Tracking the psychological and socio‐economic impact of the COVID‐19 pandemic in the UK: A methodological report from Wave 5 of the COVID‐19 Psychological Research Consortium (C19PRC) Study

Abstract Objectives The COVID‐19 Psychological Research Consortium (C19PRC) Study was established in March 2020 to monitor the psychological and socio‐economic impact of the pandemic in the UK and other countries. This paper describes the protocol for Wave 5 (March–April 2021). Methods The survey assessed: COVID‐19 related experiences; experiences of common mental health disorders; psychological characteristics; and social and political attitudes. Adults who participated in any previous wave (N = 4949) were re‐invited to participate. Weights were calculated using a survey raking algorithm to ensure the longitudinal panel was nationally representative in terms of gender, age, and household income, amongst other factors. Results Overall, 2520 adults participated. A total of 2377 adults who participated in the previous survey wave (November–December 2020) were re‐interviewed at Wave 5 (61.5% retention rate). Attrition between these two waves was predicted by younger age, lower household income, children living in the household, and treatment for mental health difficulties. Of the adults recruited into the C19PRC study at baseline, 57.4% (N = 1162) participated in Wave 5. The raking procedure re‐balanced the longitudinal panel to within 1.5% of population estimates for selected socio‐demographic characteristics. Conclusion This paper outlines the growing strength of the publicly available C19PRC Study data for COVID‐19‐related interdisciplinary research.


| Context for C19PRC Study in March 2021
A 'National Day of Reflection' was held in the UK on 23 March 2021 to mark the one year anniversary of the nation's first  lockdown, and to pay respect to the~140,000 UK citizens who tragically died due to coronavirus in the previous 12 months (UK Government, 2021a). By this time, approximately 5000 new daily COVID-19 cases were being reported in the UK (a substantial decrease from the~60,000 daily cases occurring at the peak of the second wave in January 2021-see Figure 1), and approximately 26 million COVID-19 vaccination doses had been administered since the vaccination rollout commenced in the UK on 8 December 2020 (NHS England, 2021).
During March 2021, the UK Government enacted legislative changes to permit a gradual relaxation of the harsh lockdown restrictions which had been in place across the UK since late December 2020/early January 2021 (UK Government, 2021c). These included: (i) lifting the mandatory 'stay at home' rule, allowing the public to have socially-distanced contact with individuals outside their household in outdoor spaces; (ii) children returned in-person to school and childcare facilities; (iii) care home residents were able to receive regular in-person visits from one person; and (iv) businesses offering outdoor facilities (e.g., tennis) were allowed to reopen (UK Government, 2021b). Directives to 'work from home, where possible' remained in place, however, and international travel was prohibited except for a limited number of permitted reasons (e.g., family death).
During 2020-21, the UK experienced notable social and economic disruption, attributable not only to the pandemic, but also to recent upheavals relating to the end of the Brexit transition period on 31 December 2020 (De Lyon & Dhingra, 2021). For example, the national lockdowns are estimated to have cost the UK economy £251bn in the value of goods and services (Miley, 2021), while the gross domestic product in January 2021 was 9% lower than the prepandemic level a year earlier (Office for National Statistics, 2021b).
Approximately 6.1 million people, or 19% of UK businesses' workforce, remained on furlough leave during February and March 2021 (Office for National Statistics, 2021a). Moreover, concern about the impact of recurrent lockdowns on the nation's mental health and wellbeing, a consistent issue at the forefront of public and academic debate since the earliest stages of the pandemic, persisted (Mental Health Foundation, 2021).
It is against this backdrop that the fifth UK survey wave of the C19PRC Study (hereafter C19PRC-UKW5) commenced on 22 March 2021 to coincide with the anniversary of the UK's first national lockdown. The C19PRC Study has followed a large, nationally representative sample of UK adults over multiple survey waves from the beginning of the first national lockdown in the UK, through the summer and winter of 2020, and into spring 2021 (see Figure 1). The main objective of the C19PRC Study has been to investigate trends in mental health outcomes (i.e., experiences of major depression, anxiety, and pandemic related traumatic stress) for the UK adult general population over time, and to determine how various psychological, socio-economic, and political factors have influenced these trends, whilst accounting for the wider national context in which the pandemic has been unfolding. Detailed methodological reports for these survey waves are available elsewhere . Parallel studies have also been conducted in the Republic of Ireland, Spain, and Italy (Bruno et al., 2021;Spikol et al., 2021;Valiente et al., 2021).

