Coping style and mental health amid the first wave of the COVID-19 pandemic: a culture-moderated meta-analysis of 44 nations

ABSTRACT

In the first wave of the COVID-19 pandemic, the rapid transmission of a novel virus and the unprecedented disease-mitigation measures have elicited considerable stress in many countries worldwide. Coping with pandemic stress may be differentially related to psychological symptoms across countries characterised by distinct cultural values. This study aimed to: (a) synthesise the literature by investigating the associations between some major types of coping style and psychological symptoms, and (b) investigate the moderating effects of culture on these associations. We performed a three-level random-effects meta-analysis, which included 151 independent samples from 44 countries across eight world regions (n = 137,088, 66% women, M_age= 36.08). For both problem-focused and avoidant coping styles, their hypothesised associations with psychological symptoms were robust across the countries (anxiety: rs = −.11 and .31; depression: rs = −.19 and .33; ps < .0001). For both emotion-focused and social support seeking styles, their associations with psychological symptoms were moderated by two Hofstede’s cultural dimensions: uncertainty avoidance (intolerance of ambiguity) and masculinity (concern for achievement and success). The hypothesised negative coping style–symptom associations were found only in the countries with lower levels of uncertainty avoidance or masculinity, but opposite patterns of findings were found in those with higher levels of either of these two cultural dimensions.

KEYWORDS:

• Mental health
• coping
• psychological distress
• epidemic
• coronavirus
• health crisis

The coronavirus disease 2019 (COVID-19) pandemic has swept the world, causing more than 650 million diagnosed cases and 6.65 million confirmed deaths in 240 countries and territories as of 14 December 2022 (Johns Hopkins Coronavirus Resource Center, 2022, december 14). During the first wave, which took place from December 2019 to July 2020, the highly contagious nature of this unknown virus has stressed the healthcare system of countries worldwide (e.g., Coccia, 2021), causing many governments to impose a series of drastic physical distancing policies such as national lockdowns and travel bans (e.g., Gostin & Wiley, 2020; Lindner et al., 2022). Residents of COVID-19-affected regions have deployed a variety of behavioural and cognitive strategies to cope with the mental disturbance elicited by this unprecedented health crisis.

The deployment of such strategies is largely influenced by one’s coping style, which refers to a stable pattern that distinguishes among individuals regarding the propensity of deploying certain preferred strategies for stress relief (e.g., Carver, 2011; Schwartz et al., 1999). However, these deliberate coping efforts may not always be effective; and ironically, some efforts may even exacerbate the problems that compromise mental health (e.g., Minahan et al., 2021; Yee et al., 2021). Coping style is thus proposed as a crucial factor that may explain the individual differences in the experience of psychological symptoms amid the COVID-19 pandemic.

In the present study, we performed meta-analysis for investigating the associations between diverse coping styles and the two most common mental health problems reported during the COVID-19 pandemic, namely anxiety and depressive symptoms (e.g., Shah et al., 2021; Wu et al., 2022). In addition, cross-cultural research has indicated that the effectiveness of coping effort varies by the values and norms prevalent in particular cultural contexts (e.g., Ding et al., 2021; Lai et al., 2020). This meta-analysis thus aimed to unpack differences in the magnitude of the coping style–symptom associations among COVID-19-affected countries characterised by distinct cultural values.

Types of coping style and their differential associations with psychological symptoms

A review of the coping literature identifies five major coping styles derived from three dominant theoretical perspectives: approach/avoidance, primary/secondary control, and intrapersonal/interpersonal focus. Table 1 summarises the characteristics of these coping styles and their hypothesised associations with both anxiety and depressive symptoms.

Table 1. Theoretical underpinnings and characteristics of five major types of coping style.

Coping style	Underpinning theory	Control orientation	Focus	Definition	Examples of coping strategy
Problem-focused coping	Fight or flight theory / Approach–avoidance model; Cognitive appraisal theory / Primary–secondary model	Primary control	Intrapersonal	Tendency to make direct effort to alter external stressors and/or tackle problems	planning, problem-solving
Avoidant coping	Fight or flight theory / Approach–avoidance model	Relinquished control	Intrapersonal	Tendency to escape from sources of stress or problems	behavioural disengagement, denial
Emotion-focused coping	Cognitive appraisal theory / Primary–secondary model	Secondary control	Intrapersonal	Tendency to make effort to regulate or improve one’s own unpleasant affect	catharsis, self-encouragement
Meaning-focused coping	Cognitive appraisal theory / Primary–secondary model	Secondary control	Intrapersonal	Tendency to make effort to re-interpret the significance of a stressor or to find purpose in adversity	reframing, engaging in religious activities
Social support seeking	Social support theory	Primary / secondary control	Interpersonal	Tendency to reach out to one’s social network for reaping tangible or psychological resources	requesting financial aids, asking for advice

Approach–avoidant coping

In the approach–avoidance model of coping, coping styles are categorised in terms of two fundamental motivations: fight or flight (Roth & Cohen, 1986). The former motivation is characterised by a problem-focused coping style, with a high tendency to exert direct effort to confront external stressors or problems using strategies such as direct action and planning (Carver, 2011). The latter motivation is characterised by an avoidant coping style, with a high propensity to ignore the problems or escape from the circumstances altogether by resorting to passive strategies such as behavioural disengagement and denial (Carver, 2011).

In the context of the COVID-19 pandemic, the problem-focused coping style is predicted to be negatively associated with levels of anxiety and depressive symptoms, whereas the avoidant coping style is predicted to be positively associated with levels of these symptoms. This is because individuals having a problem-focused coping style are more ready to undertake voluntary preventive measures recommended by public health authorities (e.g., Landy et al., 2022), and adopting such measures has been found to be associated with fewer psychological symptoms (e.g., Guan et al., 2021). In contrast, individuals having an avoidant coping style tend to use more passive strategies, such as blunting and substance use, that may interfere with their adherence to COVID-19 preventive measures (Al-Hasan et al., 2020; Ebrahimi et al., 2021).

Primary–secondary control coping

The cognitive appraisal theory of coping (Lazarus & Folkman, 1984) and the primary–secondary model of coping (Rothbaum et al., 1982) explain individual differences in coping in terms of two broad control styles, namely primary and secondary control. The problem-focused coping style reflects a primary control orientation, whereas both emotion-focused and meaning-focused coping reflect a secondary control orientation. Individuals characterised by an emotion-focused coping style have a strong tendency to regulate their unpleasant emotions and make themselves feel better by deploying strategies such as self-encouragement and acceptance (Folkman & Lazarus, 1988a). Those who are characterised by a meaning-focused coping style have a strong tendency to modify their thoughts or find purpose in their adverse experiences by deploying strategies such as reframing and engaging in religious activities (Folkman & Moskowitz, 2007).

