COVID-19 and Well-being

Reader’s guide

General methods and data considerations

As a starting point, this report draws on the same indicators and sources used in the How’s Life? 2020 report. This is feasible for some aspects of material well-being and economic capital in particular, as well as for data from the Gallup World Poll, used in the absence of harmonised official data sources for a limited number of indicators. However, for other outcomes, the report relies on a number of ad-hoc studies and new data collections that have emerged during the crisis. At the national level, these sources range from experimental time use studies (UK Office of National Statistics) to ‘crowdsourced’ mental health data (Statistics Canada), a Household Pulse Survey (United States Census Bureau), and the SOEP-Cov study in Germany. In other cases, existing data collections have been adapted – for example, Stats NZ introduced supplemental well-being questions in the Household Labour Force Survey in 2020, providing quarterly estimates of outcomes such as life satisfaction. At the international level, novel data collections include the Eurofound Living, Working and COVID-19 Study; the Imperial College London/YouGov COVID-19 Public Monitor; and the REpresentations, PErceptions and ATtitudes on COVID-19 (REPEAT) survey from Sciences Po. Within the OECD, existing data collections such as the Risks That Matter survey have been adapted to address COVID-19 relevant concerns.

Several of the data sources used in the Evidence Scan build on existing survey vehicles and questionnaire items, but in some cases, the absence of comparable baseline data makes it difficult to provide an accurate account of pandemic impacts. The uneven and intersectional impacts of the pandemic across the population emphasise the need for large-sample representative studies that enable data to be disaggregated and cross tabulated with confidence – which is not always possible for smaller ad hoc studies. At the same time, the process of data collection has itself been heavily disrupted by the pandemic. For example, several data producers switched from face-to-face to other survey modes. The exceptional circumstances also mean that online-only methods and unconventional sampling strategies have occasionally been adopted (e.g. convenience sampling methods used for the Eurofound study). Post-hoc adjustments to survey weights are often applied to correct for the most easily addressed sources of bias, but these methods do not fully address the non-representativeness of the data when, for example, respondents are self-selecting.

Due to the constraints of the available data, some OECD countries have better coverage than others, direct country comparisons are not always feasible, and good 2019 baseline data are often lacking. It is also not possible to apply the more rigorous data quality standards adopted in How’s Life? throughout this report. Instead, the chapters include brief boxes that describe key data sources, and results should be interpreted with these methodological details in mind.

Note on the timing of data collections, and particularly the Gallup World Poll 2020

Throughout the first 15 months of the pandemic covered by this report, well-being outcomes have been a moving target. It is rarely safe to generalise results beyond the months in which data were collected, as both disease risk and pandemic restrictions shifted – and often in different cycles among different OECD countries. All figures in this report therefore specify the month and year of data collection, as appropriate. A variety of online dynamic dashboards can be consulted to obtain information about the pandemic context in each OECD country at different points in time during 2020 and 2021. For example, the OECD Coronavirus web pages https://www.oecd.org/coronavirus/en/, and the Our World in Data COVID-19 data explorer https://ourworldindata.org/coronavirus.

For most comparative data, fieldwork dates are broadly harmonised across countries. However, as discussed in Chapter 3, Box 3.4, the fieldwork dates for the Gallup World Poll vary across OECD countries, from February-March 2020, to November-December 2020. Gallup World Poll data are used in this report to illustrate two aspects of subjective well-being (life satisfaction and affect balance), as well as social network support, feelings of safety, and trust in government. Table 1 describes some key pandemic context variables in each OECD country both during the time of the Gallup World Poll fieldwork, and during the year to date by the end of the fieldwork period in 2020. Sample sizes were approximately 1000 people in all OECD countries except Iceland, where it was 501. Prior to 2020, data in the majority of OECD countries was already collected entirely via telephone interviewing. However, in 14 OECD countries (Chile, Colombia, Costa Rica, the Czech Republic, Estonia, Greece, Hungary, Israel, Lithuania, Latvia, Mexico, Poland, the Slovak Republic and Turkey) interviews previously conducted face-to-face in 2019 were switched to telephone-based in 2020, which may result in some mode effects. In this publication, all countries with data collection method switches are marked with † in figures.