| Methodological issues relating to online survey research during pandemic
The C19PRC Study survey waves are being designed and conducted during an unprecedented time for survey fieldwork. As we have previously discussed , the four main methodological challenges or issues relate to: (i) the use of quotabased sampling to recruit participants from existing opt-in market research survey panels, as opposed to probability-based sampling; (ii) a lack of pre-pandemic data on C19PRC Study participants to assess change in core study outcomes (e.g., common mental health conditions) pre-to-post lockdown; (iii) mode of administration; and (iv) managing attrition. We will briefly summarise each of these challenges next.
With respect to surveying the UK adult population during the pandemic, two general approaches were available to researchers in March 2020. The first is that specific research teams already in the field were able to repurpose existing surveys to capture important COVID-19 related data from their study's participants. For example, many of the established UK-based cohort studies (e.g., Understanding Society; Millennium Cohort, etc.) fielded COVID-19 waves to collect new data from existing cohort members who had been recruited using probability-based sampling techniques before the pandemic (Patel et al., 2022). The superiority of probability-based samples over nonprobability quota based samples is well-acknowledged . An additional strength of this approach is the availability of 'pre-pandemic' data, often spanning many years or even decades, to detect meaningful change more readily in respondents' lives as a result of their experiences during the pandemic (Pierce, Hope, et al., 2020). Moreover, research teams working with existing cohorts were also ideally positioned to offer different modes of survey administration (e.g., telephone or web-based interviews), and to assess the likely impact of these different modes on the quality and completeness of data collection (Burton et al., 2020). Attrition in these repurposed surveys was generally managed using post-survey weighting procedures (Benzeval et al., 2021).
In the second approach, research teams set about designing new studies to collect COVID-19 specific data from general population samples. As we previously discussed , many of these studies: (i) were established hastily and limited to one or two waves of online data collection; (ii) relied on short screener-type questionnaires to measure general wellbeing or psychological distress, as opposed to gold standard instruments for assessing common mental disorders; and (iii) comprised of relatively small sample sizes recruited via opportunistic sampling methods, the composition of which did not represent the socio-demographic characteristics of the UK general adult population.
The C19PRC Study is one of the longest running newlyestablished COVID-19 surveys in the UK. Funded by the UKRI Economic and Social Research Council (ESRC), the C19PRC Study was designed to address many of the limitations of survey fieldwork encountered during the pandemic. An important feature of our Consortium's work is the production of detailed methodological papers for each survey wave in which we document the survey design and content, as well as the challenges and outcomes associated with conducting the survey wave at a specific point in the pandemic. To our knowledge, this is not common practice for other longitudinal surveys conducted during the pandemic; indeed, the absence of detailed methodological reports for other dedicated COVID-19 studies makes it challenging to compare surveys conducted during the pandemic in terms of fieldwork outcomes (e.g., retention rates). We argue this is a key strength of the C19PRC Study data, which is available for secondary use of the data via the Open Science Framework.
Here, we offer a brief summary of the characteristics of the C19PRC Study design for interested users of the data. Although the C19PRC Study recruited using quota-based non-probability sampling methods, the baseline sample was large and representative of the UK population on a wide range of socio-demographic characteristics. The collection of robust mental health data using detailed measures of common mental health conditions was prioritised at each survey wave. Concerted efforts were made to re-contact and re-engage all study participants at each wave post-baseline to encourage them to participate. Tailored communications were sent to participants to remind them about their previous engagement with the study and reassuring them of the importance of their valuable contribution to the main goal of the study (i.e., tracking the general public's experience of the pandemic over time). Approximately six-in-ten of baseline respondents returned at each follow-up wave. Levels of attrition F I G U R E 1 Graphical presentation of the number of daily COVID-19 cases and deaths in the UK, sourced from Our World in Data, 2020, aligned to the COVID-19 Psychological Consortium (C19PRC) Study survey waves. New daily deaths and cases depicted as 7-day rolling average. C19PRC-UKKW1 (baseline survey, March-April 2020); C19PRC-UKW2 (second survey, April-May 2020); C19PRC-UKW3 (third survey, July-August 2020), C19PRC-UKW4 (fourth survey, November-December 2020), and C19PRC-UKW5 (fifth survey, March-April 2021) were low, with only 15% of baseline respondents completely lost to follow-up by the fourth wave . Two specific approaches were taken to address attrition: (i) sample replenishment procedures were conducted regularly to 'top-up' gaps in quotas (with respect to age, gender, and household income); and (ii) post-survey weighting was conducted to ensure the longitudinal panel followed from baseline was representative of the UK general adult population. Finally, booster sampling was conducted by UK country to ensure that there was sufficiently large sample sizes to conduct robust betweencountry analyses.
The availability of the fifth wave of the C19PRC Study offers an ideal opportunity to further study attrition processes for a large, internet-based panel of adults recruited and followed-up during a turbulent historical event. This paper has three main aims: (i) to describe the prevalence of common mental disorders among participants in the C19PRC-UKW5 sample, as well as the sample's sociodemographic characteristics and specific experiences relating to the pandemic which were pertinent issues at the time this wave was conducted (e.g., self-isolation, diagnosis of COVID-19, and vaccination status); (ii) to examine patterns of attrition in the C19PRC Study by this fifth wave, and test whether these could be predicted by respondents' mental-health attributes, psychological characteristics, and socio-demographic factors; and (iii) to conduct and assess weighting procedures to manage attrition in the longitudinal panel.

