| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
RESEARCH ARTICLE |
1 Santa Cruz do Sul University, Santa Cruz do Sul, Brazil.
2 Center for Policy Research, Syracuse University, New York.
Address correspondence to Marilia Ramos, Av. Independencia 2293/104, Santa Cruz do Sul, Rio Grande do Sul, Brazil CEP 96815-900. E-mail: ramos{at}unisc.br
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionReferences |
|---|
Methods. The data are based on a representative sample of people aged 60 and older from southern Brazil. The baseline sample consists of 871 subjects interviewed in 1995. Among those baseline subjects, 551 responded in 1999. Cross-sectional and longitudinal multivariate regression models estimate the effect of two dimensions of social relationships (i.e., social exchanges and social integration) on depressive symptoms, controlling for demographic, socioeconomic, and health characteristics.
Results. The results indicate that unbalanced exchange increases depressive symptoms, and social integration decreases depressive symptoms. Specifically, older Brazilians who overbenefit or underbenefit from exchanges with relatives have more depressive symptoms than those with balanced exchanges. In addition, depressive symptoms are lower when an older adult who is receiving support is able to reciprocate. More intensive exchanges with relatives, being married, and satisfaction with family relationships decrease depressive symptoms, whereas living alone increases depressive symptoms.
Discussion. The results highlight the relevance of social relationships to depressive symptoms among older Brazilians. In addition, the results indicate there is a nonlinear relationship between exchange balance and depressive symptoms in this population. The implications for policy are discussed.
Older adults represent 8% of the Brazilian population, and this segment of the population is projected to increase to 24% by the year 2050 (Veras, 1994
). Southeastern and Southern Brazil, which have higher levels of economic development, are experiencing more rapid growth among the older adult population than other regions. This raises concerns among policy makers and practitioners for several reasons.
First, the standard of living among Brazilians of all ages remains low. Two thirds of the population live below poverty and are in poor health (Kalache, Veras, & Ramos, 1987), which has implications for quality of life among older adults. Second, the family is the primary source of care but familial relations are changing because of modernization. As a result, older people may not receive the care that they expect from their families. Third, there is a lack of opportunities for older adults in Brazil to participate in social activities outside of the family sphere (Neri, 1993), which limits their participation in a broad range of social relationships. Finally, there is relatively little empirical research on older Brazilians (Veras, 1994). The 60 and older age group has only recently been included in official governmental statistics (Ramos, 1992). The few studies that exist are done by physicians and tend to rely on an epidemiological approach (Ramos, 1987; Veras, 1994). Therefore, the impact of social relationships on health among older Brazilians is not fully understood.
Literature Review: Social Relationships and Depressive Symptoms—Empirical and Theoretical Considerations
The main goal of this research is to identify the effect of social relationships on depressive symptoms among older adults in Brazil. Evidence suggests depression is widespread among older Brazilians. Among people aged 60 years and older in Brazil, 48% suffer from some form of depression (Ramos & Saad, 1990). However, relatively little is known about the social factors that influence depression in this particular population, and an examination of such constructs has rarely been conducted outside of first-world countries (Jang, Haley, Small, & Reynolds, 2000).
This study uses two major theoretical perspectives to understand the relationship between social relationships and depressive symptoms in the Brazilian context. First, this study draws from social integration theory. As noted by Pillemer, Moen, Wethington, and Glasgow (2000), social integration is not consistently defined in the literature. It is often used to refer to specific interpersonal ties, such as the number of members in a person's social network or the number of social roles held by an individual. However, it also refers to social embeddedness, which is indicated by broader social ties such as marriage and contact with friends or family (Barrera, 1986). These ties facilitate social integration by providing opportunities to participate in social life and exchange social support. These forms of social integration are important to consider in this research because (a) a lack of social integration robs life of meaning, thereby generating stress and depression; (b) social integration enhances access to coping resources; and (c) social integration provides a context within which supportive relationships can develop outside the family (George, 1996).
