Relationship quality and support for family policy during the COVID-19 pandemic.

Objective: We examined how relationship satisfaction changed during the coronavirus (COVID-19) pandemic, as well as how relationship satisfaction related to public policy support. Background: Conservation of resources (COR) theory suggests that societal-level stressors (such as a global pandemic) threaten familial and individual resources, straining couple relationships. Relationship satisfaction is in turn linked with important individual, familial, and societal outcomes, necessitating research on how COVID-19 impacted this facet of relationships. Method: Drawing from an international project on COVID-19 and family life, participants included 734 married and cohabiting American parents of children under 18 years of age.
Results: Findings revealed relationship satisfaction declined moderately compared to retrospective reports of relationship satisfaction prior to the pandemic. This decline was more precipitous for White individuals, women, parents less involved in their children's lives, and those reporting higher levels of depressive symptoms. We also found that higher relationship satisfaction was associated with higher levels of support for family policy, particularly for men. At higher levels of relationship satisfaction, men and women had similarly high levels of support for family policy, while at lower levels, women's support for family policy was significantly higher.
Conclusion: The COVID-19 pandemic likely amplified facets of social inequality, which is especially concerning when considering the large socioeconomic gaps prior to the pandemic. Implications: Therapists, researchers, and policy makers should examine how relationship satisfaction may have changed during the pandemic because relationship satisfaction is linked to child and adult well-being and relationship dissolution. Further, the link between relationship satisfaction and support for family policy deserves further scrutiny.

The coronavirus pandemic, unprecedented in scope and nearly universal in influence, has left few if any individuals or societies untouched. While the full impact of the COVID‐19 pandemic will play out over upcoming years and decades, the virus is now in nearly every country, with nearly seven million attributable deaths among more than half a billion cases, as of this writing (Johns Hopkins University, n.d.). Beyond the dire health situation precipitated by the virus, research suggests that responses and restrictions to the virus have also triggered a massive increase in global poverty, with 71 million people estimated to be pushed into deep poverty in 2020, the first increase in global poverty in over 2 decades (United Nations, 2020b). Meanwhile, the pandemic has also sown the seeds of a global mental health crisis (United Nations, 2020a) and may lead to a global economic recession (The World Bank, 2021) that would likely require years and a global commitment to sustainable economic systems to recover.

Within these unique global circumstances, knowledge of how families respond to challenging health, economic, and social circumstances is critical because they are a key component to achieving sustainable and prosperous societies and a crucial institution for developing and maintaining the foundations of adult and child well‐being on which such societies rely (Richardson et al., 2020). However, we know comparatively little about how families are faring during the COVID‐19 crisis (Pietromonaco & Overall, 2021). Nor do we know much about how families falter or flourish during pandemics more generally. Of course, there is no reason to believe families' experiences with COVID‐19 will be uniform. Those couples and families with access to significant resources in terms of housing, access to mental and physical health care, and social capital, among others, may be negotiating the pandemic quite well whereas those without these resources are likely struggling (Pietromonaco & Overall, 2021; United Nations, 2020a). Additionally, while COVID‐19's effect on family functioning is unclear, much is known about how pandemics in general influence the broader public and various sociodemographic groups (Bish & Michie, 2010; Jones & Salathé, 2009; Perrin et al., 2009; Reynolds et al., 2008).

This study aims to contribute to a nascent literature on knowledge of how families were affected by the COVID‐19 pandemic. Drawing on data collected as part of a larger study of 72 countries, we employed data on 734 married and cohabiting parents of children under 18 in the United States to examine how relationship satisfaction changed during the pandemic and assessed whether these changes were similar across sociodemographic groups. We also explored, for the first time, how relationship satisfaction may be linked to support for family policy. Understanding relationship dynamics may help assess the need for social, educational, and other health interventions as well as inform reactions to future pandemics. Further, the foray into linking relationship satisfaction and support for family policy provides a new avenue of future research for family and family policy scholars as we explore how to formulate and improve government programs designed to strengthen families and ensure sustainable societies.

THEORETICAL FRAMEWORK

Stress is a prominent and widespread feature of the COVID‐19 pandemic (Boyraz & Legros, 2020). The pandemic has stressed health care delivery systems (e.g., Tanne et al., 2020), strained government resources and budgets (e.g., Khoo & Lantos, 2020), and stretched families and individuals to their socio‐emotional and financial limits (e.g., Prime et al., 2020). Despite their infrequency, novel, large‐scale societal stressors can create rippling effects at every level. While large‐scale, U.S.‐based disasters have been studied in the past (e.g., Hurricane Katrina, 9/11), knowledge surrounding couple and family relationships during pandemics is scarce and may differ from how large‐scale natural disasters impact couple outcomes, as such disasters are generally punctual (intensive) while COVID‐19 is longer in duration (extensive).

