Sampling

The Population Sampled

The sample is designed to be representative of the adult population aged 15+ of Great Britain. The survey population is estimated to have been approximately 52,748,000 according to JICPOP’s mid 2017 projections based on 2011 census data.

Overview of the PAMCo Sample

PAMCo has a multi-stage disproportional pre-selected sample design.

The three stages of sampling are as follows:

  1. Selection of sampling points
  2. Selection of addresses
  3. Selection of individual

The sampling frame for the selection of sampling points comprises all Output Areas (OAs) in Great Britain. In 2017, two OAs were selected for a sampling point (prior to 2008 only one was selected).

The sampling frame for the selection of addresses is the Postal Address File (PAF).

Interviewers were issued with the pre-selected addresses to contact. At each household contacted they enumerated eligible individuals aged 15+. The final stage of the sampling was then to select which of these eligible individuals should be interviewed. One individual was selected in most households, and up to two individuals in larger households. The rules as to how this was done are described later in this document.

Some disproportional sampling was used at the stage of selecting sampling points in order to over-sample certain ACORN areas particularly likely to contain individuals with AB social grades.

Selection of Sampling Points

The first stage of sampling is to select the sample points (or locations) where interviews will be carried out. Before this is done, certain ACORN areas are upweighted (and others downweighted). This is done so as to oversample areas where individuals with AB social grades are particularly likely to live, and therefore increase the number of interviews taken with AB participants.

The scheme used to do this is revised periodically. It was revised in 2014 so as to use the 2013 ACORN types. It was revised again for 2016 to amend the number of ACORN type categories from five to seven.

The steps taken to select the sampling points were therefore as follows:

  1. Sampling weights were applied to the adult populations of OAs by 7 categories of ACORN Types, as follows:
    Category ACORN Types Weights Applied
    1 1, 2, 3, 9, 14 1.90
    2 4, 5, 7, 8 1.65
    3 6, 10, 15, 16, 17, 18 1.35
    4 11, 12, 13, 19, 20, 21, 22, 24, 26, 29, 61, 00 1.10
    5 23, 25, 27, 28, 30, 31, 32, 33, 34, 36, 37, 50 1.00
    6 35, 38, 39, 40, 41, 42, 43, 49, 53, 54, 55, 60, 62 0.73
    7 44, 45, 46, 47, 48, 51, 52, 56, 57, 58, 59 0.63
  2. Great Britain was divided into 60 area groupings, devised for PAMCo. Ultimately these areas were individually weighted and used to piece together the Government Office Region and non-overlap ISBA TV regions. A full list of the 60 areas can be found overleaf. For the purposes of sampling, the area groupings were further sub-divided into district groupings.
  3. The weighted adult populations of each of the 60 area groupings were calculated. The target number of sampling points for each area grouping was then set. The total number of points at this stage was 2,568 per annum.
  4. Within each area grouping, OAs were ranked in descending order by PAMCo district grouping, then within district by PAMCo ACORN category, and finally within PAMCo ACORN category by NS-SEC. Prior to 2005, a system of AB scores was used instead of NS-SEC as the final means of ranking.
  5. The required number of OAs within each area grouping as specified at b) was then selected with probability proportional to weighted adult populations, using random start/fixed sampling interval procedures.
  6. Any selected OA falling into a ‘no-go’ area (as separately defined) was replaced with one in the sample District and same ACORN Type with population size closest to the originally selected OA.
  7. Modified procedures were applied in Scotland in areas 49, 50, 51, 52, 56, 57, 58 and 59. In these areas the target number of points was increased, relative to other areas, to the numbers shown in the table below.