Therefore, social integration should be negatively related to depressive symptoms in later life. This hypothesis has been supported in previous research. For example, research in the United States and other developed countries suggests that social relationships are associated with psychological well-being in later life, particularly after the onset of health conditions and functional limitations (Antonucci, 1990; House, 1981; George, 1996). Social relationships moderate distress for people facing health problems, the death of a spouse, or financial strain (Silverstein & Bengtson, 1994). In addition, in the presence of social support, older adults feel loved, are better able to deal with health problems, and have high self-esteem (Cicirelli, 1990).
This study also draws on equity theory, which is derived from social exchange theory (Blau, 1964; Dowd, 1975). This theoretical perspective focuses on the equity of exchanges among social actors (Walster, Walster, & Berscheid, 1978). Previous research on older adults in the United States indicates that receiving assistance is positively related to depressive symptoms, whereas giving assistance is negatively related to depressive symptoms (Liang, Krause, & Bennett, 2001). However, it is not clear how the balance of exchanges influences depressive symptoms because the relationship could take one of three different forms.
First, there could be a negative relationship between unbalanced exchanges and depressive symptoms. Neri (1993) suggested that being overbenefited (i.e., receiving more than one gives) can be positive, in utilitarian terms. Older adults may feel that the young have to reciprocate previous help given to them. Under these conditions, in comparison with older adults whose exchanges are balanced, overbenefited older adults should have fewer depressive symptoms and underbenefited older adults should have more depressive symptoms. Second, there could be a positive relationship between unbalanced exchanges and depressive symptoms. Some have noted that being overbenefited can contribute to a perceived lack of autonomy and feelings of dependence, which can lead to dissatisfaction, low morale, and depressive symptoms (Cicirelli, 1990; Stoller, 1985). In this circumstance, in comparison to having balanced exchanges, being overbenefited would increase depressive symptoms whereas being underbenefited would decrease depressive symptoms. Finally, there could be a nonlinear relationship between unbalanced exchanges and depressive symptoms (McCulloch, 1990; Rook, 1987). From the perspective of equity theory, the most important aspect of an exchange relationship is the balance of exchanges between social actors. Any type of exchange balance, whether it underbenefits or overbenefits the older adult, would be problematic. Thus, equity theory would predict more depressive symptoms among older adults who underbenefit and overbenefit, relative to those to have balanced exchanges.
Although the literature contains numerous studies that examine reciprocity and various measures of mental well-being (e.g., Davey & Eggebeen, 1998; Ingersoll-Dayton & Antonucci, 1988; McCulloch, 1990; Stoller, 1985), we are aware of only two studies that explicitly test the relationship between exchange balance and depressive symptoms (Lee & Netzer, 1995; Liang et al., 2001). The findings of those studies are consistent with the second explanation—overbenefiting is related to more depressive symptoms, whereas underbenefiting is related to fewer depressive symptoms. However, because both studies are based on samples from the United States, it is important to test these relationships with data from a different cultural context.
In addition, these two studies provide conflicting evidence regarding the mediating role of health status. Functional limitations often reduce opportunities for balanced exchanges within the family sphere (Dwyer, 1994). Consequently, these differences in depressive symptoms may be due to the relatively poorer health and functioning of those who overbenefit from exchanges. Lee and Netzer (1995) found that the relationship between exchange balance and depressive symptoms was not significant after self-rated health was controlled for, but Liang, Krause, and Bennett (2001) found that the relationship continued to be significant even though poor health was controlled. Our analysis will provide additional evidence regarding the nature of the relationship between exchange balance and depressive symptoms, controlling for health conditions and functional limitations.
On the basis of the two theoretical perspectives and the evidence provided by previous research, it is expected that social relationships significantly influence depressive symptoms among older Brazilians. In particular, it is hypothesized that, when demographic characteristics, socioeconomic status, and health characteristics are controlled, (a) unbalanced exchange increases depressive symptoms and (b) social integration reduces depressive symptoms.
|
METHODS |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
).
