The Innovation Panel (IP) is similar to the Understanding Society survey having a stratified and geographically clustered sample design. The IP sample covers England, Scotland and Wales but unlike the Understanding Society survey areas north of the Caledonian Canal and Northern Ireland are excluded.
In Wave 1 postcode sectors from the Postcode Address File (PAF) were ordered by Government Office Region, the percentage of household heads classified as National Statistics Socio-Economic Classification (NS-SEC) categories 1 and 2 (non-manual) and population density. A systematic random sample of 120 sectors was drawn with probability proportional to population size. Within each selected sector, 23 addresses were selected by systematic random sampling, making a total sample of 2,760 addresses from 120 sectors. The final stage of sampling was carried out in the field by interviewers. Additional detail can be found in the working papers on the sample design and weighting strategy for Understanding Society.
All household members aged 16 years or older were eligible for the adult interviews. As with the main survey fieldwork design, all persons resident at the address, including children, were defined as original sample members (OSM) to be followed throughout the life of the study. In contrast to the main general population sample, the IP does not attach absent household members living in institutional accommodation to the IP sampled households. This introduces some degree of coverage error since these persons do not otherwise have an independent chance of selection through the Postcode Address File. Despite these differences the IP has many similarities to the overall sample design of Understanding Society.
An additional 960 addresses, eight in each of the original 120 Primary Sampling Units (PSU)s, were added as the refreshment sample. Addresses were selected using systematic random sampling from amongst addresses not already selected for the IP. If a household at one of the added addresses responded, members were defined as Original Sample members, from Wave 4 onwards.
An additional 1,560 new addresses were added as a further refreshment sample. This includes an initial refreshment sample of 1,080 households (9 in each of the 120 original PSUs) and a boost refreshment sample of 480 households (4 in each PSU). The boost sample was added midway through fieldwork due to the low response rate achieved on the initial refreshment sample. If a household at one of the added addresses responded, members were defined as Original Sample members, from Wave 7 onwards.
An additional 960 new addresses, 8 in each of the original 120 PSUs, were added as a further refreshment sample. In addition, a further 455 reserve refreshment sample households were issued in July 2017. Addresses were selected in the same way as the Wave 4 refreshment sample, and members of respondent households defined as Original Sample members, from Wave 10 onwards.
An additional 1,680 new addresses, 14 in each of the original 120 PSUs were added as a further refreshment sample. In addition, a further 832 reserve refreshment sample households were issued in September 2018. Addresses were selected in the same way as the Wave 4 refreshment sample, and members of respondent households defined as Original Sample members, from Wave 11 onwards.
An additional 6,047 addresses were added as a further refreshment sample. Unlike previous refreshment samples, this was recruited by web only. Addresses were selected from the Postcode Address File (PAF), clustered in 32 areas. These addresses were then contacted by post and invited to complete the survey online.
All IP13 and IP14 participants who were in a living apart together (LAT) relationship, and those partners for whom they had provided contact details, were invited to take part in a web survey between March and May 2022. Information on the sample, contact attempts and outcome is in n_latsamp_ip. For partners who took part, a pidp_sosm was created for them in the data file of their web survey n_latsurvey_ip. Note this file is organized by the sample members’ pidp.
The sample design is described by three variables, indicating sampling stratum, primary sampling unit and design weight.
Sampling stratum and primary sampling unit (psu) are identified by w_strata and w_psu respectively. They are on individual level enumeration and response files, as well as w_hhsamp_ip. The individual level enumeration files are w_indall_ip. The individual level response files are w_indresp_ip and w_youth_ip.
The person-level design weight is a_psnenip_xd and it is on a_indall_ip, a_indresp_ip and a_youth_ip. The household level design weight is a_hhdenip_xd. It is on a_hhresp_ip and a_hhsamp_ip.
