| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
RESEARCH ARTICLE |
1 Center for Policy Research, Syracuse University, New York.
2 Department of Sociology and Anthropology, Purdue University, West Lafayette, Indiana.
Address correspondence to Janet Wilmoth, Center for Policy Research, Syracuse University, 426 Eggers Hall, Syracuse, NY 13244-1020. E-mail: jwilmoth{at}maxwell.syr.edu
 |
Abstract |
|---|
 TOP Abstract MethodsResultsDiscussionREFERENCES |
|---|
Methods. The data come from the baseline and first 2-year follow-up of the Health and Retirement Study, which were collected in 1992 and 1994. The analysis is based on 6,391 primary respondents who were aged 51 to 61 at the baseline. Descriptive statistics, cross-sectional ordinary least squares regression models, and longitudinal residualized regression models are estimated for the entire sample and by immigrant status.
Results. Living arrangements and immigrant status interact to influence depressive symptoms. The results confirm that depressive symptoms are higher among those who live alone, particularly among immigrants. Living with family or others is related to higher cross-sectional levels of depressive symptoms, especially for immigrants, and greater longitudinal increases in depressive symptoms among nonimmigrants.
Discussion. The results highlight the important influence of social integration on mental health while demonstrating that context shapes the effect of social integration. They suggest that interventions should promote social integration, particularly among older adults living alone or with family or others. However, those programs should be sensitive to the unique needs of native-born and immigrant populations.
QUALITY of life among older Americans has been a long-standing concern of gerontologists. Extant literature indicates that social integration influences physical and mental health outcomes in later life. This research, which draws from social integration theory, examines how one form of social integration (i.e., living arrangements) interacts with one dimension of social context (i.e., immigrant status) to influence a specific mental health outcome in later life (i.e., depressive symptoms).
Depression is common, but often undiagnosed and undertreated, among older adults (Koenig, 1999
; National Institute on Mental Health [NIMH], 1999). Previous research suggests a connection between depressive symptoms and living arrangements. Older adults living alone have lower levels of morale and higher depressive symptoms, mental health service use, and suicide risk (Dean, Kolody, Wood, & Matt, 1992; Florio et al., 1997; Mindel & Wright, 1982). Previous research also has linked depressive symptoms to immigrant status. Immigrants may be at a higher risk of depressive symptoms than native-born populations as a result of the stress associated with immigration and acculturation (Black, Markides, & Miller, 1998; Krause & Goldenhar, 1992; Mui, 1996, 1998; Ying, 1988).
A limited number of studies provide insight into the relationship between living arrangements and depressive symptoms for specific immigrant groups. For example, Mui (1998) examined Chinese immigrants aged 60 and older living in a Northeast metropolitan area; Tran, Khatutsky, Aroian, Balsam, and Conway (2000) focused on a group of Russian-speaking immigrants who had a mean age of 73.2 and were living in Boston. However, we do not have a clear understanding of how the relationship between living arrangements and depressive symptoms may differ between nonimmigrants and immigrants in the general population. In addition, because the majority of the research on this topic relies on participants who are aged 65 and older, we do not know how living arrangements affect depressive symptoms among adults who are making the transition from middle age into later life. Finally, relatively little is known about how living arrangements and immigrant status influence changes in depressive symptoms over time because most studies are cross-sectional.
This research examines the relationship among immigrant status, living arrangements, and depressive symptoms with a sample of middle-aged and older adults from the Health and Retirement Study. Cross-sectional and longitudinal models are used to address the following research questions: First, do living arrangements affect changes in depressive symptoms among immigrants and nonimmigrants? Second, do particular living arrangements (e.g., living alone) increase the risk of depressive symptoms among immigrants more than nonimmigrants?
Literature Review: Social Integration and Mental Health
This research draws primarily on social integration theory, which highlights the importance of social roles and attachments to well-being (Pillemer, Moen, Wethington, & Glasgow, 2000). Social integration offers opportunities to develop supportive relationships, increases access to coping resources, and provides meaning to life. This in turn minimizes stress and improves mental health, particularly in later life when there is diminished participation in the social roles that promote social integration (George, 1996). However, social integration is shaped by context, such as gender or geographic location (Pillemer et al., 2000). Consequently, it is important for us to take context into account when we are considering the relationship between social integration and mental health.
