| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
RESEARCH ARTICLE |
a School of Public Health, University of Michigan, Ann Arbor
b Institute of Gerontology, University of Michigan, Ann Arbor
Neal Krause, Department of Health Behavior and Health Education, School of Public Health, University of Michigan, 1420 Washington Heights, Ann Arbor, MI 48109-2029 E-mail: nkrause{at}umich.edu.
 |
Abstract |
|---|
 TOP Abstract Theoretical OverviewMethodsResultsDiscussionReferences |
|---|
Methods. Interviews were conducted with a nationally representative sample of older adults at three points in time: 1992–1993, 1996–1997, 1998–1999. Complete data are available for 511 elderly people. During each round of interviews, respondents were asked how often they provided emotional support to their social network members. Information on the self-esteem of older support providers was also gathered at each point in time.
Results. Initially, the findings revealed that helping others tends to bolster the self-esteem of all study participants regardless of their SES standing. However, these benefits began to taper off for lower SES elders during the course of the study. By the third wave of interviews, the salutary effects of helping others were evident only among older adults in upper SES levels.
Discussion. The results highlight the dynamic nature of the helping process and underscore the importance of taking SES into account when studying the effects of assisting others in late life.
THERE is now vast literature on social support in late life (e.g., Bosworth and Schaie 1997
; George 1996; Krause 1997; Lewis and Rook 1999; Seeman et al. 1993). However, the wide majority of these studies focus solely on assistance that is received by elderly people even though they are typically support providers as well. The lack of research on support providers is unfortunate because mental health professionals have been arguing for decades that people are especially likely to profit when they help others. Evidence of this may be found, for example, in Reissman 1965 classic article on the helper principle. Based on his work with self-help groups, Reissman found that doing something worthwhile for someone in need is a fulfilling and self-validating experience that bolsters the self-esteem of support providers.
The purpose of this study was to examine the relationship between providing support to others and self-esteem in late life. A few researchers have studied the effects of helping others (e.g., Cutrona and Russell 1987; Krause 1987; Krause, Herzog, and Baker 1992), but the present study builds on this work in two important ways. First, we examine the effects of giving support to others on changes in self-esteem over time. This is important because, as the findings presented later will reveal, assessing change in feelings of self-worth over time provides rich new insights into the underlying process that is at work. Second, existing studies focus solely on the relationship between helping others and well-being in the elderly population taken as a whole. However, some studies suggest there may be socioeconomic status (SES) variations in the helping process. For example, based on her work with women living in poverty, Belle 1990 argues that helping others when one's own resources are depleted may be a source of distress, rather than a rewarding experience. Consequently, the main thrust of the analyses that follow is to see whether providing support to others is more likely to enhance the self-esteem of upper than lower SES elders.
The discussion that follows is divided into three main sections. The theoretical underpinnings of the study are developed first. Here, an effort is made to craft a more detailed explanation for why helping others may fail to bolster the self-esteem of lower SES elders. Following this, the sample and study measures are described. Finally, results based on a nationwide longitudinal study of older adults are presented and discussed.
|
Theoretical Overview |
|---|
|
TOPAbstractTheoretical OverviewMethodsResultsDiscussionReferences |
|---|
and others fail to provide the necessary detail.
We suspect that at least part of the answer may be found by focusing on two factors. The first has to do with the outcome of the helping process, and the second involves a phenomenon known as stress contagion (Wilkins 1974).
Outcome of the helping process.
A central premise in this study is that older adults will be more likely to benefit from helping others if the situation of support recipients improves because of the assistance they have been given. Viewed the other way around, if elders provide help to others, but the circumstances of the support recipient do not improve, then it is not clear why help providers would feel they have done something worthwhile. But the impact of unsuccessful helping may go well beyond this simple realization. By assisting others, older adults are, in essence, investing part of themselves in the helping process. This means that their abilities, skills, and resources are placed in a social arena, where they are subject to evaluation by the provider, the recipient, and other social network members. If a recipient fails to respond when help is given, it may reflect poorly on the support provider. More specifically, the help giver may be forced to conclude that the assistance they have provided was inadequate, and that they lack the skill or resources that are needed to make a difference. As a result, the support provider may assume part of the blame when a positive outcome is not forthcoming (Coyne, Wortman, and Lehman 1988). It follows from this that assuming blame for the outcome of failed helping will diminish, rather than bolster, the self-esteem of older support providers.
If this rationale is valid, then the search for SES variations in the effects of the helping process can be more profitably directed toward identifying the factors that influence the delivery of effective support. At least two factors are important in this respect: the locus of the first lies within the support provider, and the second involves the support recipient.
The social support process is quite complex and involves more than the simple giving and receiving of assistance. As La Gaipa 1990 points out, people may engage in behaviors they intend to be supportive, but their efforts may not be seen as helpful by the intended recipient. Although there are likely to be a number of reasons for this, at least part of successful support providing involves the exercise of appropriate social skills. Unfortunately, effective social support skills may not be distributed evenly in the population. Evidence of this may be found in two bodies of research that have developed in a largely independent manner.
