

8. We must also decide whether outputs on two population bases are
required or sufficient for user needs.

9. Increased availability of administrative data and improved record
linkage will provide more comprehensive data and be used to supply some
census-type data more frequently. It may also be used to assess person and
item non-response and help adjust the census at an individual level.

10. The development of population and address registers may further
enhance our understanding of the population and we must take advantage of
any such developments.

11. At this stage, we have no final operational design for the 2011.
In the testing program we will try:

* Strengthened mixed and Postout/Postback enumeration,
* Using a quality assured comprehensive national address list;

12. By strengthening field methods to improve quality, we hope to
speed up processing, particularly editing and imputation.

13. The enumeration method we currently favour is delivery with mixed
collection - allowing postback and enumerator collection. We hope to
reduce face-to-face contact with respondents.

14. Population definitions will have to be as simple as possible. We
do not expect radical changes to the 2001 form design, except perhaps in
colour shading - even if the question content changes.

15. We are assessing a new role of Address Checker - to quality
assure the PAF and council address lists before the Census. The Address
checker will also revisit every property identified as vacant, second or
holiday home by enumerators and check its status.

Principles

16. GROS shares the Census objectives of ONS that: "The central
objective of the 2011 Census is to provide high quality population
statistics as required by key users ... on a consistent and comparable
basis for small areas and small population groups. Currently, these may be
expected to include counts of people, dwellings, households and families,
with a breakdown of key characteristics".

17. To achieve this, we must maximise response rates and minimise
differential response. If there is a conflict of priority, the impact on
response rates will be a key determining factor in making a decision. This
includes any post-enumeration survey or other coverage assessments
process; coverage adjustment is second best to getting the initial count
right.

18. At all times we will define, measure and manage data and process
quality and build more ways of assessing coverage and quality into the
collection operation itself through inclusion of additional questions.

Maximising Response

19. There are three key elements to tackling person non-response:

1 Identify households and communal establishments and get the
right forms to them;
2 Get completed forms back from those households and communal
establishments; and
3 have all the right people on the forms.

20. Therefore, it is essential that if we use a national address
register it must have high coverage and be quality assured to identify
where enumerators address list must be improved before the enumerator
enters the field.

21. Over the next few years, we must understand barriers to response
and address them through enumeration strategies, definitions,
questionnaire design, publicity and follow-up.

22. The explanation of who should be included on a questionnaire must
be clear and easily applied by all. The questionnaire design, content and
respondent burden must be acceptable to the public.

23. However, we must also tackle item non-response. In the 2006 Census
Test we will be assessing the cost of ensuring that Census forms are
complete for all questions both by the enumerator returning to the
household but also by making it clear to householders that enumerators
will return.

Address Registers

24. The development of a good national address register is fundamental
to potential improvements to the 2011 Census. However, we can not rely on
a national address register having 100% coverage.

25. Our address register will be built from two sources, the PAF and
from the Assessor's Portal that feeds from the 32 Local Authorities in
Scotland. In 2006, we will use Address Checkers to check and improve the
Address Register. This may change if confidence in the Address Register
improves.

26. However, even with an Address Register, the advantages of postback
are not clear. We plan to have enumerators for each area and, for
delivery, will reduce the delivery round to one attempt at contact. Our
evidence is that contacting a household at delivery has no effect on
response rate. The delivery round gives enumerators the activity which
teaches them about their area.

27. While a national address register may be an accurate list of
dwellings, we may need data at the household level, information on the
relationship between addresses and households.

28. Additionally, there is a requirement to ensure that the
appropriate questionnaires and accompanying information are sent to
addresses. Therefore, ultimately, the address register would have to
distinguish between communal establishments and private households. But we
also need to identify populations who need special information, such as
translation leaflets, large-print or Braille questionnaires.

29. Checking Address Registers will require a substantial amount of
fieldwork. It seems unlikely that many of the traditional roles of the
enumerator will be usurped. However, we can enhance their ability to help
and encourage where required by encouraging postback where no help is
required.

Targeted Enumeration

30. Even with a high quality address register and effective
enumeration strategies and publicity to maximise our ability to get
questionnaires to and from households, some sectors of society will not
respond.

31. There are a number of possible reasons for this, such as language
and literacy problems or the attitudes of some respondents. In order to
tackle this element of non-response, GROS has supported research to
identify hard-to-count groups and areas. We will develop initiatives and
procedures to improve their enumeration.

32. Research in this area will focus on who does not respond and seek
to interact with those types of people to develop enumeration strategies,
select the population definition for enumeration, improve questionnaire
design and develop foreign language material, leaflets and publicity
strategies.

33. This research is similar to that being conducted internationally
to tackle non-response to social surveys and censuses. The 2011 Census
work will be coordinated with this wider work.

Population Definitions

34. The changing nature of the population over time means that the
population bases for collection and output will be reviewed. However, even
though population mobility may change, the Census is not intrinsically a
radical approach and we need to know the limitations on population bases
that the census can successfully enumerate.

35. In 2001, the Census measured the resident, the work place, and the
school place population and hence the daytime population too. It seems
unlikely that we can step backwards from those populations but, if so, we
only could by measuring another population for which there was a clearer
demand.

Population Base for Enumeration

36. The 2001 Census enumerated fewer young men than population
estimates counted. International migration is hard to measure so the
Census results are difficult to QA. However, in England, there are
concerns that the Census `usually resident' definition was hard to
understand and let some people avoid being counted. Also, the increasing
numbers of people with multiple addresses means that usual residence is a
less relevant concept for a significant minority of people in some areas.

37. However, there a number of problems associated with the main
alternative base for enumeration - population present. They include:

* Loss of information about household structures and relationships;
* Difficulties transferring the population back to a usually resident
location (for outputs on that basis).

38. In our consultation programme, we will seek views on user needs
for population bases and for the location of specific populations such as
students etc. If the user needs are strong, we will test other population
bases in the 2006 Census test.

39. For some areas, such as those where the population varies a lot by
day of week or time of year, those two counts may not fully characterise
regular but not daily commuters. Therefore, we will determine user output
requirements for more than the current population bases. However design
and response rates factors will mitigate against more population bases.

Other Definitions

40. We will also consider requirements for outputs on households,
families, household spaces, household reference person, dwellings and
communal establishments. Definitions will be agreed where necessary

Coverage

41. It is inevitable that we will fail to get 100% coverage.
Therefore, to achieve our central objective - of providing high quality
statistics on a consistent and comparable basis - we must institute a
coverage assessment method which is sophisticated enough to estimate
differential response down to small areas and population sub-groups.

42. The original 'One Number Census' methodology was adapted to
estimate and adjust for coverage in 2001. We intend to strengthen the
enumeration method for assessing non-response and evaluate its success. In
2001, we relied on enumerator household classification. We must know if
this strategy is justified.

43. Dependency was a problem. The evidence suggests that, in 2001,
there were people who avoided both the Census and the CCS. Classically,
people missed by both have different characteristics from the population
missed by the Census but found by the CCS. Within-household dependency
may be the main problem or not - we have no measure of such dependency.

44. Therefore, we must develop new methods of quality assuring the
Census, probably by assessing individual level administrative data for
accuracy and coverage, and comparing the best to the Census and then
applying an adjustment.

45. We must also provide a measure of census over enumeration from the
2001 Census so we can decide whether we need an over enumeration strategy
in 2011. There is anecdotal evidence that children of divorced parents may
be included at both parents address, for example.

