Skilled Support Within Intimate Relationships


  • Eshkol Rafaeli,

    1. 1 Department of Psychology, Barnard College, Columbia University, 3009 Broadway, New York, NY 10027 ( *Karmanos Cancer Center, Communication and Behavioral Oncology, Wayne State University, 4100 John R Street, ROC—Room #336, Detroit, MI 48201.
    Search for more papers by this author
  • and 1 Marci E. J. Gleason

    1. 1 Department of Psychology, Barnard College, Columbia University, 3009 Broadway, New York, NY 10027 ( *Karmanos Cancer Center, Communication and Behavioral Oncology, Wayne State University, 4100 John R Street, ROC—Room #336, Detroit, MI 48201.
    Search for more papers by this author


The literature on social support within dyadic intimate relationships raises a seeming paradox: The availability of support tends to reduce distress, but its actual receipt is often unhelpful and at times engenders feelings of inadequacy, indebtedness, and inequity—unintended but potent side effects of the support transaction. Our review organizes this literature in order to solve the apparent paradox. Specifically, we theorize that, because support attempts are often unskilled and miscarried, they lead to greater costs than benefits. We identify four ways in which dyadic support can be unskillful, ways pertaining to its timing, content, process, or reciprocation. We suggest that when these are addressed, support can regain its intended goals of enhancing dyadic coping, reducing stress, and strengthening relationships.

Intuition and common sense tell us that support given within intimate relationships confers benefits on its recipients. Yet the receipt of support also runs a risk of being harmful; in fact, the harm often outweighs the benefit. Studies in social and clinical psychology, gerontology, human ecology, and health behavior have all independently corroborated this surprising finding. Unfortunately, given the broad interest in social support, the terminology and operational definitions used have grown so numerous that it is difficult to get a coherent picture of this seeming paradox. In this review, we focus on support processes within committed intimate dyadic relationships, attempt to organize the relevant social support literature, present evidence for the frequent ineffectiveness of dyadic support, and then speak to the possible causes for this miscarried or ineffective support—problems in its timing, content, process, and reciprocation. We end by discussing ways of making support the useful dyadic coping tool it is meant to be.

Defining the Scope of the Problem

Intimate partners offer support to each other (and seek support from each other) for various problems. Of course, the intuition that such dyadic support is positive is not entirely baseless: Support often provides benefits. For example, dyadic support has been shown to (positively) predict marital outcomes above and beyond constructs such as conflict style (Sullivan, Pasch, Eldridge, & Bradbury, 1998). Conger, Rueter, and Elder (1999) found couples who express greater levels of marital support report less emotional distress. Moreover, the absence of support has been shown to mediate work-family conflict on the one hand and individual distress on the other (Matthews, Conger, & Wickrama, 1996). These and other studies demonstrate the importance of sustained levels and quality of support during stressful times (cf. Revenson, Kayser, & Bodenmann, 2005; Story & Bradbury, 2004). In addition, recent research has shown that supportive enthusiastic responses to positive events (also referred to as capitalization) provide many benefits, including increased personal well-being and relationship satisfaction (Gable, Gonzaga, & Strachman, 2006; Gable, Reis, Impett, & Asher, 2004).

Yet growing evidence suggests that some dyadic support may be unskillful—and may therefore backfire (e.g., Bolger, Zuckerman, & Kessler, 2000; Fisher, Nadler, & Whitcher-Alagna, 1982; Krause, 1997). In the short run, such misguided support can worsen the negative emotions it is trying to ameliorate. In the long run, it could lead to a growing sense of dissatisfaction with the relationship (e.g., Bodenmann & Cina, 2000), to the perception that the partner is not responsive (Reis, Clark, & Holmes, 2004), and to reduction in trust (Cutrona, Russell, & Gardner, 2005). Our aim is to show that these limited benefits and potential costs have been found throughout the support literature and can be reversed with knowledge gained from relationship research about dyadic-level coping processes, specifically research about the provision and receipt of support.

Intimate dyadic relationships are the primary source of support and capitalization for most individuals, and support is very frequent within such relationships (Revenson, 1994). Bodenmann (2000) suggested that this ubiquity is best understood in terms of a stress-coping cascade—after making individual coping attempts, people turn first to those closest at hand before broadening their search for help; the closest ones are often intimate partners. Partners respond to such direct requests for help or to the apparent need for it, extend their help, and often are successful in alleviating the stress. In our view, such dyadic-level coping involves mobilization of both members of a dyad in coping with stressors or challenges that affect one or both of the partners.

Dyadic coping can involve various processes: reliance on external networks (e.g., encouragement from friends and family; Sprecher & Felmlee, 1992), personality or temperamental qualities of each partner (e.g., hardiness or resilience; cf. Bonanno, 2005), or simply humor and benign neglect (i.e., knowing when to disregard a problem or to make light of it). Additionally, contextual factors (e.g., job loss and financial strain; Vinokur, Price, & Caplan, 1996) and relational factors (e.g., relationship satisfaction; Lorenz, Hraba, & Pechacova, 2001) can play myriad important parts in determining whether dyads cope effectively. We focus our review, however, on the receipt and the provision of skillful support and argue that these are likely to be among the key processes in successful dyadic coping.

We define dyadic coping in a similar way to Bodenmann (1995, 1997), who equated it with the variety of methods partners use to cope together with a stressor. It is therefore similar, though not identical, to what O'Brien and DeLongis (1997) have termed relationship-focused coping—the efforts individuals make to maintain close relationships or to attend to the emotional needs of their partners during periods of stress. (Note that O'Brien and DeLongis, 2007, themselves use the term dyadic coping to denote the interaction of individuals’ coping patterns—for example, pursuing vs. distancing styles, avoidance vs. approach patterns, which is an entirely different matter.) Our definition of dyadic coping shares some ideas with the notion of relationship-focused coping (O'Brien & DeLongis, 2007), though in our view, dyadic coping involves acts that may be done in the service of individual or couple-level needs. As Bodenmann (1997) noted, these include positive coping acts (supportive, joint, or delegated) and negative coping acts (ambivalent, hostile, or superficial). To these, we would like to add an intermediate category—well-meant but unskilled acts, intended to be positive but failing to be.