The first literature focuses largely on children and younger adults. This work suggests that social skills and social competence are positively related to SES (Eckholdt and Lenzenweger 1990; Ramsey 1988). In particular, this research shows that compared to their lower SES counterparts, middle SES children are better at decoding facial expressions, seeing another person's point of view, and role playing at making friends and giving help (see Argyle 1994, for a review of these studies).
The only study we could locate that looked at this relationship with data provided by older adults was conducted by Hogg and Heller 1990. Their findings reveal that older women with higher levels of educational attainment score higher on measures of social competence than elderly women with fewer years of formal schooling. Indicators of empathy, as well as the ability to take the role of the other, were included among the social competence measures used by Hogg and Heller 1990. The fact that there may be SES variations in social skills makes sense for the following reason. Education is a key marker of socioeconomic status. As a number of investigators have pointed out, one of the functions of education is to build communication skills and enhance problem-solving abilities (Ross and Sastry 1999). Both are essential for the delivery of effective assistance.
The second body of research was developed by Kohn and Schooler 1983. Their impressive research program traces the lifelong implications of educational attainment. In particular, their work indicates that education determines the type of job a person holds and this occupation, in turn, has a profound implication for intellectual functioning (e.g., verbal fluency and intelligence) as well as interpersonal relationships. With respect to social ties, the work of Kohn and his colleagues indicates that individuals in low SES jobs score higher on authoritarian rigidity, and are less willing to trust others than people in high SES occupations (Kohn 1977). It seems unlikely that people who are authoritarian and view others with suspicion will be effective support providers. It is especially important to note that the effect of occupation on interpersonal relationships persists at least through the time of retirement (Schooler, Mulatu, and Oates 1999).
The upshot of the argument developed so far is that lower SES elders may try to help others, but lack the social skills and problem-solving abilities necessary to deliver assistance effectively. Under these circumstances, their efforts to help may do little more than bring their own limitations to the foreground. As a result, the self-esteem of lower SES support providers may suffer.
But part of the reason for a failed response to helping may also lie with the support recipient, as well. Research consistently shows that social networks are largely composed of people of the same socioeconomic status (Lin 1982). This means that lower SES elders are likely to be helping individuals in similar circumstances. Two key characteristics set off lower SES support recipients from their upper SES counterparts. First, support providers do not (and often cannot) do everything a support recipient may need. Instead, support providers offer advice, encouragement, and some tangible help, but frequently leave the main responsibility for resolving a problem with the recipient (Krause and Borawski-Clark 1994). Unfortunately, there is some evidence that lower SES support recipients may lack the resources necessary to follow through on the help they have been given (Hobfoll 1998). Second, the problems faced by lower SES support recipients are likely to be chronic in nature and, therefore, more difficult to eradicate (Pearlin and Skaff 1996). If the problems linger after assistance is provided, lower SES support providers may not reap the full benefit of helping others. In fact, as Gottlieb 1997 points out, prolonged support mobilization may cause support providers, "... to resent the burden that has been placed on them rather than gaining a sense of pride and satisfaction from the helpful role they have assumed" (p. 17).
But once again, the negative sequalae associated with failed helping responses may be more deleterious than they appear initially. Sociologists have argued for nearly 100 years that self-esteem is a social product formed by perceived feedback from significant others (Cooley 1902). In their insightful chapter on miscarried helping, Coyne and his associates (1988) describe several ways in which failed helping may convey negative feedback to a support provider. In particular, these researchers maintain that the continued inability of a support recipient to improve may be viewed by the support provider as a subtle accusation that the help they were given was inadequate. As a result, support providers may interpret failed responses to helping as a form of rejection.
Stress contagion.
Because social networks tend to be homogeneous with respect to SES, a number of researchers have noted that stressors affecting one lower SES person are likely to be experienced by others in the network as well (Riley and Eckenrode 1986). This is known in the literature as stress contagion (Wilkins 1974). This means, for example, that if a lower SES elder is faced with financial strain, there is a good chance that his or her significant others will be grappling with economic problems of their own. Under these circumstances, elderly support recipients may find themselves in a situation where the resources needed by significant others are the very resources older support providers need to cope with their own problems (Riley and Eckenrode 1986). Providing assistance when one's own resources are in short supply may exert an adverse effect on the helping process. Dramatic evidence of this may be found in Belle 1982a, Belle 1982b compelling research with younger women living in poverty. She reports that these women preferred to remain isolated rather than turn to their social network for assistance because doing so would create obligations they cannot fulfill. It is especially important to note that even though the common stressors facing the lower SES network members in Belle 1982b work were tangible in nature (e.g., economic stressors), it was the emotional costs of helping others that appear to exact the greatest toll. As Belle 1982b points out, helping others may lead lower SES women to experience "... betrayal, burdensome dependence, and vicarious pain" (p. 143). Simply put, Belle 1982b concludes that, "... economic stress reduced options and essentially coerced some women into relationships which they might otherwise avoid" (p. 142, emphasis added). It is difficult to see how providing help to others in relationships that are perceived as coercive can benefit the self-esteem of older support providers.
