Insecure attachment and support-seeking during COVID-19: a sequential mixed methods investigation

ABSTRACT The COVID-19 pandemic led many couples to stay at home together with minimal contact with others. As social distancing measures reduced contact with outside households, many partners could seek support only from one another. In two studies using a sequential mixed methods design, we investigated support-seeking behaviors in romantic relationships during COVID-19. In the qualitative study, semi-structured interviews (n = 48) showed differences in how and why partners seek support with an initial consideration of the role of attachment. We identified the following themes: direct support-seeking, indirect support-seeking, (in)dependence, and gender dynamics. In the quantitative study (n = 588), high COVID-19 worry, high attachment anxiety, and low attachment avoidance were associated with more support-seeking. Interestingly, we also found that when COVID-19 worry was high, individuals high in attachment avoidance were more likely to report seeking support. The paper provides a unique insight into the impact of COVID-19 on support-seeking behaviors in romantic relationships.

relationships to rely on each other more than usual, due to reduced access to their normal social support networks.
Attachment theory suggests that the emotional bond that develops between an infant and a caregiver serves an evolutionary function to ensure survival (Bowlby, 1969).The attachment behavioral system gets activated in the face of perceived threat and makes infants seek support from their caregiver to reduce their distress (Bowlby, 1969).In adulthood, one's romantic partner often becomes the primary attachment figure from whom to seek support in times of distress (Hazan & Shaver, 1994).Indeed, adults use their romantic partners as a safe haven during times of distress and ambiguity (Collins & Feeney, 2000;Heffernan et al., 2012).The COVID-19 pandemic created an environment characterized by uncertainty and fear of death, leading the attachment system to be activated.Thus, the distress and ambiguity surrounding an unprecedented global pandemic led couples to depend on their romantic attachment figures for support.Partner support has many positive outcomes for individual and relational wellbeing (Austin et al., 2022;Hilpert et al., 2018;Lane & Fink, 2015) and this support buffered the negative consequences of the pandemic on relationship satisfaction (Balzarini et al., 2023).Nonetheless, to receive support, individuals may need to actively engage in support-seeking.Support-seeking is defined as an individual's attempts to elicit support from an attachment figure or caregiver.The present research aims to add to the literature by taking a mixed-methods approach to examine whether support-seeking within romantic relationships was impacted by COVID-19 and the role of adult attachment styles in influencing these support behaviors.

Support-seeking in stressful times
A recent theoretical framework suggested that couples with the most responsive support would have the least negative consequences from the pandemic (Pietromonaco & Overall, 2020).The framework suggests that external stressors can harm dyadic processes (e.g.communication, dyadic stress, social support, and self-disclosure) within a romantic relationship and therefore reduce relationship quality.It also suggests that pre-existing vulnerabilities within the relationship, such as social status or attachment insecurity, can heighten these negative consequences.In contrast, responsive support from a romantic partner can potentially reduce the negative effects of pandemic-related stressors on relationships (Pietromonaco & Overall, 2020).More recent research has found that couples who communicate effectively and provide responsive support to one another during the pandemic are likely to be more resilient in the face of challenges and thus reduce the negative impacts of pandemic-related stress (Pietromonaco & Overall, 2022).To receive responsive support, the support seeker may need to take an active role in eliciting support from their partners (Feeney & Collins, 2015).
Several studies have examined support-seeking during the COVID-19 pandemic with mixed results.Shanahan et al. (2020) found that individuals sought more support when they were stressed.However, other studies have shown that individuals sought more support when they were less stressed (Cao et al., 2020;Chew et al., 2020).Further research found that while COVID-19 stressors led to rumination, social support buffered this consequence suggesting that participants with high support were better equipped to handle pandemic-related stressors (Ye et al., 2020).However, none of these studies have been specific to support behaviors within romantic relationships.Therefore, the present research aims to address this gap in the literature and gain an understanding of the association between the COVID-19 pandemic and partner support-seeking behaviors.

