| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
SPECIAL SECTION |
Department of Psychology and Social Behavior, University of California, Irvine, California.
Address correspondence to Karen S. Rook, Department of Psychology and Social Behavior, 3340 Social Ecology II, University of California, Irvine, California 92697-7085.
 |
Abstract |
|---|
 TOP Abstract MethodResultsDiscussionAppendixReferences |
|---|
A substantial literature has emerged in recent decades that has explored the implications of social network involvement for health and well-being in later life (Antonucci, 2001
). Many studies of older adults, as well as other age groups, have suggested that supportive relationships often help to mitigate the adverse effects of life stress (House, Umberson, & Landis, 1988). Research has also demonstrated, however, that social relationships can be a source of considerable frustration and disappointment, adding to rather than alleviating emotional distress. Negative exchanges within social networks occur less often than do positive exchanges, but they arouse considerable distress when they do occur (Rook, 1998).
The detrimental effects of negative social exchanges reported in these studies do not appear to be an artifact of participants' mental health. Pagel, Erdly, and Becker (1987) demonstrated, for example, in a short-term longitudinal study of caregivers that negative social interactions predicted caregivers' depression at a 10-month follow-up, controlling for their initial depression levels as well as for health status and demographic characteristics. Moreover, baseline depression did not predict participants' subsequent reports of upsetting interactions (cf. Morgan, Neal, & Carder, 1997). The effects of negative interactions have survived controls in other studies for a variety of demographic characteristics, health status, social desirability, and personality factors (Finch & Zautra, 1992; Rook, 1998). Moreover, Vinokur and Vinokur-Kaplan (1990) found a substantial degree of convergence in the reports of elderly husbands and wives when they were asked to describe the negative exchanges that occurred in the marital relationship. This suggests that participants' reports of the negative social interactions they experience do not simply reflect distorted or negative perceptions.
Given that negative social exchanges have the potential to detract substantially from older adults' health and well-being, it becomes important to learn what makes some older adults more vulnerable than others to such exchanges. Two basic dimensions of vulnerability may be distinguished (Bolger & Zuckerman, 1995; Kessler, 1979). First, some older adults may experience many problematic interactions with members of their social networks, whereas others may experience relatively few such interactions. Thus, they may differ in their exposure to negative exchanges. Second, some older adults may react with considerable distress to conflictual interactions with network members, whereas others may experience only mild distress in response to such interactions. In this sense, they may differ in their reactivity to negative exchanges. These two dimensions of vulnerability thus refer to the occurrence and the impact of negative exchanges.
The goal of the current study was to investigate the predictors of exposure and reactivity to negative social exchanges in the daily lives of a sample of community-residing older adults. Three categories of predictors that have been implicated as potentially important in previous research were emphasized in this study: individual differences, stressful life events, and social network characteristics.
Exposure to Negative Social Exchanges
Only a handful of published studies have provided information on the correlates of older adults' exposure to negative social exchanges. Major background characteristics such as education, income, marital status, religion, residential stability, and perceived health status have been found to be unrelated to older adults' reports of negative exchanges (e.g., Krause & Shaw, 2002; Rook, 1984; Stephens, Kinney, Ritchie, & Norris, 1987), although the level of functional impairment among physically disabled individuals has been linked to rates of family conflict (Turner, 1996). Exposure to negative social exchanges has been found to decline with age in several studies (Krause & Shaw, 2002; Lang & Carstensen, 2002; Pagel et al., 1987; Stephens et al., 1987; Turner, 1996). The evidence for sex has been inconsistent, with some studies finding no effect of sex (e.g., Krause & Shaw, 2002; Stephens et al., 1987) and other studies finding that women report more negative social exchanges than do men (e.g., Pagel et al., 1987). Thus, with the exception of age, demographic characteristics have not emerged as consistent predictors of older adults' exposure to negative social exchanges, although such findings require further replication.
Individual differences have been examined relatively infrequently as predictors of exposure to negative social exchanges in later life, despite the role they might plausibly play in precipitating such exchanges. The greatest attention, in the sparse literature that has emerged to date, has focused on the personality trait of neuroticism, with studies suggesting that more neurotic individuals experience more interpersonal difficulties in their daily lives in young adulthood and middle age (Bolger & Zuckerman, 1995; Suls, Martin, & David, 1998) as well as later adulthood (Smith & Zautra, 2002). Interpersonal sensitivity, conceptualized as heightened sensitivity to social interactions that involve rejection and criticism, was linked to greater exposure to interpersonal stressors in a recent study of older women (Smith & Zautra, 2002). In the current study, information was available about participants' self-esteem, and it is plausible that self-esteem may act as a resource to reduce the likelihood of problematic encounters with others. Individuals with greater self-esteem may feel freer to express their needs for support and companionship and to fend off intrusions and demands, thereby helping to reduce the likelihood of experiencing negative social exchanges. Consistent with this idea, an early study found that less assertive elderly women reported more negative interactions with network members than did more assertive women (Rook, 1984).
