Technical note
Published 16 December 2022
© Crown copyright 2022
This publication is licensed under the terms of the Open Government Licence v3.0 except where otherwise stated. To view this licence, visit nationalarchives.gov.uk/doc/open-government-licence/version/3 or write to the Information Policy Team, The National Archives, Kew, London TW9 4DU, or email: psi@nationalarchives.gov.uk.
Where we have identified any third party copyright information you will need to obtain permission from the copyright holders concerned.
This publication is available at https://www.gov.uk/government/publications/factors-associated-with-loneliness-in-adults-in-england-during-the-pandemic/technical-note
Datasets
Three datasets were used in this analysis to allow us to address the research questions:
| Community Life Survey, 2020/21 | Understanding Society, Wave 11 (2019-2021) | Understanding Society Covid-19 Study | |
|---|---|---|---|
| Use in research | Used to investigate the relationship between life circumstances and loneliness, and if these relationships have changed over time. | Used to investigate (a) the relationships between both protected characteristics (including age, gender, marriage or civil partnership, race, sexual orientation and disability) and loneliness, and to investigate (b) factors related to the alleviation of loneliness. | Used to examine resilience to loneliness at three time points during the pandemic, as well as the impact of pandemic-specific experiences like furlough, shielding and COVID-19 hospitalisation. |
| Geographic coverage | England | England | England, Scotland, Wales and Northern Ireland |
| Age coverage of analysis | Adults 16 years and older | Adults 16 years and older | Adults 16 years and older |
| Effective sample size | 10,917 | 23,309 | 10,129 |
| Effective sample size | April 2020 through March 2021 | January 2019 through January 2021 | The dataset used in this research has observations on loneliness at three key time points: May 2020, November 2020 and March 2021 |
| Survey mode | Push to web | Changes were made to the fieldwork approach in March 2020 due to the pandemic, which involved a shift from in-person to online questionnaires. | Push to web |
The analysis in this report uses survey weights and controls for the complex survey design of both Understanding Society (USoc) and the Community Life Survey (CLS). Weights aim to reduce bias by adjusting for differences in the probability of being selected to take part, differential levels of nonresponse and sampling error to ensure that results are representative of adults aged 16 and older living in England. Information regarding weighting is published online for Understanding Society and Community Life Survey (Chapter 7).
All results presented in this report were formally tested for statistical significance and only present odds ratios and differences in proportions that were found to be statistically significant at the 95% level. Error bars have been used to illustrate confidence intervals in charts. Results are not statistically significant when error bars overlap.
Details of the analytical methods used in each section of the report are summarised in Table 1 below.
Logistic regression
The different analytical approaches used in this report were all variations on logistic regression, a technique which allows for the examination of the relationship between a number of explanatory (or predictor) variables and a binary categorical outcome (yes/no). In this report the outcome was a measure of loneliness (either chronic loneliness measured directly, or loneliness measured indirectly). The outcome for each section of the analysis is detailed in the third column of the table below.
All the logistic regression models discussed below controlled for age, gender, and income level as standard (we refer to this as being ‘fully adjusted’). The Understanding Society model includes a variable to control for whether someone was interviewed before or after the pandemic (this was not necessary for CLS analysis, because fieldwork took place exclusively during the pandemic). In order to gain an insight into the pandemic overall, rather than just the lockdown periods, we controlled for this pandemic variable, rather than stratified by whether someone was interviewed before or during the lockdowns. This report employed two variations on logistic regression models:
- Models estimated for a subsample: Several logistic regression models were conducted on a subsample of respondents to understand the predictors of loneliness for a specific age group (i.e., the models were stratified by age group).
- Models estimated with an interaction effect: interactions can be included in logistic regression models to test the joint effect of two variables. To isolate the impact of the joint effect of variables, regression models contained no more than one interaction effect, the details of which are specified in the table below.