Similar to the problem-focused coping style, these two types of secondary control coping styles are predicted to be negatively associated with anxiety and depressive symptoms in the context of the COVID-19 pandemic. However, these associations are predicted to be modest only due to the heterogeneity nature of these two types of secondary control coping style. Specifically, some strategies of emotion-focused coping (e.g., humour, self-encouragement) target mood improvement, whereas others (e.g., emotional expression, acceptance) are more ambiguous in nature and may involve the experience of unpleasant emotions, relinquishment of control, or both (Austenfeld & Stanton, 2004). The same prediction applies to meaning-focused coping. For instance, in a study on religious coping (Lee et al., 2014), some participants used benevolent religious reappraisals (e.g., ‘God might wish to make me stronger through this experience’) while others used punitive ones (e.g., ‘God might wish to punish me through this experience’), and the use of such distinct types of meaning-focused strategy was found to be differentially associated with depressive symptoms. In light of these findings, we predict weak to modest negative associations with anxiety and depressive symptoms for both emotion-focused and meaning-focused coping styles in the COVID-19 pandemic context.

Intra-interpersonal focus

The two theoretical perspectives reviewed above emphasise coping as a personal effort. A third perspective broadens the scope in highlighting the goal and action of obtaining resources embedded in the environment (e.g., DeLongis & Holtzman, 2005; Lakey & Cohen, 2000). As a major coping style, social support seeking refers to the tendency to reach out to one’s social network to obtain both material or behavioural assistance (tangible support) and psychological resources (emotional support) rendered by others (Carver, 2011).

As the implementation of mandatory physical distancing measures has sharply reduced the availability of face-to-face social support, online social support has become especially important during the COVID-19 pandemic (e.g., Brailovskaia et al., 2021; Ni et al., 2020). Similar to previous research on social support, we predict negative associations between the social support seeking style and the symptoms of anxiety and depression in the pandemic context, but such associations were expected to be weak to modest because the meaning of social support is largely determined by the receivers’ subjective interpretations of the providers’ intention or goals (Cheng et al., 2008). In addition, culture may influence the manifestation of the benefits of social support because cultural regions vary in their relationship norms and tightness in social network ties (e.g., Gelfand et al., 2017; Triandis, 2001). Therefore, the hypothesised beneficial role of the social support seeking style may vary by cultural values. The upper panel of Table 2 summarises all the predicted coping style–symptom associations.

Table 2. Hypothesised and observed associations of five major types of coping style with anxiety and depressive symptoms (main effects) and Hofstede’s cultural dimensions as moderators.

	Main effect		Moderating effect
			Individualism		Power distance		Uncertainty avoidance		Masculinity
Style	Anxiety	Depression	Anxiety	Depression	Anxiety	Depression	Anxiety	Depression	Anxiety	Depression
	Hypothesised Association
PFC	−	−	ns	ns	ns	ns	ns	ns	ns	ns
AVC	++	++	ns	ns	ns	ns	ns	ns	ns	ns
EFC	−	−	ns	ns	ns	ns	High: −	High: −	High: −	High: −
							Low: −−	Low: −−	Low: −−	Low: −−
MFC	−	−	ns	ns	ns	ns	High: −	High: −	High: −	High: −
							Low: −−	Low: −−	Low: −−	Low: −−
SSS	−	−	High: −	High: −	ns	ns	High: −	High: −	ns	ns
			Low: −−	Low: −−			Low: −−	Low: −−
	Observed association
PFC	−−	−−	ns	ns	ns	ns	ns	ns	ns	ns
AVC	++	++	ns	ns	ns	ns	ns	ns	ns	ns
EFC	ns	ns	ns	ns	ns	ns	ns	ns	High: +	High: +
									Low: ns	Low: −−
MFC	−−	−−	ns	ns	ns	ns	ns	ns	ns	ns
SSS	ns	−	ns	ns	ns	ns	High: ++	High: ns	ns	ns
							Low: −−	Low: −−

Notes: AVC = avoidant coping style; EFC = emotion-focused coping style; High = 1.5 SD above the mean; Low = 1.5 SD below the mean; MFC = meaning-focused coping style; ns = non-significant moderating effect; PFC = problem-focused coping style; SSS = social support seeking style; − = weak to modest negative association; –− = moderately strong negative association; + =  weak to modest positive association; ++ = moderately strong positive association.

Cultural differences in coping style–psychological symptom associations

As the COVID-19 pandemic is a global health threat that has affected numerous countries worldwide (Ali et al., 2020), the between-country comparisons needed to be made using culture-moderated analysis. Hofstede’s (2001) cultural theory postulates that countries are characterised by distinct cultural values and norms. In light of these cultural differences, we predicted variation in the magnitude of associations between the coping style and psychological symptoms across countries in the pandemic context. The hypothesised culture-moderating effects are presented in the upper panel of Table 2.

As mentioned previously, Hofstede’s cultural dimension of individualism may moderate the magnitude of the hypothesised negative association between social support seeking style and psychological symptoms across countries. Studies have shown that people from countries with lower (vs. higher) individualism are more ready to render and accept social support (Chang, 2015) and are more likely to find social support effective in relieving stress (Li & Peng, 2021; Zhu et al., 2020). In this light, the hypothesised association between social support seeking style and psychological symptoms may be stronger for residents of countries lower (vs. higher) in individualism, particularly during the first wave of the COVID-19 pandemic when social activities were sharply curtailed due to the implementation of nationwide physical distancing measures for an extended period.

In addition to individualism, Hofstede’s cultural dimension of uncertainty avoidance is proposed as another relevant moderator in the COVID-19 pandemic, primarily because this dimension is closely linked with anxiety and stress related to uncertainty (Malik, 2021). In the initial wave, the implementation of a series of unprecedented drastic policies for curbing the highly contagious, novel virus has aroused widespread public anxiety (e.g., Cheng, Wang, & Ebrahimi, 2021; Randall et al., 2022). In countries low in uncertainty avoidance, residents may be more receptive to these new, practical disease-mitigation measures, because they are generally more tolerant of the ambiguity embedded in unpredictable situations (Hofstede, 2001). However, public receptiveness to the newly introduced policies with unknown consequences may be lower in countries high in uncertainty avoidance, whose residents tend to be uncomfortable with ambiguity. Although theories of coping flexibility postulate that secondary control coping (e.g., emotion-focused coping, meaning-focused coping) is more effective in relieving stress in unpredictable stressful events (Cheng et al., 2014; Folkman et al., 2000), the hypothesised negative association between the secondary control coping style and psychological symptoms are proposed to be stronger in countries with lower (vs. higher) uncertainty avoidance, especially in the initial wave of this new COVID-19 crisis when the ambiguity level was at its peak (e.g., Cheng, Wang, & Chan, 2021; Killgore et al., 2021).

