The Effect of Widowhood on Older Adults' Social Participation: An Evaluation of Activity, Disengagement, and Continuity Theories

Abstract

Purpose: This study evaluated how levels of social participation change as a result of late-life widowhood. Social participation is a multidimensional construct incorporating both formal (e.g., meeting attendance, religious participation, and volunteer obligations) and informal (e.g., telephone contact and social interactions with friends) social roles. Design and Methods: Using data from the Changing Lives of Older Couples study, analyses compared widowed persons to continuously married control participants to evaluate whether widowhood affects older adults' levels of social participation. Results: Widowed persons had higher levels of informal social participation than nonwidowed persons, whereas formal social participation levels were comparable between the two groups. Social participation levels decrease before the death of a spouse, primarily because of poor spousal health, and increase following the loss, because of increased support from friends and relatives. Implications: Maintaining continuity in the realm of social participation is a strategy older adults use to cope with spousal loss; however, not all widowed persons have the same resources to alter their levels of social participation.

Decision Editor: Laurence G. Branch, PhD

Past research has claimed that widowhood is among the most stressful of all life events and requires more psychological and behavioral adjustment than any other life transition (Barrett and Schneweis 1980–1981; D. Gallagher, Thompson, and Peterson 1982–1983; Holmes and Rahe 1967; Thompson, Breckenridge, Gallagher, and Peterson 1984). To date, bereavement studies have focused primarily on the emotional and psychological responses to widowhood (see Stroebe, Hansson, Stroebe, and Schut 2001, for review). In contrast, this article explores the social and behavioral implications of spousal loss. Because older adults have solidified and internalized a lifetime of habits, behaviors, and attitudes (Atchley 1989), the behavioral adjustments associated with late-life bereavement may be one of the most difficult challenges an elderly person faces. Upon widowhood, the survivor must relinquish the status of married person and assume the identity of widow(er). In response to this identity transition, bereaved persons may realign their social networks or alter their social activities. Social interactions outside of the marital relationship may become increasingly salient, thereby increasing bereaved persons' level of social involvement. Alternatively, social relationships may become strained if widow(er)s feel like a "fifth wheel" among married friends, thereby decreasing the bereaved persons' level of social engagement. Using the Changing Lives of Older Couples study (CLOC), we explored how the formal and informal social participation of older married and widowed persons changes over time and whether theoretical explanations from social gerontology are useful in explaining the patterns of change.

Defining Social Participation

According to Rowe and Kahn 1997, active and productive engagement in society is a central component of successful aging. Applied to bereavement, sustained social engagement could also be a critical component of successful adaptation or coping. A positive relationship between social activity and well-being is well documented (Lowenthal and Haven 1968). Higher levels of social participation are associated with reduced levels of suicide (Durkheim, 1897/Durkheim 1951), better physical health and reduced mortality (House, Landis, and Umberson 1988), and higher levels of psychological well-being (Beck and Page 1988). Literature from the symbolic interaction paradigm has contended that social participation is instrumental in creating and sustaining one's self-identity (Markus and Wurf 1987; Mead, 1934/Mead 1967; Swann 1987; Swann and Hill 1982; Thoits 1983). Given the theoretical importance of social participation, it is an important aspect of everyday life to understand. In this study, we defined social participation as social interaction with persons other than a spouse. The term social participation as used here loosely describes what other researchers have termed organizational affiliations, friendship ties, kinship networks, social connectedness, social support, or social integration (Anderson 1983; Bahr and Harvey 1980; Bankoff 1983; Durkheim, 1897/Durkheim 1951; Ferraro 1984; Thoits 1983; Vachon et al. 1982). Our measure of social participation is a multidimensional construct incorporating both formal (e.g., meeting attendance, religious participation, and volunteer obligations) and informal (e.g., telephone contact and social interactions with friends) social roles.

The Effect of Widowhood on Social Participation

The bereavement literature has generally agreed that widowed persons who have higher levels of social support also have higher levels of well-being and life satisfaction (Anderson 1983; Bahr and Harvey 1980; Bankoff 1983; Ferraro 1984; Hershberger and Walsh 1990; Lowenthal and Haven 1968; Vachon et al. 1982). These findings suggest that the more active one is after widowhood, the easier the adjustment process may be. Widowed persons often have increased contact with friends and family during the initial funeral and mourning period (Lopata 1996). However, it is unknown whether widowhood creates only a temporary or a more enduring change in widowed persons' social participation (Ferraro 1984; Hollstein 1998). The current study explores older adults' level of social participation 6 months after the loss of their spouses. It compares widowed persons with continuously married control participants to see if changes in social participation among older adults are due specifically to the event of widowhood or generally to the effect of age or the passage of time.