The role of attachment in support-seeking
Notably, individual differences predict whether and how individuals seek support in romantic relationships.Attachment theory (Bowlby, 1969) states that individuals form an attachment style as an adaptive response to the care they receive, and this style determines how individuals seek support in relationships.Attachment styles vary on two dimensions: attachment avoidance and attachment anxiety (K.Brennan et al., 1998).Avoidant attachment stems from a lack of responsive caregiving (Mikulincer & Shaver, 2016).Avoidance results in the deactivation of the attachment system, so avoidantly attached individuals tend to disregard or downplay any signs of threat and become self-reliant (Mikulincer & Shaver, 2016).Anxious attachment stems from inconsistent responsiveness from caregivers (Mikulincer & Shaver, 2016).Anxious attachment leads to hyperactivation of the attachment system and as a result these individuals focus excessively on threatening stimuli and seek excessive support and reassurance in relationships (K. A. Brennan & Carnelley, 1999;Mikulincer & Shaver, 2016;Shaver et al., 2005).Individuals who are low on both avoidance and anxiety are securely attached.Secure attachment is formed when an individual receives consistent, responsive care, and therefore feels comfortable engaging in support-seeking (Mikulincer & Shaver, 2016).
Previous research generally has concluded that individuals with a secure attachment demonstrate the most effective support-seeking behaviors (Collins & Feeney, 2000;DeFronzo et al., 2001;Feeney et al., 2008;Li et al., 2020).In contrast, avoidantly attached individuals display less frequent support-seeking (Mikulincer & Shaver, 2016;Simpson & Rholes, 2017;Simpson et al., 1992), as well as ineffective support-seeking behaviors (Collins & Feeney, 2000).In contrast, individuals with an anxious attachment seek excessive levels of support (K. A. Brennan & Carnelley, 1999;Mikulincer & Shaver, 2016), or become hesitant to seek support when they doubt its availability, which can result in employing indirect support-seeking strategies (Fraley & Shaver, 1998;Rholes et al., 2001).As such, the heightened perception of threat can lead to two converse reactions of types of support seeking for anxiously attached individuals.Pietromonaco and Overall (2020) predicted that insecurely attached individuals will be vulnerable to the negative impacts of the pandemic.Research has found that attachment insecurity was negatively associated with relationship functioning during the COVID-19 pandemic, providing support for the theoretical framework (Overall et al., 2021), however, they did not examine support-seeking behaviors despite evidence that support seeking was associated with relationship satisfaction during the pandemic (Xu et al., 2020).Therefore, we examined whether attachment dimensions moderated the association between pandemic-related worry and support-seeking behaviors -a novel research question.

The current research
The present study explored how support-seeking within romantic relationships during the COVID-19 pandemic.To our knowledge, there are currently no published papers that address the association between COVID-19 stress and support-seeking specifically within romantic relationships.As social distancing measures reduced contact with outside households, many partners could seek support only from one another; thus, romantic relationships must specifically be explored to develop a more comprehensive understanding of support-seeking during COVID-19.Additionally, given the importance of support-seeking behavior in relationship functioning (Feeney & Collins, 2015;Forest et al., 2021), it is necessary that this gap in the literature is explored.The current research addressed this relationship using two studies, qualitative and quantitative, allowing for a comprehensive understanding of the research topic (Creamer & Reeping, 2020).In the qualitative study, we examined how partners sought support during the early stages of the COVID-19 pandemic and whether individual differences in attachment styles influenced the support-seeking.In the quantitative study, we examined whether attachment styles and worrying about COVID-19 predicted support-seeking behaviors.The study p r e r e g i s t r a t i o n c a n b e f o u n d h e r e : h t t p s : / / o s f .i o / m w k n c ? v i e w _ o n l y = d78dc7f4390b48189e3ffb95a4180469.

Qualitative study
The objective of the current research was to gain an in-depth understanding of supportseeking within the qualitative component and benefit from the larger sample and generalizability of quantitative data.The objective of the qualitative study was to explore how partners sought support during the early stages of the COVID-19 pandemic lockdown (Research Question 1; RQ1).We conducted qualitative interviews with 48 participants and followed up with some of these participants one month later to assess how their supportseeking behaviors had changed during lockdown.This method allowed us to identify what couples believed had worked for them.We additionally measured individuals' attachment to consider the role of attachment in support-seeking.

Participants
Forty-eight participants took part in the qualitative study.Participants were primarily White, heterosexual women, with an average age of 36 years (Table 1).Most participants did not have children and were employed full-time or part-time.The participants had an average relationship length of M years = 10.4,SD = 10.9.

Procedure
Ethical approval was obtained from the authors' institutional review board (ERGO ID: 56087).We preregistered the hypotheses and design on the Open Science Framework.Participants were recruited via Prolific and social media; they had to be 18 years or older and living with their romantic partner in a country with social distancing measures.All participants provided written consent to participate in the study, and were told their participation was voluntary and responses would be confidential.All participants completed a questionnaire on demographic characteristics and all but one participant completed a section to measure their attachment.Using Zoom (Zoom Video Communications Inc, 2016), we audio recorded semi-structured qualitative interviews.Interviews were transcribed using artificial intelligence software then corrected by research assistants.Participants were asked "How have you asked for support from each other during the pandemic" and prompted to describe feelings toward seeking support, helpful strategies, and any changes during the pandemic.A total of 48 participants completed the first interview between 30 March 2020 and 21 April 2020 during which most countries were under stay-at-home orders.Participants who had completed the first interview within the first two weeks of data collection were invited to a second interview to identify changes in support-seeking across the pandemic.A two-week cut off for data collection was required to ensure the second interviews could take place before the study period ended.The second interviews were completed by 14 May 2020.Of the 23 participants invited to the follow-up interview, 19 replied and completed both interviews.The first interview lasted between 14 and 49 minutes and the follow-up interview lasted between 7-24 minutes.
After completing each interview, participants were eligible to win one of two £30 Amazon vouchers, and one of two £20 vouchers for the second interview.

