How gender affects patterns of social relations and their impact on health: a comparison of one or multiple sources of support from “close persons”

Abstract

Numerous studies have reported gender differences in the effects of social relations on morbidity and mortality. When studying health and associated factors, one cannot ignore that sex differences exist and methods that are not “gender-fair” may lead to erroneous conclusions. This paper presents a critical analysis of the health/social relations association from a measurement perspective, including the definitions of people's networks and how they differ by gender. Findings from the Whitehall II Study of Civil Servants illustrate that women report more close persons in their primary networks, and are less likely to nominate their spouse as the closest person, but both men and women report the same proportion of women among their four closest persons. Women have a wider range of sources of emotional support. To date, most epidemiological studies have habitually analysed support provided by the closest person or confidant(e). We compared the health effects of social support when measured for the closest person only and when information from up to four close persons was incorporated into a weighted index. Information from up to four close persons offered a more accurate portrayal of support exchanged, and gender differences were attenuated, if not eliminated, when this support index was used to predict physical and psychological health.

Introduction

Men and women are different. When studying health, illnesses or risk and protective factors among men and women, one cannot ignore that sex differences exist. Misclassification due to measurement methods that ignore the different behaviours, psychology and physiology could lead to erroneous conclusions for men or women, or both.

This paper considers the notion of “gender-fair measures”. While we know that culture, biology and psychology interact to influence behaviour, a key issue is whether we should define the entities we study to reflect diversity, or use the same set of factors to measure the construct in both men and women. Indeed, the manifestations of appendicitis or a broken leg are inherently the same for men and women. In contrast, while a core set of signs and symptoms of depression are usually observed for depressed men and women, other expressions of depressive states are gender based.

Social roles and relationships can be thought of as primarily culturally determined. It remains unclear whether social interactions fulfil the same needs or operate in the same way for women and men. The importance of the social environment to health and well-being can be traced to ancient times in the writings of Aristotle, in the biblical prescriptions of societal rules and legislation and recommendations, and has been observed across diverse societies up to the present time. While many of these accounts study men and women separately, the epidemiological literature has not sufficiently scrutinised whether the measures of social relations are “gender free”, or “gender-fair”.

The intention of this paper is to contribute to the following debate: should we use alternative ways of defining and measuring social relations for men and women? The question is not about social relations per se; the issue is whether defining and measuring any construct should be identical for men and women, when we are already informed by other disciplines that gender differences exist. However, the concern remains that if different definitions and measures are used for men and women, this makes it impossible to measure the extent of gender differences in our results. We take a provocative stance and suggest that if gender differences are not incorporated in the construction and evaluation of some instruments, we might be falsely attributing to gender nothing more than inadequate precision and divergent validity. If, for instance, an association is found between a risk factor and health for men, but not for women, this may reflect a problem with the measurement of the risk factor in women, rather than the absence of its effect in women.

The concept of social relations is multifaceted, and includes the diverse set of interpersonal relationships and exchanges that people engage in both within and between families, friendships and “group affiliation” (Antonucci, 1994). Kahn and Antonucci (1980) deconstruct social relations into three types of support: affect, aid and affirmation. Social relations may include the degree to which an individual's need for affection, esteem or approval, belonging, identity and security are met by significant others (Kaplan, Cassel, & Gore, 1977). This concept may also embrace whether he or she is cared for or loved, that he or she is esteemed or valued, and that he or she feels they belong to a network of communication and mutual obligation (Cobb, 1976). When social relations are examined from a life course perspective, some authors (Antonucci & Jackson, 1987; Bandura, 1986) claim that social relations are effective because they help people develop a feeling of competency or personal efficacy.

The term social relations cover multiple constructs, although the simple proposal by House and Kahn (1985) to differentiate structure and function as the two essential components is both parsimonious and adequate. The structure of social relations consists of the more objective characteristics of the “social network”, i.e., its size, relative composition in terms of gender and family/friend balance, network density, frequency of contacts, marital status, etc. The function of social relations, often referred to as social support, refers to the actual or perceived type of support received, provided or exchanged.

While definitions of social support and social networks vary, the ways of measuring these constructs are even more variable (Antonucci, 1985). Some studies ask about availability of support in general terms, others ask about support from specific role-defined persons, in particular spouse, confidant(e), children, and parents, while others ask about support received with no identification of its source. It is assumed that the larger the network (i.e. the structure), the greater its potential for providing functional support. Seeman and Berkman (1988) examined this assumption in an urban community sample of older adults (>65 years old) and found that network size, number of face-to-face contacts, and number of proximal ties were associated with greater availability of both emotional and instrumental support. They also found that the presence of a confidant was associated with both emotional and instrumental support, whereas the presence of a spouse was not. Unfortunately, their results are presented adjusted for sex, but neither the independent effect of sex nor the possible interactions of sex with certain types of social ties were shown. Hence, one does not know whether the findings are comparable for men and women.

Both popular culture and empirical studies tend to share the belief that women have more extensive and better social relations than men. Women, as opposed to men, are more likely to have larger and more varied networks, as well as more likely to report having a close confidante and that the confidante is someone other than their spouse (Antonucci, 1994). Women also provide and receive more support and have a wider “net of concern” than men, that is to say, they spend more time involved in responding to requests and support from other people (Kessler, McLeod, & Wethington, 1985). Furthermore, women can more readily mobilise support when in need (Belle, 1989) and men tend to have fewer emotionally intimate relationships than women. All this may mean that women would benefit more in health terms from social support than men.