As with previous work on various U.S. disasters (see Cohen et al., 2019; Freedy et al., 1992; Sattler & Smith, 2020), our theoretical framework is oriented around the conservation of resources. Conservation of resources (COR) theory postulates that threats to resources, such as a large‐scale pandemic like COVID‐19, produce stress, which in turn provokes people to seek and protect valued resources, broadly conceived as both tangible (financial assets, housing) and intangible (relationships, well‐being) assets (Halbesleben et al., 2014; Hobfoll, 2001). Conservation of resources theory further asserts that this loss, threat of loss, or failure to realize expected gains results in distress.

The onset of COVID‐19 elicited these COR processes at many levels of American society. The pandemic created threats to maintaining resources of nearly all types. Individuals and families lost jobs, loved ones, or the ability to leave their homes and pursue chosen activities. In other cases, economic well‐being was maintained but new and unexpected stressors arose, such as homeschooling children, working remotely, and the ever‐present risk of falling ill or caretaking for an ill loved one.

Conservation of resources theory also recognizes that because individuals are nested within families, which are nested within communities and societies (Hobfoll, 2001), societal events impact both the individual and their relationships with others in their families and communities (White et al., 2018). These theoretical tenets imply that a societal event such as COVID‐19 impacts both familial and individual resources. We argue that this notion strengthens the importance of examining how the onset of COVID‐19 threatened individual and familial resources. Because most American households are headed by adult partners (U.S. Census Bureau, n.d.), whose relational stress often bidirectionally “spills over” to affect the other members of the household—especially children (Flook & Fuligni, 2008)—adult couple relationships constitute one of the most salient familial relationships to study. Thus, in an effort to increase understanding of how COVID‐19 impacted both individuals and families, the current study focused on couple relationship dynamics, specifically relational satisfaction. Although relationship satisfaction is only a small part of family functioning, the quality of romantic relationships is crucial because it is associated with a wide variety of indicators of child and individual well‐being, including adult and child health and psychosocial well‐being (Karney & Bradbury, 2020).

RELATIONSHIP SATISFACTION

In the stressful context of COVID‐19, and in conjunction with the theoretical notions outlined above, it is reasonable to anticipate an overall decline in relationship satisfaction during the pandemic. Scholars have noted that stress correlates with decreased relationship satisfaction (Lavner et al., 2012), as well as magnifying the normative decrease that commonly accompanies relationship longevity (James, 2015a). This may be particularly true when stress results from the loss of resources, as outlined by COR theory (e.g., Halbesleben et al., 2009; Murray & Forti, 2009; Randall & Bodenmann, 2017). When couples encounter stress, that stress colors partners' interactions, coping mechanisms, and communication skills (Karney & Bradbury, 2020; Pietromonaco & Overall, 2021), and both COR theory and empirical studies report that couple‐ and family‐level stressors can disrupt family patterns and routines, requiring family members to adjust habits and expectations (Hobfoll, 2001; Prime et al., 2020; White et al., 2018). Adjustment to such contexts may be difficult as it requires both resilience and negotiation from romantic partners (e.g., Neff & Broady, 2011; Seiter & Lucas‐Thompson, 2020). Such relational skills may be in short supply in the overwhelming context created by COVID‐19—even if partners experience different levels of stress to the same stressor (Ledermann et al., 2010)—as partners' emotional resources are drained by new, pandemic‐related stressors (Halbesleben et al., 2014). Thus, romantic partners may struggle with heightened emotionality, increased irritability, lower positive affect, and reduced reasoning capacity (Arnetz & Ekman, 2006; Thoits, 2010; Wunsch et al., 2017). Ultimately, these factors can result in more relational conflict and poor conflict resolution (Padmala et al., 2011), which often relates to decreased relationship satisfaction (Tandler et al., 2021) and the potential separation of parents.

These notions concerning romantic relationship functioning during COVID‐19 are broadly consistent with the limited research on couple and family relationships during this time period, with important nuances. Schmid et al. (2020), for instance, observed that 40% of respondents reported negative changes in relationship satisfaction while half that many reported positive changes, with no gender differences evident in these relationship satisfaction changes. Children, when present, appeared to buffer against this decrease, although the authors observed that overall the COVID‐19 pandemic impeded parents' relationship quality and positive family processes (Schmid et al., 2020). Similarly, Williamson (2020) found that decremented relationship outcomes, while statistically significant, were not moderated either by pandemic‐related events or most sociodemographic characteristics, although couples with positive functioning were more likely to experience increases in relationship satisfaction while the opposite was true for those with lower functioning. This may be explained in part by the overall escalation in adult couple conflict observed since the beginning of social distancing measures and lockdowns across the United States (Luetke et al., 2020).