A third relevant moderator is Hofstede’s cultural dimension of masculinity, which describes the extent to which a country values achievement versus the quality of life (Hofstede, 1998). Home-based teleworking was common in the first wave of the pandemic owing to nationwide lockdowns, including business closures (e.g., Buomprisco et al., 2021). Many employees have reported some extents of impediment in their work progress (e.g., Armitage & Nellums, 2020; Pulido-Martos et al., 2021). Secondary control coping is more relevant to psychological adjustment to these stressful life changes that are largely beyond one’s control (Cheng et al., 2014; Folkman et al., 2000). The hypothesised negative association between the secondary control coping style and psychological symptoms is proposed to be stronger in countries with lower masculinity than in those with higher masculinity, because the stronger needs for achievement characterised by residents of the latter cluster of countries may be unmet in the initial wave of the pandemic.

Although the countries differ vastly in cultural values and norms, there are also country differences in the extent of the outbreak. In the initial wave, Europe was the hardest-hit continent at that period, accounting for one-fifth of the COVID-19 confirmed cases across the globe (e.g., Rodríguez-Pose & Burlina, 2021). The governments all over the world also differed in the strictness of implementing COVID-19 containment measures and health policies in this wave (Middelburg & Rosendaal, 2020). Multinational comparisons have identified that such national-level differences in COVID-19 policy implementation varied by socioeconomic factors such as gross domestic product (GDP) per capita (Wang et al., 2021). All these between-country variations should be taken into consideration when making cross-cultural comparisons in the pandemic context, and thus COVID-19 confirmed cases, deaths, and strictness of disease containment measures imposed by the government were included as covariates in the present culture-moderated analysis.

Materials and methods

The guidelines defined in both the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 Statement (Page et al., 2021) and the Meta-Analysis Reporting Standards (MARS; American Psychological Association, 2010) were followed for selecting eligible articles, conducting analyses, and reporting results in this meta-analysis. The study protocol was pre-registered in the Open Science Framework (https://osf.io/4cxdm/) before commencement of this meta-analysis.

Eligibility criteria

Inclusion and exclusion criteria were set to select eligible articles available through November 2022, and eligible articles included both published and unpublished reports to address a potential file drawer problem. The inclusion criteria were limited to empirical studies conducted during the COVID-19 pandemic that examined the associations between the coping style and the two major domains of psychological symptoms (i.e., anxiety symptoms and depressive symptoms). The exclusion criteria were as follows: (a) articles that did not contain any empirical data, (b) none of the phases of the studies were conducted during the COVID-19 outbreak, (c) studies that did not include measures of both coping style and psychological symptoms, (d) a sample size that was less than 10, (e) duplicated samples, (f) no relevant data or statistics were available for computing effect sizes, and (g) the full text was unavailable for coding relevant data. All of these criteria were defined before the coding procedures began, and articles that met the predefined inclusion and exclusion criteria were selected for this meta-analysis.

Information sources and search strategies

A three-stage search process was carried out using a pre-defined literature search strategy. First, a computer-aided search was conducted in 28 relevant electronic databases spanning two major broad domains of health and medicine (e.g., British Nursing Index, MEDLINE^®) and social sciences (e.g., APA PsycInfo^®, Sociological Abstracts). Second, additional efforts were made to identify unpublished work. Dissertations and theses were identified by searching three electronic databases (i.e., Open Access Theses and Dissertations, Networked Digital Library of Theses and Dissertations, and ProQuest Dissertations & Theses A&I), while unpublished papers were located through searching the grey literature (i.e., Google Scholar, OpenGrey, and OpenSIGLE). Finally, the search process was complemented by further searching the reference lists of full-text eligible articles.

As meta-analysis has been widely recommended to be as inclusive as possible (e.g., Cooper et al., 2019; Turkiewicz, 2017), we did not impose any restrictions regarding the demographic characteristics of respondents, country in which the study was conducted, study design, language of publication, and publication status. The search procedures were first conducted in March 2021 and then updated in July 2021 and November 2022 by two independent coders, both of whom were graduate assistants with training in meta-analysis. Ninety-one authors were approached for full-text papers, additional data, or further information for possible inclusion, and 33 had responded to our requests.

For conducting these search strategies, the Boolean search method was adopted, and truncation and wildcards were also used to broaden the scope of the search. Specifically, a set of key query terms was derived from the following keyword combinations using Boolean operators: ‘(COVID OR coronavirus disease OR SARS-CoV-2 OR severe acute respiratory syndrome) AND (cope OR coping OR stress*) AND (anxi* OR depressi*)’. This set of terms was applied primarily in searching the title, abstract, and keywords of potentially relevant articles. After the initial screening, the full-text version was retrieved and perused for eligibility. The full selection process is shown in the PRISMA 2020 flow chart (see Figure 1).

Figure 1. Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram. Source: Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ 2021;372:n71. doi:10.1136/bmj.n71.

Search and data collection process

Before starting the tasks of article screening and data extraction, duplicated articles derived from multiple database searches were omitted. All of the citations were imported into the Endnote® citation management software (version 20.4.1; Clarivate Analytics, PA, USA) to exclude duplicated citations. The same two reviewers then screened and coded all of the de-duplicated citations, abstracts, and full-text papers on their own. The coding decisions were guided by a coding manual with pre-determined categories and their coding values (see Supplementary Table S1). Titles and abstracts were initially screened for inclusion, and then full texts of the screened articles were retrieved and reviewed at a later stage.

Non-English abstracts were initially translated by Google Translate at the screening stage, and then relevant non-English text were translated by a native translator of that particular language at the coding stage. Multiple eligible articles derived from the same dataset (e.g., an unpublished dissertation was later published as a journal article) were collapsed together and coded as a single study. We coded all the relevant effect sizes from studies derived from the same dataset.

To increase the accuracy and consistency in coding, both reviewers received detailed instructions and rigorous training from the authors regarding how to apply the eligibility criteria specified in the coding manual in data extraction. The manual was pilot-tested on 10% of the articles. The two reviewers conducted the coding task independently using a standardised MS-Excel form, and then met with the first author to discuss and resolve any discrepancies in coding. The coding manual was refined through consensus. After the training, each reviewer continued to code the remaining articles using the refined set of coding categories. Any between-reviewer divergence in coding was resolved by consensus with the reviewers and the first author. Inter-reviewer reliabilities were assessed by Krippendorff’s alpha (Krippendorff, 2018), which was computed using the KALPHA macro for SPSS (Hayes & Krippendorff, 2007). Such estimates were reasonably high for the final codings (Krippendorff’s α > .77).

Coding items

The effect size was estimated by the bivariate Pearson product-moment correlation coefficient (r), which indicated the magnitude of the association between a coping style and a domain of psychological symptoms (anxiety or depressive symptoms) for an independent sample. If the Pearson r was not reported in an eligible article, other effect size estimates and statistics were converted into Pearson r using the formulas provided by Lipsey and Wilson (2001), or the authors of that article were approached in an effort to obtain the relevant data.