Theoretical Explanations: Activity, Disengagement, and Continuity Theories

For purposes of this analysis, we have adapted three prominent theories from social gerontology to proffer specific hypotheses regarding the effect of late-life widowhood on social participation. First, activity theory (Cavan, Burgess, Havinghurst, and Goldhammer 1949) claims that as one experiences disability and other age-related declines, social roles may become unattainable. As a way to preserve their self-identity in the face of these deficits, aging adults will replace lost social roles with new, compensatory activities. Thus, activity theory offers the following hypothesis: Although both widowed persons and controls will increase their social participation over time in response to age-related deficits and losses, widowed persons will exhibit higher levels of social participation compared with nonwidowed persons because of simultaneously losing the multiple social roles that were tied to the marriage or the spouse.

Second, disengagement theory (Cumming and Henry 1961) posits that aging adults will withdraw from society and vice versa. The mutual process of disengagement makes room for younger generations and prevents unnecessary disruptions in the social system caused by deaths among the older population. In the context of this analysis, disengagement theory offers the following hypothesis: Although both widowed persons and controls will slowly disengage from social activities as they age, widowed persons will exhibit lower levels of social participation when compared with similarly aged married persons because the experience of losing a spouse serves as a striking reminder of one's own mortality and how disruptive death can be for survivors.

Third, continuity theory (Atchley 1989) rests on the assumption that individuals seek to maintain role stability throughout the life course. Although the aging process may present an individual with changing normative expectations or possible disruptions in the availability of social roles, older adults will attempt to preserve continuity of attitudes, dispositions, preferences, and behaviors throughout their life course. Thus, prior behaviors and attitudes are often the single most significant predictors of present or future behaviors. In terms of widowhood, the most successful widow(er) would be the individual who is most effective in maintaining the lifestyle he or she developed earlier in life. The specific hypothesis generated from continuity theory suggests that widows and nonwidows will not necessarily differ in terms of social participation; rather, prior levels of social participation will determine current levels of social participation.

Gerontologists have debated the explanatory power of these theories since their inception (see Bengston and Schaie 1999, for review). Critics of continuity theory suggest that continuity of lifestyle and role stability is nearly impossible considering the inevitability of role loss through events such as retirement, death of loved ones, and the emptying of the familial nest (Matras 1990). Disengagement and activity theories are criticized for being far too simplistic in their explanations—can one variable really explain why older adults adapt to the aging process in the way they do (Quadagno and Street 1996)? Despite the considerable criticisms that these theories have garnered, the hypotheses stated above offer a theoretical springboard for the proposed analyses while offering a unique opportunity to empirically evaluate the utility of these theories.

Prior Research and its Methodological Limitations

Given the criticisms related to the three theories, it should not be surprising that the empirical evidence is also inconclusive. Chambre 1984 found support for continuity theory, showing that people maintain similar activities as they age. Arens 1982(–1983) and van den Hoonard 1998 found support for disengagement theory, suggesting that widows and widowers spend less time participating in recreational activities than before. S. K. Gallagher and Gerstel 1993 found support for activity theory, showing that widowed women spend significantly more time with a larger number of friends compared with married women. In a small-sample analysis of widows and widowers, Hollstein 1998 identified three typical patterns of change in social networks, with each pattern roughly equivalent to one of the three theories. Ward, Logan, and Spitze 1992 also found that patterns of social involvement vary by marital status, suggesting that married couples may not need additional forms of social support, whereas widow(er)s may seek additional social support to compensate for the loss of spousal intimacy. Given these disparate findings, the role of marital status and the effect of marital transitions in determining older adults' activity levels are unclear.

We believe the empirical literature is inconclusive for four reasons. First, findings across studies may not be comparable because analyses are based on very different types of social participation. Widowhood may affect involvement with particular types of activities and not others. Second, bereavement studies are often based on female-only samples, making it impossible to assess whether adjustment patterns vary by gender. Given the wives' predilection to maintain social networks and kinship ties for the marital couple (Goldscheider 1990; Hagestad 1986; Waite 1995; Waite and Gallagher 2000), men and women may exhibit markedly different behavioral and emotional outcomes in response to widowhood (Miller and Wortman in press; Umberson, Wortman, and Kessler 1992). Third, most research has relied on retrospective accounts of one's social involvement before widowhood. Retrospective accounts reflect the survivors' reinterpretation of their previous life, thus biasing the true levels of activity before the loss. Fourth, empirical analyses related to widowhood and social support rarely use a control group from which to differentiate the effects of widowhood from the effects of advancing age or the passage of time. Given that widowed samples are often older, sicker, and more emotionally distraught than nonwidowed samples, past studies may have attributed reduced levels of social participation among widowed persons to spousal bereavement rather than to selection traits. Without a control group and without consideration of the true levels of baseline social participation, the effects attributed to widowhood may have been overstated in past studies.