The idea that social relationships, which emerge from social integration, enhance mental health is not new (Mirowsky & Ross, 1989). However, little research has considered how the effect of social integration on mental health may vary across particular social contexts. What is beneficial social integration in one social context may be quite different in another. Consider, for example, one indicator of social integration (i.e., living arrangements) and one social status that shapes context (i.e., immigration).
Individuals living with a spouse have access to regular companionship, emotional support, and instrumental assistance. In addition, being married provides expanded opportunities for social integration (Pillemer et al., 2000). Given this, it is not surprising that older adults living with a spouse have better mental health than those who are living alone or with family (Mindel & Wright, 1982). The relatively poor mental health among older adults who live alone is often attributed to social isolation (Hughes & Gove, 1981). For immigrants, a breakdown in traditional values and family support systems can result in poorer mental health by exacerbating the stress process and increasing social isolation (Gelfand, 1994; Gelfand & Yee, 1991). Adjustments to life in a new country may be particularly difficult for older immigrants, especially because this group often contends with social isolation as a result of limited English language ability and family responsibilities (Gelfand & Yee, 1991; Tran et al., 2000).
In addition, it is important to consider that living alone is not normative among older immigrants. Older immigrants often arrive in the United States with cultural expectations for coresidence with family, which are reinforced by the immigration experience. Although living arrangements among specific ethnic groups are shaped by a wide range of factors (Olson, 2001), older immigrants are more likely to live in extended family arrangements than older nonimmigrants (Boyd, 1991; Wilmoth, De Jong, & Himes, 1997). Consequently, we contend that living alone may have a greater negative impact on mental health for immigrants than nonimmigrants because it is not consistent with cultural expectations regarding appropriate forms of social integration in later life.
Conversely, coresidence with family is more normative among older immigrants. For both immigrants and nonimmigrants, close affective ties are positively related to mental well-being (Hughes & Gove, 1981), and living with others provides an opportunity for social integration (Pillemer et al., 2000). Family members provide assistance such as transportation, health care, and emotional support that can maintain an older adult's mental well-being, especially for ethnic older adults (Gelfand, 1994). However, intergenerational conflicts can arise among coresidential family members, which can negatively affect mental well-being. This process could be magnified in multigenerational immigrant households, where a communication gap often exists between immigrants and their children, because the younger generation is more acculturated than their parents. Generational differences in proficiency of English and culture identity can cause a decline in the intergenerational relationship quality (Gelfand, 1994; Thomas, 1995). Thus, despite the fact that coresidence is common among older immigrants (Wilmoth, 2001; Wilmoth et al., 1997), coresidence may still be associated with poor mental health among immigrants because of intergenerational tensions.
Therefore, the implications of a particular living arrangement for mental health should vary by immigrant status. We cannot assume that a particular living arrangement has a universal effect on mental health. Some living arrangements may be more detrimental to the mental health than others, depending on the prevalent social integration norms of the individual's social group. These norms shape expectations about and the experience of a particular living arrangement, which should subsequently influence mental health outcomes. Given this, it is expected that (a) middle-aged and older adults living alone or living with family or others will have higher levels of depressive symptoms and greater increases in depressive symptoms over time than those living with a spouse and (b) these effects will be magnified for immigrants, particularly immigrants living alone.
|
METHODS |
|---|
|
TOPAbstractMethodsResultsDiscussionREFERENCES |
|---|
Of the 6,391 respondents included in the baseline analysis, 5,476 were interviewed in 1994; 114 had died; 564 refused, were not located, or were given to the Asset and Health Dynamics Among the Oldest Old (AHEAD) study; and 237 had missing values for depressive symptoms or the household status at the second wave could not be determined. Three lambda selection terms (one for each type of attrition) are included in the longitudinal analysis to control for selection out of the sample (Heckman, 1979). These terms were created in STATA by estimating a probit model predicting selection that was due to each cause. The footnotes in Table 4 provide detailed information about the variables included in each selection equation.
|
), is used to measure depressive symptoms in this study (Radloff, 1977). The scale used in the baseline data is based on a reliable short version of the CES-D scale that contains 11 items. Respondents indicate the presence of depressive symptoms during the past week on a scale in which 1 = all or almost all of the time, 2 = most of the time, 3 = some of the time, and 4 = none or almost none of the time.