Taken as a whole, the theoretical rationale developed up to this point suggests that lower SES elders may be less likely than older people in upper SES groups to benefit from helping others because lower SES elders may feel hamstrung by their inability to provide support effectively, and frustrated by the inability of support recipients to respond favorably. There do not appear to be any studies in the literature that explore these issues empirically. It is toward this end that the analyses that follow were conducted.
|
Methods |
|---|
|
TOPAbstractTheoretical OverviewMethodsResultsDiscussionReferences |
|---|
The sampling frame consisted of all eligible persons in the Health Care Financing Administration (HCFA) Medicare Beneficiary Eligibility List. This list contains the name, address, sex, and race of virtually every older person in the United States. It should be emphasized that elderly people are included in this list even if they are not receiving Social Security benefits. Even so, two groups of older adults are not included in this database: elders who do not have a Social Security number (this may be due to factors such as illegal immigration) and elderly people who are at least 100 years of age (HCFA does not release the names of these individuals).
A three-step process was used to draw the sample. First, 5% of the names in the master file maintained by HCFA were selected with a simple random sampling procedure. Next, 110 counties across the coterminous United States were identified as primary sampling units (PSUs). These PSUs were selected with probability proportionate to the number of persons who were retired and at least 65 years of age. Following this, eligible persons were selected at random from each PSU. Some counties (e.g., Dade County, Florida) were oversampled because they contain a disproportionately large number of eligible older adults.
Interviewing for the baseline survey took place during 1992–1993. The data collection was performed by Louis Harris and Associates, and a total of 1,103 interviews were completed. The response rate for the baseline survey was 69.1%.
During 1996–1997, a second wave of data was gathered from the elderly people who participated in the baseline survey. A total of 605 older adults were reinterviewed successfully. Excluding those who either had died or moved to a nursing home, the reinterview rate for the Wave 2 survey was 76.5%.
Finally, a third wave of interviews was conducted during 1998–1999. The disposition of the sample at Wave 3 is as follows: reinterviewed successfully (n = 530), dead (n = 249), moved to a nursing home (n = 31), refused to be reinterviewed (n = 89), could not be located (n = 85), and too ill to participate (n = 119). Once again excluding those who either had died or were living in a nursing home, the reinterview rate for the Wave 3 survey was 64.3% of those participating at Wave 1.
Cases containing item nonresponse were eliminated from this study using listwise deletion procedures. Because data from all three waves of interviews are used in the analyses that follow, the available number of cases ranges from 1,041 to 511. Based on the sample of 1,041 respondents, preliminary analyses revealed that the average age of the study participants was 72.03 at Wave 1 (SD = 5.53). Approximately 40% are men, and 53% reported they were married at the baseline survey. The respondents indicated they had completed an average of 11.36 years of schooling (SD = 3.41). Finally, 91% of the study subjects are White. These descriptive statistics, as well as the findings presented below, are based on weighted data.
Measures
The indicators used to measure support provided to others and self-esteem are contained in Table 1 . In addition, the procedures used to code these items are provided in the footnotes of this table.
|
). However, most researchers would agree that different types of assistance can be classified in the following manner: emotional support, tangible help, and informational assistance. There are three reasons why the analyses presented below focus solely on emotional support provided to others.
First, there is considerable evidence that different types of provided support correlate highly (Krause 1995). This probably reflects the fact that elders who provide one type of assistance are also likely to provide other types of support as well.
The second reason for focusing on emotional support provided to others may be found by returning to the theoretical rationale that was devised for this study. A key component of this conceptual framework specifies that lower SES elders are less likely to benefit from helping others because they lack the social skills necessary to ensure a successful outcome. Although social skills are necessary for the delivery of any kind of help, they may be especially important for emotional support because this type of assistance depends heavily on verbal communication. In contrast, other types of assistance, such as tangible support, may be provided with more limited communication skills.
Finally, the third reason for focusing on emotional support provided to others has to do with changes in social network composition that are thought to take place as people age. In her widely cited work on socioemotional selectivity, Carstensen 1992 maintains that as people grow older, relationships with more peripheral network members begin to fall off, and interpersonal contact becomes focused on a more select number of intimate ties. The factors driving this paring down of social network members are especially relevant for the present study. In particular, socioemotional selectivity theory "... predicts that, relative to younger people, older people are less motivated to engage in emotionally meaningless but perhaps otherwise functional social contact and make social choices based on the potential for emotional rewards derived from social interactions" (Lang and Carstensen 1994, p. 315, emphasis added). Viewed more broadly, this research suggests that the exchange of emotional support may become increasingly salient with age. Perhaps it is for this reason that Kulis 1992 reports that older adults are more likely to exchange emotional than instrumental support with family members. If emotional support is especially important to elderly people, and they are more likely to engage in this type of social exchange, it follows that providing emotional assistance to others should be more strongly related to self-esteem than other types of social support.