A sample of 10 counties was randomly selected, from a total of 36 counties in the Central Eastern region of that State, based on strata according to the size of population and economic activity. Fifteen percent of the census sectors in each county were selected for sampling. A minimum of eight census sectors were used for each county to capture the variability of every county, and then eight cases (households) were selected from each sector. One person aged 60 or older was interviewed per household. This yielded a baseline sample in 1995 of 871 subjects. The data collection in 1995 involved a closed questionnaire with 121 items. The baseline respondents were reinterviewed in 1999 with an expanded questionnaire. Among the 871 subjects in the baseline sample, 551 (63.4%) responded in 1999. Among those subjects who dropped out, 118 died, 51 refused, 42 were in nursing homes or hospitals or were ill or unable to hear, and 109 were not located. The multivariate analysis includes a selection bias term that controls for selection out of the sample (Heckman, 1979). This term was created in LIMDEP (which is a statistical package that models limited dependent variables) by estimating a probit model predicting selection out of the sample as a function of age, home ownership, marital status, and number of physical limitations. Table 1 presents the descriptive characteristics of the sample. The data are consistent with population data from this region, suggesting that the sample is representative.
|
). The dependent variable is an additive scale that ranges from 0 to 12, with higher numbers indicating more depressive symptoms. The mean level of depressive symptoms was 2.64 in 1995 and 2.72 in 1999 (
and.80, respectively).
We included questions from the Center for Epidemiological Studies Depression Scale (CES-D) on the 1999 questionnaire to examine the comparability of the scales (Radloff, 1977). The CES-D scale had a similar level of reliability (
). Furthermore, the correlation between the scale used in this research and the CES-D scale is.83. Therefore, our scale appears to be adequately measuring depressive symptoms in the cultural context of Southern Brazil.
Independent Variables: Baseline Social Relationships
This study measures two dimensions of social relationships: social exchanges and social integration. Social exchanges are measured by the frequency of help received from or given to relatives in three areas: emotional, financial, and instrumental. Exchange balance, which measures the concept of reciprocity, is based on two variables: help received and help given.
The construction of each type of help is explained to facilitate the understanding of the overall measures of help received and help given. Emotional help involves a question about the frequency of receiving (or giving) emotional help, such as advice. The answer options range from 0 = never to 4 = daily. Financial help is based on a question about the frequency of receiving or giving financial help (money), with the same answer options (0 = never to 4 = daily).
Unlike the other types of help, instrumental help is an additive index, which asks how frequently the subject receives from or gives to relatives the following: (a) clothes, (b) housing, (c) food, and (d) medicine or personal care. The frequency with which the subject receives (or gives) these four types of instrumental help is recoded as 0 = never to 4 = daily. Thus, the maximum score for total instrumental help received is 16 (i.e., four items times four frequency levels). The measure of instrumental help given contains one additional item: babysitting. This makes the maximum instrumental help given score equal to 20 (i.e., five items times four frequency levels). Both instrumental exchange scales are divided by the number of items in the scale to constrain their range from 0 to 4, which is comparable with the range of the emotional and financial exchange measures. For all of the exchange scales, higher numbers indicate that more help is being given or received.
The total help given variable is the sum of the emotional (0–4), financial (0–4) and instrumental help given (0–4) scales. It ranges from 0 to 12, with 0 indicating no help given in any of the three areas and 12 indicating daily help given in all three areas. Total help received is measured the same way as total help given. The main effects of total help received and total help given, along with the interaction effect between help received and help given, will be tested in the multivariate analysis. The interaction term will provide a preliminary test of the hypothesis about the relationship between exchange balances and depressive symptoms. A significant interaction will indicate that the effect of help received on depressive symptoms varies by levels of help given.
Similar to other authors (Rook, 1987; Wilmoth, 2000), we measure exchange balance by subtracting help given from help received. Although this scale theoretically ranges from –4 to +4, the actual range observed in this sample is –3 to +3. Negative scores indicate that the subjects are underbenefited, positive scores indicate that the subjects are overbenefited, and scores at or near zero indicate equitable exchange or no exchanges. Rook (1987) suggested squaring the exchange balance variable because the relationship between social exchanges and health is potentially nonlinear. We follow this suggestion by including exchange balance and exchange balance squared in the multivariate model.
