The British Population Survey (BPS) was a regularly run survey with monthly face-to-face historical data from 2008 to 2015. It aimed to capture the socio-economic and consumer characteristics of the population of Great Britain. Data includes 6000-8000 records a month covering demographics, economics, shopping preferences, durables, media, and internet use.

The BPS is conducted on in the homes of all respondents. Interviews are conducted via Computer Assisted Personal Interview (CAPI).

Content

The following variables are present in the data:

• Family: Gender, Age Group, Numeric age, Lifestage, Ethnic Origin, Marital Status, Parent of children, Parental Status, Child Maintenance, Number in household, Presence of children in household, No. of children in household, Age of children in household
• Geography: Country, Standard Region 4, Standard Region 11, Urban/Rural, Postcode Area (City), Lower Super Output Area.
• Economics: Social Grade, Qualification level, Working status of respondent, household income, chief income earner (CIE), working status of CIE, Home tenure, main shopper, main supermarket, debit card/s, credit card/s.
• Media: daily newspaper, Sunday newspaper, ITV station most watched.
• Durables: no. of cars in household, TV, Satellite TV, Cable TV, Freeview, Freesat, Landline telephone, simple mobile phone, web mobile phone, video, DVD recorder, DVD player, personal computer, laptop PC, tablet PC, games console, MP3, DAB radio, DIG camera (ex phone).
• Internet access: internet access – frequency, internet access – method, cable broadband, ADSL broadband, other broadband, non broadband, internet access – history.
• Internet use: emails, info-requests, info-products, purchases – not groceries, grocery shopping, bank a/c & finances, job search, play games online, online gaming for money, download music, download movies, download/stream TV, online dating, VOIP, social networks/blogs, other.
• Date: Year and month.
• Survey: ID, Weight.

Note that the 2015 BPS dataset has changed significantly in terms of variables included, particularly regarding durables and Internet behaviour. It now includes 145 (compared to 153) variables. However, since e.g. ethnicity is now combined into 1 variable (instead of 17), a direct variable comparison is not possible.

Representation, Quality and Bias

Samples are based on the postcode district level, using Geodemographic models for half the sample, while the other half is sampled in under-weighted profiles to increase the probability of representative selections.

The team of Interviewers are given quotas for Gender, Age, Working Status and Social Grade according to the Census statistics. The final process is to ensure, via the interview process, that no respondent is interviewed twice, over time. This methodology tries to ensure the sampling of an accurate cross-section of the British Population, and as the same methodology is used every week, it tries to ensure that trends will be equally accurate over time.

To reduce final bias, the survey includes a weighting system (specifically, by means of the Rim Weighting method). The weights are based on the Census mid-year estimates, and checks against other available population profiles such as Age, Gender, Region, Home Tenure, and Social Grade.

The dataset is quite complete, although caution should be exercised as there are a number of chain and follow-up questions in the survey which are not always applicable, hence coded as missing values (e.g. “NA”, “NULL”). However, there is a “Not Asked” code for e.g. online shopping when the individual was previously replied with no access to the internet. Furthermore, some questions have various answers that are not always usable, such as “No answer”, “Refused”, “Don’t know” etc., depending on the question. As such, the missing values reported on the data profile table below may not be entirely accurate, however it tries to be as comprehensive as possible.

Care should also be taken when linking other geographic data. For the 2008 – 2014 BPS data, LSOA coverage in GB is 64%, meaning 64% of all LSOAs (or Data Zones in Scotland) have at least 1 individual who has taken part in the survey. The average ratio is 13.28 responders by LSOA.

This dataset is a spatially generalised version of the Secure British Population Survey dataset, also available on this site.