The following variables were coded for each eligible article: author(s), publication year, publication status (1 = published; 0 = unpublished), and study period (start and end dates of the study). In addition, some participants’ characteristics were coded for each independent sample as follows: sample size, country/region in which the study was conducted, sex composition (percentage of female participants within a sample), age composition (mean or median age of a sample), education level (degree holder or not), socioeconomic status (SES; lower, middle, or upper socioeconomic class), and clinical status (mental health, physical health, COVID-19, or non-clinical samples). As some coping measures, such as the Coping Orientation to Problems Experienced (COPE; Carver et al., 1989), contain both general or situational versions, the situational specificity of coping was coded for each assessment tool (i.e., 0 = general coping; 1 = situational coping with specific reference to the COVID-19 pandemic context). Moreover, four cultural moderators (i.e., individualism, power distance, uncertainty avoidance, and masculinity) were extracted from Hofstede’s (2018) matrix of cultural data. For eligible reports that involved more than one country, separate data were coded for each country. If separate data were unavailable, the data would be omitted from the culture-moderated analysis.

As the COVID-19 pandemic had a large effect on public health and the economy, we included some relevant control variables to partial out their potential confounding effects in hypothesis testing. Specifically, the infection and mortality rates of COVID-19 were obtained from the database compiled by the European Centre for Disease Prevention and Control (2021). The number of confirmed COVID-19 cases and deaths were available on a daily basis. The daily COVID-19 data of a given country during the specific study period were extracted accordingly, and then the averaged data were divided by the population size of that country. Moreover, the degree of strictness regarding the COVID-19 containment measures during the study period was coded along a 4-point scale: 0 = no measures, 1 = recommend not leaving house, 2 = require not leaving house with minimal exceptions (e.g., allowed to leave once a week, or only one person can leave at a time, etc.), 3 = require not leaving house with exceptions for daily exercise, grocery shopping, and ‘essential’ trips. The data were extracted from the database of the Oxford COVID-19 Government Response Tracker and the coding scheme was derived from their codebook (Thomas et al., 2020). Finally, the GDP per capita, unemployment, and purchasing power parity conversion factor were extracted from the World Bank database (World Bank, 2022), which provided only annual data. For each eligible article, the figures of the country during the study period were extracted.

Measures

Table 3 lists the percentages of use for all of the frequently adopted measures of coping style and both types of psychological symptoms, respectively. For measures of the coping style, the COPE, or its abbreviated version Brief COPE (Carver, 1997), was the most frequently adopted in the pool of eligible articles. This multidimensional inventory was also the most popular measure of dispositional coping in the literature. Moreover, the Brief Resilient Coping Scale (Sinclair & Wallston, 2004), Connor–Davidson Resilience Scale (Connor & Davidson, 2003), and Simplified Coping Style Questionnaire (Xie, 1998) were also popular measures.

Table 3. List of measures of study variables and percentage of use in the pool of 138 eligible reports.

Measure	%
Coping style
COPE (Carver et al., 1989) / Brief COPE (Carver, 1997)	33
Simplified Coping Style Questionnaire (Xie, 1998)	9
Brief Resilient Coping Scale (Sinclair & Wallston, 2004)	6
Connor–Davidson Resilience Scale (Connor & Davidson, 2003)	4
Brief Resilience Scale (Smith et al., 2008)	3
Ways of Coping Questionnaire (Folkman & Lazarus, 1988b)	3
Brief Religious Coping Scale (Pargament et al., 2011)	3
Coping Inventory for Stressful Situations (Endler & Parker, 1999)	3
Multidimensional Scale of Perceived Social Support (Zimet et al., 1988)	3
Chinese Trait Coping Style Questionnaire (Jiang, 1999)	2
Coping Flexibility Inventory (Cheng, 2001)	2
Coping Style Scale (Xiao & Xu, 1996)	2
Self-developed measures of coping style	12
Other less popular measures of coping style (used only once)	15
Anxiety symptoms
Generalised Anxiety Disorder Screener (Spitzer et al., 2006)	28
Subscale of Depression Anxiety Stress Scales (Lovibond & Lovibond, 1995)	18
Subscale of Hospital Anxiety and Depression Scale (Zigmond & Snaith, 1983)	8
State-Trait Anxiety Inventory (Spielberger, 1983)	5
Zung Self-Rating Anxiety Scale (Zung, 1971)	4
Coronavirus Anxiety Scale (Lee, 2020)	3
Subscale of Patient Health Questionnaire (Spitzer et al., 1999)	3
Subscale of Patient-Reported Outcomes Measurement Information System (PROMIS®) (Cella et al., 2010)	2
Subscale of Brief Symptom Inventory (Derogatis, 1993)	2
Subscale of General Health Questionnaire (Goldberg, 1978)	2
Penn State Worry Questionnaire (Meyer et al., 1990)	2
Short Health Anxiety Inventory (Abramowitz et al., 2007)	2
Self-developed measures of anxiety symptoms	12
Other less popular measures of anxiety symptoms (used only once)	9
Depressive symptoms
Subscale of Patient Health Questionnaire (Spitzer et al., 1999)	31
Subscale of Depression Anxiety Stress Scales (Lovibond & Lovibond, 1995)	27
Centre for Epidemiologic Studies Depression Scale (Radloff, 1977)	14
Subscale of Hospital Anxiety and Depression Scale (Zigmond & Snaith, 1983)	10
Subscale of Brief Symptom Inventory (Derogatis, 1993)	6
Subscale of Patient-Reported Outcomes Measurement Information System (PROMIS®) (Cella et al., 2010)	2
Subscale of General Health Questionnaire (Goldberg, 1978)	2
Zung Self-Rating Depression Scale (Zung, 1965)	2
Other less popular measures of depressive symptoms (used only once)	6

For measures of psychological symptoms, the Depression Anxiety Stress Scales (Lovibond & Lovibond, 1995) and Hospital Anxiety and Depression Scale (Zigmond & Snaith, 1983) were commonly used to assess anxiety or depressive symptoms in the pool of eligible articles. Moreover, the Generalised Anxiety Disorder Screener (Spitzer et al., 2006) and State-Trait Anxiety Inventory (Spielberger, 1983) were popular measures of anxiety symptoms, whereas the Centre for Epidemiological Studies Depression (Radloff, 1977) and Patient Health Questionnaire Depression Scale (Spitzer et al., 1999) were popular measures of depressive symptoms.

Synthesis methods

The present meta-analysis investigated the associations between coping style and psychological symptoms. A total of 10 effect size estimates (Pearson r) were examined because the study focused on five major coping styles (i.e., problem-focused, avoidant, emotion-focused, meaning-focused, and social support seeking) and two widely reported domains of psychological symptoms (i.e., anxiety and depressive symptoms). Effect size estimates derived from individual studies were pooled using a three-level random-effects meta-analysis (Cheung, 2015). The between-study heterogeneity was evaluated using Cochran’s Q, Higgins’ I², and τ² statistics (Borenstein et al., 2009).