We begin with a brief review of the role of stress within intimate dyadic relationships. It is stress, after all, that serves as the trigger and context for support's effectiveness or ineffectiveness. We then survey and organize the existing research on the effectiveness of enacted social support within dyadic relationships, highlighting problems having to do with the timing, the content, the process, and the reciprocation of support. We conclude by discussing how the concept of skillful support fits with recent advances in both the theory and practice of relationship science.

Stress and Support in the Context of Close Relationships

The lives of most individuals are dotted with periods of stress or challenge. Some periods are characterized by acute stressful events (e.g., purchasing a home, completing major time-sensitive projects at work). Other periods are characterized by more prolonged and at times chronic stressors (e.g., financial dificulties, the transition to parenthood, or the burden caused by the declining health of one's own parents). More mundane stressors are also common components of daily life (Almeida, Wethington, & Kessler, 2002). Such stressful times and challenging events have been shown to have a powerful and negative impact on individuals and dyads (e.g., Repetti, 1993; Roberts & Levenson, 2001; cf. Karney, Story, & Bradbury, 2005). These negative effects sometimes create a fugue-like pattern by leading to more stressors, which in turn lead to more distress (cf. Hammen's, 1991, stress-generation model). Negative cycles of this sort can involve poor or unsatisfying dyadic coping processes (e.g., Atkinson, Liem, & Liem, 1986; Bolger, Foster, Vinokur, & Ng, 1996). As we detail below, we believe that absent or faulty social support skills play a large part in this cycle (cf. Bodenmann, 2005; Davila, Bradbury, Cohan, & Tochluk, 1997).

The availability of social support (much of it from intimate partners and confidants; Revenson, 1994) has long been associated with reduced distress during times of stress. Recent research suggests that the knowledge that someone is available to be supportive is indeed beneficial but that actual receipt of support may not always be (cf. Barrera, 1986; Lindorff, 2000). In fact, actual support receipt is often associated with worse rather than better psychological outcomes (e.g., Bolger et al., 2000; Fisher et al., 1982). Even when support is beneficial (in increasing positive feelings within the relationship), its effects seem to be dwarfed by those of hindrance (which has stronger and wider effects; Rafaeli, Cranford, Green, Shrout, & Bolger, 2008).

Krause (1997) reported a striking demonstration of this social support paradox. In a prospective nationwide survey of 60-year-olds in Great Britain, perceived availability of support (anticipated support) was predictive of decreased mortality risk, whereas actual support transactions (i.e., enacted support) were predictive of increased mortality risk. A possible explanation of the association between support and undesirable outcomes is that people who suffer physically or psychologically are more likely to need and elicit actual support from their close ones. The adverse association between support and outcomes, however, seems to stand even when relevant variables (such as health status [Krause, 1997] or prior distress [Bolger et al., 2000]) are adjusted for. This suggests that enacted support can indeed have a substantive deleterious effect and not simply a spurious one.

Nonetheless, committed dyadic relationships remain the source from which individuals draw the most strength and support in coping with external hardship (Revenson, 1994). Given this incorrigible tendency to seek and receive support, it seems worthwhile to turn to relationship research and to try and harness the knowledge about the costs and benefits of well-meant (if miscarried) support. This knowledge can point the way to promoting the benefits and ameliorating the adverse effects of support transactions.

The Costs of Support

Intuitively, we expect social support to be beneficial (e.g., Thoits, 1995) and to play an integral role in the maintenance of healthy close relationships (Cutrona, 1996). We expect this to be particularly true in times of stress (e.g., Cohen & Wills, 1985). As stated above, and contrary to this intuition and to the evidence regarding the availability of support, studies examining actual support transactions have revealed paradoxical findings: not only of a lack of a positive effect of support (e.g., Barrera, 1986; Bolger et al., 1996; Wethington & Kessler, 1986) but also at times of its deleterious effect on the recipient's well-being (e.g., Barrera, 1981; Barrera, Sandler, & Ramsay, 1981; Bolger et al., 2000; Krause, 1997). For example, Bolger and his colleagues (2000) used daily diaries to examine a sample of Bar examinees and their partners in the period leading up to the exam. These authors found that during a period of high stress the report of receipt of emotional support from one's partner was associated with increased feelings of depression and anxiety, even when prior distress was controlled. Moreover, even during periods of moderate stress, the reported perception of support receipt had no effect on mood. At no time was there an unqualified positive effect of reported perception of support.

There is other evidence that support and involvement efforts often misfire, even when they are guided by professionals. Such a lesson emerges from the literature on spouse-assisted therapy, an approach in which the psychological or physical symptoms of an individual are treated with the aid of the individual's partner. The results of spouse-assisted therapy programs have been mixed. For example, a review of such therapy for smoking cessation reported that interventions involving the spouse often have no positive effect and at times have an iatrogenic (i.e., harmful) effect on the target behavior (Cohen et al., 1988). Similarly, spouse-assisted programs for high-risk cardiac patients have often been most effective precisely when the “assisting” spouse is encouraged to disengage and show no interest in the symptomatic partner (e.g., Hoebel, 1976). In other words, simply using the partner as an adjunct therapist without attending to the dynamics of social support (cf. Coyne, Wortman, & Lehman, 1988) might lead to more harm than good.

Particular types of intended support seem most likely to miss the spot. Lehman, Ellard, and Wortman (1986) found that advice giving, minimization of feeling, identification with feelings, and encouragement of recovery are most often seen as unhelpful. Providers’ attempts at “cheering up” the recipient are often unhelpful as well (cf. Barbee, Derlega, Sherburne, & Grimshaw, 1998). Bass, Tausig, and Noelker (1988–1989) found that instrumental support given to caregivers of elders exacerbated the harmful effect of the elders’ functional impairment (which served as an objective index of the stress the caregivers were experiencing) on the caregivers’ strain. Kaniasty and Norris (1993) found that varying levels of support from kin did not affect the degree of psychological distress experienced following a disaster. Allen, Blascovich, Tomaka, and Kelsey (1991) showed that the mere presence of a supportive other during a stressful task can lead to heightened physiological responses (e.g., increased blood pressure) and poorer performance, perhaps due to a fear of evaluation by the supportive other. Interestingly, in the presence of one's pet—a nonevaluative other—physiological response lessened and performance improved, suggesting that non-evaluative companionship can be beneficial.