Data from the CLOC study make it possible to overcome each of the above limitations and thus provide a more complete answer to how social participation is affected by late-life widowhood. First, the CLOC data include two separate multi-item measures of social engagement (formal and informal social participation) to explore whether the effect of widowhood differs by type of activity. Second, the CLOC sample is stratified by gender. Third, the CLOC study contains an age-matched control group, which distinguishes between widowhood and age effects. Fourth, the quality and breadth of baseline (prewidowhood) data eliminates potential selection bias found in widowhood research. In sum, this article explores how older adults alter their everyday social participation following widowhood and compares these results to the predictions derived from activity, continuity, and disengagement theories.

Methods

Sample

Analyses are based on the CLOC study. CLOC is a prospective study of widowhood that used a two-stage area probability sampling technique to collect information from married individuals in Detroit, Michigan. Eligibility was restricted to noninstitutionalized English-speaking married couples in which the husband was at least 65 years old. Of those sampled, 1,532 individuals completed a baseline interview, yielding a 68% response rate.

Following the baseline interview (1987–1988), the vital status of the respondent's spouse was monitored with monthly death records from the state of Michigan. The National Death Index was used to verify all spouses' deaths. Those respondents who lost spouses during the study period were reinterviewed 6, 18, and 48 months after the deaths. Controls from the original sample, individually matched on age, race, and sex, were also reinterviewed at comparable times. Because the Wave 1 interview occurred 6 months after the death of the spouse, not 6 months after the initial baseline interview, there are significant variations in interview timing depending on when spouses died in relation to the baseline interview. Baseline assessments were more temporally distant for those who lost their spouses at later dates. To address the differences in interview timing, all analyses controlled for the duration (in months) between the baseline interview and the Wave 1 interview. This control variable was not significant in any of the analyses, nor was it significant when interacting with widowhood status.

Analyses were based on a sample of 297 older adults (217 women and 80 men) who participated in both the baseline and Wave 1 interviews. The analytic sample included 210 widowed persons and 87 nonwidowed controls. Women were oversampled in order to maximize the number of bereaved respondents during the 5-year study period. Controls were available for only some of the bereaved spouses because funding for the control sample was cut from the proposed budget and not reinstated until halfway through the data collection period for Wave 1. A final centered weight, adjusting for unequal probabilities of selection and nonresponse, was applied to all data before analyses were performed. The unweighted sample size was 333, with 249 widowed persons and 84 controls.

Measures

Dependent Variables

Two statistically and conceptually distinct dimensions of social participation were considered. Informal social participation (α = .52) is an index consisting of the following: (a) How often do you get together with friends, neighbors, or relatives and do things like go out together or visit in each other's homes? and (b) in a typical week, about how many times do you talk on the telephone with friends, neighbors, or relatives? Formal social participation (α = .71) included the following three items: (a) How many hours did you spend doing volunteer work during the last 6 months? (b) how often do you attend meetings or programs of groups, clubs, or organizations that you belong to? and (c) how often do you attend religious services? Each item was measured with a five-category response ranging from at least once a day to less than once a month. The scales were standardized, with higher values indicating higher levels of social participation. Although unstandardized scales produced similar findings, we chose to use the standardized dependent variables for ease of interpretation and comparability.

Predictor Variables

Widowhood identified those persons who became widowed between the baseline and Wave 1 interviews. Baseline formal and informal social participation were identical to the two scales described above, except they were based on baseline responses rather than Wave 1 responses.

Control Variables

Both the social participation and the widowhood literature have asserted that demographic characteristics are important determinants of late-life adjustment (Arens 1982–1983; Hershberger and Walsh 1990; House 1987; Lopata 1996). Therefore, all analyses controlled for sex (a dichotomous variable indicating female), age (measured continuously in years at the time of the baseline interview), education (a continuous measure ranging from 3 to 17 completed years of schooling), total household income at baseline (natural log of income), home ownership at baseline (a dichotomous variable where 1 = owns home), and race (a dichotomous variable indicating White).