Some researchers examined the same hypothesised predictor-criterion association with different samples (e.g., Japanese and US samples), and reported the various effect sizes of the hypothesised association in the same article. In such cases, the samples were nested within the same article. The multiple effect sizes reported in the same article were thus not independent, thereby violating the assumption of sample independence in a conventional meta-analysis. To address this issue, we followed the recommendations by Cheung (2019) and conducted a three-level meta-analysis. Specifically, in addition to the two-level structure of a conventional meta-analysis (i.e., participants at Level 1 were nested within the samples of the effect size estimates at Level 2), the effect sizes at the sample level were further nested within the articles at Level 3.

A moderation analysis was carried out to investigate when the magnitude of an effect size estimate would vary by the values or levels of a moderator variable. Meta-regression was conducted for both continuous and dummy-coded moderators. A total of 10 meta-regression analyses were performed, one for each effect size estimate. All of the moderators and control variables were entered simultaneously into each of the meta-regression models. A simple slope analysis was applied to visualise a significant moderating effect using a graph.

A p value of less than .05 was adopted as an indicator of statistical significance for hypothesis testing in all of the present analyses. Unless otherwise specified, all of the meta-analyses were performed using the metafor package (version 3.0-2; Viechtbauer, 2010) in R language (version 4.1.0, 2021-05-18; R Core Team, 2020).

Assessments of certainty and study quality

Before conducting the main analyses, outliers were first identified through the outlier and influence diagnostics method recommended by Viechtbauer and Cheung (2010) using the influence.rma.uni () function within the metafor R package. The outliers identified through this method were excluded to check whether the meta-analytic results conducted using the trimmed dataset (i.e., with outliers removed) would differ from those derived from the full dataset.

As publication bias might threaten the credibility of meta-analytic findings, this potential problem was evaluated using several graphical and statistical methods. First, a funnel plot was constructed for visual inspection of all of the effect size estimates, and an asymmetrical plot was used to reflect the publication bias. Second, Egger’s linear regression test was used as a statistical test of the presence of asymmetry in the meta-analytic data (Egger et al., 1997). Third, the trim-and-fill test was used as an alternative approach to detect data asymmetry, and it provided statistical adjustment if publication bias was detected (Duval, 2005). After a symmetrical plot was yielded from the trim-and-fill procedures, an adjusted effect size was obtained for comparison with the estimated effect size from the original data. Finally, the stepwise weight-function approach was adopted to determine whether the meta-analytic findings were at a high or low risk of publication bias (Vevea & Woods, 2005). The first three methods were performed using the Comprehensive Meta Analysis software (version 2.2.020; Biostat Inc., Englewood, NJ, the US), and the fourth was conducted using the weightr R package (version 2.0.2; Coburn & Vevea, 2019).

In addition to publication bias, another common problem was p-hacking, which refers to the practice of performing additional analyses or adding new data until non-significant findings become statistically significant (e.g., Wicherts et al., 2016). To detect the presence of such practices, the p-curve method was adopted to determine whether effect size estimates that were statistically significant reflected true effects or mere artifacts (Simonsohn et al., 2015). A p-curve analysis was conducted using the p-curve app (version 4.06; Simonsohn et al., 2017).

Finally, the risk of bias due to poor methodological quality was evaluated using study quality assessment (Johnson et al., 2015). The following five indicators were assessed: statistical power (1 = adequate, 0 = inadequate), sampling method (1 = probability, 0 = non-probability), study design (1 = longitudinal/prospective, 0 = cross-sectional), measurement reliability (proportion of reliable measures), and measurement validity (proportion of valid measures). The same two independent reviewers coded these indicators, and the inter-coder reliability of all five of these indicators was high (Krippendorff α > .73). These individual scores were aggregated to generate an overall rating, with a lower value indicating a poorer methodological or reporting quality in a particular domain for an eligible article. For evaluation of study quality, a composite score of less than 2.00 indicated low study quality, a composite score between 2.00 and 3.99 indicated moderate study quality, whereas a composite score greater than 3.99 indicated high study quality.

Results

Study characteristics

The final pool of eligible articles comprised 138 reports with 151 independent samples. All of these reports were journal articles available in or before November 2022, with the exception of a dissertation (Hammond, 2021) and an unpublished report (Ayoub et al., 2022). The studies were conducted during the first wave of the COVID-19 pandemic from January to July 2020.

In the independent samples, the sample size ranged between 25 and 13,263, with a mean number of 908. The mean age composition in the pool of eligible articles was 36.08 (SD = 12.97, range: 4–88). The samples tended to be female-dominated, with an average female sex composition of 66%. The samples were from 44 countries and territories spanning eight geographical regions worldwide. A total of 18% of the samples were from North America (Canada, the US), 1% were from Oceania (Australia), 12% were from Northern/Western Europe (e.g., Germany, the UK), 26% were from Southern/Eastern Europe (e.g., Croatia, Greece), 27% were from Asia (e.g., China, India), 10% were from the Middle East (e.g., Israel, Saudi Arabia), 5% were from Central/South America (e.g., Argentina, Brazil), and 1% were from Africa (Egypt, Morocco). Half were individualist (vs. collectivist) countries, slightly more than half (52%) of the countries were high (vs. low) in power distance, half of the countries were high (vs. low) in uncertainty avoidance, and about two thirds (67%) of the countries were high (vs. low) in masculinity.

Outlier analysis

The outlier detection methods identified 7 outliers in 5 of the following coping–symptom estimates: emotion-focused coping–anxiety association (k = 2), avoidant coping–anxiety association (k = 2), meaning-focused coping–anxiety association (k = 1), emotion-focused coping–depression association (k = 1), and avoidant coping–depression association (k = 1).

The results from outlier analyses showed that the patterns derived from both the trimmed and full datasets were highly similar, with the exception of the association between meaning-focused coping style and anxiety symptoms. The negative association between meaning-focused coping style and anxiety symptoms was non-significant for the full dataset, but it became significant after the outlier had been removed. Because of such inconsistency between the two sets of findings, the results pertaining to the meaning-focused coping–anxiety association should be interpreted with caution. The meta-analytic findings reported below were derived from the full dataset because the effects exerted by the outliers were minimal.

Main-effect analysis

A main-effect analysis was performed to examine the magnitude of the associations between the coping style and psychological symptoms reported during the first wave of the COVID-19 pandemic. The results are summarised in Table 4. In line with our predictions, both problem-focused and meaning-focused coping styles were negatively associated with anxiety and depressive symptoms (both p’s < .007), whereas avoidant coping style was positively associated with these two domains of psychological symptoms (both p’s < .0001).

Table 4. Main-effect analysis of coping style–symptom associations obtained during the first wave of COVID-19 pandemic.

