Quick start guide: Understanding Society COVID-19 Teaching Dataset

Just after the first lockdown of the COVID-19 pandemic was introduced in the UK, individuals aged 16 and over who participated in one of the two last waves from the primary Understanding Society sample were invited to complete an online survey.

The survey ran between April 2020 and September 2021 with the aim of assessing the behaviours of individuals, families, and communities over the period of the pandemic. An important characteristic of the COVID-19 survey is that it is longitudinal, which allows individuals to be traced during the pandemic.

In addition to questions about their own lives, adults answered questions about their children through a web survey. For households without access to the internet, an additional telephone survey was implemented in particular months. At later waves of the survey (waves 4, 6 and 8), children aged 10-15 years were provided with their own paper to complete the survey.

The Understanding Society COVID-19 Teaching Dataset, 2020-2021 is a simpler version of the Understanding Society COVID-19 survey data with additional information on the characteristics of the survey participants.

The Economic and Social Research Council and the Health Foundation funded the Understanding Society COVID-19 study. The COVID-19 Longitudinal Health and Wealth – National Core Study funded the Serology testing (i.e. antibody testing). The fieldwork of the COVID-19 survey was undertaken by Ipsos (for the web survey) and Kantar (for the telephone survey).

The COVID-19 Teaching Dataset, 2020-2021 are presented in two files. A cross-sectional file that consists of data predominantly from Wave 4 (collected in July 2020), and a longitudinal file that consists of data predominantly from Wave 1, 4 and 9 (collected in April 2020, July 2020, and September 2021, respectively).

The two files cover the same topics. However, the longitudinal file has more variables than the cross-sectional file to allow assessing longitudinal trends, patterns, and changes over time. The topics of both files are on physical and mental health, information on working from home, COVID symptoms and test, wellbeing and life satisfaction, information on general health, time spent on childcare or home schooling, neighbourhood cohesion, belonging to neighbourhood, social networking, volunteering, fear of crime and feeling secure. The files also have a variety of socio-demographic variables such as gender, ethnic group, religion, age, employment, qualification, size of household.

See the Appendix to the User Guide (Excel) on the documentation section if you are interested in the full list of the variables, the sources and the links of the relative questionnaire for each file.

• Do you know that there were no notable differences in levels of COVID-19 vaccination by education attainments during the pandemic (between January 2021 and September 2021)?
  The Understanding Society COVID-19 study shows that in January 2021, 19.7% of respondents with high level of education (a degree or higher) had at least one dose of COVID vaccine compared to 16.3% of respondents with less educational qualification. In March 2021, levels of vaccination were relatively similar regardless the education attainments. In September 2021, the reported percentages of COVID vaccinations increased to 96.1% (for respondents with degree or higher) and 94.31% (for respondents with less educational qualification).
• Do you know that long COVID had a relatively steady trend during the pandemic (between November 2020 and September 2021)?
  The Understanding Society COVID-19 study shows that 9.2% of respondents had long COVID in November 2020. The reported percentages of respondents with long COVID decreased to 8.6% and 8.2% in January 2021 and March 2021, respectively. However, in September 2021, estimates of long COVID rose to the same level that was reported in November 2020 (9.2%).
• Do you know that, during the pandemic (between May 2020 and September 2021), women sought medical attention for COVID symptoms experienced more than men?
  The Understanding Society COVID-19 study shows that in May 2020, 1.2% women were treated for COVID-19 symptoms compared to 0.94% men. In July 2020 percentages of treatment of COVID-19 by gender were relatively the same (0.58% for women and 0.56% for men). In January 2021, the reported percentages of treatment of COVID symptoms increased to 3% (for women) and 1.9% (for men). However, these percentages slightly decreased in September 2021 to 2.3% and 1.8% for women and men, respectively.
• Do you know that level of loneliness increased notably during the pandemic (between July 2020 and January 2021)?
  The Understanding Society COVID-19 study shows that the reported percentages of feeling lonely some of the time or often in the past three weeks prior to the survey decreased from 41.5% to 39.4% between April 2020 and July 2020. These percentages are less than what was reported in January 2021 (estimated at 46.3%).