Other pandemic‐related scholarship stresses the importance of vulnerabilities for undermining or sustaining romantic relationships during the crisis. Relationships with acute vulnerabilities prior to the pandemic struggled, while high relationship quality appeared to buffer against the worst effects (Overall et al., 2022). This observation is consistent with prior work showing that high quality relationships can act as a buffer from the worst consequences of some harmful life events and even health challenges (Pietromonaco & Collins, 2017), emphasizing the importance of studying and understanding high quality romantic relationships during the COVID‐19 pandemic.

Möhring et al. (2020) similarly found a decrease in German family satisfaction but interestingly found that fathers' decline in satisfaction was less pronounced than mothers.’ Other work has found similar gender differences in relationship satisfaction measures, with one study from Iran reporting higher scores for fathers than mothers on both relationship satisfaction and psychological well‐being (Mousavi, 2020). Other recent studies comprehensively surveil the way the pandemic has magnified stress at both the individual and familial levels (Brock & Laifer, 2020), leading to increased conflict (Luetke et al., 2020; Stanley & Markman, 2020) and less interpersonal support (Pietromonaco & Overall, 2021).

Thus, because we conceptualized COVID‐19 as a significant societal and individual stressor, we expected relationship satisfaction to decline during COVID‐19. We further expected this decline to be more pronounced for people who may have already been experiencing significant stress or are otherwise more structurally vulnerable.

FAMILY POLICY

One branch of family science research that has been garnering increasing interest is that of family policy (Bogenschneider & Wadsworth, 2019), partly because family policy expenditures in the United States are at record highs (Bogenschneider & Wadsworth, 2019; Haveman et al., 2015). Although what constitutes “family” policy varies depending on local context, policies such as universal health care, paid family leave, accessibility to counseling and elder care, and housing assistance are often thought to fall under the family policy moniker. There is good evidence that social policy, of which family policy should rightfully be seen as a branch, affects family relationships (for a review see Ooms, 2019); indeed, the United States government has spent billions of dollars over the past 20 years to boost family relationships, particularly among low‐income and racially diverse populations. These efforts have been met with varying levels of success, and many broader social‐level policies have been shown to affect family‐level relationships, both implicitly or explicitly (Berger & Carlson, 2020).

Theoretically, the COVID‐19 pandemic placed strain on many different parts of society, including the health care system, political structures, and families, among others. This strain resulted in seeking to maintain scarce resources or perhaps recognizing the need to gain access to new ones. Manifestations of this would include the payments sent to all parents, changes to the way services were delivered because of COVID‐19, and the national conversation surrounding how COVID‐19 has laid bare the inequalities that have long plagued this country. Thus, the strain and the accompanying calls to react by conserving or acquiring new resources could result in greater calls for policies designed to improve family well‐being.

To our knowledge, however, scholars have yet to ask whether individuals' relationship satisfaction influences their support for family policy. On the broadest level, this potential notion makes sense, as relationship satisfaction is an individual‐level factor whereas family policy is a societal one, and relational and societal components often interact (e.g., Hobfoll, 2001). Additionally, there are conceptual reasons relationship satisfaction may be linked to support for family policy in general. For instance, people who are highly satisfied in their romantic relationship may desire to support policies and programs that will enable others to do the same, while individuals who are not satisfied in their romantic relationships may be less inclined toward families in general and view alternatives to family policy as more likely to produce favorable outcomes for adults, children, and society. This would lead one to expect a positive relationship between relationship satisfaction and support for family policy.

On the other hand, one could envision an alternative if those with poorer relationship satisfaction have found family polices to be insufficient or feckless, in some instances even unaware of their existence. Perhaps equally plausible is that these same individuals view such policies, to the extent such policies are in place, as a buffer between them and lower standards of living whereas those with higher levels of relationship satisfaction may view family policy as unnecessary. In either of these latter cases, we may expect a negative relationship between relationship satisfaction and support for family policy.1

We therefore examined the association between relationship satisfaction and support for family policy. Due to the lack of prior research on the issue, we took an exploratory approach, though the conceptual and related empirical notions outlined above indicate a relation between the two variables is likely. We hope that this initial foray into the topic will provoke other researchers to more fully examine the predictors of individual‐level support for family policy.