To rule out the possibility that the relationship between enacted support and adverse outcomes is spurious, experimental research on enacted support was needed. A recent experimental demonstration provides this needed evidence for the costliness of visible support. Bolger and Amarel (2007) randomly assigned participants to conditions of support or no support prior to an anticipated public speech. Consistent with predictions, students who knew that they had been supported prior to the speech were more upset and anxious than were students who received no support.

Several explanations have been offered for the negative effects of support; these should not be seen as competing but rather as potentially additive factors. First, the receipt of social support might undermine a recipient's sense of efficacy, self-esteem, or autonomy. This would occur when support signals to recipients that they are incapable of coping independently with a stressful situation and that they are dependent on the provider for help (Bolger et al., 2000; Fisher et al., 1982). Second, support may paradoxically focus the recipient's attention on the stressor it is aimed to alleviate. Because of this attention, cognitive appraisal costs might accrue and could engender increased distress (Lazarus, 1991). Third, receiving support might make the recipient feel (or indeed, be) indebted to his or her partner (Walster, Berscheid, & Walster, 1973), thereby increasing stress and anxiety for the recipient as well as tension and dissatisfaction for the dyad. Fourth, it is possible that well-meant support is miscarried (Coyne et al., 1988). This suggestion rests on evidence from the family therapy literature, demonstrating that overinvolvement of relatives or other caregivers often leads to a worsening of the crisis. For example, Hooley, Orley, and Teasdale (1986) have shown that, for depressed patients, the level of emotional overinvolvement (as a key component of what is termed expressed emotion) is a strong predictor of relapse after discharge from the hospital. Fifth, support and involvement may be aversive when accompanied by criticism and interference.

Thus, supportive actions are at times ineffective or even harmful, and the above factors might be responsible for this ineffectiveness. Given this, it might be hard to imagine why a focus on support skills could help couples. However, support attempts are a natural, almost automatic response to the needs of another person, particularly those of an intimate partner, and appear to persist regardless of their effectiveness. Fortunately, the literature that identifies and explores these negative support effects also points to ways in which support can be made useful and effective. In the following text, we describe four of these ways.

The “When” of Social Support: Temporal Dynamics of Support Transactions

Almost all support transactions unfold over time and do not transpire instantaneously. This is particularly true when these transactions occur within committed relationships, rather than in more circumscribed settings (e.g., professional relationships). A better understanding by both members of a dyad of the “when” of support transactions—their timing and the way in which they unfold over time—is a necessary first step in improving their quality.

Support transactions commonly progress through several stages (cf. Bodenmann, 1995). These include early stages of identifying the stressor as well as the needs it elicits; intermediate stages of appraisal, communication, or both; and a final stage of action. Coyne et al. (1988) identified the initial stage, construction of the situation, and separated it from latter stages, including the actual enactment of support. Similarly, Pearlin and McCall (1990) conducted qualitative analyses of supportive interactions, revealing the presence of three stages: the receiver's revelation of the problem, the putative provider's appraisal of the situation and of his or her own resources, and the actual support transaction (or lack of it).

As Cutrona (1996) suggests, each of these stages can go awry. Recognizing the multiple stages, understanding each of them better, and realizing that effective coping requires a joint traversing of all stages are essential for support to be helpful. Separating support-related interactions into their stages may reveal, for example, that a gap exists in the communication stage rather than in the action stage. Resentment may be averted if the partners recognize such a gap in communication and develop a mutually acceptable vocabulary for conveying the need for help, the particular form of help sought by one partner, and the other partner's willingness to provide. Yet skillful support goes beyond the important matter of improved communication. It requires specific communication skills having to do with awareness of weaknesses or vulnerability, comfort with disclosing and hearing such vulnerability, and the development of a vocabulary for recruiting the right responses and resources.

We do not mean to suggest that the responsibility for traversing these stages lies solely with the support provider. Clearly, support-seeking behaviors matter (e.g., Collins & Feeney, 2000), as do the responses of the support recipient to the provider's actions. For example, Lane and Hobfoll (1992) point out that some behaviors on the part of the recipient—for example, anger—may alienate the support provider and are likely to reduce support.

There is a linear progression of stages in any stress-support instance but each instance is but one in a progression of many. Rather than flowing in a unidirectional way from beginning to end, the steps are more likely daisy-chained and linked by many feedback loops. For example, the successful resolution of one stressful situation should reduce the presence of the instigating problem. It may also increase the subsequent likelihood of seeking support (and alter the beliefs and expectations of such support being available).

Several pitfalls may be encountered when support providers fail to recognize the sequencing of stages. Insufficient attention to the early stages of disclosure, communication, and appraisal can lead to misguided support—the wrong help at the wrong time. For example, in couples like those studied by Roberts and Levenson (2001), a wife quickly responding to her police officer husband's first tired utterance as he returns, exhausted, from work, may miss the mark. While he may need space or time to relax, she may delve into problem solving or encouragement, setting off a cycle of well-meant, but misguided, support. In the same example, devoting a slightly lengthier time to appraising or assessing the situation may prevent such a cycle from occurring, as the actual needs in the situation, as well as the more fitting response to it, may become clearer.

Neff and Karney (2005) recently reported another pitfall, more common among male support providers. In their study, male and female partners did not differ in the proportion of positive and negative behaviors enacted in a laboratory support-provision situation, but did differ in the responsiveness of their support to stress. Specifically, male providers were less adept at providing the support specifically at those times in which their partners most needed it. Note that while Neff and Karney distinguish between skill and responsiveness, our view is that sensitivity to the timing of support is a part of the skillfulness of support and possibly a prerequisite to other, more nuanced parts, discussed later in this article.

