The Effect of Widowhood on Intergenerational Ambivalence

Abstract

Objectives

The purpose of this study is to examine: (1) the extent to which widowhood affects older adults’ ambivalence about their adult children; (2) the role of intergenerational dependence in explaining the effect of widowhood on parent-child ambivalence; and (3) temporal changes in the effects of widowhood on ambivalence.

Methods

Analyses are based on Changing Lives of Older Couples (CLOC), a prospective study of 1,532 married individuals aged 65 and older. Ordinary least squares regression models are use to estimate the direct effect of widowhood and the mediating effects of dependence on intergenerational ambivalence 6 and 18 months after spousal loss.

Results

Widowhood is associated with a decrease in ambivalent feelings toward adult children 6 months after spousal loss, which is partially explained by a reduction in the extent to which children are dependent upon their bereaved parents. However, at 18 months, widowhood does not exert any significant influence on intergenerational ambivalence.

Discussion

Our findings suggest that intergenerational ambivalence is influenced by major life events such as widowhood and sheds light on the mechanisms by which parent-child dependence contributes to intergenerational ambivalence.

Recently, a growing number of researchers have used ambivalence theory as a conceptual framework to explain the complex nature of intergenerational relationships. Compared to other theoretical frameworks that highlight either positive aspects, such as the solidarity perspective (Bengtson & Roberts, 1991; Roberts, Richards, & Bengtson, 1991) or negative aspects, such as the conflict perspective (Akiyama, Antonucci, Takahashi, & Langfahl, 2003; Rook, 1984, 1990) of intergenerational relationships, the ambivalence framework emphasizes both the beneficial and problematic aspects of the same parent-child relationship (Luscher & Pillemer, 1998; Pillemer & Suitor, 2002).

One proposition of ambivalence theory is that life transitions tend to increase tension within parent-child relationships (Fingerman, 1996; Luscher & Pillemer, 1998; Pillemer & Suitor, 2002). However, few studies have empirically examined how older adults’ experiences of stressful life events such as widowhood influence ambivalence between older adults and their adult children over time. Nor have studies explored what factors influence the relationship between life transitions and parent-child ambivalence. This paper addresses this gap in the literature by examining: (1) the extent to which widowhood as a stressful life transition affects older adults’ perceived ambivalence toward their adult children; (2) the extent to which intergenerational dependence mediates the effect of widowhood on intergenerational ambivalence; and (3) the extent to which effects of widowhood on intergenerational ambivalence change over time.

Understanding how widowhood brings changes to relationships with adult children is critical, considering the implications of these changes for older adults’ well-being. Social support is known to serve as an important resource in older adults’ coping with the distress of widowhood (Antonucci, 1990; Lin, Simeone, Ensel, & Kuo, 1979). However, the quality of social relationships may change following a stressful life transition such as widowhood. Changes in social support following widowhood may result in a heavier caregiving burden for adult children and unwelcome feelings of dependence and loss of autonomy for older adults (Talbott, 1990). Thus, mapping the diverse ways that widowhood influences intergenerational relationships, and investigating how dependence mediates the link between stress and social support, can help researchers and practitioners devise intervention plans that can maximize the benefit of social support following spousal loss.

In the subsequent section, we first lay out the theoretical framework that has guided this research. Then, we review the previous literature on the effects of widowhood and dependence on intergenerational ambivalence, followed by the description of the sample and measures used in this study. Next, we present findings and discuss implications for future research and practice.

Ambivalence theory

Ambivalence theory is particularly relevant to the current research questions in that it conceptualizes changes in positive and negative emotions as a function of life transitions. It posits that in times of life transition, people are expected to conform to the normative requirements of their new status position, while discarding the norms attached to the old status position (Coser, 1966). For example, when children become adults, they are expected to become independent from their parents. When these new expectations are not met, parents may feel ambivalent about their children’s continued dependence (Luscher & Pillemer, 1998). Studies have suggested that life transitions in old age can lead to increased ambivalence just as they do during young adulthood (Pillemer & Suitor, 2002). However, little research has empirically examined how a stressful life transition in old age such as widowhood may influence positive and negative feelings within parent-child relationships. The current study addresses this issue by focusing on widowhood as a potential source of stress that increases ambivalence in older adults’ relationships with their children.