The figures above are created using the COVID-19 dashboard on the Understanding Society web page.

Core questions and rotating modules

A key feature of the Understanding Society COVID-19 survey is that it is longitudinal, which enables individuals to be tracked over the period of the pandemic. Therefore, the questionnaires of the COVID-19 survey include core questions that were repeated in all the waves to assess trends and changes during the pandemic. The core modules were further developed in later waves by adding more questions or response options to the original sets in the first waves. The questionnaires also include non-core modules (rotating topics) that were covered in some waves to reflect how people and communities adapted to the coronavirus situation. In addition, the Understanding Society COVID-19 survey covered topics that were collected just once (i.e. presented in one wave only).

The two files of the COVID-19 Teaching Dataset, 2020-2021 include variables from each module of the COVID-19 survey as follows:

• Variables from the core modules are long-standing illness or disability, status of employment, symptoms that could be coronavirus, test for coronavirus, feeling lonely, overall satisfaction with life, and subjective wellbeing.
• Variables from the non-core modules are working at home, hours worked, time spent on childcare or home schooling, face to face contact outside household, phone contact outside household, virtual contact outside household, talk regularly to neighbours, trust people in neighbourhood, helping neighbours, people don’t get along with each other, and volunteering.
• A variable that was collected from one wave only is time in childcare.

The longitudinal file covers the same topics that are presented in the cross-sectional file, but many variables are repeated from up to three time points to allow longitudinal analysis.

All the questionnaires used for each wave of the Understanding Society COVID-19 survey are available on the Understanding Society COVID-19 Questionnaires web page. COVID-19 survey FAQs are available. Also, a detailed list of variables in the COVID-19 Teaching Dataset, 2020-2021 is available in the Appendix to the User Guide in the documentation with information on the source of variables and the links to the relative questionnaires.

Structure of data files

The variables in the cross-sectional file (c19_teaching_xw) were mainly sourced from data collected for Wave 4 (July 2020) of the COVID-19 survey, with some variables from Waves 5 (September 2020), 6 (November 2020), and 7 (January 2021). There are also variables on socio-demographics and general health measures that were sourced from the main Understanding Society survey, Wave 10 (2018-19) and other waves. These variables were added to the cross-sectional file when they were not asked in Wave 4 from the COVID-19 survey. Each variable in the cross-sectional file is measured at one point of time only.

The variables in the longitudinal file (c19teaching_lw) were mainly sourced from data collected for Waves 1, 4 and 9 (collected in April 2020, July 2020, and September 2021, respectively) of the COVID-19 survey. Since the topics of the COVID-19 survey evolved overtime, it was necessary to include some variables of key measures from other waves (specifically, Waves 2 and 3 (May and June 2020), 5-7 (September 2020, November 2020, and January 2021)). There are also variables on socio-demographics and general health status that were taken from the main Understanding Society survey, Wave 10 (2018-19) and other waves. Many variables in the longitudinal files are repeated measures from up to three time points.

The repeated variables in the longitudinal file are distinguished by wave prefixes, which indicate the waves that the variables were taken from as follows:

• ca_ = variables from Wave 1 (or taken from Waves 2 and 3)
• cd_ = variables from Wave 4 (or taken from Waves 5, 6 and 7)
• ci_ = variables from Wave 9.

Variables with no prefixes are fixed, meaning they do not change over time.

Missing data in the COVID-19 teaching data files

Missingness in the two Understanding Society COVID-19 teaching data files is due to various reasons:

• Some respondents did not answer all questions. This is either because the respondents did not know the answer or refused to respond. In those cases, responses were labelled as “Don’t know” and “Refusal” with codes of -1 and -2, respectively.
• Some questions were not applicable to all respondents. This is either due to survey routing (i.e. skipping questions that are not applicable to the respondent) or because the respondents dropped from the subsequent waves before the questions were added to the questionnaires. In those cases, responses were entered as “Inapplicable” with a code of -8 was assigned.
• On the web survey, when respondents did not provide answers even after being shown the “Don’t know” and “Prefer not to say” options, responses were entered as “Missing” with a code of -9 was assigned.
• In cases where respondents should have been asked a question but were not, the variable was labelled as “Missing” with a code of -9 was assigned.
• Changes in question wording and routing over waves (i.e. asking some questions only at certain waves) led to some missingness. In such cases, variables were labelled as “Missing” with a code of -9 was assigned.
• In uncommon cases of random or coding errors, the variable was labelled as “Missing” with a code of -9 was assigned.