The overall intensity of exchanges, which is an exchange-based measure of social integration, is the sum of all items exchanged (total help given plus total help received) and ranges from 0 to 24. This variable is expected to have a negative impact on depressive symptoms, because people with higher values are more socially integrated and therefore should be less depressed. Exchange intensity will not be included in the model that contains exchange balance, because both are a function of help given and help received.
Social integration in later life is manifested in a variety of indicators (Pillemer & Glasgow, 2000). This study includes four additional measures of social integration: contact with friends (1 = yes), contact with relatives (1 = yes), living arrangements (with spouse only, alone, with spouse and children, with children, and with others), and marital status (1 = currently married).
Finally, the analysis includes a measure of whether the respondent is satisfied with his or her family relationships (1 = satisfied). This measure will control for the respondent's subjective assessment of his or her social relationships. Although it would be desirable to include measures of satisfaction with nonfamily relationships and perceived adequacy of social support, it is not possible to do so because such questions were not included in the survey.
Baseline Control Variables
Several baseline variables are controlled because of their relationship with depressive symptoms. Age is measured by the number of years since birth. Gender is a dichotomous variable (1 = female). Race is dichotomized (0 = White). Socioeconomic status is measured with three separate indicators. Education contains four categories: did not complete elementary school, completed elementary school (reference), some high school, and completed high school. Occupation contains four categories: no waged occupation (reference category, which includes those who never or presently do not work), low (e.g., blue-collar workers, domestic workers, and small farmers), medium (e.g., clerical workers and public sector workers), and high (e.g., professionals and owners of larger businesses). Family income is measured in Reals (i.e., Brazilian currency) received per month in the household divided by the number of household members.
Another important control variable is risk behaviors that potentially have negative consequences for mental health (Idler & Angel, 1990). Risk behaviors are measured through six dichotomous items that measure current smoking and high-risk behavior related to alcohol consumption (e.g., family or friends indicate that the respondent's drinking was a problem, or that drinking interfered with work). These items are used to create an additive scale, ranging from 0 to 6, that measures the total number of risk behaviors.
The morbidity indicators were collected at baseline through the following: "Did you have (specification of the problem) in treatment in the past 6 months (0 = no and 1 = yes)?" The 13 indicators of morbidity are divided into two additive indicators to clarify effects of general types of health problems on depressive symptoms (Ferraro & Wilmoth, 2000). The chronic conditions scale includes rheumatism, bronchitis, kidney disorder, varicose veins, back problems, ulcer or gastritis problems, urinary problems, and skin problems. The serious conditions scale includes more life-threatening conditions: high blood pressure, cancer, diabetes, stroke, and heart problems.
The physical limitation measure is based on seven questions. The first six questions refer to help needed in the following areas: cooking, cleaning house, taking medicine, taking care of personal hygiene, eating, and moving (0 = no and 1 = yes). The seventh question asks whether the respondent walks without problems, even on stairs (0 = without problems and 1 = with some problem). The individual items were summed to create a scale that ranges from 0 to 7, with higher numbers indicating more physical limitations.
Because of the limited number of missing cases for the variables in the analysis, mean substitution is used to retain missing cases. For example, the variable with the highest number of missing cases is income (n = 10).
|
RESULTS |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
|
|
Cross-Sectional Analysis
Table 3 presents three cross-sectional models that predict depressive symptoms in 1995 as a function of baseline social relationships (i.e., social exchange and social integration) and the control variables. Each model measures social exchange differently.
|
It is interesting to note that, in all three models, living arrangements and marital status also influence depressive symptoms. Being married is negatively related to depressive symptoms, whereas not living with a spouse substantially increases depressive symptoms. In addition, satisfaction with family relationships is negatively related to depressive symptoms. The control variables indicate that women tend to be more depressed than men, and that chronic conditions, risk behaviors, and physical limitations increase depressive symptoms.
