Social Capital and Health: Maximizing the Benefits

Social Networks and Vulnerable Populations: Findings from the GSS

Solange van Kemenade, Ph.D., Policy Research Division, Strategic Policy Directorate, Public Health Agency of Canada; Jean-François Roy, Doctoral Candidate, Institute of Population Health, University of Ottawa; and Louise Bouchard, Ph.D., Researcher, Institute of Population Health, University of Ottawa

Continuing their research on social capital and the health of Canadians, the research team from the Public Health Agency of Canada and the University of Ottawa presents results from a further analysis of data from the 2003 General Social Survey (GSS), Cycle 17 on Social Engagement. This new research examines which aspects of social capital are important to the health of selected Canadian population groups--seniors, immigrants and members of low-income households--and examines whether the relationship between social capital and health varies with gender.

The relationship between social capital and the health of Canadians was briefly described in an earlier article. The research team applied the knowledge gained from their previous analysis to more specific analyses of the data for three subpopulations: seniors, immigrants and members of low-income households. These subpopulations are vulnerable groups that, depending on their individual and collective experiences, can experience a "disaffiliation" (i.e., a partial or complete rupture of their social links).¹ This results in a depletion of social capital stock whereby individuals that have been cut off from their social networks can no longer benefit from the resources available to those who belong to such networks. The authors set out to learn more about the links between health and various aspects of social capital in these subpopulations. The results of these analyses help identify what types of support are most beneficial to the health of Canadians.

Methodology

The GSS sample consists of a total of 24,951 individuals aged 15 years and older from all 10 Canadian provinces. Only respondents aged 25 years of age and up were included in the analysis, for a total of 21,785 in the general population, including: 4,486 seniors (those aged 65 years and older); 4,109 immigrants (those born outside of Canada, excluding those whose parents are Canadian citizens); and 3,548 members of low-income households (individuals who live in households where the total household income is below the low-income cut-off, taking into account both size of household and whether they live in an urban or rural community).

Adults under age 25 years were excluded because the researchers considered that social networks accessible to this age group are different. The resources in these networks are used quite differently than is the case for adults 25 years and over.

The operational model of social capital described in the article on page 13 was used to verify the presence of relationships between perceived health and social capital for individual Canadians. To this end, data were analyzed using logistic regression analysis models for the Canadian population and for the three subpopulations. In all analyses the effects of sociodemographic characteristics of respondents (sex, age, education, professional situation, marital status, type of household) were controlled. Given the sociodemographic circumstances of seniors, the categorization of certain variables was different in this subpopulation, specifically for age, professional situation and type of household. Results were weighted using the "bootstrap" method to control for the complex survey design (see Using Canada's Health Data on page 37).

What the Findings Show

The results of the statistical analyses are presented in Table 1 in the form of score ratios for each of the social capital indicators in the model. (See previous article for a definition of all social capital indicators.) Score ratios are provided for men and women in the general population and for each of the subpopulations studied.

Reading the Score Ratio Table

The score ratio indicates the ratio between the probability of an event--in this case, good health--in one group, to its probability in another group. A number greater than one indicates a positive association, while a number less than one indicates a negative association. Take, for example, the reciprocity index for the general population. The score ratio tells us that the respondents who responded yes, when asked if they have at least one reciprocal assistance relationship in their social network, are 1.317 times more likely to report being in good health than are those who do not have a reciprocal assistance relationship. A p-value (probability value) of less than or equal to 0.05 indicates that the result is statistically significant.

Across the Indicators . . .

A review of Table 1 reveals the following overall findings:

• In general, the size of a network of strong ties is positively associated with health in the general population, as well as in all subpopulations studied. However, the relationship is strongest for women in the general population, for immigrants and for men living in low-income households. The most notable difference appears to be for seniors, where the relationship between perceived health and the size of network of strong ties is significant only among women.
• There is a positive relationship between network ties with organizations (consisting of two or more ties) and health in the general population as well as in all the groups analyzed--except for the low-income group, where the relationship is not significant. Networks of links with organizations demonstrate the strongest associations with health for men in the general population and for immigrant women and senior men.
• Volunteerism is positively associated with health in the general population, those in low-income households and immigrant men.
• There is also a positive relationship between the reciprocity index and the health of women and men in the general population, as well as for immigrant women.
• Finally, the results indicate that there is a negative relationship between the social support index and health for the general population, and for immigrant women. In the same way, there is a negative association between the instrumental support index and the health of senior men. These results are possibly explained by a problem with the measurement of social support in the GSS data (see "Limitations of the Analysis," later in this article).