Attachment styles
Attachment was measured using the short version of the Experience in Close Relationships questionnaire (ECR-12;Lafontaine et al., 2016) which includes 12 Likertscale items with two six-item subscales: anxiety (e.g."I worry about being alone;" α = .83)and avoidance (e.g."I don't feel comfortable opening up to my partner;" α = .85).
Participants indicated their agreement with each statement on a scale from 0 (Disagree Strongly) to 10 (Agree Strongly).

Semi-structured interviews
We asked participants questions regarding support-seeking within their relationship: "How have you asked for support from each other during the pandemic?,""How do you feel about asking for support from your partner?," "How do you feel about them asking for support from you?," and "Has any of this changed due to the pandemic?."These questions were repeated in the second interview.

Analysis strategy
We conducted reflexive theme analysis (Braun & Clarke, 2006, 2019) using NVivo, version 12.0 (QSR International Pty Ltd, 2020) of the qualitative interviews.Two coders used an inductive approach in which new codes were created throughout the analysis process.
The coders separately created the initial codes before refining and jointly agreeing on the final codes to 100% agreement.To aid readability, filler words such as "like" and "you know" were removed alongside all identifying information."[. ..]" was used within quotations to provide further information in the quoted data or to remove redundant detail.Analysis of participants attachment style occurred after the interview.Data was analyzed using R, version 4.0.0 (R Core Team, 2020).

Results
A total of four themes (direct support-seeking, indirect support-seeking, gender dynamics, and (in)dependence) were identified and are described below.A total of six sub-themes were identified since main themes of indirect support-seeking and (in)dependence had three sub-themes.Additional representative quotes are provided in the supplementary material.Quotes are accompanied by participant number, gender, age, and attachment dimensions.
The means and standard deviations for the attachment variables are as follows: Attachment anxiety (M = 3.77, SD = 2.50) and attachment avoidance (M = 1.68,SD = 1.45).Individuals below five, the midpoint of the scale, were considered relatively low in attachment anxiety or avoidance with those above the midpoint considered relatively high in the respective measures.A total of 34 participants were relatively securely attached as they were low in both attachment anxiety and avoidance; 12 participants were only relatively high in attachment anxiety; and 1 participant was fearful avoidant as they were high in both attachment anxiety and avoidance.

Direct support-seeking
Participants noted that both their partner and themselves directly sought support.Some participants noted they and their partner used negative direct supportseeking strategies such as "get[ting] very rowdy and grumpy and impatient and demand[ing] to know, 'have you done this yet or have you done that?'" (#15, W, 36, secure).However, most strategies for secure participants were positive such as being "direct and honest" (#26, M, 40, secure).For example, "Me and my partner have always had the confidence of being able to talk about anything . . .I never feel like I can't say something . . .I think he is my confidant" (#13, M, 31, secure).Therefore, many participants felt their direct support-seeking behaviors were a strength and a source of relationship satisfaction.
Some participants noted only one partner directly sought support.Many of these participants believed themselves, not their partner, were "the one to initiate a conversation" (#14, W, 30, secure) when they needed support.On the one hand, some participants relatively high in attachment anxiety felt they were more direct as they required "verbal confirmation. . . .Otherwise [they will] start overthinking and worrying" (#21, W, 25, high anxiety).On the other hand, some participants noted they were not intrinsically direct individuals but had a relationship dynamic in which this was required.For example, "if I wait around for him to notice or offer support to me that usually means tension is building up.So I've learned to be like 'hey I need you to listen . . .I need you to whatever'" (#12, W, 26, high anxiety).Therefore, it appears participants had to modify their support-seeking behavior within their relationship to elicit partner support.
Furthermore, many participants high in attachment anxiety attributed changes in support-seeking to the pandemic.For example, one participant noted "we're asking for help more often, just because we're here more often" (#21, W, 25, high anxiety).An increase in direct support-seeking strategies was also mentioned in follow-up interviews irrespective of attachment style such as "we're trying to be more vocal" (#12, W, 26, high anxiety) and "less getting really stressed out and shouting at each other there and then" (#15, W, 36, secure).As such, an increase in partner availability because of lockdown and social distancing measures allowed for more support-seeking between partners.

