key: cord-308493-3fsn7awq
authors: Günther‐Bel, Cristina; Vilaregut, Anna; Carratala, Eduard; Torras‐Garat, Sonia; Pérez‐Testor, Carles
title: A Mixed‐method Study of Individual, Couple and Parental Functioning During the State‐regulated COVID‐19 Lockdown in Spain
date: 2020-07-17
journal: Fam Process
DOI: 10.1111/famp.12585
sha: 
doc_id: 308493
cord_uid: 3fsn7awq

During the recent COVID‐19 outbreak in Spain we explored the individual and relational wellbeing of people confined together with their partners and/or children during the first three weeks of state‐regulated lockdown. Adults 18 years or older (N=407) completed an online survey that included demographic, household, and employment information along with standardized measures of psychological distress (State‐Trait Anxiety, Beck Depression) and relationship functioning – either the Dyadic Adjustment Scale if there were no children in the household or a Basic Family Relations Evaluation Questionnaire (CERFB) measuring conjugal, parental, and co‐parental functions. Qualitative analyses of responses to an open‐ended question about perceived changes in couple or family dynamics during lockdown revealed 9 specific themes comprising two overarching categories: relational improvement and deterioration. The overall prevalence of improvement themes (61.7%) exceeded deterioration themes (41.0%), with increased (re)connection and conflict atmosphere cited most often. Quantitative analyses found elevated levels of state anxiety but not trait anxiety or depression during lockdown. Consistent with the qualitative results, couples having no children at home reported high levels of dyadic adjustment, but with children present CERFB parental functioning exceeded conjugal functioning, a pattern sometimes associated with child triangulation into adult conflicts. Although correlates of psychological distress (e.g., unemployment, perceived economic risk) were relatively stable across sub‐groups, predictors of relationship functioning varied substantially with household/parental status (e.g., telecommuting and employment facilitated conjugal functioning only for couples with children).

According to the World Health Organization (WHO, 2020a), a pneumonia of unknown cause was first detected in Wuhan, China in December 2019. After rapid escalation, the WHO (2020b) declared the novel coronavirus disease (COVID-19) a global pandemic. In the context of this outbreak, governments around the world have taken measures to prevent and control the COVID-19 infection. Starting in China, one such measure has been to place entire cities under mass quarantine.

Spain has been one of the worst affected European countries. The virus spread to all Spanish regions, with the communities of Madrid and Catalonia suffering the highest number of cases. On March 14th 2020, the Spanish government formally declared a State of Alarm over COVID-19 and ordered a state-regulated lockdown defined as home confinement, in which everyone, including those who were asymptomatic, had to stay confined at home for the next two weeks. The only exceptions were for basic activities like buying food or medicine, attending medical centers, or commuting to work (BOE, 2020) . Schools, universities and other education institutions were also on lockdown, going online in most cases. One week into the State of Alarm, as COVID-19 taxed the country's oversaturated health care system, Spain's government tightened quarantine even further, ordering all non-essential workers to stay home for two additional weeks with the possibility of extending the emergency measures until outbreak remission.

In addition to the biomedical and epidemiological benefits of mandatory mass quarantine, it is prudent to consider possible psychological and behavioral impacts (Rubin & Wessely, 2020) . In a prompt and clarifying article, Brooks, Webster, Smith, Woodland, et al. (2020) have reviewed scientific literature relevant to effects of quarantine on individual mental health, aiming to facilitate decision making in the current global crisis. Based on 24 quantitative and qualitative studies across 10 countries where quarantines involved direct or potential exposure to SARS, Ebola, H1N1 influenza pandemic, Middle East respiratory syndrome, or equine influenza, Brooks et al. (2020) conclude that quarantine often has negative psychological effects, including anxiety, depressed mood, irritability, insomnia, symptoms of post-traumatic stress, and emotional exhaustion. Indeed, a recent large-scale survey during the COVID-19 pandemic in China found that well over a third of the general population experienced significant psychological distress, though it is unclear how much of this was related to lockdown (Qiu, 

This article is protected by copyright. All rights reserved Shen, Zhao, Wang, et al., 2020) . In May of this year a similar proportion of American adults reported clinical anxiety or depression according to the U.S. Census Bureau (Fowers & Wan, 2020) .