METHODS

Sample

The data for this study come from the COVID‐19 Family Life Study (Ben Brik, 2020), which sought to explore the effects of the coronavirus pandemic on family life across cultures. We employed an online research firm, Qualtrics, to collect the data. Qualtrics maintains a large set of respondents from which each sample is selected, based on the desired survey characteristics. These survey characteristics, drawn from weighted data from the most recent American Community survey, functioned as our quotas. Quotas included matching the data to nationally representative data on age, biological sex, education, race–ethnicity, and marital status among parents of children under 18.

The survey, which averaged about 20 minutes to take, was made available to the participants in the Qualtrics panel who then responded to a qualifying questionnaire to assess eligibility. Of the 5082 panelists who opted to take the survey, 2044 (40%) did not meet the study's eligibility requirements (parent of a child under 18), 1893 (37%) declined to participate, two (<1%) were disallowed because the quota had already been filled, and 126 (2.5%) were removed by Qualtrics prior to receipt due to signs of insufficient attention such as speeding or providing unrealistic response patterns. This left us with 1017 parents of children under 18 who qualified for the survey and met all quality standards, of which 734 were in a relationship.

In terms of response rate, it is difficult to determine the precise rate because online panel surveys operate in fundamentally different ways than national probability samples. Our best estimate is that the response rate was around 26% (1017 responses plus two quota overfills divided by the total of the ineligibles plus the declined participations). Note that this is very roughly in line with the American Association for Public Opinion Research's raw response rate (RR1; Dutwin et al., 2014). Our 26% response rate also puts us in line with other large, national family surveys conducted in the United States in the past decade (James et al., 2021).

Data were collected between September and October of 2020 after study design and materials were approved by appropriate institutional review boards. Survey questions gauged sociodemographic characteristics, children's and parents' mental and physical health, coping strategies, relationship satisfaction, and support for family policy. Respondents could select whether to take the survey in English or Spanish. Bidirectional translation, wherein the document is translated from English into Spanish and then back into English to ensure validity, was carried out by multiple native speakers of both languages.

Variables

Our key variables in this paper are relationship satisfaction and support for family policy. Respondents were asked how satisfied they were at the time of the survey (in the middle of the coronavirus pandemic) and to retrospectively reflect how satisfied they were prior to the pandemic. Although retrospective research is not the ideal, we would like to note that prior research on family life during COVID‐19 has employed similar designs with good effects (Mousavi, 2020; Vigl et al., 2020). Relationship satisfaction was based on the ENRICH marital satisfaction scale developed by Fowers and Olson (1993), a 15‐item scale that surveys 10 different domains of relationship quality (alpha = .85) and five items that measure unrealistically positive descriptions of the relationship (idealistic distortion; alpha = .76). We followed the creators' directions for creating the score and accounting for idealistic distortion. The full scale is available online as Supplemental Table 1.

To gauge support for family policy, respondents self‐reported how strongly they support or oppose (1 = strongly oppose to 4 = strongly support) 21 different family policies, such as universal health care, paid family leave, accessibility to counseling and elder care, and housing assistance. The full set of policies is available online as Supplemental Table 2. Following current best practices on measuring support for policies (Kyselá et al., 2019), we computed an index from items indicating the same type of response to these specific family policies. Exploratory and confirmatory factor analyses suggested a single factor structure was optimal (results not shown), and statistical tests confirmed excellent reliability (α = .96).

Independent variables in all models included biological sex (1 = female), race–ethnicity (White [reference group], Black, Latinx, Asian, other), age (0 = 18–25 years, 1 = 26–30 years, 2 = 31–35 years, 3 = 36–40 years, 4 = 41–45 years, 5 = 46–50 years, 6 = 50+ years, in 5‐year increments), educational attainment (

Analytic strategy

Our analytic strategy consisted of two types of regression models. Note that we estimated each equation, described below, with and without interaction terms (moderators). When examining change in relationship satisfaction before (conceptualized as Wave 1) and during (Wave 2) the pandemic, we employed multilevel linear regression models with a random intercept, in which relationship satisfaction was modeled as a function of an individual‐level intercept (α ) and coefficients estimating the association between relationship satisfaction and each coefficient (see Equation 1): where ξ is given by the following equation: Note that we included interaction terms (represented in the final coefficient of Equation 1) to examine whether the observed drop in relationship satisfaction during the pandemic differed for various groups. Equation 1 also includes an error term comprised of two components, where ζ is a time‐constant or permanent error component that varies between individuals and ε is a transitory error component that varies both within and between individuals, to represent the effects of time nested within individuals as well as any other error. The two errors are assumed to be independent and normally distributed, with a mean of 0. Stated differently, we estimated effects on the basis of subject‐level sampling error (ζ ) plus other sources of error (ε ). Note also that i indexes individuals and t marks time periods; variables with subscript i vary between individuals and variables with subscript t vary within individuals.