Stressful life events may precipitate strains in older adults' transactions with members of their social networks, particularly when they require family members or friends to assume extended responsibilities as support providers (e.g., Kaniasty & Norris, 1993). Few formal tests of this possibility have been conducted. In a large representative sample of older adults, Krause and Jay (1991) found little evidence that life stress was linked to an increased probability of experiencing negative interactions, but they concluded that further tests are needed because their study examined a restricted range of stressors (financial strain and the death of a significant other) and negative interactions (demands and criticism).
Social network characteristics have received some attention as factors that might influence exposure to negative social exchanges. Early views centered on the idea that larger social networks and more frequent contact with network members could increase the opportunities for conflictual interaction, but empirical work has found neither network size nor frequency of social contact to predict negative exchanges (Krause & Jay, 1991; Pagel et al., 1987; Rook, 1984). Subsequent attention to the composition of older adults' social networks has proven more fruitful, with studies tending to implicate kin ties as a prominent (though not exclusive) source of friction. Social networks that contain a higher proportion of kin, relative to nonkin, appear to be associated with reports of more negative interactions (Pagel et al., 1987; Rook, 1990).
Reactivity to Negative Social Exchanges
The predictors of older adults' reactivity to negative social exchanges have received relatively little attention, even though researchers have agreed that older adults' responses to negative exchanges are likely to be characterized by substantial variability. Demographic characteristics may predict responses to negative social exchanges, but empirical data bearing on this possibility are quite limited. Theoretical perspectives on emotion regulation, for example, provide a basis for expecting age to be inversely related to reactivity to such exchanges (Carstensen & Charles, 1998; Lawton, 2001). In addition, some clues exist in the literature to suggest that women may be more upset than men by negative exchanges, although this work has been based on middle-aged samples (e.g., Schuster, Kessler, & Aseltine, 1990).
Research on younger age groups has suggested that neuroticism contributes to greater reactivity to negative interactions with others (Bolger & Zuckerman, 1995; Suls et al., 1998), but this finding was not replicated in two recent studies of older women (Smith & Zautra, 2001, 2002). A narrower personal disposition—interpersonal sensitivity—has been found to be related to older women's greater reactivity to spousal conflict (Smith & Zautra, 2001, 2002). In the current study, I expected that individuals with greater self-esteem would experience less distress in response to negative exchanges with members of their networks. Negative encounters with others are believed to detract from emotional well-being, in part, by threatening one's sense of self-worth. Greater self-esteem, accordingly, should reduce the emotional distress associated with negative social interactions by helping to counter such negative inferences about self-worth.
Life stress has also been implicated as a factor that might influence reactivity to negative social exchanges. Older adults whose coping resources are taxed by the need to deal with stressful life events may react more strongly to negative interactions with others than older adults who are not contending with such stress. This idea has been tested in several studies, with mixed results to date. Some studies have found no evidence that life stress amplifies the adverse effects of negative interactions (Finch, Okun, Barrera, Zautra, & Reich, 1989; Okun, Melichar, & Hill, 1990), and other studies have found that adverse effects of negative interactions are greater among older adults experiencing life stress (Kiecolt-Glaser, Dyer, & Shuttleworth, 1988). Further work is needed to help resolve ambiguities about the possible influence of life stress on reactivity to negative social exchanges.
Social network characteristics have begun to receive attention as possible predictors of older adults' reactions to negative exchanges. The general supportiveness or nonsupportiveness of the network may influence the distress aroused by specific negative interactions, although competing views of the direction of such influence have been offered. One perspective suggests that negative interactions that occur in the context of a generally supportive network arouse more distress (because their very rareness makes them salient) than do negative interactions that occur in less supportive networks (Rook, 1990). Consistent with this, Pagel and colleagues (1987) found that among older adults whose social networks provided a good deal of support, negative interactions with network members were strongly related to greater dissatisfaction with the network. Among older adults whose social networks provided less support, negative interactions with network members were only weakly related to network dissatisfaction. Studies of marital interaction have similarly suggested that reactivity to a spouse's negative behaviors is greater in marriages in which such behaviors are rare (Jacobson, Follette, & McDonald, 1982). A pattern of predominantly positive exchanges may increase the salience and, therefore, the impact of comparatively rare negative exchanges.
A competing perspective suggests that reactivity to negative exchanges should be reduced when the network also functions as a source of positive interactions because positive interactions will help to offset or buffer the adverse effects of negative interactions. In an early test of this latter idea, Schuster and colleagues (1990) found evidence that positive social exchanges did appear to dampen the distress-arousing effects of negative exchanges. Similar findings suggesting that the overall supportiveness of the network may help to buffer the adverse effects of negative social exchanges have been reported in several recent studies (e.g., Okun & Keith, 1998).
