The Influence of Personal Characteristics and Relationship Properties on Marital Support

Self-report data from 200 married couples were used to examine personal characteristics (i.e., extraversion, neuroticism, attachment-related anxiety and avoidance) and relationship properties (i.e., intimacy, commitment, marital conflict, and marital satisfaction) as determinants of supportive behaviours and perceptions of support availability within marital relationships. The data revealed that both personal and relational variables had important value in the prediction of support perceptions, support seeking, and support provision in marriage. Furthermore, relational predictors were found to explain more variability in spousal support than personal characteristics. Finally, results indicated that personal influences on support perceptions and support behaviours were less pronounced when relationship features were taken into account. In conclusion, the present research advocates in favour of a relationship perspective on social support in marriage.

The way spouses help each other cope with personal difficulties, and how they provide everyday support to one another is an important domain for understanding how marriages succeed and fail (see Bradbury, Fincham, & Beach, 2000). Studies on marital support consistently produced evidence for support behaviours being reliably linked to relationship functioning and changes in relationship functioning (see Bradbury & Karney, 2004). To put it simply, spouses who receive more effective support from their partner, report more positive longitudinal marital outcomes (i.e., higher levels of marital satisfaction) than do unsupported spouses. Spouses appear to be the persons most likely to be turned to for support in time of need (Dakof & Taylor, 1990) and to be the providers of nearly all types of support among married individuals (Beach, Martin, Blum, & Roman, 1993). Furthermore, support from outside the marriage does not compensate for a lack of support from a spouse (Coyne & DeLongis, 1986). Likewise, a lack of support, inequitable support transactions, and disappointed expectations concerning a partner's behaviour during support interactions are major sources of conflict that precipitate marital and personal distress in marriage (Gleason, Iida, Bolger, & Shrout, 2003).
Despite the recent surge of interest in how support operates in marriage, and despite the important recent advances in our knowledge, many fundamental issues on spousal support remain unresolved. In particular, little is currently known about the determinants of social support provision and of the perception that one is receiving adequate social support in his or her marriage. This gap in our knowledge reflects the fact that, in social support research, most of the attention has been directed to the effects of social support on a broad range of individual outcomes and, more recently, on marital outcomes (see Cohen, Underwood, & Gottlieb, 2000).
Nevertheless, the critical need for identifying the antecedents of social support has been expressed by many researchers (e.g., Gracia & Herrero, 2004;Lindorff, 2005). For example, Newcomb (1990) claimed that determining the role of social support in psychological functioning is no more important as knowing how that support came to be available. Also, Cutrona (1996) considered the study of the factors that affect social support as a prerequisite to advance our understanding of the mechanisms through which social support serves its protective function against stress. From a clinical point of view, it should be obvious that, until the field gains insights into the determinants of spousal support, the goal of designing effective supportive interventions will be difficult to achieve (Bolger, Zuckerman, & Kessler, 2000;Lakey & Cohen, 2000;Rafaeli & Gleason, 2006).
Which variables could be considered as the key determinants of support processes in marriage? Because the empirical evidence relevant to this question is currently sparse, we must turn to previous theoretical discussions. From the so-called relationship perspective on social support in marriage, it is assumed that the seeking, provision, and perception of spousal support are fostered or inhibited by two classes of variables: (1) spouses' personal characteristics, and (2) the properties of the marital relationship in which support processes occur (see Badr, Acitelli, Duck, & Carl, 2001;Reis & Collins, 2000).
Spouses can be characterised by stable individual differences in their way of thinking, feeling, and behaving towards others (McCrae & Costa, 1999). There is little doubt that a person's own characteristics are an important determinant of relationship events (Badr et al., 2001). As such, they are clearly relevant to understanding the origins of support in marriage (Reis & Collins, 2000). More specifically, spouses' personal characteristics are assumed to affect the support process in marriage by encouraging them to express the need for support to their partner, by enhancing the probability that the partner responds positively, and by shaping their perception that support is available in their marriage (Cutrona, Hessling, & Suhr, 1997;Gurung, Sarason, & Sarason, 1997). While most theorists agree that the interpretation of social support, and whether or not it is enacted or solicited may depend to a large extent on the individual characteristics of the partners, the relational perspective on social support dictates that the support given and received within a marriage is impacted by the general character of the marital relationship as well (Badr et al., 2001). In other words, both supportive interaction and the perception that support would be available if needed should be more common to the extent that relationships possess properties relevant to social support (Reis & Collins, 2000).