Indirect support-seeking
We identified a total of three sub-themes (tone of voice, use of intuition, and physical expression and behavior) within indirect support-seeking.Many securely attached participants noted a use of intuition and that they "can just sense it" (#11, W, 36, secure) when their partner requires support.However, many participants noted more objective behaviors that indicated their partner needed support.Some participants noted a change in tone of voice indicated distress, such as "he starts speaking really fast . . .maybe a little snappy . . .just a little bit short with answers" (#33, W, 29, secure) and suggested needing support.
For some participants there were positive and negative differences in physical expression and behavior which indicated partners' need for support.For example, one participant noted "I tend to be loud and frustrated, sighing and huffing and slamming things" (#25, W, 55, secure), whereas another noted "he tries to physically be in the same space as me and be nearby.That's his way of subtly hinting and overtime I've recognized that when he slowly gets closer, that's when he would like to be supported" (#3, W, 26, secure).Indeed, participants appeared to identify their own overtly indirect strategies of seeking support, such as "If I don't want to talk, I just cry" (#20, W, 29, secure) and their partners' discreet strategies such as "[they] go really quiet" (#38, M, 33, secure).Therefore, securely attached participants appeared to more readily note their partner's subtle changes in behaviors that indicated support-seeking, in comparison to their own support-seeking behaviors which appeared to be more overt.
Within follow-up interviews, participants noted changes in support-seeking as a consequence of the pandemic.For example, "there's definitely been more hugs.Just so desperate for those moments of connection because you're not able to get them from, from anywhere else" (#11, W, 36, secure).Therefore, it appears the pandemic led to an increase for some participants in the seeking of emotional support to manage the impact of social distancing and lockdown measures.

(In)dependence
We identified a total of three subthemes (comfort with partner; actively avoidance of support; does not need support) within (in)dependence which highlighted contrasting views on problem solving individually (independence), or problem solving with your partner (dependence).Some participants who were high in attachment anxiety and attachment avoidance identified they do not need support.For example, "we're both quite independent people" (#17, W, 41, high anxiety).Therefore, some partners reported no distinguishable impact of the pandemic on support-seeking.However, in follow-up interviews some participants noted this preference for independence had decreased during the pandemic.One participant said, "It feels like we're relying on each other a bit more.I said in the last interview we're both quite independent . . .I think we're a bit more dependent on each other now for looking after one another" (#17, W, 41, high anxiety).As such, negative feelings regarding support-seeking and a preference for independence may have changed across the early stages of the pandemic leading to an increase in support-seeking behaviors.
Additionally, some participants noted that although they were experiencing a difficult time, themselves or their partner were avoidant of support as they "sometimes need to be alone.That's how [they] recharge" (#2, W, 37, secure).Such behaviors were typically attributed to preference for space rather than lack of understanding or fear of appropriate support from their partner.Low feelings of stress and a desire for independence thus behaviorally presented for some participants as not seeking support.Furthermore, when typically independent partners did seek support, participants noted this often caused mixed feelings in themselves.For example, "I'm pleased he can ask . . .obviously if I'm feeling very overwhelmed it can be a bit overwhelming because the last thing you need is more things to hold" (#8, W, 26, did not complete attachment measure).Therefore, when participants identified mixed feelings, this tended to be because they felt uncertain of their ability in providing support, as their partner was often independent and rarely sought support.
Alternatively, for participants low in both attachment anxiety and avoidance, independence did not impact their support-seeking behaviors due to their comfort in supportseeking.For example, one participant noted "I'm absolutely happy to ask for support.In the years we've been together I have grown to trust her completely.So I know that if I asked her for support there's not gonna be any backbiting" (#38, M, 33, secure).Other participants noted that they and their partner were more comfortable to seek support during the pandemic than pre-pandemic.For example, "I don't even hesitate anymore, and neither does he . . .Now I find that we ask each other for different things all the time which typically we would not have done a month ago" (#36, W, 52, secure).As such, the pandemic had led to an increased willingness to seek support for both partners as opposed to independently problem solving.
When participants identified positive feelings, this tended to be regarding their own support-seeking.Whereas, when participants identified negative feelings, this tended to be regarding their partner seeking support.Therefore, it appears some partners are more comfortable seeking support than having their partner seek support, due to a lack of confidence in support-providing skills rather than an unwillingness to provide support.