The Current Study
The current study sought to build on and extend this prior work by investigating the predictors of older adults' vulnerability to negative social exchanges in their daily lives, examining exposure and reactivity as complementary dimensions of vulnerability (see Appendix, Note 1). The daily diary methodology is uniquely suited to deriving within-person estimates of exposure and reactivity that can be modeled as a function of between-person predictors of interest (Bolger & Zuckerman, 1995; Gable & Reis, 1999). Three categories of predictors were examined: individual differences (emphasizing self-esteem), stressful life events, and social network characteristics. The study also sought to extend prior research by investigating the predictors of change in exposure and reactivity to negative social exchanges over a 1-year period.
|
Method |
|---|
|
TOPAbstractMethodResultsDiscussionAppendixReferences |
|---|
); therefore, the data from the three groups of study participants were combined for analysis at T2. At the baseline assessment, the participants ranged in age from 60 to 89 years, with a mean of 70.27 years. The majority of the sample was Caucasian (88%) and female (64%). Most participants were unmarried (63%), and this unmarried group consisted largely of widowed individuals (70%). Roughly one third (31%) of the sample had a high school education or less, and 50% had attended some college or had completed a college degree. The study participants generally perceived themselves to be in good health, with the majority describing their health as either excellent (45%) or good (45%). Consistent with their good health, most lived in their own homes or apartments (92%), with the rest living either in the home of an adult child or other relative (4%) or in an apartment complex for older adults (4%); none lived in institutional settings.
Procedure
In-person interviews lasting approximately 90 min were conducted annually with each participant. The interviews asked about participants' background characteristics and health status, social network characteristics, life stress, and psychological health. In addition, for a 2-week period after the interview, participants were asked to complete a form at the end of each day that assessed their mood and any positive or negative social interactions they might have experienced during the day. Participants were shown how to use the daily checklists before being given a packet of 14 forms to complete in the next 2 weeks. Participants who failed to complete at least 12 of the daily forms (n=12) were dropped from the analysis (cf. Sheldon, Ryan, & Reis, 1996). The diary data were analyzed using hierarchical linear modeling (Bryk & Raudenbush, 1992), which can accommodate data sets with an unequal number of observations across participants. Thus, it was not necessary to impute missing data for participants with 12 or 13 days of diary data.
As part of the larger study, participants also took part in assessment protocols that evaluated their memory and neuropsychological functioning, cardiopulmonary functioning, perceptual and motor skills, and sleep patterns and disorders. These data on participants' psychobiological functioning are not examined in the current study.
Daily Diary Measures
Daily mood
Daily mood was assessed with the 20-item Affectometer 2 (Kammann & Flett, 1983), which consists of 10 positive mood items (satisfied, optimistic, useful, confident, understood, loving, free-and-easy, enthusiastic, good-natured, clear-headed) and 10 negative mood items (discontented, hopeless, insignificant, helpless, lonely, withdrawn, tense, depressed, impatient, confused; see Appendix, Note 3). The scale has been found to correlate with other measures of well-being and to have adequate test–retest reliability (Kammann & Flett, 1983). For each mood item, participants indicated on a 5-point scale (1 = not at all, 5 = all of the time) how often they felt that way during the day. A factor analysis of the mood ratings (aggregated by item over the 14-day period) revealed two distinct factors, with eigenvalues of 9.61 and 4.54, that accounted for 71% of the variance in mood ratings. The factor loadings indicated that the items formed positive and negative mood factors, respectively. The negative mood items were averaged for each day to create a composite measure of daily negative mood (
=.93). Although a mood balance scale can be constructed by subtracting the negative mood score from the positive mood score, the current study examined negative mood because two-factor theories of psychological well-being (Lawton, 1983; Zautra & Reich, 1983) and emerging empirical evidence (e.g., Ingersoll-Dayton, Morgan, & Antonucci, 1997) have suggested that negative social exchanges are most strongly linked to congruent (negative) dimensions of well-being.
Daily social exchanges
The daily checklists included 14 items that assessed positive exchanges and 6 items that assessed negative social exchanges. The negative items asked participants whether, during the day, someone had upset them, argued with them, behaved in an unkind manner toward them, refused to provide help in response to a request for assistance, made them angry, or hurt their feelings or whether they had been stuck spending time with someone they did not enjoy. Responses to these items (coded 0 = exchange did not occur, 1 = exchange did occur) were summed for each day to form a measure of daily negative social exchanges (=.81, aggregated over 14 days).
Person-Level Measures
Self-esteem
Self-esteem was assessed with the 10-item Rosenberg Self-Esteem Scale (Rosenberg, 1965), which has been widely used in studies of various age groups, including elderly people. The measure exhibited good internal consistency in this sample ( =.80).
Life stress
Twenty-six items asked whether or not the participant had experienced a variety of stressors in the past 6 months. The items represented major life events (e.g., death of a loved one, personal injury or illness, residential relocation, financial problems, criminal victimization), as well as everyday hassles and difficulties (e.g., household maintenance problems, transportation problems, problems with paperwork or red tape, problems with a service provider such as a doctor or lawyer, weather-related difficulties). The items were drawn from widely used measures of life stress (e.g., Dohrenwend, Krasnoff, Askensay, & Dohrenwend, 1978) and from measures developed to capture stressors that may be common in later adulthood (Aldwin, 1990), such as household maintenance problems or difficulties dealing with bureaucracies or service providers. Items that reflected interpersonal problems (e.g, divorce or marital separation, worsening of family relationships or friendships, conflicts with coworkers/covolunteers) were excluded to avoid a possible confound with the daily social exchange and social network measures (cf. Thoits, 1982). The resulting 17 items were scored dichotomously (0 = event did not occur, 1 = event did occur) and were summed to form a composite measure of life stress.