The second equation predicting support for family policy was modeled as a linear regression with all the predictors from Equation 1 as well as interactions between relationship satisfaction and biological sex, an error term that included all variability not captured in the ordinary least squares equation, and support for family policy (see Equation 2). All analyses were conducted in Stata 17 and utilized sampling weights to ensure the results mirrored the United States population of coupled American parents with children under age 18 on age, sex, race–ethnicity, region, educational attainment, and marital status. The 2019 one‐year file, the most recent available, of the American Community Survey (ACS) was used to construct the weights.

RESULTS

Table 1 displays descriptive statistics for all variables included in the analysis. On average, relationship satisfaction and support for family policy were relatively high, 43 (maximum = 65) and 3.22 (maximum = 4), respectively. The sample skewed slightly female at 55%. In terms of diversity, the sample closely matched the racial–ethnic distribution of coupled individuals nationally, according to the authors' calculations of the 2019 ACS (not shown but available upon request). The same was mostly true of educational attainment, although those with less than a high school degree were underrepresented (3% in the sample vs. 11% in the general population; authors' calculations of the U.S. Census Bureau's 2019 Current Population Survey's Annual Social and Economic Supplement data) while those with some college were overrepresented (36% here vs. 28%).

TABLE 1

Descriptive statistics for all variables included in the analysis. Source: COVID‐19 Family Life Study (N = 734)

	Mean	Standard error of the mean	Min	Max
Support for family policies	3.22	0.03	1	4
Relationship satisfaction	42.68	0.56	2.40	63.74
Female	0.55
Race–ethnicity
White	0.76
Black	0.06
Latinx	0.14
Asian	0.05
Other	0.00
Living together (vs. apart)	0.95
Age
18–25	0.04
25–29	0.04
30–34	0.14
35–39	0.24
40–44	0.26
45–49	0.10
50+	0.17
Educational attainment
<High school	0.03
High school	0.23
Some college	0.36
Bachelor's degree	0.24
Post–bachelor's degree	0.14
Parental involvement	15.79	0.17	5	20
Respondent is child's parent	0.84
Depressive symptoms	14.92	0.32	7	28
Number adults in household	1.64	0.03	1	4
Number children in household	2.64	0.03	2	5
Family environment (standardized)	0.00	0.02	−2.19	1.25
Lost job	0.16
Changed work arrangements due to childcare	0.12
Worked from home	0.32

Sixteen percent of the sample reported losing their job during COVID‐19, 12% were obligated to change work arrangements to accommodate their childcare responsibilities, and one third of the sample reported that they began or intensified working from home. Note that differences between this sample and national data largely reflect selection into marriage and cohabitation.

Table 2 displays the results of the multilevel models with random intercepts (Equation 1) predicting relationship satisfaction before and during COVID‐19. The results for the main effects (i.e., the noninteractive model) are displayed in the second column of the table. On average, respondents reported a decline in relationship satisfaction of 2.5 points, or about one fifth of a standard deviation (responses ranged between 2 and 64). Relationship satisfaction was significantly higher among involved parents and lower among those with increasing levels of depressive symptoms and those with more children in the household.

TABLE 2

Relationship satisfaction during COVID‐19 from the COVID‐19 Family Life Study

	Overall		Moderation
During COVID	−2.49***	(0.64)	−5.73	(4.70)
Female	−1.30	(0.96)	−0.10	(1.07)
Black	0.24	(1.40)	−1.33	(1.77)
Latinx	1.24	(1.31)	0.80	(1.42)
Asian	−0.24	(1.97)	−1.17	(2.01)
Other	0.14	(2.31)	−1.52	(2.99)
Live together	2.01	(1.67)	−0.45	(1.84)
Age	0.02	(0.32)	0.18	(0.35)
Educational attainment	0.69	(0.47)	1.27	(0.50)
Parental involvement	1.62***	(0.14)	1.43***	(0.16)
Respondent is child's parent	0.33	(1.16)	0.34	(1.15)
Depressive symptoms	−1.08*	(0.44)	−1.02*	(0.44)
Depressive symptoms # Depressive symptoms	0.02	(0.01)	0.02	(0.01)
Number adults in household	0.04	(0.54)	0.05	(0.54)
Number children in household	−1.96**	(0.72)	−1.87**	(0.71)
Family environment (standardized)	0.43	(0.83)	0.49	(0.82)
Lost job	0.16	(1.30)	0.13	(1.30)
Changed work arrangements due to childcare	−1.84	(1.39)	−1.99	(1.40)
Worked from home	1.37	(1.01)	1.40	(1.00)
During COVID × Black			3.41*	(1.73)
During COVID × Latinx			0.93	(1.55)
During COVID × Asian			1.84	(1.60)
During COVID × Other			3.21	(2.07)
During COVID × Live Together			5.14	(3.08)
During COVID × Female			−2.31	(1.15)
During COVID × Educational Attainment			−1.19	(0.62)
During COVID × Age			−0.27	(0.41)
During COVID × Depressive Symptoms			−0.20*	(0.10)
During COVID × Parental Involvement			0.37*	(0.17)
Constant	30.72***	(5.28)	32.15***	(5.37)
Observations	1468		1468
Respondents	734		734