Table 1. Gallup World Poll fieldwork dates and COVID-19 context variables

Country Code	Fieldwork start date	Fieldwork end date	Rate of excess deaths during fieldwork (compared to 2015-19)	COVID-19 deaths during fieldwork period	Rate of excess deaths from start of 2020 to end of fieldwork	Total COVID-19 deaths from start of 2020 to end of fieldwork	Stringency Index, average during fieldwork
AUS	04-02-2020	22-03-2020	5%	7	5%	7	23
AUT	24-08-2020	19-09-2020	8%	29	3%	788	37
BEL	19-08-2020	19-09-2020	4%	137	10%	9 977	54
CAN	03-08-2020	21-09-2020	11%	327	11%	9 217	66
CHL	11-09-2020	16-11-2020	14%	3 227	17%	14 819	82
COL	21-08-2020	27-10-2020	48%	15 057	26%	30 154	74
CRI	15-09-2020	04-01-2021	..	..	..	..	58
CZE	09-10-2020	28-11-2020	69%	7 694	14%	8 467	65
DNK	14-09-2020	10-10-2020	2%	42	0%	669	49
EST	14-10-2020	20-11-2020	2%	20	1%	88	37
FIN	26-03-2020	13-05-2020	8%	288	1%	299	62
FRA	07-09-2020	02-10-2020	8%	1 624	7%	32 230	49
DEU	24-08-2020	19-09-2020	6%	154	2%	9 386	54
GRC	23-09-2020	24-10-2020	5%	236	4%	574	55
HUN	19-10-2020	17-11-2020	37%	2 222	2%	3 190	51
ISL	31-08-2020	12-10-2020	2%	0	1%	10	41
IRL	17-08-2020	12-09-2020	..	..	..	..	67
ISR	24-09-2020	19-10-2020	33%	953	10%	2 209	85
ITA	24-08-2020	16-09-2020	7%	311	10%	35 707	54
JPN	07-08-2020	08-10-2020	..	..	..	..	32
KOR	25-08-2020	07-10-2020	..	..	..	..	53
LTU	08-10-2020	26-11-2020	22%	412	3%	506	53
LVA	10-09-2020	31-10-2020	1%	39	-2%	74	37
MEX	08-09-2020	18-11-2020	54%	37 518	49%	101 676	72
NLD	11-03-2020	15-05-2020	34%	5 677	14%	5 680	74
NOR	24-03-2020	04-05-2020	4%	207	-1%	208	74
NZL	17-02-2020	23-03-2020	7%	0	5%	0	26
POL	25-09-2020	24-10-2020	28%	2 145	6%	4 438	34
PRT	20-03-2020	16-04-2020	18%	714	2%	714	82
SVK	09-09-2020	09-10-2020	7%	24	1%	61	37
SVN	13-03-2020	18-04-2020	3%	74	-1%	74	76
ESP	24-08-2020	17-09-2020	15%	2 017	18%	30 663	62
SWE	30-03-2020	29-04-2020	34%	3 051	7%	3 143	62
CHE	07-09-2020	09-10-2020	3%	69	3%	1 793	39
TUR	03-10-2020	23-10-2020	..	..	..	..	71
GBR	17-08-2020	12-09-2020	2%	599	14%	56 957	66
USA	16-03-2020	08-05-2020	28%	87 499	14%	87 560	68

Note: Luxembourg was not included in the 2020 Gallup World Poll. Data relating to excess mortality are sourced from (OECD, n.d.[1]); excess deaths are measured as the increase in the number of reported deaths from all causes during the time period, compared to the average number from 2015-19 for the same period. The Stringency Index is sourced from the COVID-19 Government Response Tracker (Hale et al., 2021[2]). It runs from 0 (least stringent) to 100 (most stringent), and combines data on school closures, workplace closures, cancellation of public events, restrictions on gatherings, closing of public transport, stay-at-home requirements, restrictions on internal movement, and international travel controls.

Source: OECD calculations based on the Gallup World Poll (2021[3]) (database), https://www.gallup.com/178667/gallup-world-poll-work.aspx; OECD (n.d.[1]), COVID-19 Health Indicators (database) https://stats.oecd.org/index.aspx?queryid=104676#; Hale et.al (2021[2]), “A global panel database of pandemic policies (Oxford COVID-19 Government Response Tracker)”, Nature Human Behaviour, Vol. 5/4, pp. 529-538, http://dx.doi.org/10.1038/s41562-021-01079-8