Social network characteristics
Participants' social networks were assessed using the name-eliciting method developed by Fischer and his colleagues (McCallister & Fischer, 1978). This technique has been used in a number of studies of older adults (e.g., Finch et al., 1989; Rook, 1984; Stephens et al., 1987). A series of questions asked participants who, if anyone, performed various positive and negative functions in their lives. Participants provided the first names of any individuals who performed the function specified in each question. After the complete set of name-eliciting questions had been presented, the interviewer compiled a list of the unique names that had been generated and presented a carbon copy for the participant to review. A question about whether anyone important to the participant was missing from the list yielded few additional names (Mdn = 2.00), typically grandchildren, nieces, or nephews. Thus, the set of name-eliciting questions appeared to succeed in capturing participants' most important social ties. This method also avoided confounding the number of relationship functions performed by the network as a whole with the number of network members who performed these functions, as recommended by House and Kahn (1985).
The supportive and nonsupportive qualities of the social network were assessed in terms of the positive and negative functions performed by the network, as well as the number of people who performed these functions. Sixteen questions assessed the extent to which the network functioned as a source of emotional support, instrumental support, and companionship. The support questions asked, for example, if there was anyone the participants knew to whom they could turn if they felt depressed and needed to be cheered up, needed to discuss personal problems or concerns, needed advice about an important matter, needed assistance during times of illness, needed to borrow money in an emergency, or needed help with a task (such as help with household chores or help with transportation). The companionship questions asked to whom, if anyone, the participants could turn if they wanted to get together to go out somewhere with someone, wanted to have a friendly conversation with someone on the telephone, or wanted to get together to have a good time. The questions were scored dichotomously (0 = no one named, 1 = one or more individuals named) and were summed to create a composite measure of the number of positive functions performed by the network ( =.76).
An additional eight items asked about problematic interactions with network members. For example, items asked who, if anyone, tended to criticize the participant, acted too busy or made excuses when the participant needed help, or tried to take advantage of the participant by making unreasonable demands or manipulating situations. Fewer items assessed negative interactions because prior research has indicated that older adults report relatively few negative interactions with their network members (Rook, 1990), and repeated questioning about such interactions can cause frustration. The questions were scored dichotomously (0 = no one named, 1 = one or more individuals named) and were summed to create a composite measure of the number of negative functions performed by the network ( =.58). The modest alpha reflects the relatively small number of items and their heterogeneous content.
The detailed network assessment also yielded information about the particular network members who performed the positive and negative functions described above. From this information, measures were constructed of the number of network members who performed positive functions only, the number of network members who performed negative functions only, and the number of network members who performed positive and negative functions.
Two aspects of participants' involvement with kin versus friends were assessed. For each network member identified through the name-eliciting questions, the role relationship to the participant was assessed, and this information was compiled to determine the number of kin ties and the number of friend ties in the network. To complement these quantitative indicators of kin versus nonkin involvement, participants were asked to rate their satisfaction with kin ties and their satisfaction with friend ties (1 = not all satisfied, 7 = very satisfied).
Demographic characteristics and health status
Standard demographic characteristics (age, sex, race, education, marital status) were assessed for possible inclusion as control variables in the analyses of predictors of exposure and reactivity. Significant health problems that participants might have experienced were assessed by asking whether they had been diagnosed with any of 10 chronic conditions (e.g., high blood pressure, diabetes, arthritis; 0 = health problem not reported, 1= health problem reported). Responses were summed to form a measure of the number of chronic health problems.
|
Results |
|---|
|
TOPAbstractMethodResultsDiscussionAppendixReferences |
|---|
A comparison of participants who did versus did not complete the T2 assessment revealed no differences with respect to age or social network characteristics (size, composition, supportiveness, or nonsupportiveness). Some significant differences did emerge, however, that indicated that participants who did not complete the T2 assessment were less likely to have received education beyond the high school level (47.8% vs 71.7%), 
2(1) = 10.51, p <.01; were less likely to be married (33.3% vs 77.3%), 2(1) = 22.02, p <.0001; and reported more health problems (Ms = 2.65 vs 2.09), t(179) = 2.20, p <.05; more negative mood (Ms = 1.62 vs 1.38), t(147) = 2.18, p <.05; and more life stress (Ms = 4.92 vs 3.84), t(179) = 2.24, p <.05. The two groups of participants did not differ significantly in terms of their exposure or reactivity to daily negative social exchanges (operationalized as described below). The differences that emerged are consistent with prior research on the characteristics of older adults who drop out of longitudinal studies (e.g., Cooney, Schaie, & Willis, 1988), and they suggest that the findings of the current study may not generalize to older adults with less education, worse physical and emotional health, and more life stress.