Given the importance of the relational context according to this view, it is additionally assumed that relationship characteristics matter most, as opposed to spouses' personal predispositions (Reis & Collins, 2000). More specifically, features of couples' relationships are considered to influence the recognition of the needs for and provision of social support, and to affect the interpretation of others' actions in seeking or providing support, thereby overshadowing the influence of spousal characteristics (Badr et al., 2001;Pasch, Bradbury, & Sullivan, 1997).
Given the lack of empirical work on this topic, the present study aimed to evaluate a range of personal characteristics (including extraversion, neuroticism, attachment-related anxiety and avoidance) and a range of relationship properties (including intimacy, commitment, marital conflict, and marital satisfaction) as determinants of support processes in marriage.

Relationship characteristics and spousal support
In the present study, intimacy, commitment, marital conflict, and marital satisfaction were examined as relational determinants of actual and experienced marital support.
Most definitions of intimacy emphasise the perception that a partner is aware of highly personal self-relevant information and can be expected to be respectful, accepting, and responsive when this information is expressed or when relevant circumstances arise (Prager, 1995;Reis & Patrick, 1996). Involving emotional openness, interpersonal acceptance, and shared personal knowledge, intimacy is assumed to foster the conditions for social support in the context of marital relationships (Reis & Collins, 2000).
The second interpersonal predictor included in the current study was spouses' commitment to the relationship, which is conceptualised as spouses' intent to persist in a relationship, including long-term orientation toward the involvement as well as feelings of interdependence (Rusbult, Martz, & Agnew, 1998). Because commitment implies a motivation to preserve an existing relationship and thereby promotes several types of pro-relationship behaviour (e.g., Van Lange, Rusbult, Drigotas, Arriaga, & Witcher, 1997), commitment is hypothesised to relate broadly to supportive activities during a partner's distress.
Marital conflict -including both arguments and expression of negative sentiments toward a partner -has widely been documented as having deleterious effects on marital outcomes (see Fincham, 2003). It has previously been argued that marital conflict may undermine the willingness to ask for support from a partner and to provide support to a partner (Verhofstadt, Buysse, Ickes, De Clercq, & Peene, 2005). By the same token, levels of marital conflict in a relationship may influence the perception that support is available from that partner, if needed (Verhofstadt, Buysse, Rosseel, & Peene, 2006).
Finally, it is advocated that research analyzing potential social support antecedents should take into account spouses' global sentiment about the relationship -that is, spouses' marital satisfaction level. More specifically, partners' overall evaluations of the affective quality of their relationship is considered to be an influential determinant that may override the other judgements about the relationship (Carels & Baucom, 1999;Fincham, Garnier, Gano-Phillips, & Osborne, 1995).
To conclude, each of the relationship properties that were discussed above reflect important dimensions in marital relationships that are theorised to play a cardinal role in spousal support (Reis & Collins, 2000). Based on this idea, the current study explicitly tested the potential value of each of these relationship qualities in predicting marital support processes.
Personal characteristics and spousal support It must be mentioned that studies on the association between personal characteristics and social support do actually exist, but they generally have been conducted outside of intimate relationship contexts. Most of this research has been conducted on college students, using merely general perceptions of support without reference to specific relationships, or assessing interactions with strangers or friends (see Pasch et al., 1997;Pierce, Lakey, Sarason, Sarason, & Joseph, 1997).
A personality characteristic that has been previously linked to social support is extraversion. People that are low on extraversion tend to be more withdrawn and aloof, less energetic, and less self-confident (Clarck, Watson, & Mineka, 1994). There is evidence that these people seek support less often in times of stress (Amirkhan, Risinger, & Swickert, 1995) and perceive lower levels of support (Von Dras & Siegler, 1997). Social support has also been related to neuroticism. Individuals high on neuroticism have a negative view of themselves and others, they tend to be generally dissatisfied, and to feel inadequate (Watson & Clark, 1984). There is empirical support for the negative association between neuroticism and social support perceptions (e.g., Bolger & Eckenrode, 1991). Finally, research has demonstrated that attachment-related anxiety and avoidance are related to perceived support (Collins & Feeney, 2004). Attachment refers to individuals' generalised representations about whether close others will be responsive and supportive in times of need and whether the self is worthy of support and care (Bowlby, 1973). Researchers have identified two dimensions underlying attachment (see Collins & Feeney, 2004). The anxiety dimension reflects the degree to which individuals worry about being rejected, abandoned, or unloved by significant others. The avoidance dimension reflects the degree to which individuals limit intimacy and interdependence with others. Both attachment dimensions have been found to be predictive of social support in samples of strangers.