Longitudinal Analysis
A regressor variable approach (i.e., residualized regression) is used to predict changes in depressive symptoms between 1995 and 1999. In contrast to change score models (in which the dependent variable is 
), these models predict Y2 as a function of Y1 and X (Allison, 1990). Therefore, these models estimate 1999 depressive symptoms as a function of 1995 depressive symptoms, social relationships, and the control variables.
Similar to the cross-sectional analysis, three separate models are estimated that contain different exchange measures. The significant control variables in the longitudinal models are generally consistent with the cross-sectional results. The longitudinal models also indicate that baseline depressive symptoms have the greatest impact on change in depressive symptoms. Respondents with greater depressive symptoms at baseline are more likely to experience an increase in depressive symptoms over time.
The social exchange effects in these longitudinal models are similar to those of the cross-sectional model. In Model 1, the interaction between received help and given help is significant and negative, indicating that the change in depressive symptoms is smaller when those who receive assistance are able to reciprocate. The relationship between exchange balance and depressive symptoms continues to be nonlinear in Model 2. This indicates that respondents who overbenefit and underbenefit experience greater increases in depressive symptoms than those with more balanced exchanges. Model 3 suggests that changes in depressive symptoms are smaller among older adults who are more involved in exchanges.
Social integration tends to suppress increases in depressive symptoms. Older Brazilians who are married have smaller changes in depressive symptoms than those who are unmarried, whereas those who are living alone experience greater increases in depressive symptoms than those who are living with a spouse. In addition, satisfaction with family relationships is negatively related to changes in depressive symptoms. Similar to the cross-sectional results, women and those in poor health have more substantial increases in depressive symptoms. In fact, other than baseline depressive symptoms, health characteristics have the greatest standardized effect in these models.
|
DISCUSSION |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
; Liang et al., 2001), this research found a nonlinear relationship in which any type of unbalanced exchange was related to greater depressive symptoms. Thus, older Brazilians who overbenefit and underbenefit from exchanges with relatives have more depressive symptoms at one point in time and greater increases in depressive symptoms over time than older Brazilians who have more balanced exchanges. This relationship between unbalanced exchanges and depressive symptoms persists after health conditions and functional limitations are controlled.
This finding is consistent with the predictions of equity theory and reinforces the idea that mental health is negatively affected by a lack of interdependence among family members (Stoller, 1985; Ramos, 1992). The interaction term (i.e., Total Help Given x Total Help Received) indicated that the effect of help given on depressive symptoms varies according to levels of help received. The results also showed that exchange intensity is negatively related to depressive symptoms. Together these results point to the potential of frequent two-way exchanges between older adults and their relatives for reducing depressive symptoms among the older population in developed regions of Brazil.
An unexpected finding was that contact with friends and with relatives did not have a significant effect on depressive symptoms cross-sectionally or longitudinally. This suggests that the simple presence of social contacts, without exchanges, does not lower depressive symptoms. This may be due to cultural expectations regarding exchanges. Exchanges with family may be expected because of the obligatory character of family relationships (George, 1996; Neri, 1993; Rook, 1987).
In support of social integration theory, the models indicate that exchange intensity (which is an exchange-based measure of social integration), living arrangements, and marital status were consistently related to depressive symptoms at one point in time and changes in depressive symptoms over time. Therefore, both exchange balance and social integration are important to the mental health of older adults. In addition, older Brazilians' subjective assessments of their relationships with family members were also related to depressive symptoms. This demonstrates the importance of taking into account objective and subjective aspects of social relationships when studying mental health outcomes in later life.
It is important to acknowledge the potential for a reciprocal relationship between social relationships and depressive symptoms. A supplementary LISREL analysis (not shown but available from M. Ramos upon request) suggests there is not a reciprocal relationship between exchange balance and depressive symptoms or between exchange intensity and depressive symptoms. There is a reciprocal relationship between satisfaction with family relationships and depressive symptoms. However, these supplementary findings should be interpreted with caution. Given that there are only two waves of data, it is not possible to make a definitive statement about reciprocal effects (Bollen, 1989; Cliff, 1983). Therefore, subsequent analysis should consider exploring reciprocal relationships between social exchange, social integration, and depressive symptoms over three or more time periods.