Confounding Variables

Analyses also included several variables thought to expose potentially spurious relationships between widowhood and social participation levels. Variables considered in the analyses include individual- and couple-level health dynamics, individual personality characteristics that may make one more resilient or "protected" in times of stress (Caspi 1987; Rodin 1987), and structural barriers or constraints that may affect one's ability to engage in social activity. Widowed persons may significantly differ from nonwidowed persons on these variables; therefore, the inclusion of these confounding variables in the analyses is imperative to distinguish the effects of widowhood from those associated with age-related declines. Other couple-level characteristics such as the quality of the marital relationship or the nature of the death itself are important predictors of grief (Carr et al. 2000; Carr, House, Wortman, Nesse, and Kessler 2001), but such variables did not account for the variation in social participation following the loss and thus were not included in the analyses. Presumably, the baseline measures of social participation sufficiently captured the relevant qualities of the marriage and the nature of the death. With the exception of depression, all the following variables were measured at baseline.

To assess the role of individual- and couple-level health characteristics, analyses included three measures of respondent's health and two measures of spouse's health. Activity limitation (α = .77) was a standardized scale measuring the respondent's functional health and disability, with higher values indicating higher levels of activity of daily living impairment. Self-rated health was a dichotomous variable indicating poor or fair baseline health. The respondent's depression (α = .81) was assessed with a subset of 9 negative items from the 20-item Center for Epidemiological Studies–Depression scale (Radloff 1977), with higher values indicating higher levels of depression at Wave 1. Although the three measures of individual health were significantly correlated (activity limitation and self-rated health = .37; activity limitation and depression = .13; self-rated health and depression = .16; p < .05), each tapped into a distinct aspect of health that may influence one's social participation levels. Activity limitation described physical impediment to activity performance, depression was associated with reduced energy levels and activity involvement, and self-rated health was a global measure of overall health assessment. Two additional health measures were included to assess the dynamic role of couple-level health characteristics. Spouse health indicated those respondents who reported their spouse's baseline health as poor or fair. Provided care for spouse was a dichotomy indicating those respondents who provided in-home care for a spouse. Spouses who provided at least 1 hr of care per week were identified as the baseline caregivers.

One personality measure was included in the analysis. Extraversion (α = .53) was a standardized scale representing how emotionally connected, socially active, and outgoing the respondent was. Higher values indicate higher levels of extraversion. Other personality characteristics such as self-esteem, openness, agreeableness, conscientiousness, and emotional stability were initially included, but were later excluded because they were insignificant predictors of social participation.

Finally, three variables represent possible constraints to activity performance. Employed was a dichotomous variable indicating persons who worked for pay at baseline. Employment may enhance social participation by expanding one's social network or limit it by representing a competing use of free time. Drive automobile was a dichotomous variable indicating those persons who drove an automobile at baseline. Older adults who are able to drive themselves may be more socially engaged as they have the means for self-determined transportation. No children identified respondents who had no living children. Children may encourage their elderly parents to remain actively engaged in social activities.

Analytic Plan

The analyses used a variety of descriptive techniques to describe the patterns of social participation following the loss of a spouse. Some analyses compared widows with controls in order to differentiate the effects of widowhood from aging, whereas additional analyses compared subgroups within the widowed sample to assess the differential responses to widowhood. Using ordinary least squares (OLS) regression techniques, we compared widowed persons with controls to estimate formal and informal social participation levels 6 months postwidowhood. Estimated regression coefficients for widowhood and baseline social participation were evaluated in a test of the three competing hypotheses. If the effect of widowhood was positive, support for the activity theory would be achieved; if the effect of widowhood was negative, support for the disengagement theory would be achieved; and if the effect of baseline social participation was positive, support for the continuity theory would be achieved.

Results

Sample Characteristics

Table 1 demonstrates that the widowed and nonwidowed samples were remarkably similar in terms of major demographic characteristics. The average respondent was approximately 70 years old with nearly a high school education and an annual household income in the low $20,000s. Most owned their homes and had at least one child. More than two thirds drove automobiles, and few (<15%) were employed. Although respondents did experience moderate activity limitation levels, only a quarter to a third of respondents reported their health as being fair or poor. A few notable, but expected, differences existed between the two samples. Widowed persons had significantly higher levels of depression compared with the controls (p ≤ .001). Widowed persons were more likely than controls to have provided care for their spouses at baseline (p ≤ .001) and to report their spouses' baseline health as poor or fair (p ≤ .001).