Outputs

46. The 2011 Census programme must decide whether to publish the
results more quickly than in 2001. There is a trade-off between time,
resources and quality. By strengthening enumeration and form completeness,
we may reduce the time spent fixing errors. With a complete address list
and a method to quality assured household status classification, household
imputation methods can be simplified. Also, with a system that monitors
all addresses, households and forms, our estimate of household coverage
and the resultant adjustment can be carried out efficiently and
effectively.

47. However, comprehensive adjustment for over enumeration, and
household and within-household adjustment for under enumeration conflict
with quick output. Perhaps a solution is to produce an unadjusted followed
by an adjusted database - though this strategy has disclosure control
problems.

48. Our aim is to provide flexible outputs that are freely available
over the internet. However, we are also committed to protecting the
confidentiality of personal data throughout all census processes, from
form collection to production of outputs.

49. Research will be jointly undertaken with ONS and NISRA into output
disclosure control measures to identify or develop procedures that
maximise both data protection and the utility of the information produced.

50. As well as physical confidentiality protection measures, we will
consider the public perceptions of confidentiality raised by the
approaches we use, to avoid damaging public confidence.

Quality

51. To achieve our aim of quality outputs, we will define, measure and
manage data and process quality and build more ways of assessing coverage
and quality into the collection operation itself.

52. We will produce a comprehensive quality strategy, which will
define the quality standards for each stage of the census operation, from
data collection, through processing to the final outputs. This document
will outline how we will measure quality and ensure quality standards are
met. Finally, our quality strategy will recommend how data quality is
reported to users.

53. Part of the quality work for 2011 will focus on meaningful
measures of data quality to accompany outputs. Local Authority level
confidence intervals are not helpful to users who wish to compare small
areas within an authority.

54. Levels of imputation are misleading if there is evidence that a
question was misinterpreted by some types of respondents. The 2011 Census
will provide appropriate measures of quality at the same time as the
outputs are released.

55. Another element of quality work is the quality assurance of the
final Census results. For 2001, Local Authority level data were compared
against mid-year population estimates and aggregate administrative data.
Whilst this highlighted differences, it did not provide an explanation of
the differences or evidence of how to adjust the census results.

56. For 2011, we will consider not only how to develop this approach,
given developments in population estimates and administrative data, but
also approaches used overseas. For example, New Zealand Census uses field
information to produce early provisional estimates that are compared with
previous population estimates. Where they find large differences, they
work in partnership with local experts to attempt to explain those
differences prior to publication of the census results.

57. This approach would allow any necessary adjustments to be made to
the census results prior to publication should problems with the census
operation be identified.

58. Alternatively, if no adjustments are necessary, there is increased
time to understand and document differences prior to publication of
results. This increases public confidence in results that may at first
glance appear unusual.

Developments in Administrative Data

59. Parallel developments in the access to, use and linkage of
administrative data have the potential to contribute much to the 2011
Census exercise. Under the proposed model, access to accurate
administrative data has the potential to:

* Support linking data to addresses to improve our knowledge of the
population in particular areas and addresses;
* Reduce the need for non-response follow-up by using sources of
information about second homes for example;
* Replace or add to census topics, by linking administrative data to
those collected during the enumeration;
* Improve the accuracy of the imputation of missing or inconsistent
census responses;
* Improve our understanding of under coverage, by providing independent
sources to aid better understanding of the characteristics of
non-respondents;
* Improve our understanding of over coverage, through improved
understanding of people with multiple addresses;
* Improve the accuracy of missing person and household imputation; and
* Improve the quality assurance of the census results, through the
availability of independent sources of information against which to
compare both aggregate data and the distributions of characteristics
within the data.

60. The responsibility for taking forward administrative data research
rests within this Division. However, other areas within SE are also
developing and using administrative data sets. So we must ensure that
Census and alternative Census work are integrated so that Census needs are
met and we can take full advantage of developments as they occur.

61. In addition, we may need to expand existing research for our own
purposes, for example supplementing information held on an address
register or coverage assessment, and therefore an understanding of the
issues, and techniques surrounding record linkage, is critical.

Conclusion

62. The 2011 Census design contains some areas of innovation. It is
essential that key changes are tested adequately and proved to be viable.
Between now and the 2006 Census Test, a programme of research and testing
will be undertaken.

D) Extract from the minutes of the 2011 Census Programme Board meeting of
12 October 2004 detailing the discussion on the paper.

7. A Statistical Strategy for the 2011 Census in Scotland

7.1 Ian Mate introduced his paper and said that it had been an
ONS document originally that he had scottified and shortened.

7.2 The main points to note were that admin data should reduce
the burden of the form filler by providing an alternative source
of information that would remove the need to ask the question on
the census form. Admin data would also provide the means to QA
the census results and that eventually we would be leading towards
a triple system estimate (Census, Follow Up, Admin data) to look
at under-enumeration within households. This would probably
require an individual level database. The Chairman noted that a
lot of work was required before we get to that stage although a
number of initiatives were being looked at by other organisations
such as the DWP, but that there was no guarantee that these would
come to fruition.

7.3 Ian Mate said that the next thing to consider would be the
publication timetable. Paragraph 47 of the paper states that
adjustments for both over and under-enumeration conflict with
being able to publish data quickly and that the possible solution
would be to release 2 sets of figures (both One Number Census
figures in that the lower area data adds up to the national
totals), firstly 'straight from the field' interim figures through
an unadjusted database and then adjusted data, taking account of
admin data, to a timetable that would put us under less pressure
than in 2001.

7.4 In discussion it was agreed that users would need to be
consulted to say that we would be hard pushed to produce results
in 2011 as quickly as we did in 2001 given all the adjustments
that will be necessary and that the choices at this point would be
for them to accept a later publication date with unadjusted
interim data or with no unadjusted interim data. This would
enable users to determine what the unadjusted interim data might
comprise, what uses could be made of it and indeed what should
happen to it when the adjusted data was released.

7.5 Ian Mate said that the consultation event in November
would focus on the population base and that while this paper is
not altogether appropriate for that event, it would be amalgamated
into the main paper for that event and therefore users would be
asked to consider the possibility of 2 sets of data for 2011.

7.6 Peter Scrimgeour thought that in order to make a choice
users would want to know what sort of delays might be expected
before they would receive the adjusted data and that we are not in
a position to do so. The Chairman replied that we could if we
said that in 2001 we thought we got it right and the only possible
correction is with Glasgow City Council (7,000) but in ONS there
were greater errors (1 million (2%)) that we would risk incurring
in 2011 should we go with anything other than a single population
base. This point could also be made at the UK Harmonisation
Committee on 21 October.

7.7 Joe Fuchs said that we should try to reduce the burden on
the form filler and that this paper did not make it clear if that
was our intention. Peter Scrimgeour replied that the paper did
say that admin sources would be used to reduce the number of
questions on the form but agreed that it could be made clearer in
the principles. It was agreed that a paragraph should be added to
stress that while it was desirable to reduce the burden of form
fillers by shortening the number of questions on the form, that
this would always have to weighed up against the business needs
for questions, although the use of admin data should help to
shorten the form.

Action: Ian Mate to add a paragraph to the Statistical strategy
for 2011 paper to stress that while it was desirable to reduce the
burden of form fillers by shortening the number of questions on
the form, that this would always have to weighed up against the
business needs for questions, although the use of admin data
should help to shorten the form.

7.8 Frank Thomas said that users would no doubt want
cross-tabulations of data gathered from the Census and admin data
and that because this would require much detailed work on linkage
at household/individual level, there may be an argument that
having a bigger census would be easier and more accurate. Peter
Scrimgeour said that the linkage may not necessarily be at the
personal level but at small area level.