In addition to the linear order of stages (and the feedback that occurs between them), supportive relationships also involve cyclicity. For example, support from one partner is likely to increase the likelihood of reciprocation at a later point (Iida, Seidman, Shrout, Fujita, & Bolger, 2008). Similarly, negative affect or disappointment about the failure of a partner's supportive act is likely to lead to resentment and reduce goodwill toward providing support in return (e.g., Coyne et al., 1988; Lehman et al., 1986). Thus, supportive acts in the past or present are likely to affect future supportive acts—each support transaction does not exist in a vacuum.

In sum, the effectiveness of social support is partly dependent on its timing (i.e., when the support is provided, how the provider and the recipient traverse the stages of support, and how that supportive interaction fits, or does not fit, within cycles of reciprocation). A major pitfall related to timing may be avoided if both partners allow sufficient time or use direct communication to ensure that the provider is appraising correctly both the needs of the stressed partner and the support that will address that need.

The “What” of Social Support: Support Multidimensionality

Social support is the interpersonal analogue of individual coping (Thoits, 1986). Like coping, support is multifaceted, not unitary. Even after the “when” of support has been determined (as discussed above), individuals in relationships need to determine its “what,” as support takes varied forms in response to a range of needs.

Several attempts have been made to categorize types of support. Weiss (1974) identified several kinds of support “provisions”: advice or guidance, reliable tangible assistance, caring, social integration (i.e., companionship), and reassurance of worth (i.e., esteem support). This classiffication scheme has been used successfully by Cutrona and her colleagues (e.g., Cutrona, Cole, Colangelo, Assouline, & Russell, 1994). Barrera (e.g., Barrera & Ainlay, 1983) proposed a similar set of five dimensions and found four of them emerging in factor analyses. This set encompasses tangible support (both material aid and behavioral acts), directive guidance (i.e., advice or information), nondirective support (emotional support), and positive social interaction (i.e., companionship or network support). Several researchers (e.g., Carver, Scheier, & Weintraub, 1989) have suggested collapsing the various sets of support dimensions into two overarching categories: emotional support and instrumental support. The intuitive appeal of these dimensions is their compatibility with the two dimensions of coping (emotion- and problem-focused coping) suggested by Folkman and Lazarus (1980) in the domain of personal resources for coping. A review of dimensional models of support can be found in Cutrona and Russell (1990).

Rather than identifying one model as superior to another, we simply note the importance of a multidimensional framework for a practical (and not merely theoretical) understanding of the varieties of supportive experiences. Abundant evidence sustains this suggestion. Many individuals are likely to be thinking of support in narrow terms, using only one or two of its types. For example, as recipients of support, some individuals’ repertoire may include only reassurance seeking (a pattern particularly common and maladaptive among depressive individuals; Coyne, 1976). Along the same lines, Caldwell and Reinhart (1988) found that anxious participants are more likely to seek guidance (informational support) and less likely to seek emotional support. As providers, too, some individuals may gravitate toward guidance and advice giving. Yet as Cutrona and Suhr (1992) note, advice is often seen as noxious, particularly when given in low-control situations.

As we noted earlier, a vast literature has found that perceived availability of support (in contrast to actual supportive acts) is linked to numerous positive outcomes for the perceiver (e.g., Katz, Monnier, Libet, Shaw, & Beach, 2000; Lindorff, 2000; Monahan & Hooker, 1995; Sarason, Sarason, & Pierce, 1994). It is possible that this perception of support availability is driven, in part, by the amount of one particular type of enacted support: namely, companionship or positive social interaction. Companionship often entails simply being together with the recipient; thus, of the various types of enacted support, it can be the least active and the closest to a minimal reminder of the availability of a responsive other. It is important to note, however, that perceived availability may be unrelated to any enactments of support and instead may be more of an individual difference bias in perceiving one's social network (Kitamura et al., 2002; Lakey et al., 2002).

Carels and Baucom (1999) reported the results of an interactive support exercise in which married partners were asked to give online ratings of the supportiveness of each interaction. In this study, as in Cutrona and Suhr's (1992), spouses tended to give suggestions, advice, or help to reassess the situation, but these statements were not associated with the recipients’ experience of feeling supported. Carels and Baucom also reported that, in their sample, esteem support (statements from partners that communicated general respect and confidence in the recipients’ abilities) was the only type of support that influenced the recipients’ experience of feeling supported.

Interestingly, Carels and Baucom (1999) found that women were more attuned to the content of the supportive interaction than were men. Thus, although advice may often be useless for recipients of both genders, it may still make men feel supported because of the general tone of support or because it is embedded in a satisfying relationship. In contrast, women recipients, who are more sensitive to the type of support provided, will show more differential effects for different support types. One way or the other, informing partners regarding the range of support from simple companionship to advice could broaden the repertoire of possible supportive acts and make it easier for partners to choose effective types of support for stressful events.

The utility and skillfulness of support depends, in part, on its optimal matching with the particular coping needs of the recipient. Models of social support matching (Cohen & McKay, 1984; Cutrona & Russell, 1990) came on the heels of the finding that global indices of support often fail to show a general buffering effect on stress. In an effort to explain that surprising finding, models of matching support types with stressors or needs have suggested that support would be shown to be effective when particular forms of support interact with characteristics of the stressful situation and buffer the effects of specific stressors to which they are matched. This idea has been dubbed the specificity hypothesis.

The specificity hypothesis assumes multidimensionality of both stress and social support. A test of this hypothesis requires classifying enacted support into several kinds (as discussed above) or classifying stressful situations into several kinds, or both. Specificity can then be examined by comparing matched to mismatched support-stress pairings. Cutrona and Russell (1990) presented a dimensional model of stress that specifically addressed the matching of social support and stress. This model encompasses four dimensions of stress: its desirability, its controllability, the duration of its consequences, and the life domain in which it is occurring. According to this model, controllability is the key dimension on which stressors and support acts could be matched: Uncontrollable events require emotional support, whereas controllable events require instrumental support. One strength of this model is its agreement with Folkman and Lazarus's (e.g., 1980) notion of problem-focused coping and emotional coping, which are appropriate responses to controllable or uncontrollable events, respectively. Most matching analyses, however, have broken down support into more specific types of problem solving or emotion-focused coping.