Unfortunately, there was a change in the depressive symptoms measurement in the second wave: Only eight of the original items were asked and the response categories were simplified (see Steffick, 2000, for more information). Respondents were asked to report whether the eight depressive symptoms were present much of the time this past week (0 = no; 1 = yes). A subset of respondents was also asked the baseline depressive symptoms questions, which use a four-category response. This information allows researchers to analyze the best course of action for modifying the baseline depressive symptoms scale to create comparable measurement across the two waves. For example, using only the eight items that are common to the 1992 and 1994 surveys, Perzynski and Townsend (2002) recoded the subset's responses to the four-category CES-D measure into two variables. The first measures the presence of depressive symptoms (0 = none or almost none of the time; 1 = some, most, almost all, or all of the time). The second is a persistence measure of depressive symptoms (0 = none, almost none, and some of the time; 1 = most, almost all, or all of the time). They indicated that both measures are highly correlated with the measure that uses four response categories. Although the presence measure is more strongly correlated with the four-category measure, the magnitude of the difference between the presence and persistence correlations is modest. This suggests these measures are tapping into the same underlying construct.
Following the lead of Perzynski and Townsend (2002), we recoded the baseline measure of depressive symptoms into a presence measure and a persistence measure. After an extensive analysis of how these two recoded baseline measures are related to the original baseline measure that contained four response categories, and to the various follow-up measures at the second wave, we decide to use the persistence coding for two reasons.
First, the persistence measure only accounts for frequently occurring depressive symptoms. Infrequent depressive symptoms (e.g., those that are experienced some of the time) are not included. Consequently, the persistence measure provides a more conservative estimate of depressive symptoms than the presence approach. This is reflected in the baseline sample means for depressive symptoms, which equal 0.92 for the persistence measure and 3.41 for the presence measure.
Second, the persistence measure produces baseline means (for the total sample, by immigrant status, and by living arrangement category) that are more similar to the corresponding mean level of depressive symptoms at the follow-up. Furthermore, the persistence measure is more strongly correlated with the follow-up depressive symptoms measure. Therefore, it is a more conservative estimate of change in depressive symptoms. In addition, when we performed a diagnostic of the residuals from the longitudinal models, we found that the persistence measure is more consistent with a simple change measure (i.e., follow-up depressive symptoms minus baseline depressive symptoms). This suggests that the persistence measure is more appropriate for the longitudinal change analysis than the presence measure.
Therefore, this analysis uses a persistence coding scheme to recode baseline depressive symptoms. The baseline and follow-up depressive symptoms scales are the sum of the number of depressive symptoms reported and range from 0 to 8 (Cronbach's alpha =.78 and.85 for 1992 and 1994, respectively).
Independent Variables
There are two central independent variables in the analysis. The first is immigrant status, which is an indirect measure of social context. It is based on a self-report of place of birth, where 0 = native born and 1 = immigrant. The second is living arrangements, which is a measure of social integration. Detailed information is gathered about the characteristics of other people residing in the household, which is used to construct the living arrangement variable. The living arrangement categories include living with spouse (only or with family or others), living alone, and living with family or others only. There are no significant differences in depressive symptoms between respondents living with a spouse only and living with a spouse and family or others (not shown). Therefore, no distinction is made between these two living arrangement categories. Family and others (i.e., nonfamily) are combined because there were not enough participants living with only others to justify a separate category. Living with spouse serves as the reference category in the multivariate analysis.
The baseline control variables include the following: age (years), gender (1 = female), education, race and ethnicity, work status (1 = working), income (measured in thousands of dollars per year), and limitations in activities of daily living. Four additional measures of social integration are also included. These measures capture the proximity of social support networks and subjective satisfaction with these networks. Respondents report whether there are relatives and friends in the neighborhood (1 = yes) and rate their satisfaction with family and friend relationships (1 = very dissatisfied to 5 = very satisfied).
Only one measure of acculturation among immigrants is included in the analysis: length of time the immigrant has been in the United States, which is equal to the year of the baseline interview minus the year the respondent came to the United States. It is expected that respondents who have been in the United States longer are more acculturated and therefore will have lower levels of depressive symptoms. A commonly used indicator of acculturation is English language ability, but this measure is not available on the HRS. The data set contains an item that reports the language in which the interview was given (0 = English; 1 = Spanish). However, this is not an ideal measure because it typically applies to Hispanic respondents, who comprise over half (i.e., 51% unweighted) of the immigrants in the sample. The option of taking the interview in another language was not available for the other 49% of immigrants who might have not been native English speakers. Therefore, this measure is not used in the analysis.