As shown in Table 1 , providing emotional support to others is assessed with four indicators. It should be emphasized that the same questions were administered in all three waves of data collection. This brief composite comes from a larger scale developed by Krause and Markides 1990. A high score on these indicators stands for helping others more often. The internal consistency reliability estimates at each wave of interviews are as follows: Wave 1 (.751), Wave 2 (.713), Wave 3 (.761).
Self-esteem
Identical measures of self-esteem were also contained in each round of data collection. These indicators come from the widely used scale developed by Rosenberg 1965. A high score denotes greater feelings of self-worth. The reliability estimates for the self-esteem measures at each wave of interviews are as follows: Wave 1 (.848), Wave 2 (.833), Wave 3 (.843).
Socioeconomic status.
A measure of educational attainment is used to assess socioeconomic status. This indicator, scored continuously, reflects the total number of years of schooling that were completed successfully by study participants at the time of the baseline survey.
Demographic control measures.
The relationships between support provided to others, SES, and self-esteem were evaluated after the effects of age, sex, marital status, and race were controlled statistically. Age is scored continuously in years, while sex (1 = men, 0 = women), marital status (1 = married, 0 = otherwise), and race (1 = White, 0 = all other racial groups) are coded in a binary format.
Data Analysis Strategy
The central hypothesis in this study specifies that the effects of helping others on self-esteem will depend upon the socioeconomic status of the support provider. Stated in more technical terms, this specification calls for a statistical interaction effect between support providing and SES on feelings of self-worth. The following ordinary least squares (OLS) multiple regression equation is used to test for this interaction effect:
 | (1) |
In this equation, SE stands for self-esteem, ES denotes emotional support provided to others, EDUC represents educational attainment, and the Zi are the demographic control variables identified earlier (i.e., age, sex, marital status, and race). The bi and ci in 1 are regression coefficients, while "a" is the intercept. Following the recommendations of Aiken and West 1991, all independent variables are deviation scored (i.e., centered on their means) before 1 is estimated.
1 is solved in two steps. The additive effects of the independent variables are estimated in the first step (i.e., the effects of ES, EDUC, and the control variables). Following this, the multiplicative term assessing the interaction between helping others and education (i.e., ES x EDUC) is entered at step 2. Once these analyses are complete, an additional formula provided by Aiken and West 1991 is used to clarify the nature of statistically significant interaction effects. Cast within the context of the present study, this formula provides estimates of the impact of helping others on self-esteem at select levels of educational attainment (this procedure is described in greater detail when the study findings are presented). Significance tests are then also computed for these estimates.
Because the data for this study have been gathered at three points in time, we examine the effects of helping others and SES on changes in self-esteem across the entire range of data collection points (i.e., Waves 1–3). Doing so will allow us to determine if SES differences in the effects of helping others manifest relatively quickly, or whether some time must pass before they emerge. These analyses will be accomplished by estimating 1 three times. First, we will look at the effects of helping others and SES on self-esteem using the Wave 1 data only. These initial analyses will be performed using 1 as it appears above. Second, we will assess the impact of Wave 1 helping and SES by using the Wave 2 measure of self-worth as the dependent variable. Because the Wave 1 measure of self-esteem will also be included as an independent variable in this pass through the data, the goal of these analyses will be to assess the effects of support providing and SES on change in self-esteem over time. Finally, the long-term effects of Wave 1 helping and SES will be evaluated by using the Wave 3 measure of self-esteem as the outcome measure. Once again, the Wave 1 self-worth measure will be included in the equation as an independent variable.
Because so little is known about the help-giving process over time, it is not possible to predict how the findings will emerge in each pass through the data. Even so, it is possible to use the theoretical rationale that has been developed for this study to briefly speculate on what might happen. A central tenet in this study is that older adults will benefit from helping others only if they see improvement in the support recipient. However, because the problems faced by support recipients in lower SES groups are likely to be chronic and, therefore, difficult to eradicate, it would be unreasonable to expect to see improvement right away. But, over time, it may eventually become evident that lower SES support recipients may never improve significantly. Once this happens, the self-esteem of older support providers is likely to decline. To the extent this is true, SES differences in the effects of the helping process may not emerge until the Wave 2 or Wave 3 data are examined. Stated in more technical terms, we suspect there may be lagged effects in the interplay between providing support to others and SES on feelings of self-worth in late life.
|
Results |
|---|
|
TOPAbstractTheoretical OverviewMethodsResultsDiscussionReferences |
|---|
Effects of Sample Attrition
When the sample for this study was introduced, we reported that a number of subjects did not participate in all three waves of interviews. The loss of participants can bias study findings if those who remain differ significantly from the population they are supposed to represent. Although it is difficult to determine the extent of this problem precisely, some preliminary insight may be obtained by using select baseline data to see if subjects who were lost to follow-up differ significantly from those who participated in the Wave 3 interviews (see Norris 1985, for a detailed discussion of this approach). The following procedures were used to implement this strategy. First, a binary outcome measure was computed by assigning a score of 1 to all participants who were lost to follow-up and a score of 0 to those who took part in the Wave 3 survey. Then, using logistic regression, this binary outcome was regressed on the Wave 1 measures of age, sex, education, marital status, race, self-esteem, and emotional support provided to others. If any of the Wave 1 measures are related significantly to the binary outcome variable, then it would be reasonable to assume that sample attrition did not occur in a random manner.