Symptom	Anxiety symptom					Depressive symptom
Coping style	PFC	AVC	EFC	MFC	SSS	PFC	AVC	EFC	MFC	SSS
Main-effect analysis
Averaged r	−0.11	0.31	0.04	−0.08	−0.04	−0.19	0.33	0.01	−0.13	−0.08
Lower 95% CI	−0.15	0.26	−0.03	−0.13	−0.09	−0.23	0.28	−0.09	−0.19	−0.14
Upper 95% CI	−0.07	0.36	0.11	−0.02	0.02	−0.15	0.38	0.10	−0.06	−0.01
K	128	141	96	70	75	106	132	73	60	67
Tests for heterogeneity
Q	3602.18	4289.94	5274.76	1353.40	2074.08	3554.35	4322.49	6670.20	1645.60	1935.72
Df	127	140	95	69	74	105	131	72	59	66
p value for Q	< .001	< .001	< .001	< .001	< .001	< .001	< .001	< .001	< .001	< .001
I^2	96.66	97.31	98.36	95.83	95.95	96.36	97.01	98.83	96.38	96.64
Lower 95% CI	94.45	95.82	96.83	89.69	92.49	93.54	94.90	97.87	91.94	93.81
Upper 95% CI	97.89	98.25	99.06	98.21	97.77	97.81	98.17	99.34	98.17	98.19
τ^2	0.03	0.05	0.08	0.04	0.03	0.03	0.05	0.10	0.04	0.04
Lower 95% CI	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00
Upper 95% CI	0.05	0.08	0.14	0.09	0.06	0.04	0.08	0.19	0.08	0.08

Notes: AVC = avoidant coping; CI = confidence interval; df = degree of freedom; EFC = emotion-focused coping; I² = percentage of heterogeneity variance to total variance; k = number of effect size estimate; MFC = meaning-focused coping; PFC = problem-focused coping; Q = Cochran’s Q-statistic; r = effect size estimate (Pearson product-moment correlation coefficient); SSS = social-support seeking; τ²=  total amount of heterogeneity variance.

The social support seeking style was negatively associated with depressive symptoms (p = .02), but not with anxiety symptoms (p = .17). Moreover, the emotion-focused coping style was not significantly associated with either of these symptoms (both p’s > .22). These results provided partial support for our hypotheses.

Culture-moderated analysis

Before conducting a culture-moderated analysis, the scores of the various Hofstede’s cultural dimensions were standardised (i.e., converted into Z-scores). Only statistically significant moderating effects are described below (all p’s < .05). The full results are reported in Table 5.

Table 5. Summary of cultural-moderated analysis with COVID-19-related and socioeconomic variables as covariates.

	PFC		AVC		EFC		MFC		SSS
Variable	Cultural model	Full model	Cultural model	Full model	Cultural model	Full model	Cultural model	Full model	Cultural model	Full model
Criterion: anxiety symptom
Individualism^a	.08**	.09	.07	.05	.00	−.08	−.01	.03	.04	.07
Power distance^a	.03	.04	.04	.04	−.01	.01	.02	.10	.02	.13*
Uncertainty avoidance^a	.01	.02	−.01	−.03	.03	.13	.00	.06	.05*	.18***
Masculinity^a	−.02	−.04	.02	.02	.08	.16*	−.04	−.03	−.04*	−.04
COVID-19 confirmed cases (per 100k)^b		.04		.09		.07		.07		−.07
COVID-19 deaths (per 100k)^b		.02		−.07		−.04		.07		.02
COVID-19 containment strictness^c		.01		−.07		−.04		−.02		.05
GDP per capita^d		−.04		.04		.06		.05		.09
Unemployment^d		−.05		−.05		.00		−.05		−.01
Purchasing power parity^d		.01		−.07		.06		.12		.15*
Criterion: depressive symptom
Individualism^a	.03	.09	.05	.04	−.02	−.07	−.04	−.01	.06	.14
Power distance^a	.03	.02	.01	.02	.00	−.08	.02	−.03	.07	.08
Uncertainty avoidance^a	.02	.04	.02	.08	.03	.11	.00	−.07	.08*	.13*
Masculinity^a	.01	.00	.02	.05	.14**	.25**	.02	−.04	.01	−.02
COVID-19 confirmed cases (per 100k)^b		.05		.03		−.06		−.08		−.01
COVID-19 deaths (per 100k)^b		−.08		.05		−.03		.04		.02
COVID-19 containment strictness^c		−.01		.06		.04		−.08		.01
GDP per capita^d		−.08		−.05		−.15		−.01		−.07
Unemployment^d		−.05		−.11*		−.04		.04		−.04
Purchasing power parity^d		−.01		−.03		−.17		.13		.05

Notes: AVC = avoidant coping style; EFC = emotion-focused coping style; MFC = meaning-focused coping style; PFC = problem-focused coping style; SSS = social support seeking style. *p < .05. **p < .01. ***p < .001.

^a Hofstede’s (2018) cultural data matrix.

^b European Centre for Disease Prevention and Control (2021).

^c Oxford COVID-19 Government Response Tracker.

^d World Bank database (2022).

As predicted, significant moderating effects were found in two of the cultural dimensions, namely uncertainty avoidance and masculinity. For uncertainty avoidance, there were significant moderating effects for the associations between the social support seeking style and both domains of psychological symptoms (β = .18, SE = .05, 95% CI [.09, .26] for anxiety symptoms and β = .13, SE = .05, 95% CI [.03, .23] for depressive symptoms). The simple slope analysis showed opposite patterns of findings regarding the associations between the social support seeking style and psychological symptoms in countries with lower versus those with higher uncertainty avoidance. Specifically, the hypothesised negative associations were found only in countries with lower uncertainty avoidance (β = −.29, SE = .08, 95% CI [−.44, −.14] for anxiety symptoms and β = −.25, SE = .09, 95% CI [−.43, −.08] for depressive symptoms); but unexpectedly, positive associations were found in countries with higher uncertainty avoidance (β = .24, SE = .08, 95% CI [.09, .39] for anxiety symptoms and β = .14, SE = .09, 95% CI [−.03, .30] for depressive symptoms). These moderation effects are graphically presented in Figure 2.

Figure 2. Meta-regression plot showing the moderating effects of uncertainty avoidance on the association between social support seeking style and psychological symptoms (upper panel: anxiety symptom; lower panel: depressive symptom). The solid line represents linear predictions for the effect size estimate, whereas the dashed lines represent the upper and lower limits of 95% confidence interval. The horizontal dotted line shows a null effect (Pearson r =  0.00).

For masculinity, significant moderating effects were found for the associations between the emotion-focused coping style and both domains of psychological symptoms (β = .16, SE = .08, 95% CI [.01, .32]) for anxiety symptoms and (β = .25, SE = .10, 95% CI [.06, .44] for depressive symptoms). The simple slope analysis showed negative associations between the emotion-focused coping style and psychological symptoms in countries with lower masculinity (β = −.24, SE = .14, 95% CI [−.51, .04] for anxiety symptoms and β = −.41, SE = .16, 95% CI [−.72, −.09] for depressive symptoms), but positive associations were obtained in countries with higher masculinity (β = .25, SE = .12, 95% CI [.01, .49] for anxiety symptoms and β = .34, SE = .16, 95% CI [.03, .64] for depressive symptoms). The results are depicted in Figure 3.