Table 1 presents the weighted descriptive statistics for the total sample and by immigrant status. It is important to note that the immigrants in this sample are quite diverse, reflecting the changing history of U.S. immigration during the 20th century. A supplementary analysis (not shown) indicates that Hispanic and African immigrants have relatively low levels of education and low income in comparison to non-Hispanic White or other immigrants. Non-Hispanic White immigrants have been in the United States the longest on average, followed by African, Hispanic, and other immigrant groups. This diversity is relevant because immigrant living arrangements are influenced by education, income, and acculturation (Wilmoth, 2001; Wilmoth et al., 1997). Therefore, the multivariate analysis will control for these characteristics. However, it is beyond the scope of this particular analysis to estimate separate models for specific immigrant groups.
|
). |
RESULTS |
|---|
|
TOPAbstractMethodsResultsDiscussionREFERENCES |
|---|
|
|
Model 2, which controls for time spent in the United States, provides similar findings. Time spent in the United States is not significantly related to depressive symptoms, and controlling for time spent in the United States does not minimize the effect of living arrangements. Interestingly, the Hispanic coefficient is significant in the immigrant models. This suggests the higher level of depressive symptoms among Hispanics observed in the total sample is due to higher levels of depressive symptoms among Hispanic immigrants.
Longitudinal Multivariate Analysis
The cross-sectional models are limited in their ability to determine the causal relationship between living arrangements and depressive symptoms. The critical question is whether the interaction between living arrangements and immigrant status influences changes in depressive symptoms over time. Table 4 presents a residualized regression model predicting changes in depressive symptoms over 2 years as a function of baseline characteristics.
The first column, which presents the model for the total sample, indicates that baseline depressive symptoms have the greatest standardized effect on depression outcomes. Those who are more depressed at the baseline have higher depressive symptoms at the follow-up. In addition, immigrants have greater increases in depressive symptoms than nonimmigrants. Living alone or with family or others at the baseline is related to a significant increase in depressive symptoms over time. Furthermore, the effect of living alone is magnified for immigrants. Age, higher education, working, higher income, and being satisfied with family and friends suppress increases in depressive symptoms, whereas lower education and being African American or Hispanic magnify increases in depressive symptoms. It is important to note that the lambda for selection out of the sample as a result of death is significant, and it has a substantial effect. This suggests this selection process is related to changes in depressive symptoms.
The disaggregated model for nonimmigrants indicates that living alone or with family or others is related to increasing depressive symptoms over time. In contrast, immigrants living alone have substantial increases in depressive symptoms, but immigrants living with family or others do not. The greater increase in depressive symptoms among immigrants living alone persists after time spent in the United States is taken into account. As expected, acculturation is negatively related to changes in depressive symptoms—each additional year an immigrant has spent in the United States suppresses the increase in depression over time. It is also interesting to note that older Hispanic immigrants have greater increases in depressive symptoms than non-Hispanic White immigrants.
|
DISCUSSION |
|---|
|
TOPAbstractMethodsResultsDiscussionREFERENCES |
|---|
As predicted by the first hypothesis, the results confirm the systematic increase in depressive symptoms associated with living alone. A key component of mental health in later life is social integration, which provides opportunities for continued engagement in meaningful roles and social interaction that reinforces salient identities (Pillemer et al., 2000). Living alone poses a threat to social integration by residentially isolating the older adult.
Consistent with the second hypothesis, the results that indicate living alone is particularly salient for the mental health of older immigrants. Given that extended family households are common among immigrant populations (Wilmoth, 2001; Wilmoth et al., 1997), older immigrants may be dissatisfied with living alone because it is not consistent with cultural norms regarding later life living arrangements. In addition, older immigrants often face language barriers (Gelfand & Yee, 1991), which could increase social isolation and threaten mental health among older immigrants living alone.
Although living with family or others can increase social integration, the results suggest that coresidence can also be detrimental to mental health. Multigenerational households can strain interpersonal relationships, particularly for immigrants who may adhere to traditional cultural beliefs regarding coresidence but experience generation gaps that are due to differences across generations in English language proficiency and culture identity (Gelfand, 1994; Thomas, 1995). For nonimmigrants, coresidence is not consistent with the dominant culture's emphasis on personal independence or the older adult's preferences for residential independence.