Fig. 1 graphically illustrates the mean levels of social participation at baseline and Wave 1 for both the widowed and nonwidowed samples. Compared with the still-married persons, the informal social participation of widow(er)s was significantly lower at baseline (.01 compared with .23, p ≤ .05) and relatively similar at Wave 1 (.03 compared with −.11, p ≤ .10). On the other hand, formal social participation did not differ by marital status. Using a repeated measures t test, we found that controls exhibited a significant decline in both informal (p ≤ .001) and formal social participation (p ≤ .05), whereas the social participation of widowed persons remained relatively constant over time. These descriptive results suggest that the social participation of widowed persons significantly differs from that of their married counterparts. These results also question whether widowed persons alter their social participation in curious ways before the actual death of their spouses.

To explore whether adjustment to widowhood begins before the date of death, Table 2 presents OLS regression coefficients predicting baseline (preloss) levels of social participation. As suggested in the descriptive analyses, those persons who later became widowed had significantly lower levels of baseline informal participation than the continuously married (β_informal = −.29, p ≤ .01). Spousal health was also an important predictor of reduced social engagement (β_informal = −.31, p ≤ .001). This analysis suggests that adjustments do occur before the actual loss and reinforces the importance of considering characteristics of both the husband and the wife when assessing the daily activities of older adults.

The Effect of Widowhood on Social Participation

The multivariate analysis presented in Table 3 further explores how the social participation levels of older adults change over time and in response to widowhood. Model 1 showed the effect of widowhood controlling for demographic characteristics; Model 2 added in potentially mediating variables including the health status of both the husband and the wife, individual personality characteristics, and other structural constraints; and Model 3 controlled for baseline social participation levels.

As shown in Model 1 and replicated in Models 2–3, widowhood significantly predicted informal social participation levels, but not formal social participation levels. Model 3 showed that widowed persons had informal social participation scores 0.44 standard deviation units higher than nonwidowed persons, whereas widowed persons did not differ from controls in their level of formal social participation. Although not shown here, we ran identical models predicting social participation 18 months postwidowhood. The resulting coefficients were similar to those presented in Table 3 , suggesting that changes in social participation occurred within the first 6 months after widowhood and continued to persist across longer time horizons. In sum, widowhood yielded a sustained increase in informal activity levels, but no change in formal social participation levels.

Model 2 considered the effects of individual- and couple-level characteristics that may predispose one to certain levels of social engagement. As expected, extraverted persons participated more in social activities (β_formal = .09, p ≤ .05; β_informal = .11, p ≤ .05), whereas depressed persons participated less in informal social activities (β_informal = −.09; p ≤ .05). Driving an automobile and not having children were also important predictors of increased activity. Surprisingly, neither individual health nor spouse's health predicted level of social participation at Wave 1. However, as previously noted in Table 2 , spousal health characteristics were influential in determining baseline levels of social participation.

Model 3 considered baseline social participation as a possible predictor of future social participation. Baseline social participation was highly influential in determining subsequent social participation, revealing that persons with high social participation at baseline continued to have relatively high social participation at Wave 1 (β_formal = .58, p ≤ .001; β_informal = .31, p ≤ .001). The addition of baseline social participation in Model 3 significantly improved the fit of the models, as evidenced by the increasing of the adjusted R² from .13 to .46 for formal social participation and from .16 to .29 for informal social participation. Despite the potentially disruptive effect of widowhood, continuity prevailed as widowed persons adjusted to the loss of their marriage partners and the accompanying social opportunities tied to the marriage.

Finally, we assessed all analyses in Table 3 to determine whether adjustment to widowhood differed by sex. The Widowhood × Sex interaction term was not significant in any model, suggesting that the effect of widowhood on social participation was similar for widows and widowers. The analyses also considered whether the effects of the predictor variables differed by whether one becomes widowed or remains continuously married. Identified with a superscript a in Table 3 , widowhood interacted with income, education, and race, signifying that African American widowed persons and those with lower socioeconomic status have significantly lower levels of social participation than nonwidowed persons or widowed persons of higher socioeconomic status. Widowed persons with no children also exhibited particularly low levels of social participation following widowhood.

Given the above analyses, it is unclear if the increases in social participation among widowed persons were due to the bereaved seeking out additional activity or to the inclination of friends and relatives to offer care and support to recently widowed persons. Additional analyses using only the widowed sample shed light on whether the increased social participation was a result of passive reaction or active response. A majority of widowed persons (71%) had the same amount of interest in having contact with relatives and friends as when their spouses were alive; however, more than a third (35%) reported that friends and relatives showed more interest in having contact with them when they became widowed. Very few widowed persons expressed less interest in social participation following widowhood, either on their part (12%) or on the part of friends and relatives (6%). The results presented in Fig. 2 suggest that the predicted increase in informal social participation noted in Table 3 is perhaps attributable to being sought out by concerned friends and relatives rather than to seeking out additional forms of social activity to fill the void left by the spouse's death. Sex differences are not apparent in this analysis, meaning that men and women received increased support from friends and relatives equally. Although for men the death of their wives often represents the loss of their social coordinator, widows and widowers were likely to exhibit similar behavioral outcomes because much of the change in social participation is attributable to receiving additional support rather than to coordinating additional social activities.