Sampling Areas

The area groupings used and the target number of Sampling Points selected were as follows:

England and Wales

Area Groupings Government Office Region ISBA Area No. of Points Selected
1 South West South West 72
2 South West Southern 36
3 South East Southern 72
4 South East Southern 36
5 South East Southern 48
6 South East Southern 84
7 South East London 72
8 East of England London 84
9 Greater London London 48
10 Greater London London 96
11 Greater London London 72
12 Greater London London 72
13 East of England East of England 60
14 East Midlands East of England 24
15 South East London 36
16 East of England East of England 36
17 East of England East of England 60
18 South East Midlands 48
19 South West Midlands 24
20 West Midlands Midlands 72
21 West Midlands Midlands 48
22 Greater London London 48
23 West Midlands Midlands 24
24 West Midlands Midlands 24
25 West Midlands Midlands 48
26 East Midlands Midlands 36
27 East Midlands Midlands 36
28 East Midlands Midlands 36
29 Wales Wales and the West 12
30 Wales Wales and the West 36
31 Wales Wales and the West 48
32 North West Border 24
33 South West Wales and the West 24
34 South West Wales and the West 60
35 Yorkshire and the Humber Yorkshire 84
36 Yorkshire and the Humber Yorkshire 48
37 Yorkshire and the Humber Yorkshire 36
38 East Midlands Yorkshire 48
39 Wales Lancashire 12
40 North West Lancashire 60
41 North West Lancashire 60
42 North West Lancashire 24
43 North West Lancashire 48
44 North West Lancashire 48
45 Yorkshire and the Humber North East 24
46 North East North East 48
47 North East North East 24
48 North East North East 24

Scotland

Area Groupings Area Names No. of Points Selected
49 Aberdeen 24
50 Dundee 12
51 Edinburgh 36
52 Strathclyde, urban 96
53 Strathclyde, rural 24
54 Lothian 12
55 Central 12
56 Fife 24
57 Other Tayside 24
58 Other Grampian 24
59 Highland 24
60 Scottish Borders 12

Allocation to Month, Day of Interview and Questionnaire Version

The selected sampling points were systematically allocated to month and then start day, to ensure a balanced daily interviewing schedule. Interviewers were given ten days to complete a standard assignment, starting on the specified day.

In addition, each sampling point was assigned a code to determine the rotation order to be used to ask about the various newspapers and magazines during that particular assignment. For details of the rotation orders, see document entitled ‘The Interview’.

Selection of Addresses

Within each selected pair of OAs, 26 or 27 standard and 12 replacement addresses were selected, apart from London where 29 standard addresses were selected. There were further variations in the numbers of addresses issued as follows:

In London, the standard number of addresses issued was reduced if there were indications that a high proportion of multi-household/tenement addresses might be found.

Interviewers were allowed to use the replacement addresses issued to them only when they identified that a standard address was a:

  • Business address
  • Addresses not located, empty or demolished
  • Non-main residence

Replacement addresses were randomly selected across the selected OA and interspersed with the standard addresses.

Interviewers were instructed to use replacement addresses which were nearest, on the list of sampled addresses, to the original address to be replaced, to ensure randomness. The address letter of the original address was transferred to the replacement address used. The reasons for replacement were recorded for office inspection.

A minimum of five calls was prescribed for each potentially productive address, before attempts to interview could be abandoned. There was no restriction on the total number of calls. Interviewers were encouraged to make as many calls as possible, varying them by time of day.

Selection of Individuals

The selection of individuals was done by undertaking the following steps:

  1. All selected addresses within each selected OA were listed and the letters P, Q, R and S were consecutively allocated to them (these letters were not allocated to replacement addresses).
  2. A Contact Questionnaire for each address (not replacement addresses) was prepared specifying the address letter.
  3. For all issued addresses the Contact Questionnaire specified the one or two persons to be interviewed, as follows:

    Address Letter

    No. of adults at address P Q R S
    1 First First First First
    2 First Second First Second
    3 Second First, Third Second, Third First
    4 First, Fourth Second, Third First, Fourth Second, Third
    5 Second, Fourth First, Fifth First, Third Second, Fifth
    6+ First, Youngest Third, Fifth Second, Fifth Fourth, Youngest
  4. There were separate sheets listing replacement addresses.Interviewers were instructed to list all persons aged 15+ in a selected household, in descending order of age and to select for interview the person(s) who accorded with the specification given (that is according to the total number of adults at the address and the letter allocated to that address).At multi-household addresses, tenements (in Scotland) and residential institutions, there were special selection procedures, details of which are available on application to Ipsos MORI.
Download Information in PDF