| METHOD
A Checklist for Reporting Results of Internet E-Surveys (CHERRIES) (Eysenbach, 2004) for this survey wave is available in the Supplementary Tables Document (see Table S1).

| Fieldwork organisation overview and strategy
The survey company Qualtrics conducted the fieldwork for C19PRC-UKW5. Qualtrics partners with over 20 online sample providers to supply a network of diverse, quality respondents to their worldwide client base. To date, the company has completed~15,000 projects across~2500 universities worldwide.
As described elsewhere , at the previous survey wave, C19PRC-UKW4, which was conducted during November-December 2020, a complex booster-sampling strategy was employed to (i) recruit new respondents into the panel by oversampling in each of the devolved UK nations (Scotland, Wales, and Northern Ireland) so that sizeable country-specific sub-samples would be available to facilitate robust, between-country comparisons; and (ii) replenish the sample with new recruits according to baseline quotas (i.e., age, gender, and household income) to deal with attrition across previous waves to ensure the main longitudinal panel remains representative of the UK adult population (with respect to these characteristics). This strategy increased the panel sample from 2025 to 4949 adults. Funding budgetary constraints at this stage in the C19PRC study meant that a maximum of 3600 respondents could be re-interviewed at C19PRC-UKW5, and so a decision was made to re-contact all adults who participated in the previous wave (C19PRC-UKW4) first, as a priority. Two recruitment Phases were designed to achieve this aim (described next). In Phase 2 (8-20 April 2021), participants who had completed any previous wave except C19PRC-UKW4 (N = 1082) were recontacted and invited to participate in the fifth wave.

| Informed consent process
As in previous waves, participants were informed, that their data would be treated in confidence, that geolocating would be used to determine the area in which they lived (in conjunction with their residential postcode stem), and of their right to terminate participation at any time. Participants were also informed that some topics might be sensitive or distressing (e.g., self-harm/suicide content). Information about how their data would be stored and analysed by the research team was also provided. Participants were also informed that they would be re-contacted at a later date to invite them to participate in subsequent survey waves. Participants provided informed electronic consent prior to completing the survey and were directed to contact the NHS website upon completion if they had any concerns about COVID-19, and emotional support services if they had been negatively impacted by any of the questions asked during the survey.

| Compliance with General Data Protection Regulation (GDPR)
Participants are informed that C19PRC data will be stored confidentially in line with GDPR. When the study data is deposited with the UK Data Service and the Open Science Framework (OSF), location data is removed and replaced with relevant socioeconomic summary data (e.g., area-level deprivation and population density data). All other personal data is also removed.