7.9 Peter Scrimgeour made a comment about the paper being a
bit thin on dissemination (para 48), and thought that we could
expand it to say more about the increased flexibility of
dissemination methods, pointing towards GROS being able to tailor
products/dissemination to suit individual needs in 2011.

Action: Ian Mate to expand para 48 of the Statistical strategy for
2011 paper to say more about the increased flexibility of
dissemination methods, pointing towards GROS being able to tailor
products/dissemination to suit individual needs in 2011.

E) A Ministerial submission was made on 19 November 2004 to the Deputy
Minister for Finance, Public Service Reform and Parliamentary Business

Deputy Minister for Finance, Public Service Reform and Parliamentary
Business

FUTURE PLANS FOR THE CENSUS

Purpose and timing

1. To outline our future plans for the Census, and check you are happy
with them. Routine.

The Issue

2. Traditionally, we hold Censuses decennially. They provide
uniquely detailed information, beloved of users, particularly at small
area level. We have already started to prepare for the 2011 Census -
because one of the lessons from 2001 was that we did not start early
enough last time. Our first action on the ground would be a Census Test
in spring 2006, which is not far off

3. Involvement of users is crucial. We have already launched a
web-based consultation on the questions which we should test in 2006. We
have arranged 3 consultation events with users to look more widely at the
preparations for 2006, and the strategy for 2011. A summary of the
consultation paper prepared for the meeting is at Annex A and the full
version is at Annex B. Subject to your views, we would like to publish
that document more widely, using our website.

4. Though the Census is well-regarded and gives unparalleled
information, it has shortcomings. Our vision is to develop other sources
of information which would give demographic data of sufficient quality,
much more frequently and potentially more cost effectively. Annex C
explains this in more detail. I would like to make that vision available
to users and on our website, to promote debate on its pros and cons - in
the context of our preparations for the 2011 Census, which demonstrate
that we will not abandon the Census before a satisfactory alternative is
in place.

5. The Census is managed by the 3 Registrars General (for England &
Wales, Scotland and Northern Ireland). We keep in close touch to get
economies of scale and meet the needs of UK-wide users. In England &
Wales, ONS have already made public their intention to move down the Annex
C route; the Northern Irish are less well-advanced but have the same
general plan in mind.

Presentation

6. At this stage in the Census cycle, when the issue is not very
exciting, I suggest a low key approach - that I issue a news release
simply announcing that the consultation document (a variant of Annex B) is
on our website, and inviting responses. I think that it would make sense
to put a variant of Annex C on the website at the same time, as a
background paper. The Press Office agrees.

7. The Order and Regulations which authorise the Census need to be
approved by the Parliament. Though that is a long way off (2009/10,
probably), I would be keen to engage parliamentarians much sooner. If you
agree that general approach, I will make specific proposals.

Recommendation

8. I recommend that you approve me putting variants of Annexes B and
C on the GROS website, announcing the fact by a low-key news release which
would invite comments.

Duncan Macniven

Registrar General

19 November 2004

GROS

Ladywell House

F) Autumn 2004 Census Consultation' document
([5]http://www.gro-scotland.gov.uk/files/con...).

Consultation results
([6]http://www.gro-scotland.gov.uk/files/aut...)

G) Annex C to Ministerial submission of 19 November 2004

IMPROVING POPULATION STATISTICS IN SCOTLAND

A Position Paper by the Registrar General

Introduction

1. This paper describes ways to improve population statistics in Scotland
in line with user needs - including initial plans for the 2011 Census
and the potential to replace the traditional Census.

Background

2. Under the Census Act 1920, the Registrar General for Scotland is
responsible for taking Censuses and for collecting and publishing
population statistics in the inter-Census periods, as well as for
furthering the supply and improving the co-ordination of population
statistics.

3. "Population statistics" includes information about, and estimates
and projections of:-

* The total number of people - with key characteristics such as
their age, sex and where they live;
* The structure of the population - households, families;
* Key characteristics of people, households and families - notably
their social and economic condition, their health and welfare
and their housing.

4. These population statistics are produced for the whole of
Scotland, for local authority and health board areas and for much smaller
areas (down to about 50 households). "Neighbourhood" statistics for the
smallest areas are important to allow users to identify how circumstances
are changing at the local level (for example in Scotland's inner cities,
country towns or remote rural areas).

5. The Census has traditionally been the main source of population
statistics. They are derived also from other surveys (such as the Labour
Force Survey run by the Office for National Statistics and the Scottish
Household Survey run by the Scottish Executive) and from administrative
records (data which are primarily collected for administrative purposes -
for instance, the NHS patient list which is important for estimating
migration).

6. The conduct of the Census is devolved, and most of the other
sources on which the present population statistics system relies are also
collected for devolved purposes. So final decisions on the future of the
Census in Scotland will be taken by the Scottish Parliament. It will
however be necessary to take account of any United Nations or European
Union requirements. Many people, particularly academic and commercial
users, are interested in looking across the whole UK - and some
administrative data which could potentially improve the system is
collected for reserved purposes. So it is important not to look at the
Scottish picture in isolation

The Census and its limitations in meeting user needs

7. The main source of population statistics is the Census. While a
compulsory Census can be held every 5 years, Censuses have recently been
taken decennially. The 2001 Census was an impressive achievement. It was
completed on time and within budget. The response rate is estimated at
96.1% of the population and, with the help of a follow-up survey, it was
possible to estimate results for the remaining 3.9%. So, for the first
time, the Census results covered the entire population of Scotland (rather
than those who filled in a Census form) and are believed to be the most
reliable obtained by any Scottish Census. The results were published on
the expected timescale, with a report for the whole of Scotland in
September 2002 and the main statistical breakdown in March 2003 -
supplemented by a website which allows users to obtain the data which suit
their needs.

8. Despites its strengths, the Census has limitations in meeting
future needs:-

* it is relatively costly - £32 million over the last 10 year
cycle.
* Although by the standards of statistical sources it is highly
accurate, it is inevitably not perfect: some people will have
filled in the forms incorrectly or incompletely and there are
unavoidable inaccuracies in the process of estimating results
for people who did not fill in a form.
* Census-taking is becoming steadily more difficult, particularly
because people lead much more mobile life-styles with more than
one address and family (not necessarily in the same country).
* Increasingly, Census users are need information more frequently
than once every 10 years and for smaller areas and groups of the
population (for example, to assess the impact of government
policies quickly, ensuring that public money is spent to the
maximum effect).

9. Internationally, these trends are leading to the adoption of
alternative approaches to Census-taking. Many, if not the majority, of
the 55 most statistically- advanced countries in the world are
investigating these alternative approaches. For example, the USA is
moving to a continuous population survey, France has implemented a rolling
Census and Norway is relying entirely on register-based administrative
data. In England and Wales, an integrated population statistics system is
proposed - linking administrative and survey data with the output of the
2011 Census, to produce all subsequent population statistics without a
further major Census. In Scotland, the current policy is to prepare for a
2011 Census while developing possible alternatives.

Aim of Census alternatives

10. The aim in considering alternatives to the Census is to find a
system which:-

* Would produce the range of statistics which users need (although
not necessarily exactly the same statistics as at present);
* Allow a more frequent publication of small area statistics, instead
of basing most statistics on a decennial Census;
* Open the way to faster publication of statistics after the date to
which they relate;
* Bring together in a consistent way, demographic data from a wider
range of sources than the present Census;
* Reduce the form-filling burden on the individual citizen;
* Improve cost-effectiveness, compared with the present Census.

