Gender, social support, and well-being: Evidence from a Greek community sample

The importance of social support for psychological well-being has been aptly highlighted in epidemiological and psychological research. However, it is not clear from the existing research whether gender differences in structural (relationship status, network size, frequency of interactions with friends) and functional (support satisfaction) aspects of social support exist and -if they do- to what extent they affect males’ and females’ well-being. Hierarchical regression analyses of crossectional data from a Greek community sample showed that support satisfaction was an important predictor of well-being outcomes in males whereas several structural indicators were predictors of different well-being outcomes in females. Females’ anxiety, perceived stress, and loneliness were adversely affected by frequency of interaction with acquaintances. The results are discussed with regard to genderrole differences that may be underlying the social support effects on well-being, as well as related cultural values.

In the last two decades, multidisciplinary research programs in the social sciences have established the key role of social support for well-being (see Sarason, Sarason, & Gurung, 2001, for a review).Social support refers to perceptions or experiences of care, value and assistance from others of one's social network (Cutrona, 1996;Wills, 1991).A sizeable body of evidence has shown consistent links between inadequate levels of social support and poor physical and mental health (Sarason et al., 2001;Williams, Ware, & Donald, 1981).The magnitude of the evidence has been such that in a key article House, Landis, and Umberson (1988) proposed that social support is a fundamental construct and "insufficient social support" is a risk factor for health.
Despite the strong research interest, it has been less clear whether different facets of social support are affecting men's and women's well-being equally.Epidemiological and psychological studies have shown men's health to profit from the existence of marital relationships (e.g., Cutrona, 1996;House, Robins, & Metzner, 1982).More recent epidemiological research on social support and mental health has demonstrated stronger positive associations between social support and well-being for men than for women (e.g., Stansfeld, Fuhrer, & Shipley, 1998).However, there are studies showing equivocal effects of social support on men's and women's mental health (Paykel, 1994;Williams et al., 1981) and in some cases, gender differences in the social support and health associations are eliminated when the same data are analyzed using more detailed indices of social support (e.g., Furher & Stansfeld, 2002).The inconsistencies observed in the literature may reflect differing operationalizations of the social support construct.Epidemiological studies have mostly looked at the effects of the existence and breadth of the social network (structural aspects of social support) and have put less emphasis on psychological aspects, such as the perceived quality of support (functional aspects of social support).
The present study thus extended this line of research by distinguishing between structural (distal) and functional (proximal) facets of social support (Helgeson, 2003) and looking at the comparative effects of those on a number of men's and women's well-being indicators.Structural aspects of support (e.g., network size, frequency of interactions, type of relationships, marital status, etc.) deal with the "mere existence of social relationships" and have usually been examined with cross-sectional designs and epidemiological type surveys.Functional measures of social support are typically the concern of psychological research and have been examined with reference to psychological resources such as: emotional, instrumental, and informational support, and support satisfaction.Generally, perception of social support is a multi-componential construct and a stronger predictor of well-being and adjustment to stress than support received (Nezlek, 2001;Sarason et al., 2001;Wethington & Kessler, 1986), and in some cases, support received can have negative effects on the person (Rook, 1984).Distinguishing between structural and functional aspects of social support in men and women can help discern the underlying processes that lead to well-being outcomes (Heller & Rook, 1997).
Reflecting on the literature on gender differences in the social support-health links, Shumaker and Hill (1991) contend that the possible mechanisms linking social support with health and well-being may be different for men and women.There are documented gender differences in structural aspects of social support.Women tend to have larger and more diverse social networks (Antonucci & Akiyama, 1987), and are more likely to receive support from those relationships (Barbee et al., 1993).There are, however, studies that failed to find gender differences in network size in black adolescents in the US (Coates, 1987) and young adults in Greece (Georgas & Dragona, 1988).The latter studies however concerned collegiate samples and may not be comparable to community samples, since social support processes change across the life course (Vaux, 1985).Regarding functional aspects of social support (mainly satisfaction with perceived support), gender seems to be a lesser influential factor (Antonucci & Akiyama, 1987), a finding that supports hypotheses about the psychological makeup of the construct (Sarason, Pierce, & Sarason, 1990).
There is, however, evidence for gendered patterns of social support effects on wellbeing.Women are more likely to rely on friends and family as health resources (Kandrack, Grant, & Segall, 1991;Pretorius, 1996) and quality of community is more influential on women's than men's health and well-being (Molinari, Ahern, & Hendryx, 1998).Research has also found social interactions to be stronger predictors of older women's well-being than men's of the same age range (Lund, Modvig, Due, & Holstein, 2000).On the other hand, perceptions of social support have been found to be an equally important predictor of wellbeing in both mature men and women (Antonucci & Akiyama, 1987).
The present study tested gender differences on social support and their effects on various well-being outcomes.Well-being is a multi-faceted construct, both conceptually and operationally.Research on subjective well-being has examined general, trait-like levels of satisfaction and happiness (e.g., Diener, Diener, & Diener, 1995), and research focusing on social relations and well-being has employed measures that were focused on mental health outcomes (e.g., the General Health Questionnaire).Following criticisms for the narrow focus of such research on certain psychological and well-being indicators (Shumaker & Hill, 1991), the present study examined a wide range of well-being indicators, such as loneliness (distinguishing further between social and emotional loneliness; Davis, Morris, & Kraus, 1998), perceived stress, anxiety, and mental health.
The study extended the relevant literature on gender, social support and well-being, by utilizing a Greek community sample.There is a growing awareness of the importance of cross-cultural differences in support processes (e.g., Goodwin, 1999;Taylor et al., 2004).
Recent studies have identified the cross-cultural dimension of masculinity (a construct that reflects the extent to which roles are distributed equally between genders; Hofstede, 2001) as an important determinant of gendered patterns of social support in different cultures.
Countries high in masculinity are characterized by a clear differentiation in the way social support processes are distributed between the genders (Stevens & Westerhof, 2006a, 2006 b).Comparisons between Holland (a country low in masculinity) and the US and the UK (two countries high in masculinity) partially supported these hypotheses.Contrary to findings in the UK and the US, Dutch men and women were similar in the size and composition of core support networks, the provision of emotional support to and from the partner, and the provision of instrumental support to others, but Dutch women were still characterized by greater involvement in extensive networks (friends, relatives, etc.) than Dutch men.Given that the Greek society is relatively high in the masculinity dimension (Hofstede, 2001), and gender roles are largely segregated, we anticipated that men and women will have clear differences in social support and its effects on well-being.Moreover, along with other collectivist-oriented societies (at least at the cultural level of analysis) people in the modern Greek society explicit social support seems to be a weak predictor of well-being outcomes (Kafetsios, 2006;Taylor et al., 2004) and this constitutes the study of gendered patterns of social support in this culture both interesting and culturally distinct.