The results indicate that living with family or others is related to higher levels of depressive symptoms cross-sectionally, especially for immigrants, and longitudinally, particularly for nonimmigrants. Although immigrants living with family or others had higher depressive symptoms cross-sectionally, they did not have significant increases in depressive symptoms over time. In contrast, nonimmigrants who lived with family or others at the baseline had higher depressive symptoms both at one point in time and over time. Therefore, the relationship between coresidence and depressive symptoms appears to be more detrimental for the mental health of nonimmigrants than immigrants. This finding is consistent with the expectation that nonnormative forms of social integration may increase mental distress.
It is important to highlight the role of social integration in determining depressive symptoms outcomes. In this analysis, the higher cross-sectional rates of depressive symptoms among immigrants were primarily due to the excessively high rates of depressive symptoms among immigrants who are living alone or with family or others. The analysis included two additional measures of social integration: having relatives or friends in the neighborhood and satisfaction with family or friend relationships. Interestingly, satisfaction with relationships (particularly family relationships) was consistently related to lower depressive symptoms, but proximity to relatives and friends was not significant in any of the models. This suggests that the quality of relationships may be more important to mental well-being than the quantity of relationships.
Overall, the findings highlight the important connection between social integration and mental health outcomes. They suggest that interventions should encourage social integration, particularly among middle-aged and older adults living alone or with family or others. However, those interventions should acknowledge the unique needs of native-born and immigrant populations. This is particularly important given that immigrants experience a variety of barriers to treatment and consequently their mental health problems are often not sufficiently treated (Gelfand & Yee, 1991; Mui, 1998). Additional research on depressive symptoms within each of these groups is needed to develop appropriate and effective interventions.
Research in this area should continue to consider the impact of social integration on depressive symptoms outcomes, taking into consideration that specific forms of social integration may have different effects across various contexts. An important consideration for future research is the inclusion of more precise measures of immigrant characteristics. This analysis had limited access to detailed measures of immigrant characteristics, such as English language ability or country of origin. Country of origin is potentially important because of cultural variations in the manifestation of depressive symptoms and the willingness to report depressive symptoms. The HRS contains information about country of origin among immigrant respondents, but it is not possible to construct a detailed measure of country of origin given the limited number of immigrants in this sample. However, a general measure of region of origin (i.e., Asia, Caribbean, Central or South America, Mexico, Europe, and other) was tested. The region of origin variable was not significant in the multivariate models for immigrants, and therefore it is not included in the analysis. Despite the lack of significance in this analysis, future research should continue to take region or country of origin into account, particularly when a specific race or ethnic group is examined. Future research should also consider the characteristics of the neighborhoods in which immigrants reside to determine whether living in an ethnically homogeneous neighborhood attenuates the negative effect of living alone.
Although the longitudinal analysis presented here suggests living arrangements are causing particular depressive symptom outcomes, it is possible that a reciprocal relationship exists between living arrangements and depression. For example, individuals who are more depressed may be more likely to live alone. A next step in this line of research could involve explicitly testing for this reciprocal relationship among nonimmigrants and immigrants. Future analyses using data from the additional waves of the HRS will be able to test more fully the implications of the change in the depressive symptoms measure. They will also be able to identify the unique effects of living arrangement transitions while modeling depressive symptoms outcomes over a longer period of time. Given that the sample included in this research is at the cusp of middle age and older adulthood, the results might differ from those that use participants aged 65 and older. Analyses of data that contain older participants (e.g., AHEAD) would reveal whether particular living arrangements exacerbate depressive symptoms in the presence of cognitive and functional declines. Of particular interest is the elevated level of depressive symptoms exhibited by Hispanic immigrants in this sample, which is a heterogeneous group with varied immigration experiences. Targeted data collection among specific immigrant populations could illuminate how the diversity across and within immigrant groups shapes mental health outcomes. Such analyses, in combination with this one, will provide insight into developing interventions that can improve the mental health of native-born and immigrant populations.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Received for publication July 8, 2002. Accepted for publication March 24, 2003.
|
REFERENCES |
|---|
|
TOPAbstractMethodsResultsDiscussionREFERENCES |
|---|
hrswww/docs/userg/dr-005.pdf.This article has been cited by other articles:
 |
|
 |
|
  S. Diwan Limited English Proficiency, Social Network Characteristics, and Depressive Symptoms Among Older Immigrants J. Gerontol. B. Psychol. Sci. Soc. Sci., May 1, 2008; 63(3): S184 - S191. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