Descriptive Analyses
Table 1 summarizes the means, standard deviations, and intercorrelations of the key study variables. As shown in this table, negative exchanges were relatively rare, with a per-day average of less than one such exchange. Across the 14-day assessment period, participants reported an average of 4.40 negative social exchanges (SD = 5.15). The correlations shown in Table 1 provide some support for the hypothesized associations. Participants who experienced more negative daily exchanges reported more negative mood, lower self-esteem, more life stress, more unsupportive network ties, more negative functions performed by the network, and less satisfaction with their friend ties (but not with their kin ties). The T1–T2 correlations also suggest that participants' level of exposure to negative exchanges and their level of negative mood, as assessed with daily diary measures collected 1 year apart, were both fairly stable (rs =.58 and.70, respectively, both ps <.001).
|
Analyses of Daily Diary Data
Hierarchical linear modeling (HLM; Bryk & Raudenbush, 1992) allows researchers to analyze simultaneously both within-person variations (such as variations within individuals in the number or mood-related effect of daily negative social exchanges) and between-person differences (such as differences between individuals in levels of self-esteem or life stress; Bolger & Zuckerman, 1995; Gable & Reis, 1999). The within-subject level of analysis was used to estimate each participant's exposure and reactivity to daily negative social exchanges. The between-subject level of analysis was used to examine whether exposure and reactivity were affected by individual differences in self-esteem, life stress, or social network characteristics (cf. Bolger & Zuckerman, 1995; Sheldon et al., 1996; Suls et al., 1998). Conventional linear models cannot simultaneously analyze such within- and between-person sources of variation (Bolger & Zuckerman, 1995; Bryk & Raudenbush, 1992). The data were analyzed with the HLM 4.0 computer program (Bryk, Raudenbush, & Congdon, 1998). Each within-person predictor was centered around the mean for the person's 14 days of diary data, and each between-person predictor (excluding dummy-coded predictors) was centered around the sample mean before analysis (cf. Suls et al., 1998).
Predictors of Exposure to Daily Negative Social Exchanges: Cross-Sectional Analyses
The first HLM analyses tested the idea that exposure to negative social exchanges would be greater among individuals with lower self-esteem, more life stress, less supportive social networks, more kin ties, and less satisfying kin and friend ties. At the within-person level of analysis, each participant's exposure to negative exchanges was expressed in following model (cf. Bolger & Zuckerman, 1995): Nd = a0 + ed, where Nd refers to the number of negative social exchanges on Day d, a0 refers to the participant's mean level of negative exchanges across all diary days, and ed refers to a residual error component (or the daily fluctuations of the participant's level of negative exchanges around his or her personal mean; Almeida & Kessler, 1998).
In the between-person level of analysis, for any participant i, the within-person coefficient representing exposure, a0, is modeled as a function of self-esteem (or life stress or social network characteristics). This is summarized in the following equation, showing self-esteem as a predictor (cf. Bolger & Zuckerman, 1995): a0i = b0 + b1SEi + qi. The participant's exposure to negative exchanges (a0i) reflects an intercept (bo), an effect of self-esteem (b1), and a random effect (qi). Subsequent analyses tested the same model but substituted life stress or social network characteristics for self-esteem.
The results of these analyses (summarized in Table 2, T1 Exposure column) revealed that self-esteem was unrelated to participants' exposure to daily negative social exchanges, but greater life stress was related to greater exposure. Greater exposure to daily negative exchanges was also associated with several characteristics of participants' social networks, including having a network with more members who performed negative functions only and more members who performed both negative and positive functions and having a network in which a broader range of negative functions was performed. Neither the number of kin or friends ties, nor the level of satisfaction with these two categories of ties, predicted exposure to negative social exchanges.
|
Predictors of Reactivity to Daily Negative Social Exchanges: Cross-Sectional Analyses
The next HLM analyses examined the idea that reactivity to negative social exchanges would be greater among individuals with lower self-esteem, more life stress, less supportive social networks, more kin-dominated networks, and less satisfying kin and friend ties. At the within-person level of analysis, participants' reactivity to negative exchanges was expressed in the following model (cf. Bolger & Zuckerman, 1995; Suls et al., 1998): NMd = a0 + a1NMd-1 + a2Nd + ed where NMd is negative mood on Day d, NMd-1 is negative mood on the preceding day, Nd is the number of negative social exchanges on Day d, and ed is the random error component for Day d; a0 is the intercept, a1 is the effect of the prior day's mood, and a2 is the effect of negative social exchanges on Day d. Daily diary studies often include a control for possible carryover of distress from the prior day (e.g., Sheldon et al., 1996; Suls et al., 1998). Thus, the outcome variable, NMd, represents a residualized change in mood from the preceding day to the current day (cf. Bolger & Zuckerman, 1995; see Appendix, Note 5).
At the between-person level of analysis, the reactivity coefficient, a2i, for any participant i is examined as a function of self-esteem, as summarized in the following model (cf. Bolger & Zuckerman, 1995; Suls et al., 1998): a2i = b0 + b1SEi + si. In this equation, reactivity reflects an intercept (b0), an effect of self-esteem (b1i), and a random effect (si). The same model was tested in subsequent analyses in which life stress and social network characteristics were substituted for self-esteem. As described earlier, the analyses included controls for participants' demographic characteristics and health status.