What work is in hand

11. This section describes work which is already in hand, which will
help improve the production of population statistics.

Conventional Census

12. Planning has already started for a conventional Census in 2011 and
consultation with Census users has started. [Web reference.] Although
census-taking is likely to be more difficult in 2011 than in 2001, a
conventional Census would produce accurate results, acceptable to users.

Definitive Address List

13. The Scottish Executive and the General Register Office for
Scotland are working with local authorities in the Definitive National
Address project, which is designed to build (and keep up to date) a
complete address list from administrative data available to local
authorities - such as planning and building control information about the
construction of new properties and the demolition of old. The pace of
progress towards that goal varies from local authority to local authority
but the aim is to have a complete address gazetteer by March 2005.

Population Register

14. In parallel with the Definitive National Address project, steps
are being taken to create a population register. With the help of funding
from the Scottish Executive, Scottish local authorities are compiling a
"Citizen's Account", containing information on everyone with a customer
relationship with the relevant local authority, in order to improve public
service delivery.

15. The Registrar General for England and Wales is carrying out
detailed development work on a UK population register - the Citizen
Information Project - which was announced in January 2004. The project
aims to build a population register for use by public services across the
UK, bringing together basic information (name, date of birth, address, and
other contact information) which is already held by Government departments
such as the UK Passport Service, the Department for Work and Pensions, the
Inland Revenue, and the Driver and Vehicle Licensing Agency. The register
would be consistent with legislation covering data protection and privacy
and would act as the index to existing records held by different
Government departments. The development stage of the project will finish
in mid-2005, when Ministers will decide whether or not to go ahead with
the register.

Survey information

16. More data is becoming available from Government-funded social
surveys, covering subjects similar to the Census, cover 95,000 households
(4% of all Scottish households) and 55,000 individuals (1% of total
Scottish population). Existing surveys of households and individuals in
Scotland are not systematically coordinated and their representativeness
could be improved. The Scottish Executive and the Office for National
Statistics (who commission or carry out most of these surveys) are
discussing the scope for a single "continuous population survey",
replacing many of the existing surveys, which might be introduced in
parallel with a similar planned survey in England and Wales in 2007

Administrative Data

17. The use of information technology has made it easier to derive
aggregate statistical data from administrative records - such as the
linkage between benefits and employment records, announced by the
Department for Work and Pensions in December 2003.

What more might be done?

18. This section considers how the developments which are already in
hand could be supplemented to improve the production of population
statistics.

19. The task of carrying out a conventional Census would be simplified
by a definitive and accurately-categorised address list - because a large
temporary workforce is employed at Census time to identify properties
where people live, and inevitably some addresses are missed. A definitive
address list could give substantial savings and improve accuracy by making
it easier to target resources on areas which are difficult to enumerate
(e.g. newly-built houses, areas with transient population such as
students). But a definitive address list - updated to a stated quality
standard - could also allow a radical change to Census-taking. It would
allow the number of households in Scotland to be counted
administratively. And it would give a sampling frame which would make it
easier to be sure that other surveys gave representative results.

20. A comprehensive population register could provide statistics on
the total number of people in Scotland and their key characteristics.
Like the definitive address list, it could act as an efficient sampling
frame, allowing supplementary information to be obtained by a
representative survey of the population (instead of a universal Census).

21. A better sampling frame - from a definitive address list and/or a
population register - would allow survey data to contribute much more
powerfully to population statistics. If some of the resources hitherto
devoted to the Census were diverted to increasing the sample size of a
continuous population survey, and if the sampling frame ensured that the
results were representative, it might be possible to produce small area
statistics which are more frequent, and sufficient to meet the data needs
currently supplied by the Census. Such an approach might also make it
easier to customise for Scottish circumstances the information obtained by
the present GB or UK-wide surveys, without losing important cross-border
comparisons.

22. An increased part could also be played by administrative data.
Data collected primarily for administrative, rather than statistical,
purposes can have an important part to play. For example, information
about car ownership could be derived from the Driver and Vehicle Licensing
Agency.

Conclusion

23. A combination of administrative and survey data sources could
theoretically provide, at small area level capable of aggregation to give
national information, the data from almost all aspects of the Census. The
General Register Office for Scotland is working with other government
departments and local authorities to develop the potential of a system
which would meet the aims at paragraph 10 above.

24. For England and Wales, the Office for National Statistics has
reached a similar conclusion in its discussion paper "Proposals for an
Integrated Population Statistics System" [insert link]. The General
Register Office for Scotland will keep in touch with developments south of
the border, taking into account the interest of users who find it helpful
to make cross-border comparisons.

Creating a new system

25. A new system will not be easy to create. Key steps will be
further work on:-

* a definitive address list (in co-operation with local
authorities in the Scottish Executive) - which is essential for
a new system and very desirable for a conventional Census;
* the local authority customer database (in co-operation with
local authorities and the Scottish Executive), and the Citizen
Information Project (in co-operation with the Registrar General
for England and Wales);
* the scope and coverage for a continuous population survey (with
the Scottish Executive and the Office for National Statistics);
* administrative data which may have an important contribution to
population statistics in Scotland (with other key Government
departments).

26. In preparing for an alternative population statistics system, 3
important questions must also be answered. The first is whether the new
system can produce good-quality data, sufficient to meet the needs of
users. Users have become accustomed to the extremely high quality of data
available from the Census. The objective of the new system would be to
provide better data - in particular, more frequent data for small areas.
More attention will be needed on quantifying migration (both within
Scotland and between Scotland and elsewhere). The coverage of
hard-to-enumerate groups such as rough sleepers, travellers and other
groups with no permanent address must be carefully tested. Users may be
willing in the interests of cost effectiveness to accept slightly less
accurate data (a wider confidence interval), or drop the requirements for
little-used data (eg central heating). Key users of the Census - local
authorities, the health service, central government and academic interests
- will be invited to consultative meetings about these data quality
issues, and views of other users will be actively sought, for instance
through the Scottish Census Results On Line website.

27. Second, a careful check will be made on the privacy implications.
The independent statutory responsibility of the Registrar General to
safeguard the confidentiality of personal Census information - for
instance through strict controls to avoid any individual being identified,
and keeping the records secret for a 100 years - has the confidence of the
public. The retention of that confidence is vitally important for the new
system. So the public must be absolutely sure that, when administrative
or survey data provided for one purpose is to be used in place of the
Census as a key part of a new population statistics system for Scotland,
Census-standard confidentiality continues to apply. Wide public debate
will be encouraged and there will be detailed discussion with Government
departments holding the key data and with the Information Commissioners
who enforce data protection and freedom of information legislation.

28. Third, the costs and benefits of any change will need to be
carefully weighed. Initially, costs would be slightly higher because it
would be prudent to run the development of the new system in parallel with
the early stages of preparation for a conventional Census. There would
however be significant savings in the period when a Census would otherwise
be held.

29. Before a final decision can be taken on whether or not to replace
the present Census-based system:-

* The components of the new system need to be put in place and
tested;
* The 3 pre-requisites - data quality, privacy and cost effectiveness
- need to be concluded.

30. The Registrar General and his staff, in co-operation with others,
are working energetically on the development of the new system. At the
time of writing, it seems unlikely that a new system will be securely in
place in time to remove the need for a conventional Census in 2011
(although it might need to gather less information than in the past). But
that decision does not need to be taken until much later in the decade.
The Registrar General will continue to prepare for a 2011 Census in the
way described in paragraph 13, in parallel with the work to create a new
system. That will ensure that, nearer the time when it is necessary to
launch the 2011 Census (probably in 2008 or 2009), the right information
is available to allow an informed decision to be taken.