Note: Standard errors in parentheses. Reference categories include male, White, not living together, respondent is not child's parent, did not lose a job, did not change work arrangements due to childcare, and did not work from home.

p < .05.

p < .01.

p < .001.

The first column of Table 2 shows the results of the interactive model, where we examined whether this decrease in relationship satisfaction was similar across a variety of psycho‐socio‐demographic groups. For ease of interpretation, the relationships portrayed by each of these significant interaction terms are depicted graphically in Figure 1. Compared to White respondents, Black respondents reported a significantly less steep drop in relationship satisfaction (top left, Figure 1). While White respondents reported a decrease of about 3 points, relationship satisfaction among coupled African American respondents remained statistically flat. We see a similar pattern between men and women (top right, Figure 1), where the slope for men remained flat while women reported a drop of about 3.5 points, even after accounting for involved parenting, changes to work arrangements due to childcare responsibilities, whether someone in the household lost their job, and other control variables.

FIGURE 1

The relationship between race, sex, depressive symptoms, and involved parenting, respectively, and relationship satisfaction before and during the COVID‐19 pandemic

Data from “International Study on the Impact of COVID 19 on Family Life Across Cultures,” by A. Ben Brik, 

The pattern of some groups reporting little to no decline in relationship satisfaction during the pandemic with others reporting steeper declines continued when we examined depressive symptoms and involved parenting (bottom panel, Figure 1). In terms of depressive symptoms, those with little to no depressive symptoms did not see a decrease in relationship satisfaction, while those at increasing levels reported increasingly steep declines. Those at the highest level of depressive symptoms reported a drop of 5 points in relationship satisfaction, or .40 standard deviations. For involved parenting, we again saw that parents who were most involved in their children's lives did not experience a decline in relationship satisfaction, whereas those who were least involved saw the largest drop in satisfaction of more than 6 points, or roughly half a standard deviation.

We then examined whether relationship satisfaction was related to support for family policy (see Table 3). Because we did not ask respondents for retrospective reports of support for family policy, we analyzed the data using a simple ordinary least squares linear regression model (Equation 2). Initial analyses suggested that the relationship between relationship satisfaction and support for family policy was nonlinear, so we included the square of relationship satisfaction in the model as well. Relationship satisfaction was significantly associated with support for family policy after controlling for all other variables in the model. Each 1‐point increase in relationship satisfaction was associated with a .034 decrease in support for family policy, controlling for all other variables in the model. However, this relationship changed rapidly based on one's relationship satisfaction. For each 1‐point increase in relationship satisfaction, the slope of .036 (for those with a score of 0) increased by .01 points. In other words, there is no expected association between relationship satisfaction and support for family policy for those with a score of 36 on relationship satisfaction. Importantly, the relationship is expected to turn positive thereafter. Considering the mean score for relationship satisfaction is 43, this indicates a positive association between relationship satisfaction and support for family policy for the bulk of the respondents in our survey. Additionally, Latinx respondents reported higher levels of support for family policy, as did highly involved parents and respondents with fewer depressive symptoms, although this relationship was also nonlinear.

TABLE 3

Support for family policy and relationship satisfaction during COVID‐19 from the COVID‐19 Family Life Study