Note on terminology for the inclusion chapters

Some of the inequalities addressed in the inclusion chapters (5, 6 and 7) pertain to aspects of diversity for which data collection was already sparse before 2020. Context and country matter for how terms such as “race”, “ethnicity”, “migrant status” and “Indigenous identity” are understood - and with the exception of migrant status (when defined as people born abroad), there are no internationally comparable definitions for describing these very different aspects of diversity. Measurement approaches and regulations that underpin the collection of what is often considered sensitive data differ across OECD governments, with practices clustered in three broad categories (Figure 1). All OECD countries collect information on diversity proxies such as country of birth.1 A small majority (mostly Eastern European countries as well as the United Kingdom and Ireland) gather additional information on race and ethnicity. Finally, only a handful of countries in the Americas and Oceania collect data on Indigenous identity. By addressing these different aspects of diversity, this chapter does not imply that the situation, including the legal status, of different minority communities across the OECD is the same and that these can be directly compared. Rather, it aims to provide evidence on the well-being impacts such communities have faced during the pandemic and which can help devise locally appropriate policy solutions.

The norms around appropriate terminology are evolving even in countries that are advanced in diversity data collection. For instance, in the United Kingdom, the Commission on Race and Ethnic Disparities recommended in March 2021 that the government stop using the term “BAME” (Black, Asian and minority ethnic) because it emphasises certain ethnic groups (Asian and Black) and excludes others (Mixed, Other and white ethnic minority groups) (Commission on Race and Ethnic Disparities, 2021[4]). The government is currently considering its response to the Commission's recommendations. In Canada, the term “visible minority” is an official demographic category defined by the Canadian Employment Equity Act, and is used by Statistics Canada in their work. The Employment Equity Act defines visible minorities as "persons, other than Aboriginal peoples, who are non-Caucasian in race or non-white in colour". The visible minority population consists mainly of the following groups: South Asian, Chinese, Black, Filipino, Latin American, Arab, Southeast Asian, West Asian, Korean and Japanese (Statistics Canada, 2015[5]). However, the question of appropriate terminology is currently being reviewed in Canada, in the context of a task force on modernizing the Employment Equity Act (Department of Finance Canada, 2021[6]). This report has generally used the terminology adopted by data producers at the time of writing, while recognising that the racial and ethnic categories used are socially constructed and situational rather than static (Balestra and Fleischer, 2018[7]).2

Figure 1. Diversity collection practices across the OECD

Note: A country is listed as collecting a type of diversity data if at least one source (e.g. census, sample survey, population registry) includes relevant information. As not all OECD countries have Indigenous populations, the denominator is set at 14 for this category.

Source: Balestra, C. and L. Fleischer (2018[7]), “Diversity statistics in the OECD: How do OECD countries collect data on ethnic, racial and indigenous identity?”, OECD Statistics Working Papers, No. 2018/09, OECD Publishing, Paris, https://dx.doi.org/10.1787/89bae654-en.

Assessing the extent to which COVID-19 has affected different racial and ethnic communities is challenging. Basic statistics on the number and characteristics of COVID-19 cases are registered by national health systems, based on administrative sources such as testing and hospitalisations. Not all OECD countries consistently record diversity information (or other key socio-economic variables) in case numbers, hospital records or on death certificates, or do not always transmit these data for the compilation of national health and mortality statistics. For example, information on ethnicity or migrant status on death certificates is not transferred to the federal level in Germany; is incompletely recorded in Scotland; and not at all in England, Wales and Northern Ireland. This implies that data from census records, death registrations and hospital statistics in the latter countries have to be linked to provide information about the impact of COVID-19 by ethnicity (OECD, 2020[8]; ONS, 2020[9]). Many states in the United States have been slow to implement this practice: in May 2020, 51% of cases and 88% of deaths had an identified race (though states have been working to identify the race of deaths previously recorded without) and by September 2020, only 65% of new cases included an identified race/ethnicity code (The COVID Tracking Project, 2020[10]; NPR, 2020[11]). A year on, 39% of all cumulative cases recorded by April 2021 lacked this information (CDC, 2021[12]). American Indians and Alaska Natives in the United States and First Nations, Inuit and Métis communities in Canada, many of whom operate their own health systems, are also not officially required to report COVID-19 data. What is more, numbers of confirmed cases by ethnicity or origin are impacted by the ability of each country to reach the most vulnerable groups. For example, rates of testing among Veterans in the United States up to July 2020 have been found to be lower for Hispanic/Latino and Black communities compared to whites, for instance (Rentsch et al., 2020[13]). Hence, relative COVID-19 related risks among groups, especially those of younger ages less likely to show symptoms, are likely to be underestimated. Moreover, many population surveys, especially the non-official and experimental ones launched throughout 2020 to capture the pandemic’s psychosocial impact in real-time, often either do not contain questions on identity, or have such small sample sizes that any statements would be misleading. With these caveats in mind, and recognising that the situation is constantly evolving and varies between countries and communities, the available evidence nevertheless confirms that minority groups have disproportionally suffered from the pandemic along multiple well-being dimensions.