Before we discuss the limitations, it is important for us to reinforce that this research makes a contribution to the literature by using data from a developing country to estimate longitudinal models that take into account selection bias. Although it includes various measures of social exchange, this research is unable to address the complexity of social exchange. The baseline questionnaire does not specify the relatives with whom the older adults were interacting or exchanging (e.g., children or siblings). It also did not contain detailed information about the children with whom older adults exchange (e.g., sons or daughters, or geographic proximity of children) or the degree to which social networks are concentrated. This research did not include exchanges from friends because of data limitations that are caused by the relative rarity of friend exchanges among older Brazilians. Consequently, this research could not provide insight into the different effects of exchanging with children, extended family, friends, and neighbors.
In addition, other than the single measure of satisfaction with family relationships, the data did not include questions about negative interactions or the perceived adequacy of support. Previous research has found that older adults' perceived adequacy of support, especially the adequacy of emotional support, is negatively related to depressive symptoms (Oxman, Berkman, Kasl, Freeman, & Barrett, 1992). Furthermore, satisfaction with social support is negatively influenced by the presence of negative interactions and positively influenced by reciprocity (Krause, 1995). Having more detailed information about social relationships would enable researchers to clarify the connections between exchange balance, negative interactions, satisfaction with social support, and depression outcomes. If possible, researchers should consider pursuing cross-cultural comparisons while considering cross-cultural differences in the reporting of depressive symptoms (Mui, Burnette, & Chen, 2001; Su & Ferraro, 1997).
Finally, it is important to emphasize the contribution this research makes to policy. There is an implied emphasis on the frailty among older adults in certain policy strategies. However, the results confirm that well-being in later life is not simply related to health status. Social relationships in general, and exchange balance in particular, also have an important effect on well-being among older adults. Despite this fact, policy often does not take into account the importance of facilitating balanced exchanges. This research suggests that equitable exchanges potentially prevent older adults from feeling as if they are a burden or being exploited. This type of exchange pattern is most likely related to health outcomes in cultural contexts where utilitarian values are prevalent, such as the more developed regions of the Western societies and the region of Brazil under consideration in this research.
Positive health outcomes are also facilitated by social integration. Strong ties to social networks, by means of extensive exchange, living arrangements, and marriage, are an essential part of mental well-being among older adults. This finding is particularly important given that older Brazilians do not participate extensively in social activities outside of the family because there are limited options for social activities in the broader community. Public policy that encourages social embeddedness and supports a wider range of social activities for older adults could have a positive impact on the mental health of older Brazilians, particularly for those who are isolated because they live alone, do not have proximate kin networks, have low socioeconomic status, or are in poor health.
Overall, the results suggest that programs for older adults should focus on promoting social activities rather than solitary ones. Policies should promote a sense of usefulness among older adults, for example through voluntary work in public day care centers or hospitals and nursing homes. These policies can make older adults feel useful in a cultural context in which utilitarian values are common. These policies can also help others who have needs, such as disadvantaged children or older adults with limited family support. Therefore, such policies can enhance the well-being of the older adult population while addressing the needs of other segments of the population.
|
|
Acknowledgments |
|---|
We thank Ken Ferraro, Harry Potter, and Debra Street for comments on earlier versions of this research; Aaron Sayegh and Jessica Kelley Moore for methodological assistance; and Carmen Moraes, Clarice Scherer, Fernanda Biachi, Gilberto Iser, and Vilson Borba for their assistance during the fieldwork. Finally, we thank Tatiana Spiazzi and Elizane Wehner for their work as research assistants at the Santa Cruz do Sul University.
|
Footnotes |
|---|
Received for publication July 12, 2002. Accepted for publication January 8, 2003.
|
References |
|---|
|
TOPAbstractMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  J. M. Perkins, K. S. Multhaup, H. W. Perkins, and C. Barton Self-Efficacy and Participation in Physical and Social Activity Among Older Adults in Spain and the United States Gerontologist, February 1, 2008; 48(1): 51 - 58. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