The findings of studies examining the matching hypothesis are encouraging (e.g., Horowitz et al., 2001; Krause, 1986; Peirce, Frone, Russell & Cooper, 1996). For example, Pierce and colleagues studied how three types of social support (tangible, appraisal, and belonging support) affect the association between financial strain and alcohol involvement. Whereas tangible support buffered this association, appraisal and belonging support did not buffer it. Similarly, Krause (1986) found that, although global social support failed to modify the impact of global stressful events on an elderly population, specific types of social support buffered the impact of specific types of stressors. For example, the effect of bereavement was buffered by informational, tangible, and integration support, the effect of crime victimization was buffered by emotional support, and the effect of network crises was buffered by integration support. Not all studies agree, however: Tetzloff and Barrera (1987) examined the effect of specific types of support (parenting, emotional, and tangible), as well as of stressors of the same three types, on the well-being of divorced mothers and failed to find support for the specificity hypothesis.

In actuality, there are at least two definitions for matching: (1) when support fits well with the actual, objective needs that arise in the situation or (2) when support fits well with the needs of the support recipient as perceived by the support recipient. The encouraging findings reported above favor the first definition. But because the perceived needs (more so than the actual ones) serve as the basis for explicit requests for support, the second definition has its merits. Indeed, the last decade has brought a growing amount of evidence highlighting the utility of this definition. Dehle, Larsen, and Landers (2001) reported that a greater fit between desired and provided support on each of the dimensions identified by Cutrona and Suhr (1992, 1994) is associated with better marital adjustment, positive global marital sentiment, and less negative marital quality. Affleck, Tennen, Rowe, Roscher, and Walker (1989) reported that a formal supportive intervention for mothers of high-risk infants proved beneficial only for those who demonstrated a need for this support but was actually harmful for those who were at low levels of need for support.

In sum, support is multifaceted and can involve various types of emotional or practical assistance. Some of these types (e.g., caring, tangible assistance) are likely to be more helpful than others (e.g., advice), but the greatest benefit is likely to occur when there is optimal matching between the type of need and the type of support. Accurate communication and appraisal of the stressor by both partners can aid in achieving such matching. Communication and appraisal, in turn, require the patience to traverse the stages discussed in Point 1, above, along with a familiarity with the multifaceted forms of support discussed here.

The “How” of Social Support: Visibility, Directness, and Indirectness

Recognizing the temporal stages (the “when”) and the appropriate types (the “what”) of support are two large parts of providing skillful support. We believe that such recognition has the potential to increase the benefits of support. But there are additional and more subtle aspects to skillful support, which we refer to as the “how” and “who” of support transactions; we believe that these aspects have a greater effect on reducing the costs (rather than increasing the benefits) of support.

Major costs of support, namely, feelings of inadequacy, indebtedness, and inequity as well as increased and unwanted attention to the stressor, necessarily require the recipient of support to recognize that he or she is being supported. But supportive actions need not always call attention to the beneficiary's role as a recipient. Skillful support providers may temper “how” they deliver their help and thereby avoid many of these costs.

In particular, research conducted by Fisher and colleagues (Fisher, 1997; Fisher, La Greca, Greco, Arfken, & Schneiderman, 1997; Harber, Schneider, Everard, & Fisher, 2005) demonstrates the importance of recognizing the receiver's need not only for help but also for autonomy. Fisher and his colleagues distinguish between two kinds of support processes: directive and nondirective. Directive support occurs when the provider imposes a specific type of coping on the recipient, and nondirective support occurs when the provider allows the support provision to be dictated by the recipient. Both directive and nondirective support transactions can comprise any type of support, ranging from companionship to guidance and advice. What distinguishes them is the manner in which the support is carried out, particularly the degree to which it accounts for the autonomy desires of the recipient. For example, a provider who takes care of errands at the recipient partner's request is engaging in nondirective support; in contrast, a provider who does so without the request or explicitly against the wishes of the recipient is engaging in directive support.

Directive support runs the risk of demoralizing recipients, perhaps because it co-opts the coping process. Nondirective support tends to be more effective, perhaps because it encourages and validates the recipient's view of the situation. Fisher and colleagues (1997) find that both nondirective and directive support can be beneficial depending on the level of autonomy desired by the recipient, but most of their research suggests nondirective support is more beneficial. For instance, in a study of diabetic patients, nondirective support was associated with better metabolic control, whereas the reverse was true for directive support (Fisher et al., 1997). Interestingly, in more severe patient groups (e.g., advanced lupus) directive support was associated with less depression. This line of research demonstrates that unsolicited support may be beneficial at times; in most cases, however, the recipient of the support will be more satisfied with support that he or she has solicited.

The work of Fisher and his colleagues on directive support, as well as related work by other authors (e.g., Coyne et al., 1988; Coyne, Ellard, & Smith, 1990; O'Brien & DeLongis, 1997), helps untangle the paradox of well-meant yet ineffective and even harmful supportive acts. It suggests that teaching couples to recognize when to engage in directive and nondirective support should increase their ability to provide (match) the support desired by the recipient, not only in terms of the type (or content) but also in terms of the manner (or process).

In some instances, recipients are not able to articulate their needs while providers are aware of both these needs and of appropriate (i.e., timely and matching) acts that could assuage them. For example, Fred may recall a forgotten but impending deadline in Ginger's work or Ginger may know that Fred is feeling incapable of navigating an obligatory and dreaded family event. The frequency of such moments is unknown and probably differs from time to time and from relationship to relationship. Importantly, such moments carry the risk of becoming directive support incidents. A subtle way of avoiding such incidents is by reducing the visibility of support. After all, the harmful effects of support receipt seem to be due primarily to the support being explicit and therefore visible. Several studies suggest that maintaining the helpful core of support while stripping it of its public or manifest quality can help it regain its intended effect.