Findings from the logistic regression analysis reveal that the loss of subjects did not occur randomly. More specifically, the data suggest that, compared with elders who were reinterviewed successfully, older adults who did not participate in the Wave 3 survey were more likely to be older (b = .089; p < .001), male (b = .577; p < .001), less educated (b = -.048; p < .05), and non-White (b = -1.122; p < .001). This pattern of attrition is consistent with other studies of nonresponse in longitudinal surveys (Groves 1989). In addition, the findings further indicate that elderly people who were lost to follow-up tended to provide less emotional support to their social network members (b = -.052; p < .05). However, significant differences failed to emerge with respect to feelings of self-worth. Nevertheless, the potential biasing effects of nonrandom attrition should be kept in mind as the study findings are reviewed.
Substantive Findings
The results of the tests for the statistical interaction between providing support to others and SES on self-esteem are presented below in three sections. Consistent with the data analysis strategy presented earlier, the findings from the Wave 1 data only are examined first. Following this, the effects of support providing and SES on changes in self-worth are examined using data from Waves 1 and 2 only. Finally, the results that emerged when the Wave 3 measures of self-esteem served as the outcome measure are presented.
Table 2 contains the findings derived from estimating 1 with data from the baseline survey only. The results from the first step of this hierarchical regression analysis are provided in the lefthand portion of this table; the righthand side contains the results that were obtained after the multiplicative term was added to the equation.
|
Evidence of this may be found in Table 3 . This table contains the results of the analyses that were performed when the Wave 2 measure of self-worth served as the outcome variable. The data in the lefthand column of Table 3 reveal that, regardless of SES, providing emotional support to others is associated with greater feelings of self-worth over time (Beta = .085; p < .05). However, the size of the additive effects of help giving appears to be tapering off. In particular, the estimate based on Wave 2 data is 49% smaller than the corresponding estimate from the Wave 1 data only (i.e., .085/.166 = 51%).
|
|
Perhaps more important, the data in the righthand column of Table 4 suggest that the proposed interaction effect between SES and giving emotional support to others on changes in self-esteem is statistically significant for the first time (b = .020; p < .001). Although the multiplicative term is significant, it may be somewhat difficult to determine whether the interaction effect is in the proposed direction. Fortunately, as discussed earlier, it is possible to perform some additional calculations by hand to clarify the nature of these effects. Essentially, these computations involve deriving separate regression estimates for the effects of providing support on changes in self-esteem at select levels of educational attainment (see Aiken and West 1991, for a detailed discussion of these procedures). If the interaction effect is in the proposed direction, then the effects of help giving on feelings of self-worth should become larger at progressively higher levels of education. Although any level of schooling could be selected to illustrate these effects, the following levels of educational attainment were selected for this purpose: 4 years, 8 years, 12 years, and 16 years of schooling.
The results derived from the additional hand calculations are presented in Table 5 . As these data reveal, providing emotional support to others is actually associated with a decline over time in feelings of self-worth for elderly people with four years of schooling (Beta = -.258; p < .001). However, the potentially deleterious effects of helping others disappear completely among older adults with an eighth-grade education (Beta = -.075; not significant). In contrast, the purported benefits of helping others on changes in self-esteem begin to emerge among elderly people with 12 years of education (Beta = .108; p < .05). Finally, the data in Table 5 suggest that providing emotional support to others is especially likely to bolster feelings of self-worth over time among elderly people who are highly educated (i.e., those with 16 years of schooling; Beta = .291; p < .001).
|
Supplementary Analyses
Although the findings that have emerged so far may be thought provoking, the results can be challenged in at least three ways. The purpose of this section is to present some additional analyses that were designed to meet these problems head on. The first problem has to do with the potentially biasing effects of sample attrition over time. The second is concerned with problems involving the multiplicative term that was used to test for the proposed statistical interaction effect. Finally, the third issue involves other ways to specify the relationship between providing support to others and self-esteem over time.
Some subjects did not participate in the Wave 3 interviews, and the effective sample size declined from 1,041 to 511 over the course of the study. This makes it possible to argue that instead of reflecting substantively meaningful results, the pattern of findings that emerged reflects little more than a methodological artifact that may be attributed to the use of different cases in the three sets of analyses. An example will help clarify the nature of this problem. The Wave 1 findings reveal that all study subjects benefit from helping others; the Wave 2 analyses show a diminished effect, and significant additive effects fail to emerge altogether in the Wave 3 data. Based on our theoretical rationale, we conclude that this ultimately reflects the inability of lower SES support recipients to improve. However, one might instead argue that this finding emerged solely because of the progressive loss of subjects over time who are less involved in the helping process.