The literature cites many factors that could have negative mental health consequences during or after quarantine including fear of infection (Bai, Lin, Lin, Chen, et al., 2004; Desclaux, Badji, Ndione, & Sow, 2017) , boredom and isolation (Cava, Fay, Beanlands, McCay, & Wignall, 2005; DiGiovanni, Conley, Chiu, & Zaborski, 2004) , financial insecurity (Jeong, Yim, Song, Min, et al., 2016; Mihashi, Ostubo, Yinjuan, Nagatomi, et al., 2009) , limited access to basic supplies (Jeong et al., 2016) , and confusing public information (Blendon, Benson, DesRoches, Raleigh & Taylor-Clark, 2004; Jeong et al., 2016) . Another factor particularly relevant to the present study is quarantine duration (Hawryluck et al., 2004; Marjanovic, Greenglass, & Coffey, 2007; Reynolds, Garay, Damond, Moran, et al., 2008) . For example, Hawryluck et al. (2004) found significantly more symptoms of posttraumatic stress among people locked down for more than 10 days compared with those in quarantine less than 10 days. In Spain we collected data through the first three weeks of intensive COVID-19 lockdown.

Although the literature emphasizes psychological effects of quarantine on individuals, there are good reasons to consider implications for couple and family relationships as well (Sprang & Silman, 2013) . For example, a Global Times (2020) newspaper article reported unprecedented divorce rates in some districts of Xi'an, the capital of Northwest China's Shaanxi province, as a direct repercussion of COVID-19. Similarly, in qualitative studies, participants described Ebola containment measures in Liberia as creating mutual distrust, even between family members (Pellecchia, Crestani, Decroo, Van den Bergh, & Al-Kourdi, 2015) ; and Toronto health care workers quarantined following exposure to SARS reported disruptions in parental roles and routines, "creating stress for the entire family" (Robertson, Hershenfield, Grace, & Steward, 2004, p.404) .

From our interpersonal-systems perspective, a more general reason to widen the lockdown lens is that individual and family functioning are inextricably interwoven, especially for children and adolescents but also for adults. In fact, an enormous body of research links family conflict and dysfunction to psychological distress, physical health symptoms, and a wide variety of behavior problems (e.g., Cummings, Koss & Davies, 2015; Repetti, Taylor & Seeman 2002) . Similarly, cohesive and supportive family processes not only protect individuals from Accepted Article negative effects of life stress (Hobfoll & Spielberger, 1992) but also generate a variety of positive outcomes (e.g., Conger & Conger, 2002; Joel Wong, Uhm & Li, 2012) . Relevant to the COVID-19 pandemic, disruptions of family functioning in the wake of widespread socioeconomic stress such as the Great Recession of 2007-09 (Forbes & Krueger, 2019; Margerison-Zilko et al., 2016) , as well as natural disasters such as floods and earthquakes (Cao, Jiang, Li, Lo & Li, 2013; McDermott & Cobham, 2012) , have had multiple negative impacts on survivors' behavioral health just as stable and cohesive family relations protect against these.

While the quarantine literature emphasizes mainly deleterious effects, it is possible that positive as well as negative repercussions of lockdown could occur at the level of intimate relationships. On the one hand, home confinement can easily create conditions for conflict or estrangement as household members readjust work, school, and recreational activities; face possible contagion and financial strain; and spend virtually all of their time together in limited physical space. On the other hand, such proximity might also create opportunities for increased closeness, communal problem solving, and deeper personal relationships. The title of a May 24th New York Times article -"The Virus Has Wrecked Some Families. It Has Brought Others Closer" (Wilson, 2020)essentially captures this mixed picture. As couples and families face the demands of a new (crisis) situation, interactional discontinuities may sometimes lead to more resilient as well as deteriorated functioning (Patterson, 2002; Walsh, 2007) .

Soon after COVID-19 disrupted Spain we were able to organize a sizable on-line study of repercussions for individuals, couples, and families. Although the sample of convenience did not rigorously represent the Spanish population, we hoped to gain preliminary information about the individual and relational wellbeing of people confined together with their partners and/or children during the first three weeks of state-regulated lockdown. In addition to demographic, household, and COVID-related employment information, the survey included standardized Spanish-language measures of psychological distress and relationship functioning: The former were the State-Trait Anxiety Inventory (STAI; Spielberger, Gorsuch & Lushene, 2008; Buela-Casal, Guillen-Riquelme & Seisdedos-Cubero, 2015) and the Beck Depression Inventory (BDI; Beck, Steer & Brown, 1996; Sanz & Vazquez, 2011 

This article is protected by copyright. All rights reserved important qualitative (mixed-method) component of the study is that participants also responded to an open-ended question about perceived changes in couple or family dynamics since the beginning of home confinement.

The relationship aspects of the study were of special interest to us as couple and family therapists, and including the CERFB followed naturally from our involvement in a research and development project ("Family Relational Diagnosis in Mental Health") funded by the Spanish government (Ministerio de Economía, Industria y Competitividad, 2017). The CERFB attempts to operationalize central constructs in Linares' (1996 Linares' ( , 2002 Linares' ( , 2012 basic family relations theory, where partially orthogonal conjugal functions and parenting functions converge to create optimal (or sub-optimal) conditions for relational nurturing, a crucial determinant of child mental health.