Widowhood as a potential source of increased ambivalence

Widowhood is one of the most stressful life transitions in old age (Carr & Utz, 2002; Holmes & Rahe, 1967). Upon spousal loss, older adults become more dependent on their children for emotional as well as instrumental support (Ha, Carr, Utz, & Nesse, 2006), and adult children are expected to provide greater support to their bereaved parents (Hogan & Eggebeen, 1995). Although the increased levels of support from adult children may result in widowed parents’ positive feelings toward their children, these newly negotiated relationships, in which one provides and the other receives more help, may breach the previously existing balance of reciprocity and lead to more ambivalent feelings.

To date, no study has empirically examined the possibility of increased ambivalence following late-life widowhood. However, a number of studies have addressed related research questions. For example, Talbott’s (1990) study of the relationships between widowed mothers’ and their children suggested that widowed persons experience both negative and positive support. Using qualitative data from 55 older widowed mothers, she explored how interactions with their adult children led to the widowed mothers’ negative reactions. About half of the widowed mothers she interviewed (49%) made negative comments about their relationships with their children in addition to positive comments. The reasons for these negative perceptions were feeling unappreciated, feeling dissatisfied with the amount of help received, fears about bothering or burdening their children, and emotional dependence on their children. In particular, the imbalance in the exchange of support between parents and children, and the resulting power differential between widowed older adults and children play an important role in widows’ negative reaction to support. Zettel and Rook (2004) examined how close social relationships change over time after spousal loss, and whether widowed persons’ involvement in alternative social ties compensated for psychological distress during widowhood. They observed that widowed persons developed new social ties and rekindled or intensified old ties following spousal loss, yet they also found that these social ties did not have a significant influence on the widowed persons’ psychological well-being, implying that increased social support may not necessarily bring positive outcomes.

In sum, the literature suggests that widowhood can entail both positive and negative consequences for one’s social relationships. However, partly because of lack of appropriate data, no study has examined precisely how widowhood affects ambivalent feelings within parent-child relationships. By using prospective data of both widowed and married samples, we examine the extent to which widowed persons differ from married persons in their levels of ambivalence, controlling for the level of ambivalence prior to loss.

Intergenerational dependence as a source of ambivalence

One of the often-replicated findings in previous literature is that feelings of dependence are a key determinant of parent-child relationship quality (Luscher & Pillemer, 1998; Pillemer & Suitor, 2002). Parents’ feelings of dependence on children harms their sense of autonomy (Blieszner & Mancini, 1987; Silverstein, Chen, & Heller, 1996), and children’s dependence on parents contributes to intergenerational tension (Fingerman, 1996). On the one hand, adult children’s failure to achieve normative adult status or financial independence leads to an increase in parental ambivalence (Pillemer and Suitor, 2002). On the other hand, parental dependence on adult children can contribute to increased feelings of burden among their adult children, which may in turn influence parents’ levels of ambivalence (Talbott, 1990). A recent study suggests that parents’ or child’s conditions of potential dependence (e.g., poor health and financial difficulty) also increase the feelings of intergenerational ambivalence (Willson, Shuey, Elder, & Wickrama, 2006).

Widowhood tends to increase parents’ dependence on their children and to decrease children’s dependence on them (Ha et al., 2006). Thus, widowed parents’ increased dependence on their children may lead to increased ambivalence, whereas the reduced dependence of children on their widowed parent may lead to decreased ambivalence. Investigating intergenerational dependence following widowhood can enhance our understanding of the mechanisms through which widowhood produces ambivalent feelings in parent-child relationships. Given the proposition from ambivalence theory that increased dependence during times of life transition creates tension between parents and children, we hypothesize that the effect of widowhood on ambivalence will be attenuated after accounting for parents’ dependence on children. However, we also propose that the effect of widowhood on ambivalence will become more pronounced when the decrement in children’s dependence following widowhood is taken into account. That is, intergenerational dependence will have a mediating effect on ambivalence, yet parents’ and children’s dependence will play contrasting roles in mediating the relationship between widowhood and ambivalence. Figure 1 illustrates these potential mediating effects.

Figure 1.