In sum, evidence is found that personal characteristics (such as extraversion, neuroticism, and attachment) affect individuals' help-giving and helpseeking behaviour (Rook, Pietromonaco, & Lewis, 1994) and the level of support available from their social network (e.g., Procidano & Heller, 1983;Sarason, Sarason, Hacker, & Basham, 1985). However, as mentioned above, these findings come from research that was conducted with samples of nonintimates, or that has focused on individuals who interact with others with whom they have at most a limited history. For this reason, there may be a critical problem with the generalizibility of these research findings to support exchanged between marital partners. As Pasch et al. (1997) hypothesised, the specific qualities of marital relationships may compromise the generalizibility of results from samples of non-intimate pairs. In line with the relational perspective on social support -as outlined above -personal influences on spousal support are expected to appear less obvious when the unique relationship history that is characteristic of marital relationships is taken into account. More specifically, spouses' personal characteristics may have originally played important roles in determining partner choice and whether a relationship develops at all between two people (Pasch et al., 1997). However, after the relationship is formed, it is an open question whether these personal influences persist, or whether the relationship characteristics may have a clearer impact on support processes in marriage.
For this reason, the present study aimed to replicate the influence of personal characteristics (including extraversion, neuroticism, and attachmentrelated anxiety and avoidance) on social support in a sample of married couples. Furthermore, we examined the unique contribution of personal characteristics in explaining social support within intimate relationships beyond relationship characteristics.

Social support as both support perception and support behaviour
In this investigation, a multi-dimensional view of social support was adopted. Based on the prevailing view in the literature, we differentiated between support that is perceived to be available in the marital relationship and actual support behaviours (see Wills & Shinar, 2000). The latter refer to spouses' specific behaviours when support is exchanged within their marital relationship, including spouses' support solicitation and provision behaviours.
It has come to be generally agreed that the behavioural aspects of social support needs to be differentiated from the perception of support (see Cohen et al., 2000). Spouses' relationship-specific perceptions of support consist of spouses' expectations about the availability of spousal support, growing out of a history of experiences with their marital partner, and reflecting the unique ways in which they view this person (Sarason, Sarason, & Pierce, 1994).
A qualification needs to be made regarding the assessment of support-giving and support-seeking behaviours. Social support was originally conceptualised as purely positive in nature, resulting from the strong emphasis on prosocial aspects of relationships in research on social support. Support researchers now recognise that this in an untenable proposition, since several types of negative support behaviours have recently been identified (see Pasch, Harris, Sullivan, & Bradbury, 2004). Accordingly, the present study will focus on categories of both positive and negative behaviours of spouses offering and soliciting support (see Bradbury & Pasch, 1994 for a detailed description).
Among the support seeking categories (see Bradbury & Pasch, 1994), positive support seeking includes behaviours such as offering a clear analysis of the problem, expressing feelings related to the problem, and asking for help in a useful way. Negative support seeking includes behaviours such as making demands for help, criticising or accusing the support provider, and whining or complaining. Among the support provision categories (see Bradbury & Pasch, 1994), positive emotional support provision includes behaviours that reassure, console, or otherwise encourage the support solicitor, letting the solicitor know that he or she is loved, cared for, and behaviours that encourage expression or clarification of feelings. The positive instrumental support provision category includes behaviours such as making specific suggestions to the support solicitor, giving helpful advice, and offering to assist in the development or enactment of a course of action for solving the problem. Finally, the negative support provision category includes behaviours such as criticising or blaming the support solicitor, offering inconsiderate advice, and insisting that the support solicitor takes sole responsibility for dealing with the problem.

Participants
The sample consisted of the 400 members of 200 Belgian married couples whose participation was solicited by using a snowball sampling method. We started the survey by recruiting 50 couples in shopping areas. To participate, couples had to be: (a) currently involved in a heterosexual relationship for at least one year, (b) married for at least six months, and (c) willing to participate voluntarily in the study. For the eligible couples who expressed interest in the study, a home-visit was scheduled to complete a packet of questionnaires. In a second step, additional couples were obtained from this initial sample. For each of the initial 50 couples, a series of eligible referrals could be made within their circle of acquaintances. The referred couples were then contacted by telephone and asked for their participation in the study. In a third and final step, additional referred couples were obtained from the second sample.