Summary of aims and expectations
The study aimed firstly to examine gender differences in structural (network size, relationship status, frequency of interactions with different types of relationships) and functional facets of social support (perceived satisfaction with support), and secondly to test whether these two types of social support have the same impact on several well-being indicators (anxiety, mental health, perceived stress, loneliness) in men and women.Finally, the use of a Greek community sample allowed comparisons with findings from research that has taken place in cultural contexts with similar and different degrees of masculinity, a dimension that reflects gender role rigidity.

Hypothesis 1
Prior research certainly led to the expectation that women would have larger social networks.Therefore, it was expected that women would report a larger network of supportive persons and a higher frequency of interactions with acquaintances and friends.
We did not expect any differences in functional aspects of social support (perceived satisfaction with support).

Hypothesis 2
The second main aim of the study was to examine the relative effects of structural and functional aspects of support on well-being outcomes in males and females.Based on the available evidence, it was expected that structural aspects of social support would be more influential on women's well-being.The distinct difference of structural social support on women's and men's well-being was further supported by cultural considerations (gender role segregation in Greek society).Hypotheses as to whether functional social support (perceived satisfaction with support) would impact men's and women's well-being equally were not formulated.

Sample
Participants were 222 adults (110 males and 112 females) from an urban area of Northern Greece.This was a snowball sample recruited via student and friend networks.The distribution of participants' age is shown in Table 1 (range: 18-66 years, M = 34.6 years, SD = 12).Forty one per cent of the participants had completed secondary education, 37% had a University degree and 19% had completed post-graduate education (3% did not respond to the question).Most of the participants (73.8%) were either married or in a long-term, permanent relationship, and the majority (80%) were in full employment, equally distributed in the different age groups.Participants completed the questionnaires on a voluntary basis, after they had been approached by one of the researcher's associates.Those who expressed a wish for feedback from the research were sent a simplified version of the aims of the study and results in the address/email provided by the participants.