The analyses of the GSS data provide a wealth of information about the relationship between social capital and health. Nevertheless, in the context of public policy, follow-up work is needed to further explain the findings and to confirm them through analysis of other data banks.

Table 1: Self-Reported Health and Social Capital-Score Ratios for the General Population and Selected Subpopulations

	Score Ratios for the General Population and Selected Subpopulations
Social Capital Indicators	General Population			Seniors (65 Years and Older)			Immigrants			Members of Low-Income Households
	Total	Wo- men	Men	Total	Wo- men	Men	Total	Wo- men	Men	Total	Wo- men	Men
Size of Network of Strong Ties Outside the Household
Small -- 0 to 11	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000
Medium -- 12 to 23	1.469 ***	1.596 ***	1.326 *	1.443 **	1.590 **	1.225	1.694 ***	1.506 *	1.952 **	1.465 **	1.500 **	1.542 *
Large -- 24 to 35	1.795 ***	1.875 ***	1.730 ***	1.663 ***	1.880 ***	1.335	2.036 ***	2.182 **	1.926 *	1.885 ***	1.748 **	2.447 **
Very large -- 36 and up	1.457 ***	1.537 ***	1.360 **	1.270	1.730 *	0.920	1.931 ***	1.691	2.347 **	1.788 ***	1.555 *	2.498 ***
Size of Network of Organizations
0	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000
1	1.085	1.074	1.109	1.272 *	1.175	1.384	1.290	1.193	1.615 *	0.966	0.949	0.971
2 and up	1.601 ***	1.524 ***	1.704 ***	1.776 ***	1.665 **	1.878 **	1.698 **	1.908 *	1.700 *	1.185	1.172	1.095
Reciprocity
No	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000
Yes	1.317 ***	1.356 ***	1.236 *	1.190	1.300	1.043	1.277	1.904 **	0.762	1.268	1.350	1.109
Volunteerism
No	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000
Yes	1.247 **	1.230 *	1.263 *	1.243	1.337	1.151	1.247	0.812	2.361 **	1.524 ***	1.546 **	1.640 *
Total social support index	0.787 ***	0.757 ***	0.835 **	0.919	0.852	1.043	0.781 *	0.639 ***	1.051	0.873	0.904	0.805
Instrumental support index	1.105	1.097	1.101	0.783*	0.864	0.648*	0.957	1.134	0.721	0.993	0.886	1.233

Note: Data tables detailing confidence intervals and sociodemographic variables may be obtained by contacting the Policy Research Communications Unit.

* p <= 0.050 ** p <= 0.010 *** p <= 0.001

Source: GSS, Cycle 17.

A Closer Look at Vulnerable Populations

Seniors

Recent studies (see article on page 21) show that three types of networks are important for people aged 65 years and older: social networks, support networks and care networks.² The composition of the networks evolves and adapts at the same time as an older person's physical condition changes. While initially seniors' networks are large and diversified,² smaller care networks appear to become increasingly important as seniors develop functional limitations.

Analysis of the GSS data confirms the important relationship between social capital and seniors' health. Results show that there is a positive relationship between the size of an older person's network of strong ties and health--but only for women. Senior women who belong to a medium or large network of strong ties are more likely to report good health than those whose network size is smaller. At the same time, it is among seniors that the relationship between health and very large networks of strong ties is the weakest. This supports the thesis that small, close networks are more important for seniors needing care.

With respect to networks of organizations, results point to a positive relationship between the size of these networks and the health status of seniors; the relationship is stronger than that observed among the other subpopulations and the general population. In other words, seniors who are involved in one or more organizations are more likely to report good health than those who are not as involved.

The only group for which the instrumental support index is associated with health is seniors, and the association is negative. Seniors who had received help carrying out day-to-day activities are more likely to report poor health than those who had not had such support. This is likely due to the fact that, in the GSS data, instrumental support becomes an indicator of seniors' limited capacity to carry out activities. Secondary analyses confirm this hypothesis. So, when seniors who report activity limitations are excluded from the analysis, the relationship between instrumental support and perceived health is negligible.

Immigrants

Research supports the importance of social capital to the integration of immigrants. ^3,4,5,6 Having access to close networks of people of the same cultural origin--as well as to programs that support these networks--is associated with the social and economic integration of immigrants in the host country and with their well-being. Networks of friends and family for new immigrants in Canada represent an extremely important support, whether for finding accommodation, training or education, employment or aid.⁶ The most important sources of assistance for immigrants, according to the results of the Longitudinal Survey of Immigrants to Canada (LSIC), are parents or family members (already established in Canada), friends, organizations working with immigrants, educational institutions and health workers.