This idea was demonstrated most strongly in a diary study of couples in which one member was approaching the Bar Examination (Bolger et al., 2000). Independent reports of support provision and receipt by both partners allowed these researchers to capitalize on the possible discrepancies between the actual receipt (as indexed by the providers’ reports) and perceived receipt (as indexed by the recipients’ report). The findings suggested that skillful, invisible support reduced depression in Bar examinees: Recipients who were unaware that they were supported but whose partners reported providing support benefited from the support, especially at times of high stress. Bolger and Amarel (2007) replicated invisible support effects in an experimental study of undergraduates preparing for a stressful event. Those who reported receiving support also reported an increase in anxiety; in contrast, those who were given invisible support (i.e., support was given but was not coded as such by the recipient) reported a decrease in anxiety.

The idea behind Bolger et al.'s (2000) and Bolger and Amarel's (2007) findings are that the deleterious effects of support stem not so much from the support itself as from its visibility. The benefits of invisible support may accrue, unimpeded, for several reasons. First, lacking awareness of the support may shield the recipients from loss of efficacy and self-esteem. Second, invisible support may avoid drawing recipients’ attention to the problem. Third, it may prevent an increased sense of indebtedness and inequity. There is strong evidence consistent with these predictions. For example, a large literature (reviewed in Butzlaff & Hooley, 1998) has explored the adverse effect of expressed emotion, a family communication pattern that includes overinvolvement, criticism, and hostility among caregivers of an ill person. Though criticism and hostility may be the most pernicious aspects of expressed emotion, the construct as a whole (including both criticism and overinvolvement) is related to directive and visible guidance—direct attempts to buffer the patients (in overinvolved families) or to intervene in the patients’ problems and to control their actions (in critical families). Additionally, high emotional overinvolvement often co-occurs with high criticism and hostility (Barrowclough & Hooley, 2003). Butzlaff and Hooley concluded that family members who become involved in such ways with their loved ones contribute to a marked increase in the risk of relapse in a wide range of psychiatric conditions, including mood disorders and other disorders in which emotion regulation is of primary importance.

Similar results have been reported elsewhere. For example, in a study of cancer patients and their spouses, Hagedoorn et al. (2000) found that at times of greater distress or physical limitation, “overprotective” spousal support (which, by definition, is visible) was negatively associated with satisfaction. The same was true for protective buffering (e.g., the provider trying, often unsuccessfully, to hide his or her feelings from the recipient). Both of these styles are unskillful, for somewhat different reasons. Specifically, whereas overprotection is clearly visible (and can involve an almost coercive overinvolvement), protective buffering can often seem conspicuous as well. Protectively buffering partners silence their own worries, concerns, or problems, and this silence can be as loud as the clamor of overprotection. In addition, protective buffering deprives the support recipient of the chance to reciprocate in any way (an issue we return to in the next section).

In contrast, a more skillful style of support termed active engagement proves to be more efficacious, particularly at times of greater psychological and physical distress (e.g., Hagedoorn et al., 2000). This style involves engaging the recipient in discussion and adopting a constructive, problem-solving approach (Coyne et al., 1990). In the example given earlier, Fred may ask Ginger whether she would like to talk through the steps necessary to meet the impending deadline. Kuijer and colleagues (2000) have shown that the recipients of this style of support experience less distress and more efficacy.

Support providers may find it difficult to accept that their well-meant efforts can undermine rather than aid a needy partner, yet this knowledge may give them the opportunity to understand these problems and to circumvent them. For example, support providers might benefit from recalling times in which they felt support to be overly visible or intrusive. They might gain from learning to anticipate their partners’ times of great stress and from responding to them in a more streamlined manner. They might also benefit from discussing or rehearsing this form of “invisible support,” a discussion that could allow them to intercept resentment that might arise in the process of providing invisible support.

Work by Collins and Feeney (2000) suggests that attuned, skillful providers (i.e., ones who are able to set aside their own needs and anxiety) deliver support that is more effective. Similarly, as Katz and Joiner (2002) recently demonstrated, support that is based on a greater understanding of the partner's weaknesses and strengths (i.e., of their particular needs) is likely to be perceived as more satisfying and positive. This form of attunement or empathic ability (cf. Ickes, 1997) has been identified as a major component of dyadic, relational-focused coping (O'Brien & DeLongis, 1997). Empathic accuracy of this sort is likely to increase the optimal matching of support to need (Cutrona, 1990) and to also contribute to a more streamlined, less directive, and more invisible form of support.

In sum, even well-meant support can be carried out in a manner that reduces, and even reverses, its intended effects. Potential ways to increase the skillfulness of support would involve increasing partners’ awareness to the problems of visibility and directiveness.

The “Who” of Social Support: Equity and Efficacy

We previously reviewed an impressive array of studies questioning the benefit of receiving support. We laid most of the burden of making the receipt of support positive (for the recipient) on the provider's shoulders. This may seem unfair to anyone familiar with the research on caregiving burden, which suggests that certain support-provision roles carry serious risks for the provider. For example, caregivers often suffer from cognitive and physical decline when burdened too intensely or for too long with the role of support provider (George & Gwyther, 1986; Vitaliano et al., 2005). Caregiver burden tends to be greatest the closer the caregiver is to the recipient, putting spouses and partners at particular risk for becoming overburdened (Cantor, 1983). Perhaps most sobering for caregivers, recent research suggests that mortality increases for caregivers when their partners are hospitalized, particularly when the partners’ illness is a long-lasting, highly debilitating yet nonterminal condition such as Alzheimer's disease (Christakis & Allison, 2006). It may seem that on top of this burden, we now saddle caregivers with the added responsibility for miscarried help.