According to the theory, combining the bi-polar conjugal (harmony-disharmony) and parenting (preservation-deterioration) dimensions yields four prognostically significant quadrants:

Functionality (both dimensions high) allows for mature and balanced child development;

triangulation (parenting high, conjugal low) facilitates child involvement in couple conflicts; deprivation (conjugal high, parenting low) sustains satisfactory couple relations at the expense of child nurturing; while chaotization (both dimensions low) maximizes conditions for child psychopathology (Linares, 2002) . Because the CERFB scales apply only with children present, we used the DAS to assess conjugal relationship quality when parents had no children (couple only) or when children were no longer at home (empty nest).

More specific aims of the study were to (a) compare lockdown responses from the pandemic convenience sample to benchmarks for established measures of individual, couple and parental functioning; (b) describe via qualitative analysis the ways in which participants felt their couple and family relationships had improved and/or deteriorated during the first few weeks of lockdown; (c) identify demographic, household, and employment-related correlates of pandemic relationship functioning and psychological distress with special attention to variations across couples with children at home, couples with no children, and couples with empty nests; and (d) explore possible changes in relationship functioning over time during the first three weeks of lockdown.

This article is protected by copyright. All rights reserved A total of 407 participants recruited through Facebook and other social media platforms completed our online survey between 24 March and 7 April 2020, weeks 2 and 3 of the stateregulated home confinement. Inclusion criteria were: (a) aged 18 or higher, (b) currently living in Spain, (c) living with one's romantic partner and/or one's children (including divorced parents currently living with children in a shared custody arrangement).

The participants were predominantly female (77.0%) and ranged in age from 22 to 77 years (M = 42.7; SD = 12.7). Most were also well-educated (76.7% had university degrees) and resided in cities (67.2%) rather than smaller urbanized towns (27.5%) or rural areas (5.1%). The crucial variable of household-parental status governing which relationship measure(s) they would complete distributed as follows: partnered parents living with children (47.4%), partners in couples without children (37.3%), partners in couples whose children were not at home (9.6%), and divorced parents (5.7%). Children's ages ranged from 5 months to 51 years, and 49.7% of parents had more than one child at home.

A majority of participants (69.9%) were at least partly employed at the time of the survey, with 54% telecommuting, 7.3% working entirely on site, and 8.8% doing both. While only 7.1% had experienced COVID-related job loss, 17.0% of the sample was currently unemployed and 5.9% had retired. Occupations varied widely, with 33.4% of the sample in some way affiliated with the health professions, 11.3% working in an educational capacity, and 47.2% in general commerce or self-employed.

Some participants (14.0%) reported that a member of their household was experiencing a health problem at the time of the survey, and 11.8% indicated they were currently receiving psychological or psychiatric treatment for problems such as anxiety, depression, substance abuse, ADHD, relationship issues, or wanting "personal growth."

The project received ethics approval from the Ethics Research Committee of the School of Psychology, Education, and Sports Sciences, Blanquerna, Ramon Llull University (certificate # 1920005P). Before beginning the online survey, each participant reviewed information about the study's purpose and procedures, including assurance of confidentiality, and provided her or his informed consent.

Responses to an initial question about household and parental status distributed participants according to which standardized relationship measures they would complete later in

This article is protected by copyright. All rights reserved the survey. While all 407 participants completed the STAI and BDI, those in couples with no children or all children away from home completed the DAS (n=191); partnered parents with a child in the household did the full CERFB (n=193); and divorced parents completed the CERFB parenting and co-parenting scales but not the conjugal function scale (n=23). Although a common measure of couple functioning for participants with and without children at home would have been ideal, allocating scales as we did made the online survey more time efficient.

Fortunately, previous validation research has found high correlations between CERFB conjugal functioning and the DAS (e.g., r = .74 in Ibañez, 2016) .

The sequence of survey questions proceeded from demographic, household, and current employment information (including perceived economic risk) to the standardized measures of psychological distress and relationship functioning. Instructions throughout the survey reminded participants to focus on the lockdown period in considering their responses.

A final, open-ended question eliciting data for qualitative analysis asked, "What changes have you perceived in your couple or family dynamics since the beginning of home confinement (March 14th 2020)?" Accompanying this was an apology for not conducting a face-to-face interview and a request to answer in as much written detail as possible: "The more information you provide, the better."

We discontinued the survey on 8 April 2020, when state-regulated restrictions first began to ease.

The State-Trait Anxiety Inventory (STAI; Spielberger et al., 2008) , validated for use in Spain by Buela-Casal, et al. (2015) , is a 40-item self-report instrument that assesses anxiety as both a state (20 items) and a trait (20 items). With items in a 0-3 response format, state and trait anxiety scores range from 0-60, and benchmark cut points for adult Spanish men and women provide a basis for classifying subscale scores from "very high" to "very low" (Buela-Casal et al., 2015) . Internal consistency coefficients for the current lockdown sample were α = .93 and .84 for state and trait anxiety, respectively.