LGBTI+ populations are still not well represented in official statistics. This has significant implications for our ability to measure discrimination and to design effective policies to improve outcomes for these populations. In 2019, 15 OECD countries included a question on sexual self-identification in their nationally representative surveys and 3 countries (Chile, Denmark, the US) had started collecting data on the transgender population (OECD, 2019[14]). While this trend is growing, a majority of OECD countries identify the LGBTI+ population in an indirect way through the sex of the respondent’s partner. The limits of this approach are clear as it only captures a subset of the LGBT population (Balestra and Fleischer, 2018[7]).3 The LGBTQ2+ acronym found in this report is specifically used by Statistics Canada in order to reflect the broad scope of gender and sexual identities that exist in society. Individuals are included in the LGBTQ2+ population on the basis of self-reported sexual orientation (lesbian, gay, bisexual, or another minority sexual identity such as asexual, pansexual or queer) or gender identity (transgender, including respondents with non-binary identities like genderqueer, gender fluid or agender).

Breakdowns considered in inclusion analyses

Age and education ranges considered in the inequalities sections throughout this report were selected according to the breakdowns that are readily available in aggregate statistics, and what it is possible to compile on an internationally comparable basis.

• Specific age ranges for each indicator are reported in the respective figure or figure note.

• The education ranges refer to the highest level of education completed, and correspond to ISCED levels 0-2 for “below upper secondary” level (i.e. less than primary, primary and lower secondary); 3-4 for “upper secondary” level (i.e. secondary and post-secondary non-tertiary education); and 5-8 for “tertiary” level. For individual country-level mappings to the ISCED 2011 classifications, please see http://uis.unesco.org/en/isced-mappings.

• Indicators sourced from the Gallup World Poll form an exception and correspond to completed elementary education or less (up to eight years of basic education), completed some secondary education up to three years tertiary education (nine to 15 years of education), and completed four years of education beyond “high school” and/or received a four-year college degree. These levels are described as “primary”, “secondary” and “tertiary” in the report.

• In a small number of cases, data disaggregations lead to small effective sample sizes. Where this is the case, asterisks are used as described in the figure notes to signal small effective sample sizes. Data are not reported where fewer than 100 observations are available. ** denotes countries with between 100 and 300 observations per category; * denotes countries with between 301 and 500 observations per category. Where no asterisks are used, this indicates that more than 500 observations per category and per country are available.

Conventions for figures in the report

• In each figure, data labelled “OECD” are simple mean averages of the OECD countries displayed, unless otherwise indicated. Whenever data is available for less than all 38 OECD countries, the number of countries included in the calculation is specified in the figure (e.g. OECD 33).

• A weighted OECD average has been chosen in instances where the OECD convention is to provide this type of average. Where used, this is specified in the figure notes along with details of the weighting methodology. For example, when data are population-weighted this is done according to the size of the population in different countries, as a proportion of the total OECD population. Similarly, when OECD total sums instead of averages are used, this is indicated as “OECD Total”.

• Where trend lines are shown in the figures, the OECD averages refer to only those countries with data available for every consecutive year, since the OECD average needs to consider the same sample of countries in each year. As only countries with a complete time series and no gaps can be included, this can sometimes lead to different OECD averages for trend lines versus the latest and earliest available time points.

• Each figure specifies the time period covered, and figure notes provide further details when data refer to different time periods for different countries. Countries are referred to by their ISO codes (Table 2).

Table 2. ISO codes for countries and word regions

AUS	Australia	FIN	Finland	MEX	Mexico
AUT	Austria	FRA	France	NLD	Netherlands
BEL	Belgium	GBR	United Kingdom	NOR	Norway
BRA	Brazil	GRC	Greece	NZL	New Zealand
CAN	Canada	HUN	Hungary	OECD	OECD average
CHE	Switzerland	IRL	Ireland	POL	Poland
CHL	Chile	ISL	Iceland	PRT	Portugal
COL	Colombia	ISR	Israel	RUS	Russian Federation
CRI	Costa Rica	ITA	Italy	SVK	Slovak Republic
CZE	Czech Republic	JPN	Japan	SVN	Slovenia
DEU	Germany	KOR	Korea	SWE	Sweden
DNK	Denmark	LTU	Lithuania	TUR	Turkey
ESP	Spain	LUX	Luxembourg	USA	United States
EST	Estonia	LVA	Latvia	ZAF	South Africa

References

[7] Balestra, C. and L. Fleischer (2018), “Diversity statistics in the OECD: How do OECD countries collect data on ethnic, racial and indigenous identity?”, OECD Statistics Working Papers, No. 2018/09, OECD Publishing, Paris, https://dx.doi.org/10.1787/89bae654-en.