General Register Office for Scotland

November 2004

H) Extract from 2011 Census Business Case

2011 Census Business Case

1. Executive Summary

1.1 Although there has been a Census held in Scotland every ten years
since 1801 (except in 1941), the development of many other collections of
data about the population, and the relatively high cost of a census, mean
that the case for a Census in 2011 needs to be established.

1.2 This paper first sets out the strategic case for a Census in 2011,
and then considers a number of options for delivering it.

1.3 Following the introduction (Chapter 2), Chapter 3 describes the
many uses to which Census data is put; explains how the data produced by
the Census could not be provided from other sources - at least at present;
describes some of the particular benefits that the comprehensive nature of
Census data offer; and shows how the conduct of a Census every 10 years
helps the Government to meet domestic policy requirements and certain
international obligations. Key arguments include:

* Many of the other sources of data are not consistently collected,
and it would be very time consuming, and therefore costly, to seek
to replicate the data which can be collected by means of the Census
from these other sources. The task of cross-matching and
validating the data would be substantial. This would be an
inefficient use of Government resources;
* Because the Census uniquely covers every member of the public, the
data can be used for multivariate analysis at very local levels,
particularly important for targeting resources and policy
initiatives at deprived and needy groups;
* The Census data provides the basis for allocating resources to
local authorities and NHS Boards. Failing to carry out a Census in
2011 would lead to potential misallocation of resources. The
estimated amount of such misallocation is such that it is estimated
that one year's misallocation would cost more than the total cost
of the Census;
* The Government has obligations to provide data about the population
to the European Union and other international organisations. The
Census currently provides the only comprehensive source of that
data.

1. Background and general case for a Census

3.1 Since the Census was first conducted in 1801, the uses to which
the data collected have been put have become more sophisticated. Despite
the enormously increased demand for information about population
characteristics in recent years, no satisfactory alternative source of
such basic data has been developed yet. The Census remains the only
comprehensive consistent source of national and local data about people
and their key characteristics, and about housing.

3.2 The output of the Census is used extensively by national and local
government; by the NHS; and by the private and voluntary sectors. It
provides not only statistics on the basic characteristics of population
units - age and sex - and on the make up of households and families, but
also provides the opportunity for more sophisticated multivariate
analysis. The ability to analyse data covering a wide range of variables
at small area or postcode level is particularly valued. Currently the
Census provides the only comprehensive source of such data.

3.3 An indication of the perceived value of the Census is provided by
the number of times Census data have been accessed by various agencies.
In the first year following publication of the 2001 Census results (on 13
February 2003), the SCROL (Scottish Census Results On Line) website
recorded over 157,000 visits A wide range of different sectors seek
information from the Census database. For example, by November 2003,
there had been 15,535 visits by Central and local government
organisations, more than 45,000 by commercial organisations, and nearly
6,000 by academic organisations. GROS Customer Services unit received
2,273 requests for data in the year following publication. These requests
ranged from the straightforward reproduction of standard outputs to
complex non-standard items, such as a breakdown of the Census data by
Church of Scotland parish. GROS continues to deal with a very high level
of requests for access to 2001 Census data, via both telephone calls to
the customer services unit and the SCROL website.

3.4 This indicates a strong demand and a prima facie case for the kind
of data the Census provides. But the 10 yearly Census is an expensive
exercise, and it is important to consider whether the same outcome could
be achieved in different ways. The 2001 Census cost £34.2million, split
roughly evenly between printing and preparation (including testing);
enumeration; and coding and analysis. Allowing for inflation, a similar
Census in 2011 would cost about £43million spread over a number of years.

3.5 When the Census was first conducted, Government played a smaller
part in the lives of individuals than it does today. Annex 1 gives more
information about the social history of Scotland, and the way in which the
Census has reflected that. As the welfare state has developed, Government
agencies have had to collect and retain more data about the people for
whom services are provided. These various data sets contain at least
some, and arguably much, of the information collected through the Census.
If that information could be collated and made consistent, it would be
possible in theory to do without a separate Census every ten years.
Indeed, such an approach could provide a regularly updated or even
continuous flow of information, rather than the current 10 yearly
snapshot.

3.6 However, the data collected by the various Government agencies are
not collected on a consistent basis; there are known inaccuracies in some
of the data sets (for example, more people are registered with the NHS
than the population of Scotland); and it is not currently permissible to
exchange certain data between Government departments. Allowing such
exchanges would raise data protection issues. Furthermore, before any
data from such an approach could be used it would be necessary to
establish a clear baseline position so that the information arising from
the new system could be compared with past Census data. The only
practicable way of establishing a baseline for such a new approach would
be through a traditional Census carried out with that end in view. So
using Government administrative records to provide a "rolling census" is
an aspiration for the future, but is not practical as a replacement for
the proposed 2011 Census.

3.7 The population in each of the local authorities and NHS Board
areas is the key criterion in determining how much expenditure needs to
take place in their respective areas. The Census information enables the
GROS to produce mid-year population estimates and population forecasts
which are used for a wide range of purposes. The mid year estimates are a
key factor in allocating public expenditure among local and health
authorities. For example, the Arbuthnott Committee's report, which
determined a new, more sophisticated formula for allocating resources to
NHS Boards, considered a range of options for assessing the population in
each Board area, and concluded after careful analysis that the GROS mid
year population estimates were the best option.

3.8 But the Census information provides more than simply a reliable
population count. The detailed breakdowns that are made possible by the
Census allow authorities and central government to achieve a number of
further desirable objectives:

* Targeting resources at areas of greatest need for certain services
* Planning for future population and demographic change
* Understanding the demographic effects of policy developments
* Identification of small areas with particular needs
* Meeting parliamentary, public and media demand for population
related information
* Helping business to plan for population change to improve its
competitiveness
* Meeting international obligations.

3.9 Government needs an effective means to allocate and target
resources, particularly in the context of the Efficient Government agenda,
under which every opportunity must be taken to provide services more
efficiently. In 2006-07, local authorities in Scotland are expected to
spend some £9billion in revenue expenditure, of which some £8.3billion
comes from central government allocations (including non domestic rate
income). Scottish NHS Boards are expected to spend some £9.3billion,
virtually all of which is allocated by central government. Between them
they account for more than two thirds of Scottish Executive managed public
expenditure. A reliable source of population data is essential if
allocations are to reflect need.

3.10 GROS has conducted an exercise to estimate how much resource would
have been misallocated between NHS Boards over the ten year period from
2001 to 2011 if no Census had been carried out in 2001, and population
estimates had been extrapolated from the 1991 figures. Details are in
Annex 2, but overall, the misallocations would have comprised a total
rising from £21million to £31million a year, with some Boards
consistently receiving more, and others consistently less than their
appropriate share. Though the misallocated money would not be misspent,
it would be spent sub-optimally and the cumulative effect of this would be
very significant. It is reasonable to assume that the position for local
authorities would be similar. This indicates that the total cost of the
Census would be less than the total amount of one year's misallocation to
the NHS and local authorities.

3.11 As well as avoiding misallocation of resources at the local
authority and NHS Board level, the Census information enables government
at all levels to identify trends in demographic change, not only within
Scotland as a whole, but also within smaller areas. That is an aid to
policy making, allowing for more finely tuned policies targeted more
appropriately. In particular, the Census is the only nationally
consistent source of data relating to areas smaller than local
authorities, even down to post code level. This allows the identification
of pockets of deprivation and narrowly defined areas in need of special or
more intensive services, important in pursuing the Government's equality
agenda. It also facilitates the identification of small population
sub-groups, such as ethnic minorities, in a sufficiently specific way to
enable their needs to be targeted effectively. Similarly, the business
sector benefits from more accurate population and demographic data, which
feeds in to their research and development of new products to meet the
needs and demands of Scots in the future.