Although there is no way to address this problem conclusively, it is possible to gain some preliminary insight through the following supplementary analyses. In particular, we repeated the Wave 1 and Wave 2 analyses using only those respondents who participated at all three data collection points. If our substantive interpretation of the findings is invalid, a different pattern of results should emerge in the supplementary analyses. For example, these supplementary analyses may reveal that the proposed interaction between SES and helping others may be significant in the Wave 1 and Wave 2 analyses, and that the gradual tapering off of the additive effects fails to emerge in the data.
The additional analyses (not presented here) do not support the alternative interpretation of our results. Instead, the findings are virtually identical to those presented in the previous section. More specifically, a statistically significant interaction between education and helping others failed to emerge in either the Wave 1 or Wave 2 analyses. Moreover, the additive effects reflect the same pattern discussed above. In particular, the additive effect of helping others on self-esteem at Wave 1 was statistically significant (Beta = .159; p < .001). But by Wave 2, this effect had dissipated entirely (Beta = .042; not significant). Once again, these supplementary analyses suggest that, at least initially, all elderly people regardless of their socioeconomic status benefit by helping others, but that over time these benefits tend to be enjoyed by upper SES elders only (tables containing the results of these supplementary analyses are available upon request from the first-listed author). The fact that sample attrition does not appear to exert an adverse effect on our findings is consistent with a small, but intriguing cluster of studies which suggest that the loss of subjects over time may not alter substantive study findings appreciably (see, for example, Mihelic and Crimmins 1997).
Research by Krause and Borawski-Clark 1995 suggests that older adults in lower SES groups tend to provide less support to their social network members than elderly people in higher SES groups. This may create a problem in the present study for the following reason. If the correlation between education and providing emotional support to others is very large, it will be difficult to disentangle the interaction between SES and help giving from the impact of SES on providing support to others (Thoits 1982). The following strategy was used to explore this possibility. The Wave 1 measure of providing emotional support to others was regressed on the following baseline study indicators: age, sex, education, marital status, and race. The results of these additional analyses (not shown here) reveal that even though education is related to helping others, the size of the relationship is not substantial (Beta = .171; p < .001; a table containing the results of these analyses is available upon request from the first-listed author). Although the type of confounding discussed here is always a matter of degree, the magnitude of this relationship suggests that the interpretation of the interaction effects presented in Table 4 and Table 5 is probably not unduly affected.
Finally, data on providing emotional support to others was gathered at all three waves of interviews. This makes it possible to evaluate an alternative model that has not been discussed up to this point. More specifically, these data allow us to see whether the impact of change in providing support to others on changes in self-esteem over time also differs for upper and lower SES elders (see Finkel 1995, for a discussion of this type of specification). The analyses using the Wave 1 and Wave 2 data will be described here to show how this alternative specification was evaluated. Self-esteem at Wave 2 served as the dependent variable. Self-esteem at Wave 1, emotional support provided to others at Wave 1, as well as age, sex, education, marital status, and percentage. White served as independent variables. In addition, and most importantly, this equation also contained the measure of emotional support provided to others at Wave 2, as well as a new multiplicative term that was created by multiplying emotional support to others at Wave 2 by education. This model tests the hypotheses that there are SES variations in the impact of change in emotional support provided to others over time on change in self-esteem. A comparable equation was also estimated with the Wave 3 data by using the Wave 3 measures of emotional support provided to others in place of its Wave 2 counterpart.
The data from the first two waves of interviews (not shown here) failed to provide support for the notion that lower SES elders are more vulnerable to the effects of change in the amount of support provided to others. In particular, the unstandardized regression coefficient associated with the multiplicative term representing the joint effects of Wave 2 support provided to others and education was not statistically significant (b = .017; not significant). Similarly, we were also unable to uncover a significant interaction effect between changes in emotional support provided to others and education on changes in self-esteem using data from Waves 1 and 3 (b = .288, not significant). Tables containing the results of these analyses are available from the first-listed author.
|
Discussion |
|---|
|
TOPAbstractTheoretical OverviewMethodsResultsDiscussionReferences |
|---|
The pattern of findings we have observed is consistent with the theoretical rationale that was developed for this study. The ability of support recipients to respond favorably to the help they are given figures prominently in this respect. In particular, we argued that lower SES support providers may lack the social skills necessary to deliver assistance effectively. Moreover, because lower SES support recipients are likely to be grappling with chronic problems, it may not be reasonable to expect them to improve right away. Finally, based on the work of Belle 1990, we argued that the help-giving process may be further complicated for lower SES elders because the potential coercive nature of ties with significant others may erode the benefits associated with the help-giving process. Taken together, these three factors suggest that, as time passes, it may become increasingly evident to older support providers that the help they have given has not (and probably will not) make much difference. At this point, lower SES support providers may begin to question their ability to help others, and they may even feel rejected by the very individuals they are trying to assist. When this occurs, feelings of self-worth among support providers are likely to decline.