Research on caregiver burden tends to focus exclusively, however, on individuals who are caring for chronically or severely ill patients, a support process that is unique in its demands. In contrast, a broader literature examining the effects of support provision acts has frequently shown them to be beneficial for the provider. For example, in a study of more than 500 older adults (Krause & Shaw, 2000), participants reporting that they provided support to others demonstrated higher self-esteem both immediately and at delayed assessments. Additionally, Williamson and Clark (1989) provided experimental evidence that individuals who anticipate helping someone with whom they wish to form a communal relationship experience increases in self-esteem and positive mood. In fact, recent work has even suggested that those who provide support live longer than those who do not do so, even when adjusting for relevant health and lifestyle variables (Brown, Nesse, Vinokur, & Smith, 2003).

These findings are consistent with the theoretical model of Weiss (1974) presented earlier—a model identifying types of support that are beneficial to the recipient. In addition to the five types that were listed above, Weiss identified a sixth “support provision” or type: the creation of opportunities for the recipient to provide support in return. As such, Weiss's theoretical model, which has since gained considerable confirmation (e.g., Cutrona et al., 1994), was one of the first to note the importance of reciprocation for the individuals involved. Giving support allows a person to demonstrate competence and efficacy, to equalize a relationship characterized by imbalance in neediness, and to draw attention away from one's own problems. These benefits occur whether or not the provision occurs within an intimate relationship (e.g., Clary & Snyder, 1999). Within relationships, however, giving support often leads to a state of supportive equity. This state might be particularly positive, not only for personal outcomes (e.g., positive mood) but also for relationship outcomes such as intimacy and satisfaction (Antonucci & Jackson, 1990; Gleason, Iida, Shrout, & Bolger, 2008; Walster et al., 1973). Importantly, Gleason and colleagues did not find any differences between men and women in the importance of supportive equity on intimacy and mood (Gleason, et al., 2008; Gleason, Iida, Bolger, & Shrout, 2003), which is in line with support research suggesting that men and women react similarly to support (Neff & Karney, 2005; Porter et al., 2000) and that equity is equally important to men and women in their judgments of relationship satisfaction and commitment (Sabatelli & Cecil-Pigo, 1985).

Consistent with equity theory, research has demonstrated that giving support might in fact buffer the deleterious effects of receiving it (Gleason et al., 2003). These results suggest that supportive equity, the state of both receiving and giving support, is beneficial for both individual mood and dyadic outcomes. Indeed, the findings suggest that the negative effects of receiving support on individual mood appeared only when the recipient failed to reciprocate the support. In further studies of couples in which one member is facing a stressor, supportive equity not only eliminated the negative effect of support receipt on mood but increased positive mood and relationship intimacy (Gleason et al., 2008). Even individuals who suffer from a chronic illness appear to benefit from providing support and from supportive equity: In a study of couples in which one partner suffered from multiple sclerosis, supportive equity days were associated with increased self-esteem, and provision of support was associated with increased well-being (Kleiboer, Kuijer, Hox, Schreurs, & Bensing, 2006).

The discussion of reciprocated support (along with the earlier discussion of indirect and invisible support) resonates with points raised by O'Brien and DeLongis (1996) regarding the personality context within which stress, coping, and support occur. Specifically, these authors highlighted the dual motives of agency and communion (individuals’ desire to be both self-sufficient and connected to their partners). Most individuals attempt to balance these two motives in their lives. Either of these motives can run amok and lead to ineffective support and poor well-being (Helgeson, 1994). For instance, work by Mashek and Sherman (2004) has demonstrated that both low levels of closeness (too much agency) and high levels of closeness (too much communion) are associated with lower levels of relationship quality. In skillful support transactions within committed relationships, however, both agency and communion can be fostered—and this is particularly true if the processes of reciprocation and equity, and of invisibility and indirectness, are attended to.

As with invisible support, reciprocated support might be a difficult concept for couples to accept. After all, it seems contrary to the logic of buffering the stressed person as much as possible from any additional responsibilities or stressors. Nonetheless, this research suggests that couples can benefit from creating ways in which both partners continue to feel involved, supportive, and present, even amidst major stressors. In fact, even extremely burdened caregivers (those who are caring for chronically ill partners) benefit from reciprocity: Reid, Moss, and Hyman (2005) found that caregivers who experienced higher levels of reciprocity from a care receiver feel less burdened than those that do not experience reciprocity.

Couples who find ways to “check in” (e.g., by having routine discussions about both partners’ agendas) are likely to maintain a sense of reciprocity. Results suggest that this increased reciprocity brings with it more manageable levels of distress, maybe because of a greater sense of perspective about the major stressor at hand (Fergus, Gray, Fitch, Labreque, & Phillips, 2002; Gleason et al., 2008; Stephens & Clark, 1997). In other words, learning to maintain reciprocated support at times of stress is likely to result in better personal and relationship outcomes.

In sum, whereas the receipt of support can have negative effects, receipt coupled with provision tends to be beneficial for both partners in a relationship. This finding, consistent with equity theory (Walster et al., 1973), suggests that the skillfulness of support can be improved by enhancing the reciprocal, equitable nature of a couple's dyadic coping behaviors.


By attending to the when, what, how, and who of support—its timing, content, process, and reciprocation—couples can increase the benefits and reduce the costs inherent even in the most well-meaning support attempts. By doing so, couples will use more positive dyadic coping, a powerful predictor of marital satisfaction and functioning (cf. Bodenmann & Cina, 2000; Sullivan et al., 1998).

Conclusions and Limitations

This review involved several premises. We began with a reminder that stressors affect both personal and dyadic well-being. We continued by reviewing the social support literature, the reading of which led us to conclude that unskilled support often fails to benefit receivers and is at times even detrimental to them. Finally, and fortunately, we noted how the same literature identifies several ways for making support skillful and for harnessing it into being an effective way to deal with both individual- and couple-related stressors.

Our focus here was on understanding (and promoting) skillful support and effective coping in committed dyads. Some of the ideas reviewed here may apply to support within other relationships (parent-child, friendship, even therapist-client). The dynamics of close relationships, however (Coyne & DeLongis, 1986; Cutrona, 1996; Revenson et al., 2005), make committed dyads unique from others: The commitment, the reciprocity, and the social sanctioning given to these relationships suggest that they warrant this sort of privileged attention.