| Quality control
Qualtrics deliver high-quality survey data from online survey panels and conduct multiple validation checks on the C19PRC survey data.
First, the survey is piloted ('soft launch'; n = 50) prior to the fieldwork going live ('full launch') to rectify sequencing/coding errors and omissions prior to the full launch. The soft launch also calculates the median survey completion time, providing an opportunity to tailor the content to ensure the median survey time does not exceed 30 min; this is important to minimise respondent burden and maximise participation over time. A soft launch for C19PRC-UKW5 was conducted using a new general population sample of (n = 51) on 22 March 2021, and the median survey completion time was 19 min 34 s. Participants in the soft launch are excluded from the final sample for the survey wave. Table 1 provides an overview of the C19PRC-UKW5 survey content (see Supporting Information S1 for details of all measures administered).

| Ethical approval
Ethical approval for the project was provided by the University of Sheffield (Reference number 033759).

| Data analysis plan and weighting procedures
Five sets of analyses are presented.
First, re-contact rates at C19PRC-UKW5 were calculated for Phase 1 and Phase 2, and patterns of respondent participation in previous waves by phase were described and compared.
Second, the socio-demographic, mental health, and COVID-19 related characteristics of the cross-sectional sample surveyed at C19PRC-UKW5 are presented.
Third, a binary logistic regression model was estimated to assess the extent to which participation at C19PRC-UKW5 could be predicted by a range of socio-demographic factors, mental health conditions, and psychological factors assessed at the previous wave.
Fourth, the process and outcome of post-stratification survey weighting for this longitudinal panel is detailed. As per previous waves, survey raking or sample-balancing was conducted using the 'anesrake' package in R (Pasek & Pasek, 2018). Raking is one common method of adjusting survey data to ensure that the distribution of the characteristics of a given sample closely mirror the known population distribution. In practice, this means the baseline sampling quotas for age, gender, and household income, as well as the baseline proportions achieved for ethnicity, urbanicity, household composition, and being born or raised in the UK, were imposed on the sample of respondents returning from baseline at C19PRC-UKW5, and the raking algorithm was conducted to produce, and iteratively adjust, a weight value for each case in the sample until the sample distribution aligned with the population distribution for the chosen characteristics (DeBell & Krosnick, 2009 mental health treatment (current or past treatment for mental health problems vs. other); loneliness (score of ≥6 on the Loneliness Scale); neuroticism (total score on the neuroticism subscale of the Big-Five Inventory-10); paranoia (total score on the Persecution and Deservedness Scale); Conspiracy mentality (total score on the Conspiracy Mentality Questionnaire); hopefulness (total score on the Brief-H-Pos Scale) and COVID-19 anxiety (total score on single item indicator).
In addition, these variables (same categorisation as above, unless otherwise specified) were used to describe the attrition analyses for longitudinal panel (recruited at baseline) participating in C19PRC-     Figure 2 illustrates the outcome of recruitment of C19PRC-UKW5,

Phase 1 and Phase 2. The median survey completion time for both
Phases was 31 min 28 s.

| Outcome of recruitment at Phase 1 and Phase 2, by participation in panel to date
At Phase 1, 3867 adults were eligible to participate in C19PRC-UKW5 having completed the survey at the previous wave, and 2377 were successfully reinterviewed (61.5% recontact rate).

| Attrition analysis for baseline entrants only by C19PRC-UKW5
By this fifth wave, 1162 of the 2025 adults recruited at baseline (57.4%) were re-interviewed. Almost six-in-ten (N = 677; 58.3%) of those who were re-interviewed had participated in all four previous C19PRC Study waves; three-in-ten (N = 351; 30.2%) participated in any three previous waves, one-in-ten (N = 117; 10.1%) in any two previous waves, and a small number (N = 17; 1.5%) returned having only participated at baseline. As presented in the third column of Table 4, the raking procedure successfully re-balanced the characteristics of responders at this fifth wave (N = 1162) to the baseline proportions for gender and age (exact rebalance), household income (within 1.1%), household composition and urbanicity (exact rebalance), ethnicity (within 0.1%), and status relating to being born or raised in the UK (within 1.5%).
Applying this weight for all analyses of the C19PRC-UKW5 survey data completed by this longitudinal panel of adults recruited and followed from baseline is recommended to account for attrition over survey waves on core study outcomes.
The characteristics of the C19PRC-UKW5 longitudinal panel (N = 1162) by gender and age group were examined (see Table S3).