Measures
All scales were translated into Greek (by the author) and blindly back-translated by a Greek graduate student with some items modified to enhance the naturalism of the translations (Van de Vijver & Leung, 1997).

Social relations
Relationship status (i.e., married, divorced, steady relationship, occasional relationship, no relationship) was re-coded into two categories (participants in relationships or not).
Frequency of interactions.Participants were requested to report the average daily amount of interactions (in hours) they had had separately with acquaintances, good friends, and partner.
Social support.Social support was assessed with the short-form social support questionnaire (SSQ6; Sarason, Sarason, Shearin, & Pierce, 1987).For each of the six questions participants were required to list all persons who could provide support of the type described in the question (min.0, max.9), and also indicated how satisfied they were overall with that level of support (on a six-point scale).Hence, the scale provided a quasistructural measure of social support (number of persons available for support), and one perceived global satisfaction measure.The two parts had good internal consistency (a = .92and a = .87respectively).

Psychological well-being
UCLA Loneliness Scale (Russell, Peplau, & Cutrona, 1980;Russell, 1996).The scale consists of 20 questions and was developed to measure feelings of loneliness and social isolation.The development of the scale was based on a series of studies that looked at how lonely persons describe their feelings (Cronbach's alpha = .87).
Social and emotional loneliness.Two paragraphs introduced in Davis, Morris, & Kraus (1998) were also used to assess subjective feelings of emotional and social loneliness.
Respondents were asked to indicate the extent to which they agreed with each of the two the statements on a five point scale (1 = 'not at all', to 5 = 'very much').
General Health Questionnaire (Goldberg, 1978).The GHQ (20 item version) measures participants' current mental health.The scale assesses depression, state anxiety, somatic symptoms, and social dysfunction.Its correlation with the Beck Depression Index is particularly strong (r = .72).Items concern situations with which the individual had to cope over the last few weeks and influenced his/her psychological health.The internal consistency of the scale was satisfactory (Cronbach's alpha = .75).A higher score signifies poorer psychological health.
Trait anxiety was measured with the Trait form of the State-Trait Anxiety Inventory (Spielberger, Gorsuch, & Lushene, 1970).This is a 20-item scale that assesses participants' vulnerability to anxiety experiences (Cronbach's alpha = .84).
Perceived stress.Six questions required of participants to report how frequently they had experienced a list of stressful situations (unexpected negative event, nervousness, being overwhelmed, becoming angry, anxiety to do things, accumulated difficulties) during the last month.All items were rated on a 5 point scale (Cronbach's alpha = .81).

Results
Three sets of analyses were carried out.First, gender differences were tested through ANOVA models, entering age and relationship status as covariates.Secondly, product moment correlations between social relations and well-being variables were computed separately for males and females.Finally, hierarchical regression models were computed for males and females separately, to predict facets of well-being and perceived satisfaction with support.
Social support.More detailed findings on social support with this sample are presented in another paper (Kafetsios & Sideridis, 2006) and only part of it will be described here, as it pertains to the aims of this study.The inter-correlation of the two parts of the Social Support Questionnaire was similar for both genders (r = .39for males and r = .36for females).This is comparable to the correlations reported for student samples in the US (r = .34;Sarason, Levine, Basham, & Sarason, 1983) and in Greece (r = .35;Georgas & Dragona, 1988).

Gender differences in Social Support and well-being
ANOVA models tested for gender differences in social support and well-being (with age and relationship type as covariates, see table 1).There were some gender differences with regards to social support.Females reported somewhat higher levels of support satisfaction (F (1, 217) = 4.96, p < .05)but the total number of supportive persons reported was similar in both men and women.However, women differed in some other quantitative support indicators, reporting almost double average frequency of contact with acquaintances (F (1, 217) = 5.78, p < .05)and higher frequency of partner contact F (1, 217) = 5.44, p < .05).
There were also some interactions between age and gender effects on social support indicators.Younger men reported significantly lower levels of satisfaction with support (F (1, 217) = 9.07, p < .01)than any other group.Older and younger females' support satisfaction did not fluctuate as much.This effect remained significant even after controlling for relationship status.These differences in quality of support were not reflected in gender differences with the number of supportive persons reported.
Frequency of interactions with partners was predicted solely by relationship status (being attached or not) and there was no gender by age interaction effect observed.Younger men and women reported interacting more frequently with acquaintances (F (1, 217) = 3.71, p < .05)than older men and women.
Regarding trait anxiety, females were significantly more anxious than males (F (1,217) = 11.92,p < .01).Women reported higher levels of perceived stress than men (F (1, 217) = 2.81, p = .07)and this tended to be particularly true for younger females (F (1, 217) = 3.53, p = .06).As it can be seen in table 1, there were no gender differences in GHQ scores and loneliness.