The findings from this study are noteworthy for two reasons. First, this is, to our knowledge, the first time that the relationships among providing support to others, SES, and self-esteem have been evaluated with data provided by a nationwide survey of older adults. Second, this study appears to be the first to use multiple waves of longitudinal data to show how a pattern of findings involving these constructs emerges and unfolds over time.
A word is in order at this point about the impression our work may create about lower SES life. It is not our intention to provide an overly bleak view of existence for elderly people with relatively little education. Instead, it is important to take two factors into account so that our research can be placed in a proper perspective. First, the assertion that lower SES elders may lack sufficient social skills is not meant to imply that every lower SES person is socially incompetent. Far from it. Instead, we merely wish to point out that, when viewed in the aggregate, there is some evidence that proportionally more lower SES elders may have problems with social skills than their upper SES counterparts. Second, even though lower SES elders may be vulnerable in certain respects, they also possess unique sources of strength. It is important to take these sources of resilience into account as well. For example, recent research reveals that religion may be an especially potent coping resource for older adults with relatively little education (Krause 1998).
Rather than providing definitive answers, the research presented in this report merely takes a modest first step in evaluating what is likely to be a very complex process. Consequently, it may be helpful to identify some potentially important ways to further extend the thinking that has been developed so far. Three areas to consider in future research are examined briefly below. The first deals with the need to examine other social structural factors (i.e., race and gender) in conjunction with socioeconomic status. The second involves the potential benefits of using qualitative methods to delve more deeply into the meaning and practice of social support in different SES groups. Finally, the third area focuses on assessing the complex interplay between support received, as well as support provided to others.
Although SES has emerged as a potentially important construct in the present study, other social structural factors are likely to come into play as well. Race and gender figure prominently in this respect. More specifically, a small but compelling body of research indicates that older African Americans receive and give more assistance to their significant others than do elderly Anglos (Silverstein and Waite 1993). In addition, a substantial number of studies reveal that older women are more involved in the lives of their social network members than elderly men (Antonucci and Akiyama 1987). Taken as a whole, this research suggests that focusing on the intersection of SES, race, and gender is likely to lead to a more complete understanding of the impact of helping others in late life.
So far, most survey research on social support in late life is based on scales that have been developed for White, middle-class respondents. As a result, researchers know relatively little about the meaning and practice of social support among lower SES elders. Studies focusing on SES differences in support may, therefore, fail to yield valid results if assistance is exchanged in relatively unique ways in lower SES groups, or if the same supportive behaviors take on a different meaning among older adults who are not well-to-do. A few compelling qualitative studies of social support in late life have appeared in the literature (e.g., Stephens 1976), but researchers have yet to use the insights they provide to devise good quantitative measures. Employing qualitative research methods to develop scales that assess social support among lower SES elders should be a top priority in the future.
Social exchange theorists have been arguing for decades that one cannot fully understand the impact of providing social support without also taking into account what has been received as well (Dowd 1975). Although the relationship between giving and receiving assistance is quite complex, this perspective generally holds that elderly people are most likely to benefit when social exchanges are balanced, and they are able to reciprocate when others do things for them. However, as a recent study by Ikkink and van Tilburg 1998 reveals, the proper estimation of balanced social relationships is fraught with difficulty. Even so, a careful consideration of what is received as well as what is provided is likely to provide valuable insight into the helping process among older adults.
Those wishing to pursue further research on the helping process in late life would also be well advised to consider the limitations in our study. Four shortcomings are discussed briefly, as follows. The first involves the need to empirically evaluate key linkages in the theoretical rationale that was devised for this study. The second is concerned with deriving a better understanding of the temporal lag that may be at work in the social support process. The third limitation has to do with testing causal assumptions embedded in our research. Finally, the last shortcoming involves the need to consider ways that elderly people can help others beyond providing emotional support.
The theoretical rationale developed for this study was not evaluated fully because measures of key constructs were not available in the data. For example, we argued that lower SES support recipients are likely to be confronted by chronic stressors that are not amenable to change. However, measures of chronic strain experienced by significant others were not obtained in our surveys. In addition, the theory devised for this study is based on the notion that feelings of self-worth among lower SES support providers are likely to suffer when support recipients fail to improve. Unfortunately, a measure of support recipient improvement was not included in the study protocol either. Finally, we proposed that the lack of social skills may adversely affect the quality of support provided by lower SES elders. However, measures of social skills or relational competence were not included in our work. Given the lack of complete data, it is important to clearly spell out what our study does and does not do. We have been able to demonstrate that SES and providing emotional support to others affects change in self-esteem over time. However, we have not been able to show empirically why or how these effects arise. Although this is typical when a new area of inquiry is first explored, a high priority should be placed on measuring the intervening constructs in our conceptual framework so that a more complete understanding of the underlying process may be obtained.