The mean ages for the men and the women were 42.81 (SD = 12.60, range = 22-78), and 40.62 (SD = 12.57, range = 19-78), respectively. On average, the men and the women had completed 13.08 (SD = 2.53, range = 6-16) and 13.26 (SD = 2.47, range = 6-16) years of education, respectively. The couples had an average of 1.47 children (SD = 1.28, range = 0-6). The average length of their relationships was 16.92 years (SD = 11.80, range = 1-54). About half of the male and female participants were white-collar workers (50% and 56%, respectively). Twenty-six percent of the men and 10% of the women were blue-collar workers. Seventeen percent of the men and 10% of the women were self-employed. Five percent of the men and 9% of the women were students; 2% percent of the men and 15% of the women were unemployed.

Procedure
Data were collected using home surveys. As agreed in the first telephonic contact, couples were visited at home by one of the research assistants. Both partners then completed measures concerning several personal and relationship properties. In the present research, the Dutch version of all the instruments was used. At the end of the session, couples were debriefed more fully about the global aim of the study and thanked for their participation. Couples were assured that their data would be analysed anonymously and confidentially, and they were asked to grant written permission to use the collected data for scientific goals.

Measures
Perceived Spousal Support. The Quality of Relationships Inventory Support-scale (Pierce, Sarason, & Sarason, 1991;Verhofstadt et al., 2006) was used to assess spouses' perceived availability of social support from their relationship. The seven items were rated on a 4-point Likert scale (1 = not at all, 4 = very much) and spouses' scores were obtained by computing the mean of their responses across all items in the scale. The alpha reliabilities for the support-scales used in this study were high (α = .85 for husbands, and α = .86 for wives).
Spousal Support Behaviour. Spouses' support solicitation and provision behaviours were assessed by a questionnaire based on the Social Support Interaction Coding System (Bradbury & Pasch, 1994; with permission from the authors). This social support interaction questionnaire (Verhofstadt, 2005) included 54 items and each item was rated on a 9-point Likert scale (1 = very unlikely, 9 = very likely). Five subscales were used indicating the likelihood of (a) Positive Support Seeking (e.g., clear analysis of problem, recognises partner as an aid, agrees with suggestions of helper), (b) Negative Support Seeking (e.g., rejects help, criticises helper, makes demand for support), (c) Instrumental Support Provision (e.g., offers specific plan or assistance), (d) Emotional Support Provision (e.g., reassurance, expression of care for help-seeker, understanding), and (e) Negative Support Provision (e.g., criticises, minimises problem). Subscales scores were computed separately for husbands and wives by computing the mean of their responses across all items in the scale. Cronbach's alphas ranged between .73 and .87 for husbands, and between .76 and .85 for wives, indicating good internal consistency for the subscales. For each spouse, two composite scales were created from these five subscales indicating the likelihood of positive support behaviour (sum of scales a, c, and d) and negative support behaviour (sum of scales b and e) during exchanges of spousal support.
Intimacy. We used the Emotional Intimacy Subscale of the Personal Assessment of Intimacy in Relationships (PAIR; Schaefer & Olson, 1981), which has 6 items (e.g., "I sometimes feel lonely when we're together", "I often feel distant from my partner") on a 5-point Likert scale (0 = strongly disagree, 4 = strongly agree). The items were scored as such that a higher score on this scale reflects a higher level of intimacy within the couple. The internal consistencies for this subscale were .73 for husbands and .77 for wives.
Commitment. The Rusbult et al.'s (1998) seven-item Commitment Level Scale was used (Van Lange et al., 1997). Participants reported their degree of agreement with each item (e.g., "I want our relationship to last for a very long time", "I am committed to maintaining my relationship with my partner") using 9-point Likert scales (0 = do not agree at all, 8 = agree completely). The internal consistencies were .70 for husbands and .76 for wives.
Marital Satisfaction. Marital satisfaction was assessed with the ten-item Satisfaction Subscale of the Dyadic Adjustment Scale (DAS; Spanier, 1976).