Gender dynamics
Gender dynamics was an overarching theme that impacted support-seeking.Men were noted to be reluctant to seek help due to "traditional, old fashioned (values) . . .so [they] will internalize loads of stuff" (#4, W, 46, secure).Nonetheless, the impact of gender dynamics was not one-sided with women noting reluctance in support-seeking.For example, one participant noted "if it's emotional support I'm after then I'm okay asking.If it's a practical task and I wouldn't be able to do it myself there's this gender dynamics going on "no I don't need a man to help me'" (#11, W, 36, secure).As such, traditional values of gender appeared to have been influential to some men, thus they did not seek support.Whereas, some women seemed to oppose traditional gender values so did not seek practical support.
Alternatively, due to the pandemic, some partners were more willing to seek support.For example, one participant said "he's quite masculine . . .very stereotypical gender roles in his head. . . .now he's been a lot more open with the fact that other people could help you . . .that's definitely changed that he's willing to ask for support if he needs to" (#33, W, 29, secure).Indeed, within follow-up interviews some participants also identified changes in support-seeking that countered gender stereotypes.For example, "we've talked a little more . . .All the men in his family are very "don't talk about emotions' and so that's something we kind of worked on for a long time and he's been a lot more communicative" (#5, W, 36, high anxiety).It therefore appears the pandemic assisted in changing support-seeking behaviors, especially for men who often became more willing to seek support.

Discussion
The qualitative component highlighted specific ways partners engaged in support-seeking and how these behaviors have been affected by the pandemic.The inclusion of the qualitative component fills a clear gap in the literature regarding how individuals seek support from romantic partners during COVID-19.
Most participants in the qualitative interviews reported that both them and their partner engage in direct support-seeking behaviors, which were noted to have increased during the pandemic in follow-up interviews.Some participants additionally noted that they were more direct than their partner -for anxious individuals, this was due to needing reassurance or believing their partner did not realize they needed support.An increase in direct support-seeking was attributed to an increase in partner availability due to lockdown and social distancing measures that allowed partners to learn new ways of communicating their needs.Direct support-seeking was mainly associated with positive feelings, which is unsurprising given it elicits more helpful forms of support than indirect support-seeking (Don & Hammond, 2017); this was most notably discussed by participants relatively secure.This is in line with previous research which identifies that individuals high in attachment anxiety desire support and assistance from their partner but use ineffective coping strategies and support-seeking behaviors to elicit this care (Simpson & Rholes, 2017).
Additionally, participants high in both attachment avoidance and anxiety typically noted a use of negative strategies to seek support as opposed to the more helpful use of positive strategies.As individuals high in avoidance are often reluctant to seek support whereas those high in anxiety often seek more support (Collins & Feeney, 2000), it may appear those high in both attachment dimensions utilize negative strategies to navigate these two contrasting support needs: seeking support appeals to attachment anxiety and using a negative strategy appeals to attachment avoidance.
Participants also engaged in indirect support-seeking behaviors including changes in tone of voice, intuition, and physical behaviors or expressions.Participants reported that their partners also engaged in indirect support-seeking -many relatively secure participants seemed adept at recognizing their partner's indirect bids for support, an example of sensitive caregiving.Furthermore, in line with previous research (Addis & Mahalik, 2003;Armstrong & Kammrath, 2014), we found that gender dynamics influenced participants' perceptions of their own and their partner's support-seeking behaviors.Some women felt hesitant to seek support due to wanting to be more independent and be capable without the assistance of a man.Additionally, some women reported that their partners, who were men, were hesitant to seek support due to existing stigma surrounding vulnerability.Therefore, gender roles seem to be influential in support-seeking behaviors, suggesting both genders face stigma in support-seeking, albeit for different reasons.

Quantitative study
Where the qualitative study focused on how and why individuals were seeking support with a preliminary consideration of attachment, the quantitative study used a larger dataset to more systematically examine the role of attachment in support-seeking during the pandemic.Additionally, in line with the literature on support-seeking during stressful times, we also assessed the role of COVID-19 worry.We hypothesized individuals with high levels of COVID-19 worry would use more support-seeking strategies than those with low levels of COVID-19 worry (H1).Furthermore, due to the use of hyperactivation strategies, we expected individuals high in attachment anxiety would seek support more often than those low in attachment anxiety (H2).Conversely, due to deactivation, we expected individuals high in attachment avoidance would seek support less often than those low in attachment avoidance (H3).We did not expect there to be a significant interaction between COVID-19 worry and anxious attachment in predicting support-seeking from partners (H4).Heightened perception of threat and support can impact amount and type of support-seeking for some individuals with high anxious attachment, but not for others (Fraley & Shaver, 1998;Mikulincer & Shaver, 2016;Vogel & Wei, 2005) thus, it is not clear if there will be an interaction effect.However, we expected there to be a significant interaction between COVID-19 worry and avoidant attachment (H5) in predicting support-seeking from partners: we expected that when COVID-19 worry was high, this would trigger an increased use of deactivation strategies for avoidantly attached individuals, and they would thus seek less support from their partner.
Participants had to be 18 years or over and in a romantic relationship, living together or separately, for at least 6 months.Participants completed questions about demographics, COVID-19 worry, support-seeking behavior, and attachment style.A total of 588 eligible participants completed the questionnaires from 3 December 2020 to 24 February 2021: during this time UK participants experienced a tiered system of local restrictions prior to a full national lockdown on January 4th.Participants recruited via Prolific received £1.00 per 10 minutes.Participants recruited via the institution participant pool received two research credits.Using G*power 3.1 (Faul et al., 2009), we conducted a sensitivity analysis based on a desired power level of 0.9, p = .05and five predictors -this gave us sufficient power to detect a small effect size of f 2 = 0.03.