Mediating effects of intergenerational dependence

Temporal change in ambivalence

Finally, we examine the extent to which the effects of widowhood on intergenerational ambivalence change over time. A life transition is not a one-time event, but a process that requires different coping efforts across time (Stroebe & Schut, 1999). The life transition of widowhood may be most stressful during the initial stage (Zisook & Shuchter, 1991). At this stage, the surviving spouse must cope not only with the emotional distress of losing a loved one, but also with various practical matters. The bereaved spouse must attend to funeral arrangements and assume household responsibilities that were once shared by the deceased spouse (Utz et al., 2004). The schism between widowed persons’ need for support and their adult children’s ability to meet these needs may be greatest at this phase. Likewise, widowed individuals’ feelings of dependence and ambivalence may be heightened at this stage. As time passes, widowed persons may adjust their expectations to adult children’s availability to provide support. Or, their actual need for assistance declines, and thus, they have fewer requirements for support. In either case, their ambivalence is likely to decrease.

Previous studies examining the impact of life transitions on positive and negative social interactions have used cross-sectional data (e.g., Krause & Jay, 1991), and thus, could not investigate how the influence of widowhood changes over time. Furthermore, previous studies with longitudinal data often do not have comparison groups of married persons and thus cannot distinguish the effect of aging from the effect of widowhood. Using prospective data of spousal bereavement and by utilizing a comparison group of still-married persons, the current study investigates the effects of widowhood both at an early stage (6 months after spousal loss) and a later stage of bereavement (18 months after spousal loss). Examining temporal changes is important because it can guide intervention plans at different stages of widowhood. Given that people adjust to social roles as time passes, we expect that the impact of widowhood and intergenerational dependence on ambivalence would be greater at 6 months after bereavement than at 18 months after bereavement.

In sum, three research questions are addressed in this paper. First, we examine whether widowed persons differ from married persons in their levels of perceived intergenerational ambivalence 6 and 18 months after spousal loss, controlling for baseline levels of ambivalence. Consistent with the ambivalence hypothesis, we predict that widowed persons will experience higher levels of ambivalence than their married counterparts. Second, we examine the extent to which parents’ perceived intergenerational dependence mediate the relationship between widowhood and ambivalence. We hypothesize that the positive association between widowhood and intergenerational ambivalence will be mediated by parents’ increased dependence on children and children’s decreased dependence on parents after spousal loss. Finally, we examine how these relationships vary over time (i.e., 6 months and 18 months after widowhood). We predict that the effect of widowhood will be greater at an earlier stage than at a later stage.

Potential influence of contextual factors

We control for pre-widowhood marital contexts (e.g., marital quality and spousal health at baseline) that may confound the relationship between widowhood and parent-child ambivalence. Those who had higher marital quality may have relied on their spouse for emotional support prior to spousal death. This reliance may result in increased expectations of support from children or dependence on them following spousal loss which may require greater adjustments from both the surviving parent and the child, and thus create greater ambivalence (Willson et al., 2006). On the other hand, those who had a spouse with poor health may have been already relying on their children for support prior to spousal death; thus the transition to widowhood may not significantly change the quality of the parent-child relationship or the degree of intergenerational ambivalence.

We also control for demographic (i.e., gender, socioeconomic status, and race) and health characteristics (i.e., functional and mental health) of the respondents that previous literature has found to be important correlates of social support. Studies have shown that women are more likely to receive support from children than men (Umberson, 1992). Although findings are mixed, researchers have found that income and race play an important role in explaining different levels of support (Krause & Rook, 2003; Raschick & Ingersoll-Dayton, 2004; Willson, Shuey, & Elder, 2003). Health characteristics are considered because the effect of widowhood may be confounded with the effect of poor mental or physical health, which can lead to greater ambivalence by increasing caregiving burden for adult children.

Methods

Sample

The analyses are based on data from Changing Lives of Older Couples (CLOC), a prospective study of a two-stage area probability sample of 1,532 married men and women (for the initial report on this study, see Carr et al., 2001). To be eligible for the study, respondents had to be English-speaking members of a married couple in which the husband was aged 65 or older. All sample members were community dwelling and were capable of participating in a two-hour face-to-face interview. Approximately 68% of those contacted for an interview participated, which is consistent with the response rate from other Detroit area studies. Baseline face-to-face interviews were conducted between June 1987 and April 1988.