The Beck Depression Inventory (BDI; Beck et al., 1996) , validated in Spanish by Sanz & Vázquez (2011), consists of 21 self-report items measuring the presence and severity of

This article is protected by copyright. All rights reserved depression. As with the STAI, benchmark cut points for the Spanish population permit classifying BDI scores as reflecting minimal, mild, moderate or severe depression (Sanz & Vázquez, 2011) . Reliability for the lockdown sample was α = .85.

Participants with no children at home completed the Spanish version of the Dyadic Adjustment Scale (DAS; Spanier, 1976, 2017), a 32-item questionnaire measuring general couple relationship quality. The DAS also has consensus, satisfaction, affectional expression, and cohesion subscales, but because these were highly intercorrelated we used only the DAS total score (α = .94) in the main analyses. Although clinical cut points for the DAS are not 

Participants with children at home completed the Basic Family Relations Evaluation Questionnaire (CERFB; Ibáñez et al., 2012; Vilaregut et al., 2019) , a 25-item parent-report instrument inspired by Linares' (2002 Linares' ( , 2012 theoretical ideas about relational nurturing. The original CERFB includes a 14-item parenting function scale (α = .92), measuring the quality of parent-child relations (e.g., "I feel that my children return my affection"), and an 11-item conjugal function scale (α = .91) reflecting the quality of how parents relate to each other as a couple (e.g., "My partner knows how to treat me"). Responses are on 5-point Likert scales ranging from 1 (never) to 5 (always). To further assess collaboration between the two parents with regard to childrearing, we included 16 additional items from a preliminary co-parenting scale (α = .85) currently undergoing validation (e.g., "We make a good team as parents"). Clinical norms for CERFB scales are not yet available but, as with the Spanish DAS, descriptive statistics from validation studies provide tentative benchmarks for evaluating levels of parental and conjugal functioning in the lockdown sample (Campreciós, 2015; Ibáñez et al., 2012; Roca et al., 2020; Vilaregut et al., 2019) . Interpretation may be complicated, however, because identifying Linares' patterns of functional and dysfunctional parenting requires taking both dimensions (and ideally the views of both parents) into account.

This article is protected by copyright. All rights reserved Given the exploratory nature of the study, we approached the research aims with the basic premise that combining qualitative and quantitative forms of evidence provides a better understanding than either method does by itself (Creswell & Plano Clark, 2011) . Indeed, both qualitative and quantitative data figure prominently in the resultsand despite homage to validated quantitative measurement methods (above), our most direct evidence of actual lockdown "effects" on couple and family functioning came from participants' qualitative (written) descriptions of what had changed. In the spirit of mixed-method research, we then used dichotomous variables representing the presence or absence of specific qualitative themes in participants' responses to explore quantitative associations with other study variables.

After dropping 23 written responses reporting no couple/family change (e.g., "Everything continues as usual") and 53 blank responses, we used Braun & Clarke's (2006) method of thematic analysis, assisted by ATLAS.ti software for Mac (v. 8), to code 329 descriptions of change. The total qualitative data set consisted of 13,226 words, with individual responses ranging from 2 words (e.g., "Closer now") to 353 words. The thematic analysis involved identifying interesting data features, or codes; clustering codes and searching for potential themes; and finally, naming and defining the themes. To facilitate accuracy and trustworthiness, two authors served as co-coders in an ongoing consensual review process, and the full team reviewed emerging results to reach on the final thematic configuration.

On the quantitative side (using SPSS Statistics, v. 20), preliminary analyses justified creating two composite variables that would simplify later examination of psychological distress and couple relationship functioning: composite distress was a z-score combination of STAI state, STAI trait, and BDI scores, which intercorrelated highly with all r s > .66. Similarly, the composite measure of couple functioning combined z-transformations of DAS total scores and CERFB conjugal function scores, which were not available for the same participants here but had correlated highly in previous research. As dependent variables, the two composites helped clarify multivariate and moderated influences on central study constructs.

We approached the main study aims by first examining descriptive statistics for psychological distress (STAI, BDI) and relationship functioning (DAS, CERFB) in the lockdown sample with an eye toward areas of possible discrepancy and/or alignment with benchmarks for the broader population. In light of sampling limitations both here and in the standardization studies, however, such comparisons can only be approximate, with conclusions about lockdown

This article is protected by copyright. All rights reserved effects necessarily tentative. Next, having defined qualitative change themes (as described above) and coded their presence/absence across participant responses, simple tabulations and cross-tabulations illuminated the prevalence of various improvement and deterioration themes in the full sample and across the four main participant groupings: partnered parents with children at home, partners in couples with no children, parents with all children away from home, and divorced parents.