COVID-19 worry
COVID-19 worry was measured in the using a shortened version of the COVID Stress Scale (Taylor et al., 2020).The original questionnaire includes 36 items but to reduce participant fatigue we shortened it to eight items which we selected based on face validity.We added two additional items that pertained to long-term pandemic worries: "I'm worried about the long-term impact of the virus on the economy" and "I'm worried about the impact of the virus on my finances (or job)."On a scale from 0 (Not at All) to 10 (All the Time) participants answered a total of ten questions on how often they had COVID-related worries; α = .83.Higher aggregate scores indicated greater worry.

Attachment style
As in the qualitative study, attachment orientation was measured in the quantitative study using the short version of the Experience in Close Relationships questionnaire (ECR-12;Lafontaine et al., 2016).This measure includes 12 Likert-scale items across two six-item subscales: anxiety (e.g."I worry about being alone;" α = .83)and avoidance (e.g."I don't feel comfortable opening up to my partner;" α = .85).Participants indicated their agreement with each statement on a scale from 0 (Disagree Strongly) to 10 (Agree Strongly).

Support-seeking
Support-seeking was measured in the quantitative study using the Stress Communicated by Oneself (SCO) subscale from the Dyadic Coping Inventory (Bodenmann et al., 2018).On a scale from 1 (Very Rarely) to 5 (Very Often), participants responded to four questions regarding how often they sought support (e.g."I tell my partner openly how I feel and that I would appreciate his/her support;" α ≥ .71;Tavakol & Dennick, 2011).Higher aggregate scores indicated greater support-seeking.

Quantitative analysis strategy
We conducted a regression analysis to test the main and interaction effects.Data were analyzed using R, version 4.0.0 (R Core Team, 2020); the package interaction, version 1.1.3(Long, 2019); and the package lm.beta,Version 1.5-1 (Behrendt, 2014).Standardized beta coefficients were converted to Cohen's f 2 (Cohen, 1988) to assess the relative strength of the predictors whereby f 2 ≥ 0.02, f 2 ≥ 0.15, and f 2 ≥ 0.35 represent small, medium, and large effect sizes, respectively.To determine the relative strength for the individual levels of the predictors, we applied the following thresholds: β ≥ 0.10, β ≥ 0.30, and β ≥ 0.50 for small, medium, and large effect sizes, respectively.

Exclusions
To handle missing data, we used pairwise deletion in the analysis when participants had not responded to >20% of items on each scale: as a result, one participant was removed from the analysis.Simple mean imputation was used when combining the scales if less than 20% of items were missing, n = 5.Data from these five participants were retained in the dataset to allow for accidentally missing an item.

Regression analyses
As shown in Table 2, our hypothesis that COVID-19 worry would be positively associated with support-seeking behavior (H1) was supported: participants high in COVID-19 worry reported more support-seeking behaviors than those low in COVID-19 worry (f 2 = 0.009, p = .004).There was also a significant positive association between anxious attachment and support-seeking behavior (H2): participants high in attachment anxiety reported more support-seeking behaviors than those low in attachment anxiety (f 2 = 0.017, p < .001).There was a significant negative association with a large effect size between avoidant attachment and support-seeking behavior (H3): participants low in avoidant attachment reported more support-seeking behaviors than those high in avoidant attachment (f 2 = 0.441, p < .001).These associations remained in the moderation analyses.As expected, there was no significant interaction between attachment anxiety and COVID-19 worry when controlling for attachment avoidance; those high in attachment anxiety sought more support than those low in attachment anxiety at both low and high levels of COVID-19 worry (H4).As expected, there was a significant but small interaction effect between COVID-19 worry and avoidant attachment (H5; see Figure 1; f 2 = 0.007, p = .014,)when controlling for attachment anxiety.We explored this interaction effect further by examining the effects of the predictors at each level.When COVID-19 worry was high, those who were high in attachment avoidance reported less support-seeking behaviors than those who were low in attachment avoidance, β = −0.18,SE = 0.02, t = −9.50,p < .001.When COVID-19 worry was low, those who were high in attachment avoidance reported less support-seeking behaviors than those who were low in attachment avoidance, β = −0.25,SE = 0.02, t = −12.82,p < .001.At low levels of attachment avoidance, there was no difference between those high and low in COVID-19 worry, β = 0.00, SE = 0.02, t = 0.25, p = 0.81.At high levels of attachment avoidance, there was a small effect whereby those who were high in COVID-19 worry sought more support than those low in COVID-19 worry, β = 0.07, SE = 0.02, t = 3.65, p < .001.Therefore, individuals low in attachment avoidance sought more support than those high in attachment avoidance.However, when COVID-19 worry increased, individuals low in attachment avoidance did not report a difference in support-seeking behaviors, whereas those high in attachment avoidance reported a significant increase in support-seeking behaviors.