Although continuity prevails in their behavioral adjustments, most widowed persons considered increased social activity an effective way to combat the psychological distress associated with the loss. Fig. 3 shows that a majority of widowed persons (87%) said that they "tried to keep busy or tried to get involved in some activity as a way to cope with the negative effects of widowhood." When comparing the demographic characteristics of these two groups, those respondents who reported increasing their activity tended to be younger and have higher incomes, and were more likely to own their home, be White, and drive an automobile. If increased social participation is beneficial to the successful adaptation to widowhood, older widow(er)s who are not as economically well off are perhaps structurally disadvantaged in their ability to successfully cope with the potentially devastating losses associated with widowhood.

Discussion

Summary and Conclusions

This research explored how the social participation of older married and widowed persons changes over time and whether theoretical explanations from social gerontology are useful in explaining the patterns of change. We found that social participation consists of two distinct subtypes—formal and informal—that respond differently to widowhood. Widowed persons, both men and women, exhibited a sustained increase in informal social participation following spousal loss, whereas formal social participation levels did not fluctuate over time or in response to widowhood. A majority of widowed persons admitted to using increased social participation as an active coping strategy to deal with the negative effects of widowhood. These behavioral patterns suggest that social participation, particularly informal activities such as telephone contact and getting together with friends, is an aspect of daily life that is deeply affected by late-life bereavement. Formal social participation may not provide the emotional and instrumental support that informal social participation does, thus explaining why formal social participation levels remain constant after widowhood.

Four broad themes, which encompass broader issues related to late-life bereavement, characterize the major findings of these analyses. The first theme holds that a desire for continuity dominates the adjustment process following widowhood. Widowed persons overwhelmingly expressed a desire for continuity when asked about how much contact they preferred to have with others compared with how much contact they had before the loss. Analyses demonstrated that prior social participation is, by far, the best predictor of subsequent social participation. Despite the disruption caused by widowhood, survivors displayed tremendous resiliency when adjusting their levels of social activity and community involvement. In both preferences and actual behaviors, older adults demonstrated a sense of continuity when faced with the unfortunate experience of becoming widowed.

The second theme contends that widowhood is a process rather than a discrete event or static state. Much of the past research has characterized widowhood as a discrete event defined by the spouse's date of death or as a marital status defined as being widowed rather than married, single, or divorced. Past research has largely been based on cross-sectional or retrospective data, whereas the current analyses used prospective, longitudinal data from a widow–control sample. Our findings, using the methodologically unique CLOC study, insinuate that widowhood is a process of adjustment that begins before the actual date of death and extends well beyond the initial mourning period associated with spousal bereavement. Future widow(er)s and those with ill spouses may have a sense-of-death forewarning and devote their time and energy to their spouses in their final days or months, thus reducing their preloss levels of social engagement. Following the actual death, widowed persons continue their process of adjustment, generally exhibiting a sustained increase in extrafamilial social activities. The use of prospective, longitudinal data is imperative to evaluate the processual nature of widowhood.

A third theme concerns the interrelatedness of social actors. Although only the survivor assumes the widowed label, the effects of widowhood are not experienced in isolation. We found that the increased social participation among the widowed sample was perhaps a function of others rallying around the bereaved, rather than of the survivor's seeking out additional social activity. Consistent with past research, older adults receive an outpouring of support from friends and relatives during times of perceived need (Hogan and Eggebeen 1995). Lopata 1996 argued that social involvement increases after widowhood because friends, relatives, and neighbors unconditionally offer emotional and instrumental support to the recently bereaved. As childless widow(er)s exhibit particularly low levels of social integration, kin may be particularly crucial to the adjustment process by encouraging elderly parents to remain socially engaged. Increased social participation, whether passively or actively acquired, is a reflection of how multiple actors experience and influence the transition to widowhood.