The relative contribution of social relations to health and mortality has been reported from numerous studies using different definitions of social relations, different health outcomes and diverse samples. The magnitude of the reported effects varies according to the methodologies employed; nonetheless, there appears to be a consistent association between inadequate levels of social relations and poor physical and mental health. The apparent consistency and magnitude of these findings are such, that House, Landis and Umberson (1988) proposed that “insufficient social support” should be considered an important risk factor for ill-health and mortality.

Many explanations have been proposed to describe how and why social relations impact upon health. Social support may have both direct effects on health or may buffer the negative effects of life events and chronic stressors (Cohen & Wills, 1985). Social support may act directly to increase our sense of control over the environment. In this way, it may dampen physiological arousal, strengthen immune responses, and promote healthy (or occasionally unhealthy) behaviour. In addition, social support may act indirectly, to alter the appraisal of threatening events and may provide both emotional support and tangible resources to deal with life crises.

Supportive networks are considered to provide a health benefit, and their absence to be detrimental to health, but less is known about the health consequences of negative interactions with network members. Rook (1990) has proposed three forms of negative interactions: unwanted or aversive contacts, ineffective support and social pressure to adopt or maintain unhealthy behaviours.

Many of the early studies of social relations and health were limited to men, and some to white men only (Welin et al., 1985). When analyses are performed separately by gender, research has often found a weaker or non-existent association between social support and ill health or mortality for women. Many studies show an advantage for men that is not observed for women (House, Robbins, & Metzner, 1982; Kaplan, Salonen, & Cohen, 1988; Schoenbach, Kaplan, Fredman, & Kleinbaum, 1986), although some studies report an equivalent effect of social support on mortality in both men and women (Berkman & Syme, 1979; Orth-Gomer & Johnson, 1987).

In a comprehensive review of social support and physical health, Shumaker and Hill (1991) examined the available evidence for gender differences and discussed the possible factors that may account for these observed gender differences. They contend that the associated factors are related not only to the definitions of support and health that are used, but also to the possible mechanisms linking social relations to health that may differ for men and women.

While the majority of published reports have examined the effect of social relations on physical health and mortality, an extensive literature also exists on the effects of social relations on mental health and well-being. Cross-sectional studies show a clear negative association between levels of support and psychiatric disorders in both men and women, in community and patient samples (Aneshensel & Stone, 1982; Lin, Simeone, Ensel, & Kuo, 1979; Williams, Ware, & Donald, 1981). Longitudinal studies suggest that lack of support while individuals are depressed predicts poor outcome (Brugha, Bebbington, MacCarthy, Sturt, Wykes, & Potter, 1990; Fondacaro & Moos, 1987; Paykel, 1994). Individuals with better support, or with larger networks, or who are married report better mental health and less psychological distress. The positive effect for marital status on mental well-being is stronger for men than for women. Furthermore, several authors report that network interactions are more strongly associated with women's mental health than with men's. Antonucci and Akiyama (1987) note that for both men and women, quality of social support has a stronger effect than quantity, however the magnitude of the combined effects of quality and quantity has a greater impact on women's mental health. Likewise, Kessler et al. (1985) and Schuster, Kessler and Aseltine (1990) found that depression in women was correlated with both lack of emotional support and negative interaction with partner, relatives and/or friends, while for men only negative interaction was associated with depression. Due to the cross-sectional design of most of these studies, the direction of causation cannot be inferred.

In the first analyses of social support using the Close Persons Questionnaire (Stansfeld, Fuhrer, & Shipley, 1998b; Stansfeld & Marmot, 1992) among British Civil Servants in the Whitehall II Study (Marmot et al., 1991), we found that low confiding and poor emotional support from the person identified as closest by the participant, predicted psychiatric disorder in men, though not in women. Similar gender differences were obtained for the SF-36, a measure of physical and mental functioning (Martikainen, Stansfeld, Hemingway, & Marmot, 1999; Stansfeld, Bosma, Hemingway, & Marmot, 1998a). The absence of a protective factor for women was unexpected. Confiding in the closest person without receiving accompanying emotional support conferred greater risk of psychiatric disorder in women than men, and may partly explain the observed gender difference in the effectiveness of confiding/emotional support. However, this seemed unlikely to be the full explanation.

A critical analysis of our earlier Whitehall II findings included the examination of the definition of respondents’ networks and, if and how, those differed by gender. Women report more close persons in their primary networks and are less likely to nominate their partner as their closest person (Fuhrer, Stansfeld, Chemali, & Shipley, 1999; O’Connor & Brown, 1984). Thus, women should have a wider range of sources of emotional support than men. In order to examine whether measurement of support from only the closest person or up to four close persons generates different effects on health, we conducted further analyses on the Whitehall II sample. In a first analysis, we found that by incorporating support from up to the four closest persons, low confiding/emotional support was predictive of “psychiatric disorder” in women (Fuhrer et al., 1999); this result differs from the absence of an association observed when using only information about the first close person. In contrast, one notes the consistency of the associations observed in men using either the first close person or up to four close persons. These results are concordant with the literature from social psychology delineating different patterns of social relations in men and women (Knipscheer & Antonucci, 1990).

The objective of the present paper is to examine different approaches to measuring social relations to ensure “gender-fair” comparisons when analysing the impact of social relations on health. This is accomplished by comparing gender differences in the components of the Close Persons Questionnaire when assessed for the closest person only vs. weighted scores that combine information on up to four close persons, including the spouse. We then contrast the two ways of scoring the Close Persons Questionnaire by modelling each component's ability to predict satisfaction with relationships, adjusted for age, marital status, grade of employment and the other components in the model. Results of the comparison are also examined to see whether the measures render different findings in their longitudinal relationship to two health outcomes.