Discussion
The quantitative study aimed to examine support-seeking behaviors during the pandemic as well as whether support-seeking during this time differed by attachment dimension.We found that insecure attachment was associated with how individuals seek support from their partner during the pandemic.Our findings are in line with Simpson et al. (1992) who found that avoidantly attached individuals were less likely to seek support from their partner.Our findings regarding attachment anxiety support and build upon previous research, identifying that anxiously attached individuals appear to seek high levels of support (K. A. Brennan & Carnelley, 1999;Mikulincer & Shaver, 2016), regardless of their COVID-19 worry.In addition to exploring how support-seeking had changed due to the pandemic, we explored whether attachment dimensions were associated with support-seeking behaviors.In line with our predictions, high COVID-19 worry, high attachment anxiety, and low attachment avoidance were associated with more support-seeking behaviors (H1-H3).
Furthermore, as expected, anxious attachment did not moderate the impact of COVID-19 worry on support-seeking from partners (H4) whereas avoidant attachment did (H5).There was a significant interaction between COVID-19 worry and avoidant attachment; the results showed that highly avoidant individuals were more likely to report seeking support under high levels of COVID-19 worry.This interaction was in the opposite direction than we predicted, however, these results support Girme et al. (2015) who found partners who provided high levels of support (as opposed to low levels of support) may overcome highly avoidant individuals' defenses.Therefore, higher levels of partner support were associated with more positive outcomes for individuals high in avoidance, such as decreased distress (Girme et al., 2015).Whereas Girme et al. focused on the partner's perspective and support provision, our research considers the support seeker's perspective and may complement this understanding of avoidant individuals and the acceptability of partner support.Our findings, combined with Girme et al., suggest that as avoidant individuals experience more stress, their partner may provide more support to overcome their defenses and in turn allow them to seek more support.This relationship between support provider and a highly avoidant support seeker should be explored in further research.

General discussion
The present research highlights the impact of individual differences in support-seeking during the COVID-19 pandemic.The qualitative study highlights specific ways that individuals sought support from their partner and how these changed during the course of the pandemic.Individuals typically used direct strategies of support-seeking and indicated differences in approaches to, and feelings toward, support-seeking.Such differences appeared to be associated with differences in attachment styles.Therefore, the quantitative study formally considered the role of attachment in support-seeking behavior.Specifically, we found that high COVID-19 worry, high attachment anxiety, and low attachment avoidance were associated with more supportseeking behaviors -this was in line with predictions (H1-H3).Notably, our findings demonstrate that individuals with an avoidant attachment become were more likely to report seeking support when they face high levels of COVID-19 worry.Indeed, such findings are supported in the qualitative study whereby a number of avoidant individuals reported positive direct support-seeking due to their partner's supportive responses.Considering the importance of support-seeking for both relationship and individual functioning, these findings provide further support to Overall's et al. (2021) framework suggesting those with an avoidant attachment could fare worse during the pandemic.