Participants reported their degree of agreement with each item (e.g., "Do you ever regret that you married or lived together?", "In general, how often do you think that things between you and your partner are going well?") using Likert scales (0 = all the time, 5 = never). The items were scored as such that a higher score on this scale reflects a higher level of marital satisfaction within the couple. The internal consistencies for this subscale were .83 for both husbands and wives.
Marital Conflict. The Quality of Relationships Inventory Conflict-scale (Pierce et al., 1991;Verhofstadt et al., 2006) was used to assess the extent to which the marital relationship is a source of conflict. The 12 items (e.g., "How much do you argue with this person?", "How angry does this person make you feel?") were rated on a 4-point Likert scale (1 = not at all, 4 = very much) and spouses' scores were obtained by computing the mean of their responses across all items in the scale. Alpha reliabilities for the Conflictscales were high (α = .87 for husbands, and α = .89 for wives).
Extraversion and Neuroticism. The NEO Five-Factor Inventory (NEO-FFI; Hoekstra, Ormel, & De Fruyt, 1996) was used to measure the personality factors of Extraversion and Neuroticism. The NEO-FFI included 60 items, and each item was rated on a 5-point Likert scale, ranging from strongly disagree to strongly agree. Cronbach's alphas for the Extraversion Scale were .60 for husbands and wives. Cronbach's alphas for the Neuroticism Scale were .66 (husbands) and .70 (wives).
Attachment. The 18-item Adult Attachment Scale (Collins & Read, 1990) measures attachment-related anxiety and avoidance. Items are rated on 5-point Likert scales (1 = not at all characteristic, 5 = very characteristic). Cronbach's alphas for the Anxiety Scale were .60 for husbands and wives. Cronbach's alphas for the Avoidance Scale were .60 (husbands) and .65 (wives).

Do relational characteristics predict spousal support?
A series of multiple regression analyses were performed to determine the potential value of relationship characteristics in predicting support perceptions, positive support behaviour, and negative support behaviour. In each regression model, spouses' reports of relational characteristics were entered into a regression equation as predictors. Separate regressions were carried out for husbands and wives.
Prior to each regression analysis, collinearity diagnostics were performed using the variance inflation factors (VIF) as criterion. No multicollinearity was evident since the VIF for the predictors ranged between 1.04 and 3.22 (< 10) (Cohen, Cohen, West, & Aiken, 2003).
As hypothesised, husbands' and wives' support perceptions and behaviours -positive and negative -were significantly predicted by relational characteristics (see Table 1). The explained variance ranged between 22% and 60%.
As predicted, the level of intimacy in the relationship was strongly linked with both husbands' and wives' support perceptions and actual support behaviours. Higher levels of intimacy in the marital relationship resulted in higher levels of perceived support, more positive support behaviour, and less negative support behaviour, in wives. Likewise, husbands reporting higher levels of intimacy reported higher levels of perceived support, and higher levels of positive support behaviours.
Counter to our expectations, spouses' level of commitment seemed unrelated to spousal support. As reported in Table 1, this finding was similar for husbands and wives, and for each social support component.
Also contrary to our expectations, marital conflict was not significantly related to husbands' support behaviour. The level of marital conflict in a rela- tionship was, however, a significant predictor of wives' behaviour during supportive interactions. In addition, higher levels of marital conflict were associated with lower levels of perceived support in both husbands and wives.
Results only partially supported our hypotheses for marital satisfaction that was not significantly related to husbands' perceived support and wives' support behaviour. In contrast, higher levels of marital satisfaction significantly predicted higher levels of perceived support in wives, and more exchange of positive support behaviours in husbands. Husbands who were more maritally satisfied reported to display lower levels of negative support behaviours.

Do personal characteristics predict spousal support?
Next, multiple regression analyses for predicting spousal support from spouses' personal characteristics were conducted. As predicted, husbands' and wives' social support perceptions and behaviours -positive and negative -were significantly predicted by their own personal characteristics, resulting in an explained variance that ranged from 7% to 28% (see Table 2).
For husbands and wives, attachment-related anxiety provided significant contributions to the explanation of variance in support perception and supportive behaviour, in the predicted direction. Higher scores on the anxiety dimension of attachment resulted in significantly lower levels of perceived support, as reported by both husbands and wives. Higher levels of attachment-related anxiety were also associated with higher levels of negative support behaviours in husbands and wives.