| DISCUSSION
The C19PRC Study is a dynamic, longitudinal survey, which contains a broad array of socio-political, economic, and mental health mea-  is large and diverse, and many new measures were introduced at this wave (e.g., prolonged grief disorder; perceived burdensomeness and thwarted belongingness; mania) which provides a unique opportunity to explore nuanced research questions relating to mental health experiences 1 year into the COVID-19 pandemic in the UK.
Our Consortium has advocated previously that researchers conducting COVID-19 related survey research should be transparent with respect to methodologies and recruitment practices for survey fieldwork conducted (largely online) during this time. Indeed, we have previously debated two core potential weaknesses in our study protocol, that is, the opt-in, non-probability-based web panel of adults recruited according to pre-determined quotas, and a reliance on a single mode of survey administration (internet-based survey) (see . Here, we focus on four additional methodological issues that critics might highlight as additional potential weaknesses: (i) average retention rates; (ii) the non-routine sample replenishment to account for attrition (conducted at the third and fourth waves only); (iii) the generation of weights for the longitudinal panel (followed from baseline) to adjust for attrition against quotas determined to recruit the baseline sample; and (iv) sample diversity/representativeness.
We take the opportunity here to address each of these in turn.
The evidence indicates that retention rates over the five survey. During April 2020 and March 2021, nine survey waves were administered to these participants and the proportion of respondents providing either full or partial web-based interviews dropped from 65% to 46% during this time period (Understanding Society, 2021).
Against these studies, we argue that the C19PRC Study competes well with respect to retention rates.
All longitudinal studies face challenges relating to attrition over time, and experts disagree as to whether attrition is more or less problematic when respondents are surveyed frequently (e.g., multiple times in a 1-year period) compared to when longer gaps occur between contact (e.g., several years between survey waves) (Laurie, 2008 -13 of 16 used to address research questions of a longitudinal nature (see Shevlin et al., 2021). Moreover, consistent with other established and reputable panel studies (e.g., the American National Election Study), we have engaged in sample replenishment procedures. Funding constraints dictated that sample replenishment was only feasible at specific C19PRC survey waves, and a strategic decision was taken to undertake this process at the third and fourth waves only (July-August 2020 and November-December 2020), which is the midpoint in the panel study (running between March 2020-November 2021). This process ensured that the cross-sectional sample at these waves was sufficient to conduct meaningful analyses (N > 2K), but not so large as to impede continued follow-up of all study participants moving forward in subsequent survey waves.
Our Consortium adopted the strategy of accounting for attrition in the longitudinal panel (recruited at baseline) by generating weights to re-balance data for this sample re-interviewed at post-baseline waves to the socio-demographic characteristics of all adults participating in the baseline sample. Given that the core C19PRC Study outcomes were mental health conditions, we felt it was a suitable approach to account for the fact that respondents with better health status tend to continue to participate in post-baseline survey waves (Radler & Ryff, 2010). We recognise, however, that alternative approaches may be more suitable for specific types of quantitative analyses, and secondary users of the C19PRC Study data may wish to generate additional and/or alternative weights if using the data to address specific research questions.
Finally, whilst it has been demonstrated that the C19PRC Study cross-sectional samples and the longitudinal panel are representative of the UK adult general population overall, the composition of the sample may not directly mirror the national population with respect to specific sub-groups (e.g., distribution of gender across the age groups). Quota sampling methods at baseline was not designed to be interlocking across gender, age, and household income, however, and secondary users of the data should be cognisant of this. We also acknowledge that the recruitment from existing online marketing research panels likely excluded from the C19PRC Study specific subgroups of the population who may be particularly vulnerable to widereaching effects of the pandemic (e.g., the homeless, those not connected to the internet, etc.).
In conclusion, however, the C19PRC Study data is a large, content-rich, survey data resource. We strongly encourage secondary use of this survey data by researchers and stakeholders interested in addressing pertinent research questions which will contribute to the existing evidence base on the impact of the COVID-19 pandemic on the lives of citizens of the UK now and into the future.
T A B L E 5 Outcome of the raking weighting procedure conducted at C19PRC-UKW5, March-April 2021, for the longitudinal panel recruited at baseline and followed-up at this survey wave (N = 1162)