Finally, we employed a variety of univariate, multivariate, and moderation analyses to identify demographic, household, and employment-related correlates of pandemic relationship functioning and psychological distress, again with attention to variations across household/parental subgroups. In addition to CERFB scores, DAS scores, and the two composites, these correlational explorations included the qualitative change themes and a rough approximation of the Linares CERFB parenting styles (functionality, triangulation, etc.). Also of interest were possible changes in relationship functioning over time as the lockdown progressed, including system-symptom links between relationship quality and individual distress. 13.4% of lockdown participants but only 4.6% were in the moderate or severe range. When asked about their perception of economic risk, just over half of the respondents were at least moderately concerned with economic risk during the pandemic (53.5%) but less than a quarter were very concerned (15.7%) or extremely concerned (6.6%). In general, the lockdown experience appeared to generate moderate to high levels of situational anxiety and uncertainty but not much chronic distress among the adults who completed the on-line survey.

This article is protected by copyright. All rights reserved CERFB norms provide no firm basis for locating parenting styles in the Linares' quadrants, a rough comparison of lockdown means with standardization data suggests that at least half of the reports in our sample would be most consistent with the triangulation style, where parents tend to maintain harmony by involving children in their conflicts. Such classification is approximate at best, however, especially without reports from both parents or confirmation from a child.

To summarize, with no children in the household, the quality of couple relationships during the COVID-19 lockdown appeared no worse and possibly better than would have been the case without lockdown. With children present, however, our data raise the possibility that preservation of family harmony may have sometimes occurred at the expense of relational nurturing.

Two overarching thematic categoriesperceived improvement and perceived deteriorationemerged from our qualitative analysis of participants' free-form descriptions of how family and couple dynamics had changed during the lockdown. Table 2 shows the prevalence of these themes in the subsample of 329 respondents who indicated that some degree of lockdown-related family change had occurred. (Note that a given response could include more than one theme; in fact, 24 cited both improvement and deterioration.) Interestingly, the overall prevalence of improvement themes (61.7%) exceeded the prevalence of deterioration themes (41.0%), with increased connection/cohesion (44.7%) and conflict (21.9%) cited most often. Relative to the DAS and CERFB data above, this appears more consistent with the dyadic adjustment of couples having no children at home than with the CERFB parenting picture. Table 2 also indicates (via chi-square comparisons) that some themes distributed less evenly than others across subgroups defined by household composition and parental status. For example, the theme of family (re)connection was more prevalent for parents with no children and those with children at home than for parents who were divorced or had children not at home. The most striking differences, however, involved deterioration themes such as couple/family distance (most common with children away and negligible with children at home), conflict atmosphere (less frequent with emancipated children) and unbalanced needs (most likely with children with children at home).

Overall, participants tended to use more words when describing deterioration themes than improvement themes, with unbalanced needs (r = .24, p < .01), negative expectations (r = .21, p < .001), and conflict atmosphere (r =.14, p = .01) entailing the thickest descriptions. On the improvement side, longer responses were associated with balanced needs (r = .26, p < .001) and teamwork spirit (r = .16, p = .003).

As noted above, we created two composite dependent variables representing psychological distress and couple functioning to facilitate identifying correlates of individual and relational wellbeing during lockdown. The distress composite is available for the full sample while the latter includes all participants except 19 who had divorced. The couple composite is useful because the survey entailed different (though highly correlated) measures of couple functioning depending on whether or not the respondent had a child at home. Thus, although it was not possible to compare these subgroups directly, the z-score composite permits examining differential (moderated) prediction of conjugal functioning.

This article is protected by copyright. All rights reserved Table 3 provides an overview of predictor variables associated with various measures of individual, couple, and parental functioning. Point bi-serial correlations in the first few rows, where dichotomous dummy variables represent household/parental subgroups, reveal few group differences in composite individual or couple functioning apart from the small group of divorced parents and parents with a preschool-age child reporting more psychological distress.

A strikingly different picture of couple functioning emerges in relation to pandemicrelated employment, where stronger associations appear when respondents have children at home (CERFB conjugal function) compared to when they do not (DAS total score). Of particular interest is how couple adjustment relates to telecommuting and (un)employment. General linear model (GLM) moderation analyses examining these association across three groups of couples (no children, child at home, empty nest) found significant Group x Telecommute (F = 5.23, p = .006) and Group x Employed (F = 4.44, p = .012) interaction terms, with group means suggesting that telecommuting and employment were positively related to couple functioning when respondents were parents (regardless of whether children were at home or emancipated) but in the opposite direction when they were not. Table 3 also highlights correlates of individual distress and relationship functioning that did not vary appreciably (interact) with household/parental status: Psychological distress was generally higher among women, recipients of psychological treatment, unemployed respondents, those in business professions or perceiving economic risk, and those with a health problem at homebut lower among health professionals. Among the CERFB family relationship measures, parental functioning showed significant correlations with some of the same predictors and was positively related to education; and if anything, marital and co-parental functioning was better among younger couples.