Implications for research, theory, and practice
This study highlights the importance of qualitative research to go beyond averages and gain a rich understanding.This is particularly important because we believe no studies have currently examined the association between COVID-19 and support-seeking in romantic relationships, so our qualitative interviews provide rich explanations to fill the gap in the literature.Furthermore, whereas many researchers consider the support seeker to be a passive recipient, this research considers the support seeker to play an active role in the support process in line with Feeney and Collins (2015).Therefore, our research has important practical implications: through considering the support seeker as an active role, this research lays a foundation for designing interventions to help support seekers elicit beneficial support in times of distress (Forest et al., 2021).For example, our research suggests the role of gender is important, thus interventions may need to directly address gender stereotypes to improve acceptability of support-seeking.
Indeed, stereotypical values dictate that men who adhere to societal norms should be independent, stoic, and not seek support (Addis & Mahalik, 2003;Forest et al., 2021).However, our research highlights a departure from stereotypes in if, and how, men and women seek support.Some women noted a reluctance to seek support to show independence, whereas some men noted an increased willingness to seek support during the pandemic.When considering these findings in the context of the current literature, it appears that although there has been an increase in some gender-stereotype behaviors during COVID-19 (e.g.childcare responsibilities; Shockley et al., 2021;Waddell et al., 2021), other behaviors, (i.e.support-seeking), may have strayed from gender stereotypes.Our findings suggest researchers and clinicians need to consider changes in gender stereotypes for both men and women, with more research needed after COVID-19 to assess if these gender-related changes in support-seeking behavior remain outside of this novel situation.
In addition, our findings provide empirical evidence that aligns with attachment theory and support the guiding theoretical framework for relationships during COVID-19 (Pietromonaco & Overall, 2020).Specifically, this study identified that individuals with an avoidant attachment will engage in less support-seeking, in line with previous research (DeFronzo et al., 2001;Simpson & Rholes, 2017), but we believe this is the first study to show increased support-seeking of these individuals when COVID-19 worry is high.These findings advance current theories by identifying this interaction.In addition, we found that anxious attachment positively predicted support-seeking regardless of COVID-19 worry, in line with previous research (Mikulincer & Shaver, 2016).Studies have shown that high concern about COVID-19 positively predicts support-seeking (Shanahan et al., 2020); however, we believe this is the first study to empirically assess the relationship between COVID-19 worry, attachment dimensions, and support-seeking in romantic relationships.These findings expand current theory regarding the impact of COVID-19 on romantic relationships, since it incorporates a wider range of factors that may influence the consequences for couples.

Strengths, limitations, and future directions
There are several strengths to this research.First, the sequential mixed-methods design incorporating both qualitative and quantitative data allowed us to develop a deeper understanding of the ongoing impacts of the pandemic on individuals in relationships and the various ways in which support may be sought during the pandemic, or during other stressful periods.Second, as these data were collected during the pandemic, they provide valuable in-depth information on how couples sought support under these unpredictable circumstances.Such intra-disaster research is important given most major stressor research is concerning post-disaster adaptation (Bonanno et al., 2010).
However, this study has some limitations that should be considered when interpreting the findings.First, only one partner in each couple was involved in the research and therefore we only gain one person's perspective -therefore the partner's response to support-seeking was not considered.This also means this study did not assess the role of agreement or disagreement between partners regarding the threat of COVID-19 -a factor that could be influential for both COVID-19 worry and support-seeking behaviors.Future research could assess both partners' perspectives on the impact of the pandemic to gain a more nuanced understanding.
Second, the qualitative component did not include any individuals high in avoidance.This may be due to the interview methodology which required participants to discuss their romantic relationship, but individuals high in avoidance prefer not to open-up (Fraley & Bonanno, 2004) so may be less likely to partake in research.Additionally, individuals high in avoidance avoid stressful stimuli and instead focus on themselves, so may not wish to discuss the pandemic (Mikulincer & Shaver, 2016).
Moreover, our sample was predominately White and heterosexual, potentially limiting the generalizability of our findings to LGBTQIA+ and ethnic minority groups, which have been disproportionally impacted by the pandemic (Boserup et al., 2020;Konnoth, 2020).Future research should explore how these groups sought support during the pandemic.In addition, our research did not assess the impact of cohabitation on support-seeking during the pandemic, since the opportunities to seek support may have varied between couples who lived together during lockdown and those who did not -this could be a direction for future research.
Additionally, within Study 1, participants noted a range of support-seeking behaviors which the 4-item support-seeking scale in Study 2 did not capture.This scale assessed support-seeking; however, does not distinguish between emotional and practical support behaviors that were identified in Study 1. Indeed, evidence suggests that avoidantly attached individuals prefer practical support to emotional support in highly stressful circumstances (Alonso-Arbiol et al., 2002;Girme et al., 2015); however, the quantitative study could not investigate such differences in types of support seeking.At face value, one item appears to address indirect support seeking, with the other items primarily focus on direct support seeking.Nonetheless, it is of note that neither of the major scales used to assess support seeking in romantic relationships would have enabled a comprehensive exploration of support seeking: at face value neither the Coping Orientation to Problem Solving Inventory (COPE; Carver et al., 1989) nor the Berlin Social Support Scale (BSSS; Schulz & Schwarzer, 2003) include items on indirect support seeking.Developing a suitable scale to explore a range of support seeking behaviors is therefore an area for future research.

Figure 1 .
Figure 1.Effects of COVID-19 worry at different levels of attachment avoidance.

Table 1 .
Means and standard deviations for demographic variables.

Table 2 .
Regression analyses for COVID-19 worry and attachment dimensions as predictors of supportseeking behavior with and without moderation effects.