The results of these analyses indicated that, contrary to expectation, life stress was associated with less, rather than more, reactivity to daily negative social exchanges (as shown in Table 2, T1 Reactivity column). Self-esteem, in contrast, was strongly associated with less reactivity to such exchanges. In addition, participants whose social networks included more individuals who performed positive exchanges exclusively, as well as individuals who performed both positive and negative exchanges, exhibited less reactivity to negative exchanges. Participants whose networks performed a broader range of positive functions also exhibited less reactivity. Thus, the overall supportiveness of participants' networks appeared to play a role in buffering the mood-depressing effects of negative social exchanges experienced in the course of their daily lives. Neither the number of kin nor the number of friends in the network was associated with reactivity, but greater satisfaction with both kin and friend ties was related to less reactivity.
In addition, although the effects of the control variables are not shown in Table 2, age and marital status were significantly related to reactivity in these analyses, with older and married participants exhibiting less reactivity (Bs for age ranged from -.04 to -.06, all ps <.001; Bs for marital status ranged from -.23 to -.37, all ps <.01). The number of chronic health problems was unrelated to reactivity in these analyses, as was sex.
Predictors of Exposure to Daily Negative Social Exchanges: Longitudinal Analyses
A relatively unique feature of this study was the availability of daily diary data collected at two points in time, 12 months apart. The final analyses accordingly sought to examine whether changes in self-esteem, life stress, or social network characteristics over the 1-year period were associated with changes in either exposure or reactivity to daily negative social exchanges. To analyze predictors of change in exposure to daily negative social exchanges, the T1 and T2 diary data were merged into a single file, with a variable appended to each day's data that designated the time (year) of assessment (0 = T1, 1 = T2). The within-person level of analysis then tested the exposure model presented earlier, with the time variable added to the model.
Time in itself was not expected to influence exposure to negative social exchanges, but time-covarying changes in the key predictors examined could have influenced exposure. Participants who experienced a decline in self-esteem or an increase in life stress from T1 to T2, for example, might have experienced more negative social exchanges in their daily lives. In the longitudinal between-person analyses, the time effect accordingly was treated as the outcome, modeled as a function of each T2 predictor, controlling for its T1 counterpart. For example, a significant effect of self-esteem at T2, controlling for self-esteem at T1, would indicate that a change in self-esteem was associated with a change in exposure to negative social exchanges from T1 to T2. Parallel analyses substituted T1 and T2 levels of life stress and social network characteristics for T1 and T2 self-esteem. These analyses included the controls described earlier for demographic characteristics and health status.
The analyses revealed few significant predictors of change in exposure to negative social exchanges. No significant effects emerged for self-esteem or life stress, but an increase in satisfaction with friends from T1 to T2 was associated with reduced exposure to negative exchanges, as shown in Table 3 (T1 – T2 Change in Exposure). In contrast, an increase in the number of positive functions performed by network members was associated with an increase in exposure to negative exchanges from T1 to T2.
|
Predictors of Reactivity to Daily Negative Social Exchanges: Longitudinal Analyses
The analyses of predictors of change in reactivity to daily negative social exchanges tested the reactivity model presented earlier, with the time variable and an interaction term (Time x Negative Exchanges) added to the model. The interaction term estimated the extent to which the impact of negative social exchanges differed at T1 and T2 and, accordingly, was treated as the outcome variable in the longitudinal between-person analyses. Time in itself was not expected to influence reactivity to negative social exchanges, but changes from T1 to T2 in the predictors examined could have influenced reactivity. For example, an increase in life stress could have increased reactivity to negative social exchanges. This would be represented by a significant effect of T2 life stress, controlling for T1 life stress, in predicting the Time x Negative Exchanges interaction effect derived from the within-person analyses. Parallel analyses substituted T1 and T2 levels of life stress and social network characteristics for T1 and T2 self-esteem, with controls included for participants' T2 demographic and health status characteristics (see Appendix, Note 6).
The analyses revealed that only one of the predictors examined was related to change in reactivity to negative exchanges. Increased satisfaction with friends was associated with reduced reactivity over the 1-year period, as shown in Table 3 (T1–T2 Change in Reactivity). None of the T2 control variables was associated with change in reactivity to negative social exchanges, although a separate analysis that examined change in the number of chronic medical conditions revealed that an increase in such conditions was associated with increased reactivity to negative social exchanges (B =.04), t(115) = 2.50, p <.01.
|
Discussion |
|---|
|
TOPAbstractMethodResultsDiscussionAppendixReferences |
|---|
Predictors of Exposure to Negative Social Exchanges
Older adults who experienced more life stress and whose networks were less supportive (as reflected both in the membership of the network and in the functions performed by network members) reported more negative social interactions in the course of their daily lives. These associations emerged in cross-sectional analyses that included controls for participants' age, sex, marital status, and number of chronic health problems and that examined a measure of life stress pruned of stressors of an interpersonal nature. Participants who reported more chronic health problems also were somewhat more likely to report experiencing more daily negative exchanges.
Contrary to expectation, self-esteem was unrelated to the number of negative exchanges reported. In addition, neither the kin–nonkin composition of the network nor the degree of satisfaction with kin and nonkin ties were related to exposure. It was expected that participants with greater self-esteem would be better able to express their needs to others and to resist unreasonable demands by others, which in turn should contribute to fewer disappointments and conflicts with others. The nonsignificant finding for self-esteem may mean either that self-esteem does not facilitate social exchanges in this way in later life or that the kinds of negative daily encounters that participants experienced were unrelated to their ability to express or assert themselves. In addition, the nonsignificant findings for the measures of kin versus nonkin involvement suggest that neither greater involvement with kin nor less satisfying kin ties created the conditions for more negative exchanges in the course of participants' daily lives, at least in the particular 2-week period captured by the daily diary assessments.