Attachment-related avoidance was another significant predictor of support perceptions and behaviours. More avoidantly attached wives perceived less support in their relationship, and seemed to behave less positively during support interactions. In both the male and the female subsample, higher scores on attachment-related avoidance predicted higher levels of negative support behaviours.
In general, and contrary to our predictions, spouses' level of neuroticism and extraversion seemed unrelated to their level of perceived support and the amount of positive and negative support behaviours they display. Only for males, higher scores on neuroticism predicted higher levels of negative support behaviours.

Are personal characteristics less influential on support in the context of marital relationships?
Finally, hierarchical multiple regression analyses were performed to test whether the contribution of personal characteristics to spouses' support perception and support behaviours would reduce in magnitude and/or become insignificant, after relationship characteristics were taken into account. Separate regressions were carried out for husbands and wives and for support perception, positive and negative support behaviours.
In the first block, spouses' reports of relationship characteristics were entered. In the second block, spouses' personal characteristics were entered (see Table 3).
The regression models in which relationship and personal characteristics were entered as joint predictors of support were found to be significant for all dependent variables and the explained variance ranged between 24% and 62%. As expected, spouses' support perceptions and behaviour were significantly predicted by relationship characteristics, when entered first. Overall, the unique contribution of personal characteristics, entered in second step, in explaining social support beyond relationship characteristics was about 2%. Except for husbands' negative support behaviour and wives' level of perceived marital support, the variance explained by personal variables beyond relationship characteristics no longer reached the significance level. A possible post-hoc explanation for this pattern of findings is that both classes of determinants of spousal support overlap to some extent. Therefore, canonical correlation analyses were performed to determine the correspondence between both sets of determinants. These analyses revealed a significant association between both classes of determinants, Wilks's Lambda = .69, F(16, 587) = 4.67, p < .001, Wilks's Lambda = .81, F(16, 587) = 2.52, p < .001, for husbands and wives, respectively. The analyses revealed four canonical correlations for husbands and wives. The pair of canonical variates that accounted for the significant relationships between the two sets of variables revealed a canonical correlation of .52 for husbands and .34 for wives, that is, 28% and 12% overlapping variance between both classes of determinants.

Summary of results
The present research aimed to further our understanding of potential determinants of spousal support by examining the contribution of personal  Table 3. Summary of hierarchical multiple regression analyses to predict husbands' and wives' support perception and behaviour from relational and personal characteristics. characteristics and relationship properties to perceptions of support availability and supportive behaviours within marital relationships.
As expected, support processes within marriage were found to be substantially affected by the general characteristics of the marital relationship. Relationship qualities significantly predicted the process of seeking and providing support, as well as the level of support that is perceived to be available in the marital relationship. More specifically, spouses who experienced more intimacy in their relationship felt more supported by their partner, and were more positive and less negative in soliciting and providing support. These results are in line with prior research on the association between intimacy goals and support-related variables (Sanderson & Evans, 2001). The association between the level of intimacy spouses experienced in their relationship and support in their relationship proved to be consistent across gender, and across the different components of spousal support that were included in the current study.
In contrast, our findings indicated that spouses' level of commitment did not affect any support-related variable that was included in the study. On the face of it, the pattern of results concerning intimacy and commitment appears to be inconsistent since they are both cardinal relationship features (Reis & Collins, 2000). A conceptual explanation for these contrasting findings is that intimacy and commitment refer to different underlying dimensions of marital relationships. Intimacy refers to the dimension of so-called "partner responsiveness"or how partners are perceived to respond to the many public manifestations of the self (Baumeister, 1995). In contrast, commitment refers to the nature and the extent of the "interdependence" in a marital relationship, which is primarily a structural feature of the relationship (Kelley, 1983). Partners can be very interdependent and have a stable relationship without being responsive to the other spouse, and can be very responsive without having a very interdependent relationship (Lewis & Spanier, 1982). An examination of how support processes may vary across these kinds of relationship types is certainly a new avenue by which we can expand our focus in future research.
Marital satisfaction affected husbands' support behaviour as well as wives' support perceptions. This finding makes sense because marital satisfaction -referring to the partners' overall evaluations of the quality of their relationship -has been previously described as a global variable pertaining to the couple's relationship as a whole that has a generalised influence on relationship communication behaviour across several situations (Bradbury & Fincham, 1991;Sanford, 2003). Our results on support perceptions coincide with those of other studies in which higher ratings on marital satisfaction were found to be associated with more support-related feelings (Pasch et al., 1997), higher ratings for partner supportiveness (Cutrona & Suhr, 1994), and perceptions of support helpfulness (Carels & Baucom, 1999).