The last two rows in the top panel of Table 3 show potentially important associations involving lockdown duration, couple functioning, and psychological distress. First, marital functioning for couples with children at home systematically improved with days in lockdown, which was not the case for parenting functions, psychological distress, or for couples without children at home. This cross-sectional temporal trend remained after statistically controlling for parallel changes over days in other demographics such as urban location, employment status, or having a preschool child at home. Note also that effective marital functioning with a child at home correlated positively with telecommuting and having a job but not with psychological

This article is protected by copyright. All rights reserved distress. In this way, couple functioning in the context of active parenting during lockdown was unique among the relationship measures.

The strong associations between relationship quality and individual distress in Table 3 are consistent with a large couple and family relations literature, but we do not know if they are in any way unique to quarantine. An indirect way to approach this is to ask whether the linkage between relationship quality and individual functioning systematically changed over the course of quarantine, as would be indicated by statistical interactions involving lockdown days. GLM analyses along these lines, with psychological distress as the dependent variable, did in fact show suggestive relationship-quality x lockdown-duration interaction terms for DAS total dyadic adjustment (F = 5.37, p = .022) and CERFB co-parenting (F = 4.50, p = .035) in the direction of system-symptom linkages strengthening as the lockdown progressed. The lockdown study period was relatively short, however, so this intriguing evidence is tentative at best.

Finally, because the CERFB results in Table 3 do not speak directly to the Linares model of relational nurturing that guided development of the instrument, we were interested in how parenting styles representing functionality, triangulation, deprivation and chaotization might have operated during the lockdown. This seemed important because the apparent parentalconjugal discrepancy in our sample suggests that some degree of triangulation may have been in play. Although the CERFB standardization data offers no cut points for defining Linares quadrants with any precision, we attempted to approximate the quadrants in a relative manner by splitting the conjugal and parental distributions at their medians. Interestingly, participants in the high-parenting/low-conjugal (triangulation) quadrant reported less distress and fewer deterioration themes than other participants (Supplementary Table S1 , online). Triangulation represented in this manner was also more common among university graduates and health professionals.

Because the most direct evidence of lockdown effects on family relationships comes from participants' free-form reports of improvement and deterioration, we were interested in which qualitative themes were most and least likely to come from which participants. The bottom panel of Table 3 shows strong associations between perceptions of relational improvement/deterioration and some but not all of the standardized individual, couple, and parental functioning measures. In particular, the qualitative reports aligned most solidly with

This article is protected by copyright. All rights reserved psychological distress and with the dyadic adjustment of partners with no children in the household. With children at home there were no significant correlations for CERFB conjugal functioning and only marginal ones for parenting and co-parenting. Of the specific qualitative themes, family (re)connection dominated criterion connections on the improvement side while conflict and distance did so on the deterioration side. Table S2 , online) indicate that respondents reporting improvement were more often employed and had no emancipated children or preschoolers at home. Deterioration themes, on the other hand, were associated with receiving psychological treatment, the presence of preschoolers, and coping with health problems in the household. As for markers of specific qualitative theme categories, improvements in (re)connection were more prevalent among health professionals and respondents with no preschoolers; education and employment predicted improved communication; expressiveness themes correlated with telecommuting and low perceived economic risk; city dwellers were high on teamwork; younger participants described more conflict and fewer distance themes; females reported more conflict; health professionals described fewer negative expectations; an empty nest predicted more distance and less conflict; health problems in the household portended distance themes; and having children at home meant less distance and more unbalanced family member needs as noted above.

This mixed-method exploration of the recent COVID-19 lockdown in Spain adds depth and complexity to the quarantine literature by highlighting repercussions for relationships as well as individuals in the direction of improved as well as compromised functioning. Although survey respondents experienced moderate to high levels of situational anxiety during the first three weeks of state-regulated lockdown, they also reported high dyadic (couple) adjustment relative to a norm group and cited more instances in which their couple and family dynamics had improved rather than deteriorated. This contrasts with previous studies that have emphasized predominantly negative psychological effects of mass quarantine on individuals (Brooks et al., 2020) , and occasionally also on couple and family relations.