Social support and psychological health in men and women
Table 2 presents the product-moment correlations between social support and psychological health indicators, separately for men and women.The three variables measuring frequency of social interactions were logged to achieve normalization.
Age.In both males and females, age was associated with lower frequency of contact with good friends and a higher frequency of interaction with partner.However, older males reported higher support satisfaction and a lower frequency of interaction with acquaintances, something that was not observed in females.Older males and females had lower anxiety perceived stress, and older males had lower mental health (as measured by the GHQ).
Relationship status.In both males and females, being in a relationship was negatively related to being socially and emotionally lonely, anxious, and stressed.
Social support.The size of supportive network was almost unrelated to all psychological well-being variables apart from loneliness, to which it was negatively related in both men and women.This finding is in accordance with other psychological research findings indicating that size of support network is generally unrelated to psychological health in US and Greek samples (Holahan & Moos, 1982;Georgas & Dragona, 1988;Nezlek, 2001).
Frequency of interactions with good friends was negatively associated with loneliness and psychological malaise (GHQ scores) in women only.Women's interactions with acquaintances were positively associated with trait anxiety.It is interesting that interactions with partner were equally beneficial for the psychological health of both sexes.
Frequency of contact with the partner was negatively associated with loneliness, social and emotional loneliness, anxiety, and perceived stress in both genders but was not associated with GHQ.
Satisfaction with support, an indicator of functional social support, was negatively related to loneliness (social and emotional), trait anxiety and perceived stress in men but not in women.

Regressing well-being outcomes on structural and functional aspects of social support
In order to clarify the relationships between social support and psychological health outcomes, a series of hierarchical regression models were examined separately for males and females (table 3).
In these models, age and relationship status were entered first, followed by structural support indicators (frequency of interactions with acquaintances, friends and partner, size of supportive networks), and, last, satisfaction with social support.This order makes sense on a theoretical basis as it allows teasing out the effects of structural and functional aspects of social support.As it can be evidenced in table 3, a general pattern of gender differences in the associations between structural and functional social support and well-being outcomes prevailed.The use of hierarchical regressions allowed separating and comparing the effects of structural (second step) and functional (third step) support on well-being.As it can be evidenced in table 3 satisfaction with support predicted several aspects of men's well-being, whereas structural aspects of social support (such as frequency of interaction with acquaintances, good friends, and partner) influenced women's well-being.
In men, satisfaction with support was a significant independent predictor of perceived stress, mental health, loneliness, and emotional loneliness (and it was also highly associated with anxiety and social loneliness).In women, however, support satisfaction was an independent predictor of loneliness only.In women, frequency of interaction with good friends was an independent predictor of perceived stress, loneliness, and anxiety.In women, frequency of interactions with partner had positive effects as it was negatively associated with anxiety, perceived stress, and emotional loneliness.