The longitudinal analyses indicated that exposure to daily negative social exchanges was fairly stable (cf. Krause & Shaw, 2002) over a 1-year period (r =.58, as indicated earlier) and that the predictors examined were related only modestly to changes in exposure. An increase in satisfaction with friends over the 1-year period was associated with reduced exposure to negative social exchanges. This may reflect a process of selective disengagement from friends with whom interactions had become strained, consistent with a central premise of socioemotional selectivity theory that older adults proactively manage their social lives by concentrating on those relationships that offer the greatest emotional rewards (Carstensen & Charles, 1998). Alternatively, the link between increased satisfaction with friends and reduced exposure to negative exchanges could indicate that strains in existing friendships had been resolved in the intervening year through processes other than relationship disengagement or dissolution.
The other significant predictor of a change in exposure to negative social exchanges involved the number of positive functions performed by the social network; an increase in the positive functions performed by network members during the 1-year period was associated with an increase in exposure to negative exchanges. This finding is contrary to what might have been expected, but it could reflect an increase during the year in the support needs of some participants, requiring network members to perform a broader range of support functions and, in the process, contributing to tensions in the network (Kaniasty & Norris, 1993).
The relatively sparse findings for the longitudinal analyses of exposure to negative exchanges may reflect not only the stability of the measure of exposure but also limitations of the specific predictors examined. For example, the measure of life stress may have omitted important stressors that could have precipitated an increase in negative exchanges, although a supplemental longitudinal analysis that used a less restrictive measure of life stress (one that included interpersonal losses and conflicts) replicated the nonsignificant finding. In addition, some of the predictors examined were themselves fairly stable, limiting their potential to account for changes in exposure to negative exchanges. It is also possible that circumstances in participants' lives could have led to short-term increases (or decreases) in negative social exchanges at some point during the course of the year, but these changes in patterns of social interaction might have dissipated by the time of the 1-year follow-up assessment. Future longitudinal investigations of the predictors of exposure to negative social exchanges would benefit from examination of a broader range of predictors and also from more frequent and closely spaced assessments.
Predictors of Reactivity to Negative Social Exchanges
The analyses of reactivity suggested that somewhat different dynamics may underlie older adults' affective responses to negative social exchanges in their daily lives. Life stress appeared to increase participants' likelihood of experiencing negative exchanges (as indicated in the exposure analyses), but it did not appear to exacerbate the distress aroused by these exchanges. In fact, the cross-sectional data suggest that negative exchanges were less, rather than more, distressing when they occurred in the context of other life stress. This finding, although not expected, mirrors the results of a daily diary study of married couples in which daily stressors were found to be less upsetting when they occurred on days characterized by more, rather than less, stress (Bolger, DeLongis, Kessler, & Schilling, 1989). The researchers interpreted this finding as evidence that participants had reached an emotional plateau, such that the occurrence of further stressors (within the range studied) would not add to further distress. In contrast, stressors experienced singly, rather than in combination, were found to have a greater impact on negative mood, presumably because emotional habituation had not yet occurred (Bolger et al., 1989). Similar habituation processes may account for the inverse association between life stress and reactivity to daily negative exchanges in the current study, although further replication is needed in view of the counterintuitive nature of the finding.
Self-esteem and the supportiveness of the network appeared to operate, as anticipated, to reduce participants' reactivity to negative interactions with others. How these intrapersonal and interpersonal resources function to cushion the adverse effects of troubling social interactions is unclear, although it is possible that they encourage more benign appraisals of the interactions or compensatory responses that help to restore mood. For example, access to a broadly supportive social network provides opportunities to seek support about specific problematic interactions that have occurred with others (Okun & Keith, 1998), to experience affirmation of self-worth, and to participate in mood-enhancing companionship (Larson, Mannell, & Zuzanek, 1986). The fact that married participants exhibited less reactivity to negative social exchanges is consistent with this interpretation, as is the evidence that greater satisfaction with family and (especially) friend ties was related to reduced reactivity to negative social exchanges.
The effect of age per se was not a central focus of the study, but age was found to be inversely related to reactivity to negative social exchanges. This finding converges with evidence from an emerging body of research on emotion regulation in later life, which suggests that as people age they become more skillful at managing their emotions to limit emotional distress (Carstensen & Charels, 1998; Lawton, 2001).
The longitudinal analyses that sought to investigate the predictors of change in reactivity identified relatively few significant associations. An increase in satisfaction with friends was related not only to reduced exposure to negative exchanges, as noted above, but also to reduced reactivity to such exchanges. This may reflect the resolution of tensions in participants' friendships over the 1-year period, or it may indicate that more satisfying friendships helped to buffer the mood-depressing effects of negative exchanges experienced in other relationships. The longitudinal analyses also revealed that an increase in the number of chronic health problems was associated with increased reactivity to negative social exchanges. Thus, declining health appears to amplify the adverse effects on mood of negative social exchanges.