The results also suggest that quarantine ramifications for couples and families are more complex than those for individual children and adults. One aspect of this complexity is that positive as well as negative forms of change appear more likely at the level of close relationships

This article is protected by copyright. All rights reserved (with opportunities for collective coping, interpersonal need satisfaction, etc.) than at the level of individual adjustment. Positive individual change (e.g., post-traumatic growth) is certainly possible, of course, but the quarantine literature has not emphasized this. Another indication of greater complexity is that measures of couple and parental functioning varied with household composition and parental status in ways that measures of individual wellbeing (psychological distress) did not. For example, conjugal relations during lockdown appeared more harmonious when there were no children in the household, and moderation analyses indicated that COVIDrelated employment variables (e.g., telecommuting) predicted successful couple functioning in different ways depending on parental status.

In general, the relationship implications of lockdown were clearer for participants with no children at home than for those actively engaged in parenting. For the former, dyadic adjustment scores were higher than those from a benchmark comparison group and correlated highly with both the absence of psychological distress and the presence of improvement themes in participants' descriptions of how relationship dynamics had changed. For active parents, on the other hand, benchmark CERFB comparisons were more ambiguous, with parenting quality tending to exceed conjugal quality, and qualitative improvement/deterioration themes did not map so clearly onto relationship functioning.

Although tentative, the apparent prevalence in our sample of a parenting style Linares (2002) associates with triangulation raises the possibility of a subtle downside to family relations during lockdown. If greater appreciation of parent-child than parent-parent (conjugal) bonds does in fact invite triangulation of children into adult conflicts, this could undermine child development in ways that would not be apparent without direct assessment of child functioning.

The fact that provisional triangulation correlated with less psychological distress and fewer family conflict themes seems also to underscore the subtlety of this ultimately dysfunctional pattern. As noted earlier, however, the CERFB norms as yet provide no firm basis for identifying Linares parenting styles, so without converging reliability and validity evidence we must offer this interpretation cautiously. If nothing else, the possibility of increased risk for triangulation during lockdown provides a hypothesis for future research.

What might it mean that participants without children at home more often reported increased couple cohesion during lockdown? From an interpersonal-systems perspective, the relative absence of third-party involvements with children, relatives, friends, or colleagues,

This article is protected by copyright. All rights reserved coupled with sustained proximity and more time for shared conjugal activities, provides a plausible explanation. A more psychological explanation, from the author of a recent survey of U.S. couples conducted around the same time as ours, is that people simply want more closeness in their important relationships and turn to partners for support under COVID-19 stressso in effect they are getting what they want (G. Lewandowski, as cited in Bonos, 2020) .

The current data are cross-sectional and can only indirectly address any changes that may have occurred over time during the relatively brief (two-week) study period. Such change was of interest because at least one prior study found increased symptoms of psychological distress among individuals quarantined at least 10 days (Hawryluck et al., 2004) . This was not the case in the present study, where STAI, BDI and composite measures of individual distress did not changeyet the CERFB measure of conjugal functioning actually showed an opposite trend of improvement over lockdown days. In other words, participants challenged by managing children at home, whose conjugal relations were probably not as good on average as those of other participants, nonetheless tended to report more couple-level resilience in the third week of lockdown than in the second. Our rough indicator of triangulation decreased over days as well, implying the possibility of positive spillover for children. Intriguingly, moderation results also suggest that the correlation or linkage between individual distress and couple relationship functioning may have strengthened as the lockdown progressed. The direction of influence in this would be ambiguous, however.

In addition to supplementing quantitative analyses, the qualitative themes add texture and detail to how participants experienced relational improvement and deterioration during lockdown. On the deterioration side, some of the themes echo aspects of previously identified stressors linked to quarantine at the individual level. For example, experiencing loneliness and couple/family distance connects to boredom and isolation (Cava et al., 2005; DiGiovanni et al., 2004) ; and negative expectations incorporates fear of the disease (Bai et al., 2004; Desclaux et al., 2017) , economic uncertainty (Jeong et al., 2016; Mihashi et al., 2009) , and information overload (Blendon et al., 2004) .

Although a few participants cited instances of both improvement and deterioration, most followed one of the two paths in ways that sometimes resonated with our clinical experiences during the first few months of the pandemic. As if to illustrate the improvement pathway, a stayat-home mother who had come to therapy distressed by a disobedient, disrespectful and

This article is protected by copyright. All rights reserved occasionally violent child reported that she, her husband, and the difficult son had been having "so much fun together" (for a change) since the lockdown began, adding that "while we were playing, several picture frames fell of the wall and broke without even bothering me!" More often, however, we encountered deterioration themes in the form of a lockdown family crisis.

Here a useful strategy has been to contextualize the current problem situation by exploring how family members had (successfully) handled similar situations or had more satisfying relations before the lockdown began. The qualitative improvement themes suggest possible areas to explore.

Our study has many limitations, with sample representativeness foremost among them.