Table 1: Summary of models
| Section | Dataset | Outcome Variable | Model description |
|---|---|---|---|
| Predictors of loneliness – analysis of USoc and CLS | USoc Wave 11 (2019-2021) CLS 2020-2021 | Chronic loneliness | Fully adjusted |
| Predictors of loneliness measured indirectly | CLS 2020-2021 | Indirect loneliness measure | Fully adjusted |
| Relationship quality: extending the original model | USoc Wave 11 (2019-2021) | Chronic loneliness | Fully adjusted + additional household size-pandemic interaction and relationship quality control |
| Different risk factors at different life stages | USoc Wave 11 (2019-2021) | Chronic loneliness | Fully adjusted, stratified by age |
| Risk of chronic loneliness for people with protected characteristics | USoc Wave 11 (2019-2021) | Chronic loneliness | Fully adjusted |
| Factors predicting resilience to chronic loneliness during the pandemic | USoc COVID-19 study | Resilience to chronic loneliness | Fully adjusted |
Table 2: Predictors of chronic loneliness in USoc 2019-2021
| Odds Ratio (95%CI) [footnote 1] | Odds Ratio (95%CI)[footnote 2] | |||||
| 65+ (reference) | ||||||
|---|---|---|---|---|---|---|
| 50-64 | 2.60 (2.06-3.28) | 2.61 (2.07-3.29) | ||||
| 35-49 | 3.35 (2.56-4.36) | 3.35 (2.57-4.38) | ||||
| 16-34 | 3.76 (2.76-5.12) | 3.76 (2.76-5.12) | ||||
| Male sex (reference) | ||||||
| Female sex | 1.28 (1.11-1.48) | 1.28 (1.11-1.48) | ||||
| No long-term illness or disability (Reference) | ||||||
| Have long-term illness or disability | 2.76 (2.36-3.23) | 2.76 (2.35-3.23) | # No caring responsibilities (Reference) | |||
| Caring responsibilities | 1.11 (0.92-1.34) | 1.11 (0.92-1.34) | ||||
| Married/Civil Partnership/Living as a couple (Reference) | ||||||
| Widowed, separated or divorced | 3.39 (2.78-4.12) | 3.39 (2.78-4.12) | ||||
| Never been married | 3.45 (2.76-4.31) | 3.46 (2.77-4.32) | ||||
| Employed (Reference) | ||||||
| Unemployed | 1.41 (1.18-1.69) | 1.42 (1.19-1.70) | ||||
| Education – Degree (Reference) | ||||||
| Other higher degree | 1.00 (0.77-1.28) | 0.99 (0.77-1.28) | ||||
| A-Level or equivalent | 1.05 (0.85-1.30) | 1.05 (0.85-1.30) | ||||
| GCSE or equivalent | 1.01 (0.81-1.27) | 1.01 (0.81-1.27) | ||||
| Other qualification | 1.29 (0.96-1.75) | 1.29 (0.96-1.75) | ||||
| No qualifications | 1.09 (0.81-1.46) | 1.09 (0.81-1.46) | ||||
| Income level - highest income quintile (reference) | ||||||
| Second highest income quintile | 1.11 (0.88-1.40) | 1.11 (0.88-1.40) | ||||
| Middle income quintile | 1.28 (0.99-1.64) | 1.28 (0.99-1.64) | ||||
| Second-to-bottom income quintile | 1.37 (1.05-1.78) | 1.37 (1.05-1.78) | ||||
| Bottom income quintile | 1.21 (0.93-1.57) | 1.21 (0.93-1.56) | ||||
| Interviewed pre-pandemic (reference) | ||||||
| Interviewed during the pandemic | N/A | 0.94 (0.80-1.10) |
Limitations
The Community Life Survey and Understanding Society measure loneliness in slightly different ways, meaning the estimates of people who are chronically lonely vary and results are not directly comparable. Similarly, the report measures loneliness directly, which can be problematic if some respondents are more likely than others to describe themselves as lonely even when they feel similarly about their circumstances. The CLS includes a measure of indirect loneliness, combining responses to multiple questions related to loneliness (such as isolation) which is less likely to be impacted by this. Measures of indirect loneliness need to be carefully constructed to ensure they capture the intended underlying concept – in this case, loneliness.