By the same token, the extent to which the relationship is a source of conflict affected wives' support behaviours, as well as husbands' support perceptions. This type of finding is consistent with the results of the few previous studies that have examined the interplay between conflict and support in marriage (Julien, Chartrand, Simard, Bouthillier, & Bégin, 2003;Pasch & Bradbury, 1998;Verhofstadt et al., 2005).
In the present study, spouses' personal characteristics as well, proved to be significant determinants of spousal support. However, they explained a much smaller portion of the variance in spousal support (explained variance ranged between 7% and 28%), than relational predictors did (explained variance ranged between 22% and 60%). These findings thereby provide empirical support for the assumption that processes operating within ongoing intimate relationships are more influential on spousal support than spouses' interpersonal predispositions (Reis & Collins, 2000).
Our findings concerning personal determinants of spousal support are only partially consistent with research from non-marital relationships. As predicted, individual differences in attachment were significantly related to spouses' support perceptions, as well as their supportive behaviours. In accord with recent research (Collins & Feeney, 2004), spouses who have a general tendency to worry about being rejected or unloved by others, felt less supported by their partner and tend to behave in less positive and more negative ways towards their spouse during support transactions. This finding makes conceptual sense, especially in combination with the results indicating a link between intimacy and social support in marriage, as described above. As is the case with intimacy, spouses' attachment style or how a person sees himself in relation to others (Bowlby, 1973(Bowlby, , 1980 also refers to the dimension of "partner responsiveness", although conceptualised and assessed in more dispositional terms. What is surprising is the apparent lack of impact of personality traits such as neuroticism and extraversion upon marital support in this study. More specifically, the extraversion and neuroticism of support recipients did not directly predict the amount of support they perceived in their relationship. In addition, individual differences in extraversion and neuroticism did not correlate with the amount of support seeking and provision behaviours. Our findings are not consistent with past research that suggests such personality features to impact support processes in non-marital relationships (e.g., Bolger & Eckenrode, 1991;Von Dras & Siegler., 1997). Therefore, our results require replication and clarification in further research on support in intimate relationships.
Of particular interest was that personal characteristics did not provide a unique contribution in explaining spousal support beyond relationship characteristics. In our opinion, these findings do not question the importance of some individual sources of support in marriage. They rather imply that, as noted previously by Pasch and colleagues (1997), personal characteristics of support providers and recipients may be less influential in close dyadic relationships, as they may in some cases be overshadowed by the qualities of the relationship between the provider and recipient.
Additional analyses on these data revealed an overlap between the personal and relational determinants, which might have contributed to this pattern of results. This is consistent with recent research showing that personal and relational constructs are intertwined in various and intricate ways (White, Hendrick, & Hendrick, 2004). Although not the major aim of the present research, these findings require some consideration. Some variables included in this analysis describe relationships with particular partners, whereas other variables are measured across relationships, as general social predispositions. Their common characteristic -referring to interpersonal properties relevant to social support -is a factor that might have contributed to their relatedness. Alternatively, since individual differences shape perceptions of support, they may also influence other reports of relationship properties. We therefore analysed if relationship influences on social support persisted when personal factors were taken into account. These results revealed that relational characteristics still made a significant and substantial contribution to the prediction of spousal support, over and beyond personal variables (explained unique variance ranged between 15% and 51%). Finally, because both relational and personal characteristics were both assessed via self-report measures, the possibility that common method variance contributed to their overlap cannot be ruled out.

Limitations and strengths
There are several limitations to this study that suggest future research. The first potential limitation of the present study concerns our choice of the sample in which to investigate the determinants of spousal support. In the current study, we used a sample of white, middle-class, non-clinical couples, thereby limiting the generalizibility of the results. An important goal for future research will be to replicate these findings with couples drawn from more diverse samples (ones that include, for example, couples who are seeking marital counselling, or homosexual couples).