3.12 Increasingly, everything that Government does is subject to
scrutiny in Parliament, by the media, and by members of the public, all of
whom look to Government to provide accurate and timely information. While
expressing concern about the need to protect individuals' data from
unauthorised and inappropriate use, they nevertheless assume that
comprehensive information, particularly on matters like population
numbers, will be readily available. The Census assists in meeting
Government's obligations for this; and the fact that the Census is well
established and has a good record in keeping sensitive data confidential,
while producing useful information, helps to alleviate any concerns about
data protection. At present, any alternative means of providing
equivalent information would be likely to raise concerns about the
confidentiality of sensitive data.

3.13 Scotland has both intra-UK and international obligations to
provide certain demographic data. Census information is used in policy
development and implementation by UK Government departments with
responsibilities that cover Scotland. Censuses traditionally take place
simultaneously in Scotland, England and Wales, and Northern Ireland. If
any one or two of the Censuses did not take place, not only would there be
a need to find other ways of providing the necessary data, but there would
be a significant risk of inconsistencies between the data from different
parts of the UK. That in turn could result in distortions of policy,
disadvantaging one or more parts of the UK. Even if Censuses were carried
out in all parts of the UK a similar, although lesser, risk could arise if
the questions asked were greatly different in the different Censuses.
That is why the Registrars General within the UK and the three Census
Departments work closely together to ensure consistency wherever
appropriate, while allowing for differences between the countries where
necessary. For example, there has been a perceived need for different
data about ethnicity in the English Census than in the Scottish and
Northern Irish Censuses; the secondary school examination systems, which
lead to formal qualifications, differ; and the Scottish Census includes
specific questions on the Gaelic language.

3.14 The statistics generated from Censuses are also used by the
European Commission to allocate funding to member states. Although there
is not currently a legal requirement to provide this information, the UK
(like other member states) has agreed to do so, and the agreement states
that:

"...only population and housing censuses conducted at regular
intervals.... will permit the collection of periodical and reliable
statistical data, at different geographical levels, on the
population....".

3.15 The strategic case for a Census is strong. It supports policy
making, development and implementation; it ensures more accurate
allocation of resources in line with needs; it provides useful information
for the business sector in designing and developing new products and
markets; and it helps Scotland to meet its obligations both within the UK
and internationally. While in theory much of the information emerging
from the Census could be derived from other sources, in practice there are
obstacles to that approach which could not be resolved before 2011 when
the next Census would fall due. The absence of a Census would be likely
to result in an increasing level of misallocation of resources, less
available support for policy making and implementation, and difficulty in
meeting international obligations.

4.3 Option 1a

4.3.1 A Census similar to that carried out in 2001 would be a
traditional Census taking place on a single day and seeking to enumerate
every household in Scotland. The intention would be to have a 3 page per
person form, including questions covering basic information about the
number of people in each household and their characteristics in terms of
age and sex as well as other data such as ethnic origin, religion, type of
accommodation and the facilities available in the house.

4.3.2 This option has a number of benefits. Following the model of
previous Censuses would ensure a high degree of continuity, make for
easier planning and enable comparisons to be drawn and trends to be
detected reasonably straightforwardly. Furthermore, since the Censuses in
England and Wales and Northern Ireland are likely to be on similar lines,
it would enable UK wide data to be collated. A full Census ensures that
consistent data is available across virtually the whole population, which
allows analysis to be carried out at postcode level, a facility which is
much valued.

4.3.3 However, there are some potential disadvantages. In order to
enumerate the whole population on a single day, the Census requires
substantial expenditure to cover, among other things, a large number of
field staff who are employed around the date of the Census. This
expenditure is not spread evenly over many years but is inevitably
concentrated at various points in the development of the Census work.
This means that there is a premium on sound financial planning and
forecasting and an early commitment to budget. Second, a form comprising
three pages of questions will appear onerous to some of those being asked
to complete it. Obtaining a substantial enough return rate of completed
forms is a major task, and one that appears to be becoming harder in the
current climate. Third, although most of the population is relatively easy
to contact there is a minority which is harder to locate.

4.4 Option 1b

4.4.1 An enhanced Census would expand the number of questions, resulting
in a form of about four pages per person instead of three. In preparation
for the Census, Government Departments are asked to indicate what
questions they would like included. This option would allow more of those
requests to be met. Most of the other benefits associated with Option 1a
would continue to apply - there would still be continuity with earlier
Censuses, and an ability to collate data across the UK where appropriate.
Data would still be available at postcode level.

4.4.2 But the disadvantages would also apply, and in some cases would be
greater. The cost would still be concentrated at various points in the
process. But those costs would be greater. Not only would there be the
extra cost of an extra page (which could increase printing costs by up to
20%-25% and also processing costs), but the costs of analysing the data
emerging would also be commensurately higher.

4.4.3 Furthermore a longer form would be likely to meet more resistance
from those asked to complete it, with the potential for either reducing
the comprehensive nature of the information, or requiring additional
expenditure on follow up work by enumerators or other means to chase up
those forms not completed.

4.5 Option 1c

4.5.1 A mini Census would seek to collect only the data which is
required for a basic head count. It would be a much shorter form asking
for name, age, gender ethnicity and household relationships. Printing,
processing and analysis costs would be less than with a full Census,
although the costs associated with preparing for the Census, getting
Parliamentary approval, issuing and collecting the form would still be as
large. The benefit of this approach, apart from the cost savings, would
be that people asked to complete the form would find it a less onerous
task.

4.5.2 However, adopting a mini Census approach would mean that the
information collected would be very limited. There would be many areas
where comparable figures across the UK would no longer be possible; and
the value of the information to researchers, policy makers and others
would be severely constrained. The consequence would be likely to be that
they would seek other, less comprehensive sources of information which
could have an adverse effect on the reliability of their plans and might
lead to the reduction in the burden on the public being partly offset by
the institution of new sample surveys and other data collection exercises
as well as by less optimal allocation of public expenditure. There would
also be a big increase in the costs of data matching work required to pull
these other sources of information together.

4.5.3 Annex 4 describes the steps which GROS is taking to test public
reaction to questions which might appear in the Census. It also discusses
the questions which might be in a mini Census, the range of topics in a
traditional full Census, and some of the questions that might be included
in an enhanced Census.

4.6 Coverage of the population

4.6.1 The five main coverage options are:

* No Census
* A sample Census
* A full Census carried out on traditional lines
* A Census comprising a short set of questions directed at the whole
population, and a longer set answered by a sample
* A Census based solely on other data sources.

4.6.2 For the reasons set out earlier, failing to carry out any Census
would lead to costly gaps in the information available to public and other
authorities, and would put Scotland at risk of failing to meet
international obligations; and a Census based on other sources of data is
not practicable at present. These options are therefore not considered in
detail here, although Annex 3 sets out their characteristics.

4.6.3 The benefits and drawbacks of a traditional full Census are
discussed in paragraphs 4.3.1 to 4.3.3. This section considers further
the case for:

* a sample Census (Option 2a); and
* a part sample, part full Census (Option 2b).

4.7 Option 2a

4.7.1 A sample Census would seek to enumerate only a proportion of the
population and extrapolate national results from the sample. Sampling is
a widely used technique to obtain information without contacting everyone
in the population of interest. The printing and enumeration costs would
be less than for a full Census.