A fourth and final theme assumes that there are multiple pathways to individual bereavement outcomes. Our findings imply that the patterns of behavioral adjustment following widowhood depend on the psychological, familial, and socioeconomic resources of the survivor. Some widowed persons—whether due to race, class, or some other constraint such as the availability of transportation or the presence of children—exhibit significantly lower levels of social participation than widowed persons with more favorable characteristics. The traditional theories of aging, as well as the general bereavement literature, implicitly assume that all persons have equal resources as they age. Our findings question the universality or widespread availability of the economic, intellectual, and interpersonal resources needed to successfully adjust to old age generally or the event of widowhood specifically. Because the pathway to widowhood and its ensuing grief may be structurally determined and diversified by the particular resources available to the survivor, bereavement research must consider both the material and the interpersonal resources available to widow(er)s in order to understand the adjustment process.

Evaluation of Activity, Disengagement, and Continuity Theories

In addition to providing an empirical analysis of the behavioral adjustments related to late-life widowhood, we also explored the utility of three theories from social gerontology to explain why older adults adjust their social participation levels. Despite its criticism as an explanatory model, continuity theory prevails as the most applicable theory (among the three theories considered) for the longitudinal study of social participation among older adults. Both married and widowed persons are likely to exhibit levels of social participation in the future similar to those they displayed earlier in the life course. Additional results, using the widowed-only sample, indicated that older adults preferred to maintain similar pre- and postloss levels of contact with relatives and friends. These findings offer convincing support for the existence of and desire for continuity, throughout the life course as well as in the face of widowhood. However, our analyses, which are limited to persons older than 65, cannot capture whether individuals may have selectively narrowed their social networks before age 65 (Carstensen 1992). Developmental research has suggested that social networks are highly adaptive throughout the life course, with preferences for certain social partners beginning as early as age 3 (Macoby 1990) or as early as adolescence (Dunphy 1963). If this is the case, have widowed persons really maintained continuity throughout the life course or have they merely exhibited stability between the two waves of data collection?

Given our finding that widowed persons have increased levels of informal social participation, do the analyses offer empirical support for the activity hypothesis? Activity theory implies that the person experiencing the losses will actively replace the lost social roles with compensatory social roles, rather than passively acquire increased social support as suggested by our analyses. Thus, although the direction of change is reminiscent of activity theory, we argue that our findings offer little empirical support for activity theory. Nor do the empirical results offer convincing support for the disengagement hypothesis, as widowed persons do not exhibit significant declines in social participation following the deaths of their spouses, as hypothesized. Nevertheless, soon-to-be widowed persons or persons with sick spouses do have reduced levels of social participation, suggesting that older adults may experience a social withdrawal in times of spousal illness or impending spousal death.

Although this exercise in theory testing is largely inconclusive, it is not altogether futile. None of the theories, as hypothesized in the context of widowhood, are wholly supported nor wholly refuted. This limited and tempered support provides further testament to the inability of the classical theories of social gerontology to adequately explain the empirical realities of aging (Bengston, Rice, and Johnson 1999). A developmental or lifespan perspective may prove useful in assessing late-life changes (Antonucci 1989; Baltes, Reese, and Lipsitt 1980). For example, Laura Carstensen 1987(1991, 1992) theory of socioemotional selectivity and Toni Antonucci and her colleagues' (Antonucci 1990; Antonucci and Jackson 1987; Kahn and Antonucci 1980) social convoy model seem to be more appropriate in understanding the changing social relations among older adults. Socioemotional selectivity theory suggests that levels of social participation in late life are a culmination of strategic selections made throughout the life course to form social networks that maximize social and emotional gains and minimize social and emotional risks. The social convoy model states that people age throughout the life course, experiencing the stress of transitions and turning points with a highly selected group of people from whom they derive a basis for self-identity as well as emotional and instrumental support.

Despite activity, continuity, and disengagement theories' inimitable presence in social gerontology, their explanatory power has fallen short in trying to explain how or why older adults alter their social participation in the face of widowhood. These classic theories of aging cannot capture the diversity in reactions or the processual nature of the human life course, whereas developmental perspectives such as socioemotional selectivity theory and the social convoy model may be more successful in understanding the ongoing adjustments that unfold over the entire life course. As opposed to the grand theories of social gerontology, middle-range theories that consider unique life trajectories may provide the most promising avenue for future theory building related to late-life stress and adjustment.