Results of current analysis of the GSS support the above findings that social capital is a major determinant of health for immigrants to Canada, and reinforce results of the LSIC:

• There is a positive association between the size of networks of strong ties and reported good health among immigrants. The association is more pronounced for immigrants than for any of the other groups studied, as well as for the Canadian population overall.
• There is also a positive association between the number of ties with organizations and immigrants' self-reported health. Immigrants with a high number of ties to organizations perceive their health to be good--a trend that is in keeping with results for the Canadian population overall.
• Except for the general population, immigrant women are the only group for which the results of the analysis indicate a significant relationship between reciprocity in social networks and perceived health. Immigrant women who say they had at least one reciprocal support relationship within their social networks were more likely to say they are in good health than their peers without such a relationship.
• The results show that for immigrant men, volunteerism and perceived health are strongly linked. Immigrant men who volunteered in the year preceding the survey are more than twice as likely to say they are in good health as their peers who had not participated in volunteer activity.
• The situation is different for the social support index and self-reported health. Immigrants who say they had received at least one form of social support in the year preceding the survey are more likely to report poor health. This result is likely explained by a limitation in measurement of social support inherent in the GSS data which is discussed in greater detail at the end of this article. This negative relationship holds for the overall Canadian population as well.

Members of Low-Income Households

Some researchers hypothesize that poverty and increasing social inequality engender negative stress which, in turn, can have a negative impact on the physical and emotional health of individuals.^7,8 Recent studies show that when solutions and the ability to adapt to stress are limited, people become more vulnerable to a whole range of diseases that affect the immune system and the hormonal system.⁸ In this context, as in many other difficult life circumstances, networks can serve as moderators in attenuating adverse living conditions, helping people to remain healthy or to increase their resilience.⁹

Results of the GSS analysis feature the following key points:

• For members of low-income households, there is a positive association between perceived good health and the size of networks of strong ties. Those with medium or large networks are more likely than those with smaller networks to report good health. The relationship is stronger for men than for women in this group.
• Counter to the general trend, the perceived health of people living in low-income households is not related to networks of ties with organizations.
• The positive relationship between volunteering and perceived health is significantly stronger for those living in low-income households than is the case for the general population or all other groups studied. In fact, only among immigrant men is there a stronger association between volunteer participation and health.

Limitations of the Analysis

Even though the findings from the analysis of the GSS data do not allow for the establishment of causality, the current study demonstrates that social capital indicators that are closest to a network approach (i.e., networks of strong ties and networks of ties with organizations) are significantly linked to the perceived health status of Canadians. At the same time, the data available in the GSS present some important challenges.

The first is a problem often encountered by researchers when they analyze secondary data (that they did not collect): how to operationalize a model that relies on a theoretical framework that is different from the one used to guide the development of the database used. In general, researchers faced with this challenge have access to fewer data on which to carry out their analysis.¹⁰ This was true in the current case, where the adoption of a theoretical framework on social capital based on the network approach limited the choice of indicators of social capital available for analysis. While the resulting analyses have clearly produced interesting results that merit attention, the data available in the GSS are not sufficient to complete a full analysis of social capital based on a network approach.

Useability of GSS data posed challenges as well. The GSS , Cycle 17 does not use the more complex social support measurement tools usually used in large databases, such as that of the National Population Health Survey (NPHS), which include measures of respondents' perception of the availability of social support resources. Instead, the GSS measures social support with six variables that identify respondents who have used a form of social support in the year preceding the survey. Since it measures use of social support instead of the perception of its availability, the social support index derived from the GSS data becomes, by definition, an indicator of poor health. The integration of more sophisticated tools to measure social capital in the large databases (such as Resource Generator)¹¹ would be helpful in informing public policy.

Another limitation of the findings is related to use of the indicator of "perceived health." While it is a reliable indicator of mortality,^12,13,14 the variable is not as objective as a composite variable of health, such as the Health Utility Index (HUI).

Next Steps . . . What the Findings Mean

The results of these analyses provide a first empirical pan-Canadian snapshot of the relationships between social capital--as operationalized in the model described here--and the health of women and men in three vulnerable population groups.

Even though these analyses offer much food for thought, more research is required to explore and better understand the meaning behind the results. For example, why are older men the only group of men for which networks of strong ties are not significantly related to perceived health? Why are networks of ties to organizations most important to the self-reported health of seniors and immigrants? What is the nature of the relationship between volunteerism and good health?

The next few articles take a closer look at the role of social capital in the context of specific vulnerable populations, and how social capital research is informing the development of policies and programs.

Researchers may be contacted through <bulletininfo@hc-sc.gc.ca>.