	Overall		Moderation
Relationship satisfaction	−0.036*	(0.02)	−0.026	(0.02)
Relationship Satisfaction × Relationship Satisfaction	0.001**	(0.00)	0.001**	(0.00)
Female	0.248**	(0.08)	0.896***	(0.26)
Black	0.190	(0.14)	0.208	(0.15)
Latinx	0.199*	(0.09)	0.171	(0.09)
Asian	0.194	(0.13)	0.191	(0.12)
Other	0.155	(0.15)	0.154	(0.14)
Live together	0.235	(0.27)	0.281	(0.27)
Age	0.020	(0.03)	0.027	(0.03)
Educational attainment	−0.020	(0.04)	−0.024	(0.04)
Parental involvement	0.035*	(0.01)	0.031*	(0.01)
Respondent is child's parent	−0.059	(0.11)	−0.059	(0.11)
Depressive symptoms	−0.126**	(0.04)	−0.131***	(0.04)
Depressive symptoms × Depressive Symptoms	0.004**	(0.00)	0.004***	(0.00)
Number adults in household	−0.050	(0.06)	−0.046	(0.07)
Number children in household	0.012	(0.08)	−0.003	(0.07)
Family environment (standardized)	0.104	(0.08)	0.095	(0.08)
Lost job	0.077	(0.12)	0.091	(0.12)
Changed work arrangements due to childcare	0.017	(0.12)	−0.011	(0.12)
Worked from home	−0.021	(0.09)	−0.008	(0.08)
Female × Relationship Satisfaction			−0.015**	(0.01)
Constant	3.651***	(0.57)	3.294***	(0.62)
Observations	711		711

Note: Standard errors in parentheses. Reference categories include male, White, not living together, respondent is not child's parent, did not lose a job, did not change work arrangements due to childcare, and did not work from home.

p < .05.

p < .01.

p < .001.

Motivated by work showing that women tend to be more relationship focused than men and often take on the bulk of family tasks, we next conducted an exploratory analysis examining whether the association between relationship satisfaction and support for family policy differed for women and men (in the final column of Table 3). The results of the significant moderated relationship are again displayed graphically in Figure 2. For women, the linear relationship was not significant; however, modeling the quadratic relationship showed that women at either end of the relationship satisfaction scale reported high levels of support for family policy, with those in the middle reporting significantly lower levels, revealing a u‐shaped curve. For men, higher levels of relationship satisfaction were associated with increasingly high levels of support for family policy. In terms of predicted values of support for family policy, further analyses (not shown but available from the first author) showed that, at lower levels of relationship satisfaction, women reported significantly more support for family policy than men. However, the difference was no longer significant when relationship satisfaction was about 50, or approximately the top one third of the distribution.

FIGURE 2

Relationship satisfaction and support for family policy, by gender

Data from “International Study on the Impact of COVID 19 on Family Life Across Cultures,” by A. Ben Brik, 

Measures of model fit indicated that both the multilevel model and linear regression model fit the data reasonably well, with estimated R 2s of approximately .34 and .26, respectively, with slightly higher values in the models with the significant interaction terms.

DISCUSSION

This paper contributes to the nascent but growing literature on family relationships during the COVID‐19 pandemic (Brock & Laifer, 2020; Luetke et al., 2020; Möhring et al., 2020; Mousavi, 2020; Overall et al., 2022; Pietromonaco & Overall, 2021; Saasa & James, 2020; Schmid et al., 2020; Stanley & Markman, 2020; Williamson, 2020; Yendork & James, 2020). Most evidence suggests that relationship quality generally declined during the crisis, which is consistent with COR theory, as the pandemic stressed many individuals' and families' economic and emotional resources (e.g., Prime et al., 2020). However, to what extent and for whom relationship quality decreased returned nuanced results.

Our results suggest that, on average, relationship satisfaction declined about 2.5 points or one fifth of a standard deviation. We regard this as a moderate effect size,3 considering that trajectories (not overall levels) of relationship satisfaction tend to be quite stable over time (James, 2015a; Lavner & Bradbury, 2010; Proulx et al., 2017). A decline of this size, complete with the limitations of the retrospective data, suggests that COVID‐19 likely did negatively influence relationship satisfaction for most individuals. Note, however, that the retrospective nature of the data means we cannot be certain of the size (i.e., the 2.5‐point drop); the actual drop could be different due to respondent recall issues, the fact that the crisis may have heightened existing relational issues, and so forth.

Not all groups experienced this same decline, however. We found that White participants reported significantly more decline in relationship satisfaction than Black participants. In all other instances, however, when we found significant differences in how much a given group declined, the most privileged groups (i.e., men, those with better mental well‐being, those experiencing positive family dynamics) tended to decline less. Regarding gender, the decrease in satisfaction was more precipitous for women than men. Individuals with more depressive symptoms reported steeper drops than those with fewer symptoms, with individuals reporting no or only few symptoms reporting no decline. Similarly, parents who reported being heavily involved in their children's lives reported no decline whereas the drop was quite steep for parents who reported being the least involved.