Limitations and Future Research

Although the CLOC data have provided both empirical and theoretical findings related to old age adjustment, no study is without limitations. Our measure of social participation is an improvement on earlier studies, in that it captures statistically and conceptually unique realms of social engagement and community involvement. By separating formal and informal social roles, we were able to distinguish the unique effects of spousal loss on two broad types of social participation. Nevertheless, future research may benefit from evaluating the effects of spousal loss on even more specific types of social participation. In addition, future research should consider including a measure of the subjective meaning derived from each type of activity, as past research has suggested that rate of contact may not capture the emotional quality of the relationship or activity (Rosow 1967). Perhaps the quality or value attached to an activity affects one's willingness to adopt or abandon that role upon widowhood. Future studies may also want to explore the perceived stress associated with particular behavioral changes, as this may expand understanding of the mechanisms that create psychological distress following widowhood. The loss of marriage represents not only the loss of a significant other, but the loss of the instrumental arrangements between the couple (Miller and Wortman in press; Umberson et al. 1992). The stress and strain produced by changes in the survivor's instrumental and social spheres may elucidate the differential vulnerabilities to psychological distress following widowhood.

Future research should also consider the role of couple-level characteristics that determine one's level of social integration. We have found that the partner's health status acutely influences the social activities of older adults. This finding reiterates the importance of considering individual behaviors within the context of related social actors (Elder 1998). Using couple-level data, future research should explore the patterns of social participation across the life course, paying particular attention to transitions in marital status and the underlying health characteristics of both husband and wife. Consideration of both the husband and the wife is necessary in order to understand the dynamics of the marital relationship and the processual nature of widowhood.

Practical Applications and Implications

Finally, the results derived from the preceding analyses should be extended to devise effective care and support strategies for the recently bereaved. A clear majority (87%) of widow(er)s admit to actively keeping busy or getting involved as a way to cope with feelings of grief and loneliness; therefore, increased social activity is one way older adults can potentially reduce the perceived psychological distress associated with widowhood. Although prior research has consistently found that higher levels of social integration yield favorable physical and mental health outcomes (Durkheim, 1897/Durkheim 1951; House et al. 1988; Lowenthal and Haven 1968; Thoits 1983), we caution against assuming that activity for activity's sake is the most effective way to provide support for the recently bereaved. Senior centers play a vital role in keeping older adults engaged in the community. However, the vast majority of older adults do not take advantage of such formal integration efforts (Carstensen 1991). Instead, older adults, particularly those experiencing a devastating or stressful loss such as widowhood, tend to rely on the convoy of lifelong social relations from whom they derive support, self-definition, and a sense of stability and continuity (Antonucci and Jackson 1987; Kahn and Antonucci 1980). Our analyses illustrate that widowed persons, in both their intentions and their behaviors, prefer to maintain the relationships and activities that have been most meaningful and rewarding throughout the life course. Thus, simply providing widowed persons with an abundance of new activities will not be as effective as having others (e.g., family members, friends, religious communities, and neighbors) assist them in achieving a continuity of lifestyle and stability in social relations.

Intervention efforts should aim to minimize the disruption in social roles rather than maximize the availability of compensatory social activities. One such proposal is to offer affordable and flexible transportation to older persons, especially to those who no longer drive. The use of the telephone, the Internet, and other means of communication that allow older adults to maintain contact despite geographic distance or physical impairment (Litwak and Longino 1987) may also be invaluable to successful adaptation to widowhood. Although support via communication technology may be a more effective support strategy for future cohorts of older adults, any programmatic or policy intervention should seek to minimize disruption in daily activities rather than to construct a new life with new activities and new acquaintances. Furthermore, intervention efforts should target those persons most likely to benefit from the outreach efforts. Our findings imply that persons with low socioeconomic resources and those with no children may be particularly vulnerable to social isolation upon widowhood. Also the soon-to-be widow(er) and those persons with ailing spouses may need as much support as recently widowed persons. Outreach efforts, therefore, could be directed at married couples experiencing the final stages of a spouse's life rather than only at those persons who have recently lost spouses.

The findings from this study not only expand our knowledge about the daily social adjustments that accompany late-life bereavement, but also add to the ongoing debates in social gerontology regarding the utility of activity, disengagement, and continuity theories to explain the variations in how individuals adapt to the aging process. In addition, this article provides multiple suggestions for future research and offers practical implications that may help families and practitioners offer more effective care and support to the recently bereaved.

The Forum

Book Reviews

Practice Concepts

The Changing Lives of Older Couples study was supported by National Institute on Aging (NIA) Grants P01-AG05561-01 and R01-AG15948-01. Rebecca Utz's participation was supported by an NIA predoctoral traineeship. We thank the anonymous reviewers for their thoughtful and detailed comments on earlier versions of this article. This paper was presented at the 53rd Annual Scientific Meeting of The Gerontological Society of America, November 2000, Washington, DC.

References