4.7.2 However, while for some questions a sample response might provide
a reasonable approximation for policy makers and others to work with,
conducting only a sample Census would lose the comprehensive picture that
Censuses provide, and thus reduce substantially the quality of the
information. In particular, it is likely that the response rate from
groups of the population that are difficult to enumerate (people who have
a mobile or unsettled lifestyle) could fall to such a low level that
effective statistical analysis of these groups would become impossible.

4.7.3 People in these groups are among those who benefit most from
government initiatives informed by the Census and their characteristics
may differ significantly from those of the mainstream population, thus
rendering extrapolation from the achieved sample dangerous. Furthermore,
it would not be possible to compare accurately results from a sample
Census with those from earlier Censuses; there would be difficulties in
collating information across the UK; and the only comprehensive and
reliable source of data at a level below that of the local authority would
be lost.

4.7.4 Although some costs would be reduced by this approach, others
would increase. The need to identify a representative sample for the
Census would require investment; and the cost of analysing the data would
be increased because of the need to identify how far the results can be
extrapolated to the population as a whole.

4.8 Option 2b

4.8.1 A variation of the sample Census would involve a mini Census form
distributed to the whole population, supported by a fuller form directed
at a sample. This would avoid some of the drawbacks associated with the
mini Census in that the information deficit would be reduced; and, unlike
the solely sample Census, there would still be comprehensive data on the
basic headcount variables. But this approach would share many of the
drawbacks of both those options.

4.8.2 Furthermore, much of the cost saving which would be associated
with either of those two options would be lost - the enumeration savings
associated with the sample Census would no longer be available, and the
savings on analysis costs of the mini Census would also be lost. Indeed,
costs could actually increase as it would be necessary to print and
process two distinct types of Census form.

4.9 Data collection

4.9.1 Under any of the options, there are various available methods of
collecting the data. The use of enumerators to ensure optimal delivery
and collection of forms is a well established feature of the Census. That
involves recruiting a team of people who will distribute and then collect
the forms from people, and, if necessary help them to complete the forms.

4.9.2 In 1991 and earlier the approach was to expect enumerators to be
the prime means of distributing and collecting the forms. In 2001 more
use was made of the postal service for form collection. However,
enumerators still played a key role in chasing up those forms which were
not returned promptly.

4.9.3 With major advances in technology since then, further options are
now possible. Many people are used to providing information - even
sensitive information - over the internet, or over the telephone. Allowing
people to complete the Census in these ways could be attractive to some.
The main benefit would be the potential for a better response rate if
people found these approaches more convenient.

4.9.4 On the other hand, it is arguable that those who find it most
difficult to complete the paper based form, or who are reluctant to do so
for whatever reason, would also be least likely to be prepared or able to
use alternative means to complete it. So the benefit in response rate
might be limited. It would not be possible to rely solely on the web or
on the telephone, since some people do not have access to these
facilities, and some would continue to have doubts, however unfounded,
about the security of providing data in this way.

4.9.5 So there would be significant additional costs in providing the
different kinds of form, and in integrating the various completion modes.
It would also be necessary to ensure that any such alternative options had
adequate security processes, to protect the confidentiality of the Census.
In addition there would be additional costs in recycling duplicate paper
and internet forms for some individuals and households at the processing
stage.

4.9.6 The Government has a policy of encouraging the availability of
electronic means of interaction between Government and people where
possible and appropriate. Society nowadays expects that online completion
is available for most services and this expectation is likely to be
greater in 2011.

4.9.7 There is therefore a presumption that a web based option for
completing the Census should be offered. This will need to be offered as
an alternative to, rather than instead of, a paper based form. Apart from
the need to establish a help line for people having difficulty in
completing the form, and perhaps to provide for telephone based form
completion for people with certain disabilities, for example those with
sight problems, wider use of the telephone would not be worth pursuing,
because any small additional convenience to the public would be outweighed
by the substantial additional cost, and the potential loss of
confidentiality.

4.9.8 There will undoubtedly remain an important role for the
enumerators. While the forms could be issued by post, and people can be
asked to return them by the same method, enumerators will still be
required to deliver forms to households missed by the post and to chase up
forms which are not returned. Furthermore, there are various groups in
society - those who have no permanent residence, and those who change
their residence frequently for example - who are difficult to trace to get
them to complete a form. In addition many kinds of "communal
establishments" e.g. hospitals, care homes, prisons need special attention
to ensure that the population living there are enumerated properly. By
making more use of the postal service, more of the enumerators' time can
be freed up to deal with these more difficult matters.

4.10 Data processing and statistical analysis

4.10.1 The main options for data processing - data capture and coding -
are in-house processing and outsourcing.  The data have to be coded in
such a way that they can be readily used, without breaching
confidentiality, for the various purposes desired by Government and other
agencies - a process which is obviously most efficiently and securely done
once, under GROS's control.

4.10.2 The intention for 2011 (like 2001) is to outsource this work under
a UK-wide contract and the procurement process is already well-advanced.
Partnership with the chosen contractor will enable efficiencies and
quality improvements to be explored.

4.10.3 Once the data are coded, processed etc. they are available for
statistical analysis. This task is carried out mainly by GROS itself,
although certain other users may be permitted to access part of the data
for analysis under strict conditions put in place to ensure the
maintenance of confidentiality and the integrity of the information
generated. In each Census, GROS seeks to improve the quality of the
output by developing more sophisticated analyses and output arrangements.

4.10.4 If it did not do so, other agencies would not only be disappointed
but would seek to achieve the same end themselves, probably in
inconsistent ways, and resulting in a cost shift from GROS elsewhere,
higher cost to the public purse overall, and inefficiency in the use of
public resources. But there is a balance to be struck between the
sophistication and extent of the analysis and the associated costs.

4.11 Outsourcing

4.11.1 GROS needs to manage the whole process of preparing, conducting
and coding and analysing the data arising from the Census. But it does
not make sense for all the work to be done in house. Because the Census
is a ten yearly event, some of the work associated with it requires
additional staff for relatively short periods. For example, a large
number of enumerators is required to help conduct the Census. GROS plans
to outsource work where that is appropriate and where it provides best
value for money. It would be inappropriate to outsource the core work on
Census planning - the staff of GROS are experienced, and work on the
Census is an integral part of their duties.

4.11.2 But GROS is already testing the market for discrete elements of
the work - such as printing, the distribution of forms, publicity and data
capture, and the contact centre to deal with public questions about
completing the form. This avoids the need for GROS to employ staff
directly to cover some of these functions, but allows control to remain
firmly in the Department.

4.12 Charging

4.12.1 GROS recognises the need to consider the scope for charging for
the information it provides as a result of the Census. Much of the
information cannot be charged for - it is provided to other Government
agencies (where to charge would simply be to cost shift). Charges have
previously been levied for some paper publications, but now publication is
increasingly on the web. Indeed because the Census provides such basic
information about the nation, which has been provided by the public, it
would be inappropriate to charge for access to it.

4.12.2 However, where specific analyses or data sets which would not
otherwise be produced by GROS are requested, a charge is made. GROS
proposes to continue with this policy, treating each such request on a
case by case basis. The Department will seek to maximise income within
these policy constraints.