The limited findings from the longitudinal analyses of reactivity may be partly attributable to the stability of the predictor variables examined, because reactivity per se exhibited only modest stability from T1 to T2 (r =.35). For example, although self-esteem was associated with reactivity in the cross-sectional analyses, its stability in this sample over the 1-year interval (r =.71, as indicated in Table 1) undoubtedly limited its ability to predict changes in reactivity. Future longitudinal studies of reactivity to negative social exchanges would benefit from a focus on potential antecedents that exhibit greater variability over time.
Limitations of the Study
Several limitations of the study should also be noted. The sample is not representative of the elderly population, and participant attrition from T1 to T2 further limits the generalizability of the findings. The participants who were omitted from the analyses tended to have fewer psychosocial resources, worse emotional and physical health, and greater life stress than did participants who were included in the analyses. Although it is difficult to know how such selective attrition may have affected the results, it is plausible that it could have worked against finding significant associations by contributing to truncated variability in some of the key predictors of exposure and reactivity. If so, then the current findings may underestimate the effects of the predictors investigated. Whether or not this is the case requires replication studies.
In addition, some of the measures were limited, which may have compromised tests of the hypothesized associations. For example, the measure of life stress was a simple checklist that assigned equal weight to major and minor stressors. This could have led to an underestimation the effects of life stress on exposure or reactivity to negative social exchanges. Similarly, the assessment of daily negative exchanges was brief and may have overlooked negative exchanges that are important in this population. Two of the six negative exchange items used affective wording (asking participants whether someone had upset them or had made them angry or hurt their feelings), and an assessment strategy that prunes references to affect from the item wording would be preferable in future research. In addition, the daily diary assessment strategy in the current study did not yield information about the specific individuals with whom negative exchanges occurred and therefore did not permit tests of source-specific effects (Okun & Keith, 1998). Moreover, the selection of the particular 2-week interval for conducting the daily diary assessment was arbitrary and may have yielded an atypical sample of participants' customary patterns of exchange with others. This problem is intrinsic to most diary studies, but it raises questions about the accuracy of the resulting estimates of exposure and reactivity. In addition, ambiguities exist concerning the causal direction of the associations reported (cf. Bolger & Zuckerman, 1995; Sheldon et al., 1996; Suls et al., 1998). Negative mood may cause daily social interactions to be perceived as negative or may even contribute to difficulties with others (cf. Suls et al., 1998), although the inclusion of a control for the prior day's mood helps to reduce this concern to some extent (Suls et al., 1998; see Appendix, Note 7).
For all of these reasons, the current study must be considered a preliminary investigation of the factors that affect older adults' vulnerability to troubling interactions with members of their social networks. Despite these limitations, however, the current study illustrates a possible approach for gaining insight into this problem by distinguishing between exposure and reactivity as two complementary aspects of dimensions of vulnerability and by modeling within- and between-person processes simultaneously (Bolger & Zuckerman, 1995). Such an approach, when combined with the survey approaches that have been more common in this literature, should help to shed light on the distinctive dynamics that underlie older adults' experiences of positive and negative interactions with others.
|
Appendix |
|---|
|
TOPAbstractMethodResultsDiscussionAppendixReferences |
|---|
), loneliness and the likelihood of heart disease (Sorkin, Rook, & Lu, 2002), and the effects of involvement in a volunteer role (Rook & Sorkin, in press). An additional study analyzed the diary data to examine the relative strength of associations between emotional health outcomes and positive versus negative social exchanges (Rook, 2001). ). ). Of the 22 effects shown in Table 2, all were replicated except for 2 that involved the prediction of exposure to negative exchanges: the effect of life stress became marginally significant, t(114) = 1.74, p <.08, and the marginal effect of network members who performed positive functions exclusively became nonsignificant. The longitudinal analyses were repeated as well, with controls included for the participants' T1 and T2 depression levels. The effects shown in Table 3 were replicated except that change in satisfaction with friend ties as a predictor of change in exposure to negative exchanges became marginally significant, t(114) = -1.81, p <.07. On balance, the substantive findings for exposure and reactivity show relatively little change even in these stringently controlled analyses.
|
Acknowledgments |
|---|
Received for publication October 28, 2002. Accepted for publication March 15, 1999.
|
References |
|---|
|
TOPAbstractMethodResultsDiscussionAppendixReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  C. K. Chen, M. Sabir, S. Zimmerman, J. Suitor, and K. Pillemer The Importance of Family Relationships With Nursing Facility Staff for Family Caregiver Burden and Depression J. Gerontol. B. Psychol. Sci. Soc. Sci., September 1, 2007; 62(5): P253 - P260. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. J. August, K. S. Rook, and J. T. Newsom The Joint Effects of Life Stress and Negative Social Exchanges on Emotional Distress J. Gerontol. B. Psychol. Sci. Soc. Sci., September 1, 2007; 62(5): S304 - S314. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. Stevens and G. J. Westerhof Partners and others: Social provisions and loneliness among married Dutch men and women in the second half of life Journal of Social and Personal Relationships, December 1, 2006; 23(6): 921 - 941. [Abstract] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