Figure 3. Meta-regression plot showing the moderating effects of masculinity on the association between emotion-focused coping style and psychological symptoms (upper panel: anxiety symptom; lower panel: depressive symptom). The solid line represents linear predictions for the effect size estimate, whereas the dashed lines represent the upper and lower limits of 95% confidence interval. The horizontal dotted line shows a null effect (Pearson r = 0.00).

Post-hoc moderation analysis

As meta-analysis has been widely recommended to be as inclusive as possible, no restrictions were placed on study and sample criteria to obtain as many eligible reports as possible. The current pool of eligible reports yielded considerable between-study heterogeneity that might influence the meta-analytic findings. Hence, post-hoc (unplanned) moderation analyses were performed to further explore these potential influences. Additional meta-regression analyses were conducted to examine the potential moderating effects of four sample characteristics (sex composition, age composition, education level, and SES), clinical status, COVID-19 (vs. general) coping, and study design. The results are summarised in Table 6.

Table 6. Summary of post-hoc moderation analysis by moderator variable.

	Association (r) of psychological symptoms with
Moderation: higher (vs. Lower) level	PFC	AVC	EFC	MFC	SSS
Sample characteristics
Higher (vs. lower) percentage of females	stronger	ns	stronger	ns	ns
Older (vs. younger) age composition	ns	ns	stronger	ns	weaker
Degree (vs. non-degree) holder	stronger	weaker	weaker	ns	weaker
Middle/upper (vs. lower) SES	weaker	weaker	ns	ns	stronger
Clinical status
Mental health (vs. no) diagnosis	ns	stronger	stronger	ns	ns
Physical health (vs. no) diagnosis	ns	ns	ns	weaker	ns
COVID-19 (vs. no) diagnosis	ns	weaker	ns	ns	ns
Measure characteristics
COVID-19 (vs. general) coping	ns	ns	weaker	stronger	stronger
Study characteristics
Longitudinal (vs. cross-sectional) design	ns	ns	ns	ns	weaker
Higher (vs. lower) study quality	ns	ns	weaker	ns	weaker

Notes: AVC = avoidant coping style; EFC = emotion-focused coping style; MFC = meaning-focused coping style; ns = inconsistent or non-significant moderating effect (p < .05); PFC = problem-focused coping style; SSS = social support seeking style. For clarity sake, the table only shows significant moderation effects after Bonferroni correction of 50 comparisons (i.e., .05 / 50 = .001, such that significance level is p < .001).

The post-hoc moderation analyses revealed a tendency of the association between the problem-focused coping style and both domains of psychological symptoms to be stronger for samples with more females, a higher education level, or a lower SES. The association between the avoidant coping style and psychological symptoms was generally stronger for samples lower in education level or SES, with a mental health diagnosis, or without a COVID-19 diagnosis. A stronger association between the emotion-focused coping style and psychological symptoms was generally found in samples with more females, who were older, with a lower education level, or with a mental health diagnosis; or in studies adopting a general measure of coping. The association between the meaning-focused coping style and psychological symptoms tended to be stronger for samples without a physical health diagnosis (excluding COVID-19), or for studies adopting a situational measure of COVID-19 coping. A stronger association between the social support seeking style and psychological symptoms tended to be found in samples who were younger, with a lower educational level, higher in SES; or in studies adopting a situational measure of COVID-19 coping.

Assessments of certainty and study quality

For certainty assessment, a statistical power analysis showed that the present meta-analysis was sufficiently powered to detect effects of the magnitude of the associations between the coping style and psychological symptoms.

For publication bias detection, only two unpublished reports has been identified because some unpublished papers have already been published when we wrote this article, probably due to the expedited review and publication processes adopted by many journals for COVID-19-related articles. The lack of variability made it inappropriate to conduct publication bias analyses that compared published versus unpublished work. Instead, the p-curve analysis was more apt to identify the problem of p-hacking, which was more prevalent among published work. The results showed no evidence of p-hacking in any of the coping style–symptom associations.

For study quality assessment, the overall quality was rated moderate (M = 2.39; SD = 0.76), indicating that the pool of eligible reports generally had moderate methodological quality. Specifically, most (91%) of the reports included a sample that was large enough to yield adequate statistical power. However, the reliability of one-third of the measures was unknown while evidence for scale validation was missing in 32% of the measures, mostly translated versions or general coping measures adapted for assessing situational coping in the COVID-19 context. More severe methodological concerns were found in sample representativeness, with only 3 samples drawn from probabilistic sampling; and also in study design, with 11% investigated temporal changes using a longitudinal or prospective design.

As shown in Table 6, the moderating effects of study quality were significant for the associations between two coping styles (i.e., emotion-focused coping, social support seeking) and psychological symptoms. We included study quality as an additional covariate and reran the culture-moderated analyses, and a similar pattern of findings was obtained.

Discussion

The present culture-moderated meta-analysis revealed universal findings regarding the hypothesised negative associations between the problem-focused coping style and both anxiety and depressive symptoms, as well as the hypothesised positive associations between the avoidant coping style and these two symptom domains reported in the initial wave of the COVID-19 pandemic. However, the pattern of findings for the coping styles of both emotion-focused coping and social support seeking is more complex across countries with varying levels of uncertainty avoidance and masculinity, respectively.

In countries with higher levels of masculinity, unexpected positive associations are found between the emotion-focused coping style and the two domains of psychological symptoms, but there are strong negative associations between the problem-focused coping style and psychological symptoms. Such contrary findings may be explained by the specific cultural norms emphasised in these countries that value achievement and success. Residents of these countries tend to be more goal-oriented and to display more instrumental behaviours than those of the countries lower in masculinity (Cyr & Head, 2013). In this light, the problem-focused coping style, with a high propensity of confronting and tackling problems proactively, may fit well with the cultural norms of achievement valued in countries higher in masculinity. However, the use of emotion-focused coping is largely irrelevant to deal with problems directly, and thus may not fit well with the achievement-oriented cultural norms in these countries. Previous studies have revealed that individuals’ coping deployment that fails to align with the specific norms of their cultural regions is associated with psychological problems (e.g., Yeh, 2003). This mental health issue may be especially prominent during the initial wave of the COVID-19 pandemic in which many working adults had no prior experience in home-based teleworking, and experienced a range of work-related stress (e.g., ‘technostress’, environment-oriented problems) that disrupted teleworking during the lockdown period (e.g., Armitage & Nellums, 2020; Persol Research and Development, 2020).