6. Conclusion

6.1 This paper has sought to demonstrate the case for holding a Census
in 2011. The arguments are strong:

* There is no other way that could be developed in time to produce
the same range of data to the same level of accuracy as a Census;
* The UK has obligations - both internationally and to various users
in the UK - to make information available which can currently only
be provided by the Census;
* The data produced from the Census uniquely allows for multivariate
analysis of data at a local level, which in turn facilitates
targeted policy making focussed on population subsets with
identifiable special needs;
* The cost of the Census is justified by the identified misallocation
of resources to local authorities and NHS Boards which would be
likely to occur if population numbers (and their characteristics)
were estimated by some other means;
* England and Wales and Northern Ireland will be conducting Censuses
in 2011. In the interests of consistency and continuity across the
UK and over time a Scottish Census is also desirable.

Annex 2 - Cost-benefit anaysis for 2011 Census ( provided at A in this
document).
Annex 3 - Options

1. Annex 3 - Options

1.1 Introduction

9.1.1 This Annex describes briefly a wider range of options in relation
to the Census, beyond those which are assessed more fully in paper. For
each option, there is a short description, an indication of the costs that
would arise in comparison with the base case of a traditional Census, the
benefits of the option, and the drawbacks. The options are:

* No Census
* Simple head count
* Mini Census
* Traditional Census
* Sample Census
* Part full, part sample Census
* Enhanced (4 Page) Census
* Ambitious Census
* On-line Census
* Telephone Census
* `Census' based on administrative data
* Rolling Census.

1.2 No Census

Description: No Census held in 2011. Any data required obtained by
extrapolation
from 2001 data
Costs: £0
Benefits: No requirement for additional staff and work associated
with conduct of a Census.
Drawbacks: Increasingly inaccurate population data; Inconsistent
comparisons with England; Failure to meet international obligations; Loss
of detailed information to support policy making.
1.3 Simple head count

Description: A very basic population count on a fixed date, providing
information on numbers sex and age only.
Costs: would be the same as for a full Census. There would be
additional costs for policy makers and others in finding other sources for
the data not provided by the Census. Savings on printing since the form
would be very short. Savings since there would be very little data to
code and analyse but the cost of setting up the operations for the
processing and coding is the same regardless of number so savings are not
likely to be so great. But the costs of preparation and enumeration would
be the same as for a full Census. There would be additional costs for
policy makers and others in finding other sources for the data not
provided by the Census.
Benefits: Financial savings. Likely better response rate for a
simple form. Ability to meet international obligations.
Drawbacks: Information of interest and use to policy makers and
others not available, leading to inconsistency.
1.4 Mini Census

Description: Short Census, asking fewer questions than in 2001, but
more than in the simple head count.
Costs: As for the simple head count, there would be savings on
printing and coding and analysis (although less than for the simple head
count), but the substantial costs of preparation and enumeration would
remain.
Benefits: As for simple head count, and some of the additional
information desired by users would be provided.
Drawbacks: Much of the information of interest and use to policy
makers and other users would not be available, leading to inconsistency
and inefficient duplication of work to obtain data.
1.5 Traditional Census

Description: A Census about the same length as in 2001 (about 3 pages
of questions), providing a range of information; distributed to every
household, and with enumerators taking the lead in collection.
Costs: Total cost about the same in real terms as in 2001 - about
£43million in today's prices, spread over about 7 years but peaking in
2010-2012.
Benefits: Continuity with earlier Censuses; availability of wide
range of data at postcode level; ability fully to meet international
obligations; many people will recall earlier Censuses, and their
familiarity should help with return rates.
Drawbacks: Three pages remains a substantial amount of form filling
which may reduce return rate; some desirable questions could not be
asked.
1.6 Sample Census

Description: A Census of whatever length sent only to a representative
sample of the population.
Costs: Some savings on printing, distribution and collection; but
additional costs for selecting the representative sample, and for
extrapolating the sample data to give a nationwide picture.
Benefits: Need to contact only a proportion of the population,
saving enumerators' time. Possible ability to gather data relating to
more areas of interest.
Drawbacks: Extrapolated data would be less accurate than those
gathered from the whole population. Ability to analyse at post code level
would be lost.

1.7 Part Full, Part Sample Census

Description: A basic head count Census for the whole population, with a
wider range of questions sent to a sample.
Costs: Small savings on printing and distribution costs; but
additional costs for selecting the sample and for extrapolating sample
data.
Benefits: Full headcount would mean that all international
obligations were met; sample survey would allow more areas of interest to
be covered.
Drawbacks: No ability to analyse sample data at postcode level, and
extrapolated data would be less accurate than those gathered from the
whole population.
1.8 Enhanced (4 page) Census

Description: As the traditional Census but with an extra page of
questions.
Costs: Additional printing, preparation, coding and analysis
costs.
Benefits: Ability to gather data on a wider range of matters.
Drawbacks: A longer form might encounter more resistance from those
asked to complete it.
1.9 Ambitious Census

Description: As the enhanced Census, but not limiting the size of the
Census to 4 pages, enabling all questions potential users want to be
asked.
Costs: Additional printing, preparation, coding and analysis
costs.
Benefits: Data gathered for a much wider range of matters;
possibility of replacing some other data collection exercises.
Drawbacks: Likely to be a very long form with potential major
implications for response levels; major additional task to ensure no
duplication with other data collection exercises.
1.10 On-line Census

Description: Any size of Census to be completed purely on line rather
than on a paper form.
Costs: Extra costs in designing and testing an on-line form;
savings in collection costs; savings in coding and analysis costs.
Compared with a traditional Census there would be more costs before the
Census is conducted and fewer afterwards.
Benefits: Checking and editing could be done automatically at the
point of data entry, improving speed and accuracy; analysis could be
started more quickly; in line with Government's policy to encourage
electronic access to services etc.
Drawbacks: Significant proportion of population still not on line;
many users may not be comfortable with using the technology; need new
procedures to detect mistakes; need to provide reassurance about security
of data etc.
1.11 Telephone Census

Description: Conduct the Census by phone.
Costs: Additional costs for establishing and staffing telephone
service centre (one call to complete a census form is approx. 40 minutes -
very costly) to conduct the Census; savings on printing, distribution and
collection.
Benefits: Many people now use the phone to conduct business; quicker
results; consistency of completion of data.
Drawbacks: Not everyone has a phone; difficulty of confirming
identity of respondent on the phone; need to provide reassurance about
confidentiality of data; possible data quality implications particularly
for some questions; cost.
1.12 Figures based on administrative data

Description: Use other sources of administrative data to provide the
information collected by the Census.
Costs: Large costs to identify the best sources of the data, to
ensure it is accurate and consistent, and to create links between
different data sources. Savings on preparation, printing distributing
sand collecting Census forms.
Benefits: Makes efficient use of many existing data sources; avoids
having to contact the whole population every 10 years.
Drawbacks: Not currently a practical proposition - would probably
require data sharing legislation; would need a baseline position to
calibrate the other data sources, which would require a Census type
survey; data confidentiality issues would need to be resolved, since data
collected for one purpose would be used for another. Data matching would
also be difficult to carry out and not consistent.
1.13 Rolling Census

Description: Instead of holding one Census every ten years, take a
Census of say one tenth of the population every year.
Costs: Overall higher than for a single Census, but costs would
be constant each year rather than peaking every ten years.
Benefits: Maintain expertise among enumeration etc staff, instead of
having to retrain every ten years; provide more dynamic data
collection.
Drawbacks: Major methodology change would be expensive and likely
cause delay; the population does not remain static for ten years, so
mobile people could be more likely to be missed leading to an incomplete
picture; unlikely to be able to produce as useful data at postcode level.

I) PDF of full 2011 Census Business Case

<<business case.pdf>>

show quoted sections

Richard Taylor left an annotation ()

The attached PDF contains a better formatted version of the information than has been possible in the plain text.