The findings reported in this study are dependent upon the timing of the observations. As discussed above, the data were gathered in an initial baseline interview, followed by a second wave four years later, and a third wave of interviews two years after that. As a result, we lack adequate insight into three key issues. First, it is not possible to tell precisely when SES differences in the relationship between helping-others self-esteem first emerge. Second, we cannot determine when SES differences in the effects of helping others peak or become most pronounced. Finally, we do not know if these SES differences eventually dissipate, and if they do, how long it takes for them to taper off.
But great care must be taken in approaching the causal lag dilemma. We have embarked on the study of an issue that has, to the best of our knowledge, never been investigated previously. Consequently, nothing is known about exactly how the relationship between SES and providing support to others unfolds over time. Even specifying precisely when effects first appear, peak, and initially begin to dissipate is not enough because the line (possibly a curve) linking these key data points would not be revealed completely. As a result, our understanding of the underlying process would be incomplete. For example, even though it is important to know when effects first begin to dissipate, these data alone would not adequately describe the precise nature of the subsequent downward trajectory. This means that we would not know whether the impact of helping others dissipates quickly or abruptly once the effects begin to subside, or whether the tail of the curve depicting this relationship is long and tapers off slowly. The same is true when effects first begin to emerge; we do not know whether the effects of providing support rise sharply, or whether they emerge slowly over more extended periods of time. In order to describe the process fully, a number of studies are needed to evaluate a range of different time lags. It is for this reason that Gollob and Reichardt 1987 argue that, "... one must study many different lags to understand causal effects fully" (p. 82). All we can say at this point is that statistically significant SES differences in helping others arise sometime between four and six years after providing support to others is assessed. Clearly, a number of studies using a full range of time periods are now needed to fully map out the nature of the relationship between helping others, SES, and self-esteem.
Although assessing a range of temporal lags is important, it may be helpful to speculate briefly on why significant SES variations in the helping process appear to take some time to emerge. At least three factors may be involved. First, as noted earlier, support recipients may be grappling with chronic problems, and as a result, they may not respond favorably to the assistance they have been given. However, some time may pass before it becomes apparent to older support providers that the assistance they provided has not, and probably will not, have the desired effect. Second, a good deal of the assistance provided by older adults is given to family members. Because family ties at all SES levels are characterized by a deep sense of commitment (Argyle 1994), it may take some time before older support providers are willing to face the possibility that the help they have given family members may not be effective. The third reason why it may take some time before SES differences emerge in the helping process has to do with how older adults respond when their self-esteem is threatened. More specifically, there is some evidence that elderly people may take steps to shore up faltering feelings of self-worth by turning to various defensive coping responses (e.g., denial, readjusting personal goals and aspirations—see Brandtstadter and Greve 1994). It is important to note that these defensive coping styles are more likely to be used by lower than upper SES individuals (Argyle 1994). To the extent that this is true, the differential use of defensive coping responses by lower SES elders may extend or prolong the time it takes for the negative sequalae of the helping process to become manifest.
Even though the data for this study were gathered at three points in time, the temporal ordering between providing support to others and self-worth is based on theoretical considerations alone. As a result, it is possible to propose a different causal ordering in which self-esteem shapes the subsequent willingness of older adults to help others (Baumeister 1993). Clearly, the causal assumptions in the analyses presented above need to be evaluated rigorously with the latent variable modeling procedures described by Kessler and Greenberg 1981.
Finally, our study focused solely on emotional support provided by older adults to their significant others. However, this hardly exhausts all of the ways in which elderly people may help their social network members. More specifically, elders may provide tangible and informational support to the people they know, as well. Assessing SES variations in the impact of providing tangible and informational support to others would add greatly to our understanding of the helping process in late life.
When sociology first emerged as a discipline, the nature and function of social relationships was a primary concern (Durkheim 1951). A good deal of this early work was focused on how social structural factors influence interpersonal ties. However, as time passed, the study of social support became increasingly dominated by psychologists. Due in part to the orientation of psychologists, interest in social structure and social support waned. In fact, the shift in emphasis has become so pronounced that some researchers claim that the study of social support has become "overly psychologized" (Lieberman 1986). Viewed at the most general level, the intent of the present study was to redress this imbalance in the literature by highlighting the central role played by socioeconomic status in shaping the interpersonal exchanges of elderly people. We hope the theoretical perspective we have devised, as well as the findings we present, encourages other investigators to delve more deeply into the potentially important influence of the wider social context on other key determinants of health and well-being in late life.
|
Acknowledgments |
|---|
Received for publication December 14, 1999. Accepted for publication April 26, 2000.
|
References |
|---|
|
TOPAbstractTheoretical OverviewMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  T. L. Gruenewald, A. S. Karlamangla, G. A. Greendale, B. H. Singer, and T. E. Seeman Feelings of Usefulness to Others, Disability, and Mortality in Older Adults: The MacArthur Study of Successful Aging J. Gerontol. B. Psychol. Sci. Soc. Sci., January 1, 2007; 62(1): P28 - P37. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