One might also question whether the use of snowball sampling may have affected the current study's results. Snowball sampling has a number of deficiencies, including problems of representativeness (see Atkinson & Flint, 2001). More specifically, our sample may be biased because the couples were not randomly drawn, but were dependent on the subjective choices of the respondents first accessed (see Griffiths, Gossop, Powis, & Strang, 1993). Furthermore, snowball samples are often biased towards the inclusion of indi-viduals with interrelationships, that share similar and unique characteristics not shared by the wider population (Griffiths et al., 1993). To partially address these problems of selection bias, a large sample of married couples was used in the present study. In addition, the sample of couples was recruited in several discrete referral chains (i.e., 50 couples) with fewer links, rather than in a large single chain. Although we cannot evaluate whether the population from which we have drawn our couples is the population to which we wish to generalise, the sample obtained was diverse in relationship length and age of participants.
We do, however, have a number of reasons to believe that snowball sampling was an appropriate way of sampling married couples. First, it is an extensively used sampling strategy in the marital research area. Second, it has been found to be economical, efficient, and effective in various studies (Atkinson & Flint, 2001). Third, the additional main value of snowball sampling in the present research was as a method for obtaining respondents where some degree of trust is required to initiate contact (see Atkinson & Flint, 2001). In other words, relationship matters are often considered 'private' and outside the 'public' realm of research (Browne, 2005), and the more 'sensitive' the phenomenon under study the more difficult sampling is (Faugier & Sargeant, 1997). In sum, because we do not have the data at present to test how the sampling strategy may have influenced the present study's results, the usual recommended caution should be exercised in generalising our results.
Another limiting factor for this research is that we did not account for the interdependence among couple members when analysing our data. Because the major aim of the present study was to evaluate the unique value of a range of personal and relationship characteristics in predicting spousal support, a (hierarchical) multiple regression approach was used. Although this analytic strategy was considered as the most appropriate one for the research questions under consideration, it did not allow us to simultaneously analyse both husbands' and wives' data. Testing a model that takes the relationships between male and female partners into account (e.g., general estimating equations, as suggested by an anonymous reviewer), would definitely allow a more thorough exploration of the similarities and differences between partners' support behaviours and perceptions. 1 Although this approach was -----beyond the scope of the present study, it is certainly a new avenue by which we can expand our focus in future research.
Finally, the present findings are limited by the reliance on self-report to assess actual support behaviours and relationship properties. Because measures of communication behaviours -including social support -and relationship quality share method variance, our findings await replication in observational research on support transactions in couples.
An important strength of the present study is the multi-component approach when evaluating and interpreting the influence of several determinants of social support. The present findings indicate that support behaviour and support perceptions do not necessarily converge, either in the impact that the personal determinants have on them or in the impact that relational characteristics have on them. This disjunction between support behaviour and perception illustrates that they each have unique relevance for understanding social support.

Conclusions and implications
From the present investigation of personal and relational determinants of social support in marriage, the following conclusions can be drawn. First, both spousal and marital characteristics have important value in the prediction of support processes in marriage. Second, relational predictors explain more variability in spousal support than personal characteristics. Finally, personal influences on spousal support are less pronounced when relationship features are taken into account (i.e., no unique effect), whereas relationship features do have a unique effect on spousal support -over and beyond personal characteristics.
The first lesson from the present findings is that relationship characteristics compromise the generalizibility of results on personal characteristics and social support from samples of non-intimate pairs to support in couples (Pasch et al., 1997;Verhofstadt, Buysse, Ickes, & De Corte, 2007). This implies that existing theory pertaining to the intraindividual antecedents of social support in non-intimate relationships may not be simply generalised to how support operates in marriage.
From a clinical point of view, our findings imply that researchers and practitioners should focus on characteristics of support recipients, support providers, as well as on characteristics of relationships in which support transactions take place because they all affect the probability that supportive exchanges occur, as well as the effectiveness of such exchanges. As outlined by Rook and Underwood (2000), interventions may be designed to fit an individual's relational circumstances. Similarly, in planning intervention programs, the supportive capacities of potential support providers could be taken into account to optimize the delivery of support. In the same way, when clinicians design support interventions, it may be useful to screen out potential participants who lack a receptivity to support and therefore, may require a program that is tailored to their individual needs (Cutrona & Cole, 2000).
To conclude, the relational approach on social support taken in the present study allowed us to explore personal as well as relational antecedents of spousal support. What counts is the larger pattern of results that furthers our understanding of the nature of social support in close relationships. More specifically, our results reinforce the claim of many contemporary theorists that social support is an interpersonal, rather than an intrapersonal phenomenon (cf. Badr et al., 2001;Reis & Collins, 2000).