The self-selected sample of convenience was predominantly female, well educated, employed, interested in the topic at hand (family relationships), and probably more functional and economically advantaged than the adult Spanish population at large. It is entirely possible that different results (e.g., more evidence of individual distress or relational deterioration) would obtain in other sectors of the population or in other countries or cultures. Another limitation is that time-limited, cross-sectional survey data shed little light on enduring effects of quarantine, on how adaptations to lockdown changed or evolved over time, or on what happened during reopening, when home-confinement restrictions began to ease. We plan to address these and other questions through follow-up interviews with study participants. Finally, it is difficult to obtain reliable and valid representations of family functioning from the self-reports of only one family member, and shared method variance (e.g., positive or negative response set) could account for some of the quantitative association between individual and couple/family variables.

Despite these limitations, our exploratory study of the recent state-regulated COVID-19 lockdown in Spain helps to break new ground by looking beyond the individual to understand psychosocial impacts of quarantine experiences. The ramifications of these experiences for couples and families are clearly important and probably more complex than for individuals. Notes. Table entries are Pearson or point-biserial correlation coefficients reflecting associations between predictor variables and measures of individual, couple and parental functioning. Left-most column includes prevalence proportions for dichotomous predictors and qualitative themes. a Z-score composite of STAI state and trait anxiety measures and BDI depression available for all participants. b Z-score composite of total dyadic adjustment (DAS) and CERFB marital functioning scores, excluding divorced parents. c Dyadic Adjustment Scale (DAS) total score. d Family relations (CERFB) scores for parents with at least on child at home. Two-tailed significance levels: *** p < .001, ** p < .01, * p < .05, † p < .10.

Survey of stress reactions among health care workers involved with the SARS outbreak

BDI-II. Beck Depression Inventory-Second Edition. Manual

The public's response to severe acute respiratory syndrome in Toronto and the United States

Real Decreto 463/2020, de 14 de marzo, por el que se declara el estado de alarma para la gestión de la situación de crisis sanitaria ocasionada por el COVID-19

Our relationships are relationships are actually doing well during the pandemic, study finds

Using thematic analysis in psychology

The psychological impact of quarantine and how to reduce it: rapid review of the evidence

Cuestionario de ansiedad estado-rasgo

Validación y aplicabilidad clínica del Cuestionario de Evaluación de las Relaciones Familiares Básicas (CERFB) en los trastornos de la conducta alimentaria (Doctoral Dissertation)

Family Functioning and Its Predictors among DisasterBereaved Individuals in China: Eighteen Months after theWenchuan Earthquake. PLoS ONE, e60738

The experience of quarantine for individuals affected by SARS in Toronto

Resilience in Midwestern families: Selected findings from the first decade of a prospective, longitudinal study

Designing and conducting mixed methods research

Prospective Relations between Family Conflict and Adolescent Maladjustment: Security in the Family System as a Mediating Process

Accepted monitoring or endured quarantine? Ebola contacts' perceptions in Senegal

Factors influencing compliance with quarantine in Toronto during the 2003 SARS outbreak

The Great Recession and Mental Health in the United States

A third of Americans now show signs of clinical anxiety or depression, Census Bureau finds amid coronavirus pandemic. The Washington Post

Chinese city experiencing a divorce peak as a repercussion of COVID-19. Global Times

SARS control and psychological effects of quarantine

Accepted Article This article is protected by copyright. All rights reserved

Family stress: Integrating theory and measurement

Validación del Cuestionario de Evaluación de las Relaciones Familiares Básicas (Doctoral Dissertation)

Propiedades psicométricas del Cuestionario de Evaluación de las Relaciones Familiares Básicas (CERFB)

Mental health status of people isolated due to Middle East respiratory syndrome

Asian Americans' family cohesion and suicide ideation: Moderating and mediating effects

Identidad y narrativa. La terapia familiar en la práctica clínica

Del abuso y otros desmanes. El maltrato familiar, entre la terapia y el control

Terapia familiar ultramoderna. La inteligencia terapéutica

Health Impacts of the Great Recession: A Critical Review

The relevance of psychosocial variables and working conditions in predicting nurses' coping strategies during the SARS crisis: an online questionnaire survey

Family functioning in the aftermath of a natural disaster

Accepted Article This article is protected by copyright. All rights reserved

Predictive factors of psychological disorder development during recovery following SARS outbreak

Propuesta de Resolución Provisional del Procedimiento de Concesión de Ayudas a Proyectos de I+D+I

Construcción del Cuestionario de Evaluación de Relaciones Familiares Básicas y Coparentalidad (CERFB-CoP). Poster session presented at XIX Jornades ANPIR

Understanding Family Resilience

Social Consequences of Ebola Containment Measures in Liberia

A nationwide survey of psychological distress among Chinese people in the COVID-19 epidemic: implications and policy recommendations

Risky families: Family social environments and the mental and physical health of offspring

The authors appreciate the assistance of Dr. Michael Rohrbaugh in preparing the manuscript for publication.