[12] CDC (2021), COVID-19 Hospitalization and Death by Race/Ethnicity, https://www.cdc.gov/coronavirus/2019-ncov/covid-data/investigations-discovery/hospitalization-death-by-race-ethnicity.html#footnote01 (accessed on 11 February 2021).

[4] Commission on Race and Ethnic Disparities (2021), Foreword, introduction, and full recommendations, https://www.gov.uk/government/publications/the-report-of-the-commission-on-race-and-ethnic-disparities/foreword-introduction-and-full-recommendations#full-recommendations (accessed on 17 June 2021).

[6] Department of Finance Canada (2021), Budget 2021: A Recovery Plan for Jobs, Growth, and Resilience, Annex 4: Gender, Diversity, and Quality of Life Statement, https://www.budget.gc.ca/2021/report-rapport/anx4-en.html (accessed on 17 June 2021).

[3] Gallup (2021), Gallup World Poll, https://www.gallup.com/analytics/318875/global-research.aspx (accessed on 18 June 2021).

[2] Hale, T. et al. (2021), “A global panel database of pandemic policies (Oxford COVID-19 Government Response Tracker)”, Nature Human Behaviour, Vol. 5/4, pp. 529-538, http://dx.doi.org/10.1038/s41562-021-01079-8.

[11] NPR (2020), As Pandemic Deaths Add Up, Racial Disparities Persist — And In Some Cases Worsen, https://www.npr.org/sections/health-shots/2020/09/23/914427907/as-pandemic-deaths-add-up-racial-disparities-persist-and-in-some-cases-worsen?t=1608509692531&t=1612899321921 (accessed on 9 February 2021).

[8] OECD (2020), “What is the impact of the COVID-19 pandemic on immigrants and their children?”, OECD Policy Responses to Coronavirus (COVID-19), OECD Publishing, Paris, https://dx.doi.org/10.1787/e7cbb7de-en.

[14] OECD (2019), Society at a Glance 2019: OECD Social Indicators, OECD Publishing, Paris, https://dx.doi.org/10.1787/soc_glance-2019-en.

[1] OECD (n.d.), COVID-19 Health Indicators (database), https://stats.oecd.org/index.aspx?queryid=104676# (accessed on 19 September 2021).

[9] ONS (2020), Updating ethnic contrasts in deaths involving the coronavirus (COVID-19), England and Wales: deaths occurring 2 March to 28 July 2020, https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/articles/updatingethniccontrastsindeathsinvolvingthecoronaviruscovid19englandandwales/deathsoccurring2marchto28july2020#main-points (accessed on 11 February 2021).

[13] Rentsch, C. et al. (2020), “Patterns of COVID-19 testing and mortality by race and ethnicity among United States veterans: A nationwide cohort study”, PLOS Medicine, Vol. 17/9, http://dx.doi.org/10.1371/journal.pmed.1003379.

[5] Statistics Canada (2015), Visible minority of person, https://www23.statcan.gc.ca/imdb/p3Var.pl?Function=DEC&Id=45152 (accessed on 17 June 2021).

[10] The COVID Tracking Project (2020), The COVID Racial Data Tracker, https://covidtracking.com/race (accessed on 9 February 2021).

Notes

← 1.  In general, collecting migration-related information on the foreign-born population and their children is a crude method for capturing diversity. Although such data are relatively readily available and often considered as ‘objective’, their use as proxy for ethnicity or race is problematic. The country of birth of a person neither takes account of the diversity of the country of origin of the individual or the parents (e.g. ‘white’ people in the United Kingdom that were born in former British colonies) nor does it capture cultural affiliation, or the inherently self-perceived aspect of belonging to an ethnic group. This view is also reflected in the UN Principles and Recommendations for Population and Housing Censuses, which state that country of birth or citizenship as well as questions on religion and language should not be taken as providing proper ethnic data.

← 2.  People may change how they identify themselves over time or they may identify themselves differently in different environments, which can be important for the interpretation of data and the dynamics of race and ethnicity.

← 3.  People may change how they identify themselves over time or they may identify themselves differently in different environments, which can be important for the interpretation of data and the dynamics of race and ethnicity.