Additionally, our focus was on dyadic coping—the processes used by couples to cope with external stressors. In such situations, partners may be support recipients or providers and at times embody both roles. They are not, however, the source of the stress. In contrast, when stressors stem from discord within the relationship, other processes (e.g., communication or conflict resolution skills) may be of much greater relevance than the components of skillful support we have outlined. We would explicitly expect skillful support not to be sufficient to ameliorate relational discord.

Yet skillful support is likely to make a difference in a wide range of situations. Adopting an approach presented by Feeney (2004), we would expect skillful support to be important both for alleviating distress in the face of stressors or setbacks (i.e., creating “safe havens”) and for aiding in goal pursuit (i.e., serving as a “secure base”). Feeney, who used an attachment theory framework, posited that support within committed couples that is sensitive and responsive rather than intrusive and controlling allows for the support recipient to explore and to strive toward goals in ways that enhance both relationship and personal well-being. We agree with her analysis and believe that our operationalization of skillful support is likely to apply equally well to both safe haven and secure base situations.

As Story and Bradbury (2004) note, dyadic coping is best seen as an ongoing flow of interpersonal exchanges, occurring within several contexts. These include individual characteristics (e.g., personality traits: Robins, Caspi, & Moffitt, 2002; attachment styles: Collins & Feeney, 2000), couples’ characteristics—e.g., relationship satisfaction: Cutrona, 1996; and broader social circumstances: Conger, Ge, & Lorenz, 1994; Wade, Howell, & Wells, 1994. These important factors were outside the scope of this review but have been reviewed by others (e.g., Karney et al., 2005) and certainly merit further investigation. For example, the effects of these personal, dyadic, and environmental factors on the likelihood of support provision (of any kind) have only recently begun to be explored (Iida et al., 2008). It remains to be seen which factors are specific predictors of providing skillful support. Further, future research on dyadic support would do well to concentrate on observing or capturing (e.g., through daily diaries) ebbs and flows in support transactions as they unfold over time; such research is less dependent on individuals’ global perceptions of their relationships and the support received therein and more sensitive to nuanced fluctuations, even within the same relationship.

The concept of skillful support complements both theoretical and applied current approaches to dyadic processes. As a theoretical construct, it fits well with two recent organizing constructs proposed by Reis et al. (2004) and by Cutrona et al. (2005). Reis and his colleagues suggest that the overarching concept of perceived partner responsiveness to the self is key to understanding close relationships. In particular, they suggest that individuals who feel that their partner responds to their needs will be happier and in healthier relationships, and that this perception is driven, in part, by the dyad's actual supportive history. In a similar vein, Cutrona and her colleagues’relationship enhancement model suggests that the perception of partner support, as well as its immediate consequence, trust, mediate the effects of (actual) consistent supportive responses on relationship satisfaction and stability and ultimately on the individual's physical and mental health.

The literature we reviewed is in agreement with both models. As we have shown, dyadic support skills are central to effective support provision and to dyadic coping with stress. When supportive acts are unskillful, they will not be experienced as responsive (Reis et al., 2004) or as consistently helpful (Cutrona et al., 2005). In contrast, when they are skillful (e.g., when emotional needs are met with emotional support or when support is given in a nondirective, noncritical way: e.g., Cutrona, Shaffer, Wesner, & Gardner, 2007), they will. The ability to give effective, helpful support is often lacking and might be most lacking for the neediest couples, those facing the greatest stressors. As Reis et al. and Cutrona et al. note, the costs of unhelpful or miscarried support extend beyond the unfortunate fact that the immediate stressor is not buffered and are likely to cause downstream effects on perceived responsiveness, on trust, and ultimately on both dyadic and individual well-being.

Relationship researchers, clinicians, and policymakers are interested in strengthening committed relationships, for the sake of the partners, their children, and at times, the institution of marriage itself. Interventions developed for these purposes have shown the greatest promise when they were constructed on a foundation of sound relationship science. Recent, and widely used, examples of such interventions include relationship enhancement or prevention programs teaching couples emotional regulation skills (e.g., Shapiro & Gottman, 2005) or communication and problem-solving skills (e.g., Markman, Floyd, Stanley, & Storaasli, 1988). Both of these popular types of interventions produce encouraging benefits (cf. Christensen & Heavey, 1999). The successful manner in which they have been pursued and the enthusiasm with which they have been disseminated may be due, in part, to the strong evidence that the processes underlying them (emotion regulation, communication) are indeed strong predictors of well-being.

In contrast, our review indicates that dyadic social support has a patchy record of effects. Perhaps this record of ineffective—and at times costly—support discouraged attempts to implement interventions focused on harnessing social support. It may have been reasonable to doubt whether support would be a fruitful and worthwhile target of intervention. But our review, and other recent work, suggests that support can be done skillfully and, when done so, can positively affect both individual and dyadic-level outcomes, including perceived partner responsiveness, trust, relationship satisfaction and commitment, and ultimately individual health (cf. Cutrona, 1996; Cutrona et al., 2005; O'Brien & DeLongis, 1997; Revenson, 1994). Several groups (Kuijer, Buunk, de Jong, Ybema, & Sanderman, 2004; Rogge, Cobb, Johnson, Lawrence, & Bradbury, 2002; Widmer, Cina, Charvoz, Shantinath, & Bodenmann, 2005) have been developing promising interventions that seek to strengthen dyads and dyadic coping by addressing skillful social support (either alongside other topics [Widmer et al., Rogge et al.] or by focusing on a particular aspect of support [equity; Kuijer et al., 2004]). We have also been developing an intervention program, the Skillful Support Intervention, based on this literature (interested readers can find additional information about this program at

We hope the skillful support framework offered here can aid both basic relationship scientists and applied relationship practitioners redouble the effort to help intimate partners achieve greater levels and quality of support.