Relationships among social support indicators in men and women
Since satisfaction with support is a multifaceted construct (Sarason et al., 2001), it made sense to look into how it relates to the other social relations variables.The bivariate correlations showed that in men size of supportive network was unrelated to age or relationship status, but satisfaction with support was positively related to age and marital status.In women, size of supportive network was inversely associated with age (r = -.22,p < .10), and perceived support was unrelated to age.Associations between frequency of interaction with friends and both aspects of social support were found only in women (r = .20 and r = .22,p < .05)suggesting that for women friend relationships function as sources of social support.In males, frequency of interaction with partner was positively related to satisfaction with support (r = .34,p < .01)whereas interactions with acquaintances was negatively related to support satisfaction (r = -.32,p < .01).
Two multiple regression analyses (one for each gender) were employed to help clarify the relative contributions of age, relationship status, frequency of interactions with friends and acquaintances, and the size of supportive network.For both regressions, co-linearity checks suggested that the three structural aspects of social support were to a large extent unrelated and could enter in the regression as separate predictors.In men, support satisfaction was significantly predicted from relationship status (β = .21,t = 2.11, p< .05),frequency of interaction with acquaintances (β = .28,t = 3.26, p< .01),and supportive network (β = .43,t = 5.19, p< .001).The overall model had a good fit (R 2 = .35,Model F (2 , 101) = 10.16,p < .001).In women, the overall model was significant but did not have as good an explanatory power (R 2 = .18,Model F (5 , 82) = 3.62, p < .01).The only significant predictor that emerged from this model was size of supportive network(β = .39,t = 3.26, p< .01).

Discussion
The study reported here examined gender differences in structural and functional social support indicators, and their effects on men's and women's well-being.The measurement of different aspects of social support and well-being provided evidence for gender differences in the patterning of perceived social support and their psychosocial effects in a Greek community sample.There is a scarcity of research on gender effects on social support and well-being (Matud, Ibanez, Bethencourt, Marrero, & Carballeira, 2003), and the present study demonstrated the differential effects of structural and functional aspects of social support on men's and women's well-being.
Findings on gender differences in structural and functional social support were mixed.In accordance with previous research in Greece (Georgas & Dragona, 1988) and in other cultures (Barbee et al., 1993;Pretorius, 1996), there were no significant differences in the main structural support indicator (supportive network size) or frequency of interactions with good friends.However, females reported somewhat higher satisfaction with social support (an indicator of functional support) and higher frequency of interaction with acquaintances, a finding that partly supports findings for the differential patterning of social interactions in men and women (Antonucci & Akiyama, 1987;Barbee et al., 1993).
indicators on women's well-being.Moreover, men's agentic orientation can partially account for the perceived support effects on men's well-being.Perceived support is a global indicator of psychological resources that for many, reflects intrapersonal processes (such as selfesteem, internal working model models etc. Sarason et al. 1990), and may be more relevant to the way that males are socialised to perceive supportive relationships.
Cultural considerations may add the explanatory value of the above findings.Greek men seem to rely mostly on the (emotional) support received from their spouse.Women, on the other hand, may not find the emotional support needed to sustain adequate levels of well-being in their spouses and turn to the wider social network available for support.These assumptions may be particularly true of Greece, a society that is characterised by relatively higher masculinity and lower individualism.These results supported expectations that higher levels of masculinity in the Greek society and differences in gender role distribution also result in differences in social support processes (Stevens & Westerhof, 2006a).
Moreover, evidence that being in a close relationship functions as a significant resource against social loneliness in men but not women, also supported expectations for gender-role differences in Greece.It should be noted however, that in both genders, links between different aspects of social support and well-being indicators were admittedly weak and at places inconsistent, a finding that supports arguments on the function of explicit social support in collectivistic countries (Kafetsios, 2006;Taylor et al., 2004).
The employment of a community sample allowed examining correlations between age and social relations and well-being indicators in men and women.Older men and women reported lower levels of anxiety and perceived stress, but older men also reported lower mental health.Finally, the study observed that women had elevated trait anxiety scores and perceived stress.Closer examination suggested that it was particularly the younger, female, participants who had the highest scores on the trait anxiety scale and this is consistent with previous research in Greece (e.g., Georgas & Giakoumaki, 1984).Future research should examine the evolving role of women in modern Greek society (Kourvetaris, 1999) and how this may impact on social support processes and well-being.

Limitations and future research
The results should be interpreted with the following limitations in mind.Firstly, the findings are based on cross-sectional data, thereby disqualifying any causal statements.It is possible for example that size and satisfaction with support networks can be affected by current state of psychological health (e.g., depressed persons are more likely to underreport the sources of social support).Secondly, the study is descriptive in nature and is increasingly important to research why such effects occur.Future research could examine the emotions

Table 1
Demographics, social relations and well-being in men and womenNote: Table presents marginal means and SD after controlling for the effects of age and relationship status (covariates) * p < .05** p < .01*** p < .001

Table 3
Hierarchical regressions of support indicators on well-being in men and women