In countries with higher levels of uncertainty avoidance, unexpected positive associations are found between the social support seeking style and anxiety symptoms. These unpredicted findings may be attributable to the specific study context where the etiology and treatment of the atypical virus are largely unknown. When feeling uncertain, people tend to rely on others for support, endorsement, or assurance (e.g., Hill & Hamm, 2019; Paralkar & Knutson, 2021). Such a tendency of social support seeking may thus be especially strong in individuals from countries with higher uncertainty avoidance. However, engagement in social activities for accessing social support or tangible services have been limited in the first wave of the pandemic in which extensive physical distancing measures have been implemented nationwide. In addition, social support networks tend to shrink during the pandemic as the number of deaths and hospitalised people increase (Kaniasty et al., 2020). Although information support can still be obtained online in the digital age, the strong need for social support may not be fully gratified due to the prevalent problems of fake news and misinformation (e.g., Germani & Biller-Andorno, 2021; Siebenhaar et al., 2020), which in turn elicited greater psychological symptoms in the initial wave of the pandemic (e.g., Cheng et al., 2020; Rocha et al., 2021). Therefore, the positive association between the social support seeking style and anxiety symptoms obtained in countries with higher uncertainty avoidance may reflect the failure of gratifying the heightened need for social support among their residents in this highly uncertain period.

Theoretical, research, and practical implications

The COVID-19 pandemic has had a large effect on many aspects of life and elicited enormous stress, but the present culture-moderated meta-analysis indicates that such influences tend to vary considerably among diverse cultural regions in the initial wave. Such intricate findings imply that both context and culture should be considered when studying the associations between various coping styles and mental health issues. The new findings thus call for more nuanced analyses that provide further insights regarding how and why certain coping styles are differentially associated with major symptom domains in both pandemic and cultural contexts.

Although cultural theories postulate multiple cultural dimensions that influence people’s thoughts and behaviour, most cross-cultural studies have only focused on a single dimension of individualism (e.g., Triandis, 2001; van Uchelen, 2000). The present study demonstrates that only focusing on a single cultural dimension may miss the ‘big picture’, especially in the context of the COVID-19 pandemic. The popular dimensions of individualism may be less relevant in the first wave because of the nationwide mandatory implementation of disease containment measures in this specific pandemic context. Specifically, between-country differences in individualism are minimised owing to a sharp curtailment of face-to-face interactions, and the social ties have become weaker for most if not all people.

Surprisingly, this meta-analysis has shown cultural differences in the two cultural dimensions that have been less frequently studied – uncertainty avoidance and masculinity – during the COVID-19 pandemic. These findings might be attributable to the core cultural values highlighted in these two dimensions (i.e., intolerance of ambiguity, achievement) that match well with the major concerns (i.e., unprecedented nationwide mitigation measures of a novel virus, teleworking issues) emerged in this particular stressful context. In light of these new cultural findings, a more comprehensive cultural perspective should be adopted when making cross-cultural comparisons to supplement existing cross-cultural research that focused on only a single, popular cultural dimension of individualism, especially in the context of unusual or exceptional stressful encounters.

Importantly, the intricate cultural findings identified in this meta-analysis may provide additional insights into cross-cultural counselling (Lee & Zalkalne, 2017). Specifically, mental health professionals should be aware of not only the clients’ coping style and mental health problems but also their cultural backgrounds and the specific stressful context they are in. Such cultural sensitivity and contextual understanding foster the delivery of culturally appropriate intervention strategies that best meet the clients’ fundamental needs (Gundel et al., 2020; King & Stewart, 2019). Case studies have indicated that people from diverse cultural backgrounds differ considerably in their core concerns, source of stress, preferred coping style, and mental health issues (e.g., King & Stewart, 2019). In line with these cultural findings, the present results suggest that when dealing with clients from countries with high uncertainty avoidance, greater effort may be expended to work with their anxiety arisen from their intolerance of ambiguity in an unknown stressful encounter. When dealing with clients from countries with high masculinity, however, more attention may be paid on working with achievement-related issues.Hofstede’s (2001) cultural framework may serve as a reference that systematically guides the intervention design such that functional counselling relationships can be established with clients from diverse cultural backgrounds. The treatment efficacy can be enhanced if the clients’ coping tendency and their core cultural-specific issues faced in a stressful encounter are all taken into account (Gundel et al., 2020).

Limitations, research directions, and concluding remarks

There are some limitations to the present meta-analysis. Specifically, this meta-analysis showed considerable cross-cultural differences in the emotion-focused coping style, but such differences are absent in another type of secondary control coping, namely the meaning-focused coping style. It is reasonable to infer that meaning-focused coping style is closely related to religion, which is a specific subset of culture rather than culture in general (Edara, 2017). As shown in multinational research, countries are also considerably different in the extent of religiosity. Countries such as Thailand and Oman have the highest religiosity, while countries such as Japan and Sweden have the lowest religiosity (Crabtree, 2010). However, religiosity is not reflected in any of the cultural dimensions proposed in Hofstede’s (2001) cultural theory, which has been the most widely researched in the cross-cultural literature. To address this unexplored issue, researchers should explore the potential influence of national religiosity when making cross-cultural comparisons.

Although our study investigated 44 countries and territories across eight geographical areas, the number of countries is not equally distributed among the geographic areas. Studies from Oceania and Africa are under-represented, with each of these areas representing only 2% of the pool of eligible articles. In addition, all of the studies included in the meta-analysis are conducted in the first wave of the COVID-19 pandemic, and public sentiment and policy implementation tend to vary across different waves of health crises (e.g., Cheng & Cheung, 2005). Most of the studies have adopted a cross-sectional design, and thus no directions of causal precedence between the study variables can be inferred. Therefore, our findings may not be generalisable to other waves of the COVID-19 pandemic. Future effort should be expended to broaden the scope of the culture-moderated meta-analysis by including more countries across various pandemic waves when more such studies are available.

In conclusion, this culture-moderated meta-analysis investigated some theory-driven hypothetical associations between multiple major coping styles and two important domains of psychological symptoms in the first wave of the COVID-19 pandemic. Robust, universal findings are observed for problem-focused and avoidant coping styles, both of which are driven by fundamental motivations. The cultural dimensions of uncertainty avoidance and masculinity play an influential role in accounting for cross-country differences in the magnitude of the coping-symptom associations for two major coping styles: emotion-focused coping and social support seeking. These two less frequently studied cultural dimensions may be influential because of their relevance to the specific contextual characteristics of the initial wave of the pandemic. At that time, unprecedented nationwide disease control measures are imposed and home-based teleworking are required. Multi-wave comparisons should be performed in the future for a more comprehensive evaluation of the explanatory utility of the various cultural dimensions regarding the associations of coping styles and psychological symptoms in this global health crisis.

Acknowledgement

We thank Linus Chan, Boris Chung, Yuhe Hu, Janice Leung, Ming-yeung Shiu, Roxanne Wong, and Mingyi ZhangFeng for research and clerical assistance.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This research was supported in part by Hong Kong Research Grants Council General Research Fund [grant number 17400714], and the University Research Committee, University of Hong Kong Seed Fund for Basic Research [grant number 202011159134] to Cecilia Cheng, and Hong Kong Research Grants Council General Research Fund [grant number 16601020] to Kin Fai Ellick Wong. The funders had no role in the study design, data synthesis or interpretation, manuscript writing, or the decision to submit the paper for publication.