This yields additional empirical evidence to what many have suspected—the COVID‐19 pandemic is likely to have amplified (Bambra et al., 2020) societal inequalities, although systematic reviews show that at least for income, pandemics can both amplify and alleviate inequality, dependent on the sector of population examined (Sayed & Peng, 2020). Such a response again aligns with COR theory, as less privileged groups have fewer resources in the first place as well as experience more threat to or loss of those resources than those in more privileged groups. This is concerning, because if more privileged groups weather the COVID‐19 storm comparatively unscathed while the disadvantaged fall further behind, the gap between rich and poor may increase, with concomitant psychosocial, economic, demographic, and educational consequences. Given the United States was already facing historical levels of societal inequality prior to the crisis, the importance of public policy in addressing the issue is hard to understate, especially if, as research suggests, workers suffering from the consequences of the 2008/2009 recession are more likely to face some of the worst consequences (e.g., job loss) of the COVID‐19 recession (Casarico & Lattanzio, 2020).

These differences in relationship satisfaction matter for many reasons, most directly because they are tied to adult and child well‐being (Amato, 2010; Karney & Bradbury, 2020). One contribution of this paper is that it pushes the argument even further by linking relationship satisfaction to support for family policy, at least in terms of the 21 policies referenced here, including universal health care, paid family leave, accessibility to mental health counseling, elder care, and housing assistance. To our knowledge, this is the first paper to do so. We found strong evidence that the two are positively related (higher levels of relationship satisfaction were typically linked to higher support for family policy), although the association in our analyses was nonlinear. While we cannot speak to pre‐pandemic levels of family policy support in this paper, we do show that this link is robust to a variety of control variables. This leaves open an interesting question: If relationship satisfaction and support for family policy are linked, could increased relationship satisfaction lead to higher levels of support for family policy? Also of question is whether relationship satisfaction could have increased support for various family policies, ultimately resulting in structural improvements that benefit families of all kinds.

Crucially, asking this question presupposes that the link is causal, at least to some degree, a topic which we cannot establish here given the cross‐sectional nature of our data. Further, it is unclear whether the relationship is direct or mediated by some other mechanism. As articulated earlier, we can envision reasons why this would be a direct though nonlinear relationship, as people who find great satisfaction in their family relationships may be more likely to view policies designed to help families more favorably, and those struggling in their relationships may feel a desire for family policy to help improve their situation and likewise support family policies at high rates.

Notably, we found evidence of both, at least among women, although the reason support for family policy was different for men and women remains unclear. One possibility is that women may be more supportive of needs of family policy because on average, they are more attuned to familial and relational needs than men.

Much more work remains before this relation can be firmly established. But it does suggest the possibility that relationship satisfaction may be an unanticipated avenue through which support of family policy may be garnered. Additionally, if there is a relationship between relationship satisfaction and support for family policy, researchers may also rightfully ask if, as relationship quality on the whole decreases (Amato et al., 2007; Glenn, 1998; James, 2015b), does support for family policy decline as well? Some policies, such as universal health care and affordable childcare, are beneficial for economic and social reasons. However, the path linking relationship satisfaction to support for economic and social policies remains unexplored. This avenue of research may be especially important to pursue as the COVID‐19 pandemic is likely to lead to a major mental health crisis (United Nations, 2020a). Subsequent research should examine these associations longitudinally to assess both the robustness and directionality of these findings.

This paper, although a solid step in building on a small but growing literature on relationship satisfaction during the COVID‐19 crisis in the United States, has several important limitations. The sample was not a probability sample, although it does reflect the characteristics of the United States in terms of age, education, race–ethnicity, biological sex, and marital status. Yet, national quota samples are not equivalent to nationally representative ones (James et al., 2021). Additionally, participants were required to be in a preestablished committed relationship, indicating that levels of relationship satisfaction may be higher than individuals whose satisfaction may have declined enough during the pandemic that they ended their relationship prior to the launch of our survey. Further, the data are cross‐sectional, so future research should explore whether the associations examined are causal, selective, or even bidirectional. Although our discussion of these findings is based in previous empirical studies, we believe that future research will further push our understanding in new ways.

Despite these and other limitations, this paper contributes to the growing literature on family relationships during COVID‐19. A once‐in‐a‐generation pandemic deserves scrutiny because health crises can trigger economic, social, and educational ones. This paper presents evidence that “familial” ought to be added to the previous list. Although some relationships appear to be faring well during the coronavirus storm, others, particularly the less privileged and disadvantaged, appear to be struggling. Government policy makers should take notice and enact policies that will counteract these negative effects, especially as deepening inequalities continue to upend the social order.

Supplementary Table 1 Items from the Enrich (Fowers and Olson, 1993) scale used to measure relationship satisfaction.

Supplemental Table 2. Family policies included in the Covid19 Family Life survey.

Click here for additional data file.