Participants completed a longitudinal, online survey distributed through social media platforms, university and professional mailing lists, and university press releases. To reach a diverse sample, the survey was presented in 12 languages: Arabic, Bangla, German, English, Spanish, French, Hindi, Italian, Mandarin, Persian, Swedish, Turkish. First timepoint (T1) took place between 9th April and 20th May 2020, with the survey remaining online for five weeks in each language at T1. Subsequent timepoints (T2-T6) took place fortnightly after T1.

The number of people participating in the survey per timepoint was n = 6675 at T1, n = 2105 at T2, n = 1832 at T3, n = 1504 at T4, n = 1253 at T5 and n = 1169 at T6. Participants opting out of certain questions led to some missing data (i.e., if they had no one in their close circle: n = 1199 at T1; or if they did not reveal their country of residence: n = 41 at T1). Table 1 shows the sociodemographic characteristics of the study population.

Ethics statement

The study was approved by the ethics committee of the University of Nottingham School of Psychology. All participation was in line with the General Data Protection Regulation (GDPR) and the Helsinki Declaration of 1975, as revised in 2008. Participants provided written informed consent and were assigned an anonymous ID for analysis.

Patient and public involvement

The public were consulted, engaged and informed at all stages of the research wherever possible. Due to time constraints while setting up the survey, we could not formally involve the public in a focus group. Instead, a convenience sample of members of the public living in a diverse range of countries (i.e., Bangladesh, England, France, Germany, India, Iran, Italy, Spain, Sweden, Turkey and USA) were consulted to provide informal feedback on our survey items, namely those assessing demographics, vulnerability to Covid-19 and adherence to guidelines, to ensure the questions reflected pandemic experiences in their countries. In addition, the public were involved in the data collection process through both participating in the study and helping  to disseminating the survey to others. The results were shared with the public at multiple stages of the study through blog posts, social media activity and media interviews.

Materials & Procedure

The T1 survey was longer than surveys administered at T2-T6, though the variables reported in this study were collected at all timepoints except for some demographic questions (i.e., age, gender, education). Full survey items can be found at:


Participants reported their age, gender (options: man, woman, non-binary, prefer not to say), highest educational attainment (options: no schooling completed, primary education, secondary education, university undergraduate degree, postgraduate degree), number of people in their household (dichotomised as solo vs cohabiting with others), and work/study status (dichotomised into active vs  inactive with work/study).


Participants indicated how vulnerable to the Covid-19 disease they considered (a) themselves and (b) loved ones using continuous scales, where 1 = Not vulnerable at all, 50 = As vulnerable as an average person, and 100 = Extremely vulnerable.


Participants rated how well they had been following the general advice of keeping distance from others as applied in their local area on a continuous scale, where 0 = I have not been following the advice at all, 50 = I have been following the advice exactly, and 100 = I have been doing more than what is advised. In addition, we asked people how others in their close social circle (i.e., people they would turn to for advice/comfort during challenging times) and people in their country had been following these guidelines. These items were adapted from pre-pandemic research examining normative and empirical expectations [27].


Participants completed the 7-item short Warwick-Edinburgh Mental Wellbeing Scale (WEMWBS) considering their feelings in the past week [28]. WEMWBS measures wellbeing as a single-factor construct, comprising affective-emotional, cognitive-evaluative and psychological aspects. Since the Persian version of WEMWBS was not available, native speakers proficient in English translated and back-translated it for this survey. Short WEMWBS is a well-established scale with good content and construct validity, strong internal consistency (0.91), high test–retest reliability (0.83), and relatively low social desirability bias [28]. WEMWBS has been successfully adapted to many cultures and languages [29]. Additionally, we asked participants to rate how depressed, anxious, angry and lonely they had been feeling in the past week; analyses of these item are reported in Supplementary Materials (SM). Both WEMWBS and these mood items were answered on a 5-point Likert scale, where 1 = None of the time, 2 = Rarely, 3 = Some of the time, 4 = Often, 5 = All of the time.


Using the Oxford Covid-19 Government Response Tracker (OxCGRT) dataset [30], we obtained a stringency metric, which was used as a co-variate in all of our analyses. OxCGRT recorded the stringency of a range of Covid-19 measures applied in over 180 countries (and states within the US) from public gathering restrictions and school/workplace closures to social distancing and mask mandates. Adding this variable was critical due to the high variability across countries and states in terms of the prevalence of Covid-19 [31] and the measures taken to curtail its spread. Using the timeseries data in OxCGRT, we obtained a stringency score per participant by calculating the rolling average of the overall stringency score in the participant’s region within the 14 days preceding the date of their survey completion.

Statistical Analyses

Analyses were conducted using RStudio 1.3.959, packages car, nlme and tidyverse [32]. For each hypothesis, we report analyses examining T1 only and change over the 6 timepoints. For all analyses, linear mixed-effects models were conducted, with wellbeing as the outcome variable and the participants’ country of residence as a random effect to account for the fact that participants are nested within countries. For change over time analyses, we also included timepoints as a random effect, random slope and in interaction with the predictor variables of that model (intra-class correlation analyses in SM). Hypothesis 1 models had age (levels: split into categories by every 10 years), gender, and household as predictors. Hypotheses 2–3 models included age, gender, household status, education and the stringency of measures used in the participants’ country or state [30] as covariates. Including these covariates partially addresses the fact that our samples are not representative of the population structures. Tables S1314 show descriptive statistics of all variables used in this study, and Fig S1 shows how key variables of wellbeing, perceived vulnerability to the disease and adherence to pandemic guidelines vary across countries.

For ease of visualisation and to account for non-linear effects, we converted continuous predictors into categorical variables depending on data spread. All findings were replicated with continuous variables (Tables S7-12). For Hypothesis 2, perceived self-vulnerability and loved ones’ vulnerability variables were categorised using a median split. For Hypothesis 3, self-adherence categories were created based on the 25% and 75% quantiles: low adherence (score < 49), medium adherence (scores 49–79), and high adherence (score > 79).

To examine social alignment, the stringency variable was categorised using a median split and two ‘compliance’ variables were created that indicate how similar people’s adherence behaviour was to the perceived adherence of their close circle’s (close circle compliance) and country’s (country compliance). For these compliance scores, we took the absolute difference between participants’ self-adherence and the perceived adherence of (a) their close circle, and (b) fellow citizens. We then categorised these compliance scores into high vs low (median split). For example, a person who strongly adheres to the distancing guidelines would have a high adherence score, yet may still have a low compliance score, if their close circle and/or fellow citizens were reported to have low adherence  to the guidelines.

Additionally, we ran the same models using an aggregate mood variable as the outcome variable, which comprises 4 custom-made mood items on anxiety, depression, loneliness and anger (Tables S1621). Distinctly from the wellbeing scores reported here, these mood items aimed to capture unique aspects of mental health and were not derived from standardised scales.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit The Creative Commons Public Domain Dedication waiver ( applies to the data made available in this article, unless otherwise stated in a credit line to the data.


This article is autogenerated using RSS feeds and has not been created or edited by OA JF.

Click here for Source link (