Unit of analysis

The Understanding Society COVID-19 Teaching Dataset, 2020-2021 represent adults aged 16 years old and over (as of April 2020) from households who participated in at least one of the last two waves of the main study Understanding Society.

Sample design

The sample of the COVID-19 survey consists of all participants from the primary Understanding Society samples, which include the Understanding Society General Population Sample (GPS), the Ethnic Minority Boost Sample (EMBS), the Immigrant and Ethnic Minority Boost Sample (IEMBS), and the former British Household Panel Study (BHPS). The sample consists of all people living in households in the United Kingdom who participated in wave 8 and wave 9.

The primary Understanding Society samples (hence, the COVID-19 study sample) used the postal addresses as the sampling frame. A clustered and stratified random sampling approaches were used in England, Wales and Scotland, whereas unclustered systematic random sampling approaches were used in Northern Ireland. Oversampling approaches were applied in Northern Ireland and other places in Great Britain where high immigrant and ethnic minority populations exist.

The COVID-19 survey respondents consist of all adults aged 16 years old and over (as of April 2020) in each sampled household. The sample includes all those who have valid postal addresses inside the United Kingdom, and were mentally and physically fit to consent informally to participate in the study. Those who were eligible to participate in the survey but refused were not included in the sample.

All those who were invited to participate in the first COVID-19 survey (wave 1-April 2020) remained eligible in wave 2 (May 2020), wave 3 (June 2020) and wave 4 (July 2020) regardless of whether they took part in past Understanding Society surveys. From wave 5 (September 2020) onwards, only those who participated in at least a partial interview of any of the first four web surveys were invited. From wave 6 (November 2020) onwards, those who only completed the first Understanding Society COVID-19 survey (conducted in April 2020) and had not completed any other surveys since were not invited to participate in the study anymore.

The Main Survey User Guide (PDF) includes more information about the main Understanding Society survey design. You can also read more on Understanding Society sample design in Lynn et al. (2018) (PDF), McFall et al. (2012) (PDF), Berthoud et al. (2009) (PDF), and Lynn (2009) (PDF).

Are there survey weights?

Each data file of the Understanding Society COVID-19 Teaching Dataset, 2020-2021 has a survey weight variable to account for sampling (betaindin_xw in the cross-sectional file, and ci_betaindin_lw in the longitudinal file).
Using these weights when conducting data analysis is crucial for producing results that reflect the UK adult population to key demographics. The weights adjust for features of the sample design (unequal selection probabilities) and correct for non-response bias (since not everyone invited to participate in the survey took part in all waves).

For more information on how the weights of the Understanding Society COVID-19 Teaching Dataset, 2020-2021 are produced, see the main Understanding Society COVID-19 Study User Guide (PDF).

Accessing the data from the UK Data Service

The Understanding Society COVID-19 Teaching Dataset, 2020-2021 are available from the UK Data Service after a quick registration.

For more information about data access levels and conditions, see our Access levels and conditions page. Information about the terms and conditions of data access are available on the Types of data access page.

You can learn more about the Understanding Society COVID-19 survey and the Understanding Society COVID-19 Teaching Dataset, 2020-2021 from:

• The Understanding Society: COVID-19 Study Teaching Dataset, 2020-2021 on the UK Data Service data catalogue. The dataset comes with documentation including the questionnaire and user guide.
• The Understanding Society COVID-19 survey website.
• The COVID-19 dashboard on the Understanding Society web page. You can use this tool to create charts to see trends over the period of the pandemic using variables from the Understanding Society COVID-19 study.
• Our online data exploration tool Nesstar. You can explore online the survey questions and responses and do some basic analysis.