Value-added work is carried out on the data to make it easier for researchers to use.
In addition to longitudinal consistency checks on key variables, such as date of birth, a range of derived variables are produced so that users do not need to create them themselves.
Derived variables are created where information needs to be combined from multiple data files or across different waves of the survey. For example, country of birth is collected only when a participant first joins the study. As new samples have been introduced over time and new household members enter the study at different waves, this information is not available in the same wave for every respondent. To simplify analysis, information from across waves is combined into single derived variables.
These derived variables are included in the datasets, together with supporting syntax to help users work with the data. This includes user-developed syntax for derived employment histories. Two key derived data files available with the main survey dataset are xwavedat, which contains time-fixed information, and xhhrel, which provides identifiers linking family members and households over time.
Additionally, income variables are imputed and derived, survey weights are produced and free text responses are coded. In line with our data access policy, free text responses are never released. Instead, those coded using standard classifications are made available to data users, for example, SOC & SIC codes of current and last jobs.
Further details about the main survey are provided in the user guide, including sections on Derived variables, Derived income variables and survey weights.
Along with variables, some datasets are created to make it easy for data users to use the data. These combine information collected in multiple files and across multiple waves.
- Calendar year datasets
- Partnership history dataset
- Fertility history dataset (GGP dataset)
- Pregnancy and early childhood – Understanding Society (PEACH)