CLOC researchers identified spousal loss by reading the daily obituaries in three Detroit-area newspapers, and by using monthly death records provided by the State of Michigan. The National Death Index (NDI) was used to confirm the deaths and obtain causes of death. Of the 319 respondents who lost a spouse during the study, 86% (n = 276) participated in at least one of the follow-up interviews, conducted at 6 months (Wave 1), 18 months (Wave 2), and 48 months (Wave 3) after the death. The primary reasons for nonresponse were refusal to participate (38%) and ill health or death at follow-up (42%). Controls from the original sample of 1,532 were selected to match the widowed persons along the dimensions of age, race, and sex. The matched controls were interviewed at the three follow-up interviews at roughly the same time as their corresponding widowed participants.

In this paper, two sub-samples of the CLOC are used. The first sample includes 266 persons (188 widowed persons and 78 married controls; not every widowed person was matched with a control due to a temporary cut in funding) who participated in the baseline and the 6-month follow-up (Wave 1) interviews, and who gave valid responses to questions asking for positive and negative social interactions with their children. The second sample includes 338 persons (160 widowed persons and 178 married controls) who participated in the baseline and the 18-month follow-up (Wave 2). For both samples, we included only those who have at least one living child at baseline and the respective follow-up interview. Sample sizes are weighted in order to account for the different probabilities of selection and nonresponse at the initial baseline interview.

Measures

Ambivalence

Ambivalence is defined as the extent to which older persons perceive their relationships with children as both positive and negative. We used the following computational formula that previous research has found suitable for capturing both the similarity and the intensity of positive and negative components (Thompson & Zanna, 1995; Willson et al., 2006):

$$\text{Ambivalence}=(\text{positive}+\text{negative})/2-\mid \text{positive}-\text{negative}\mid$$

Positive interactions with children (α = .70) is assessed with the following questions: “How much do your children make you feel loved and cared for?” and “How much are they willing to listen when you need to talk about your worries or problems?” Negative interactions with children (α = .48) is assessed with the following questions: “How much do you feel they make too many demands on you?” and “How much are they critical of you or what you do?” For both measures, response categories are: a great deal, quite a bit, some, a little, and not at all. Higher values indicate higher positive and higher negative social interactions with children.

It should be noted that the reliability coefficient for the negative social interactions measure is quite low, due most likely to the somewhat diverse content of the two items (criticism vs. demands). This low reliability may make it harder to detect the significant relationship between independent and dependent variables by attenuating the correlations between variables, though the broader range of content may increase the validity of the measure in covering the dimensions of negative interaction.

Independent variables

Widowhood is a dichotomous variable, where 1 indicates that the respondent became widowed during the study period (after the baseline interview), and 0 indicates that the respondent is still married at the follow-up interviews.

Pre-loss marital context

Two variables are considered in order to capture marital contexts prior to widowhood. Marital quality at baseline (α = .88) is assessed with ten items: “How much does your (husband/wife) make you feel loved and cared for?”; “How much is (he/she) willing to listen when you need to talk about your worries or problems?”; “Thinking about your marriage as a whole, how often do you feel happy about it?”; “Taking all things together, how satisfied are you with your marriage?”; “How much do you feel (he/she) makes too many demands on you?”; “How much is (he/she) critical of you or what you do?”; “How often would you say you and your (husband/wife) typically have unpleasant disagreements or conflicts?”; and “How often do you feel bothered or upset by your marriage?” Respondents were also asked to state how much they agreed with the following statements: “There are some serious difficulties in our marriage” and “My (husband/wife) doesn’t treat me as well as I deserve to be treated.” Response categories for the first 8 questions are a great deal, quite a bit, some, a little, and not at all; response categories for the final two statements are very true, somewhat true, a little true, and not true at all. Higher values indicate higher marital quality. Spouse’s health at baseline is assessed with the question, “How would you rate your (husband’s/ wife’s) health at the present time?” Responses of ‘fair’ and ‘poor’ are coded 1, indicating poor health; ‘excellent,’ ‘very good,’ and ‘good’ are coded 0, indicating good health.

Control variables

The analysis controls for demographic and health variables. Demographic variables include gender (1=female, 0=male), race (1=white, 0=black), age (in years), education (in years), and the availability of a child (1=respondent has at least one living child; 0=respondent does not have a child). Two variables are used to capture physical and psychological health characteristics. Functional difficulty at baseline (α = .77) is a four-item scale indicating how much difficulty the respondent has (a) bathing by oneself; (b) climbing a few flights of stairs; (c) walking several blocks; and (d) doing heavy housework, such as shoveling snow or washing walls. Response categories are: a little, some, a lot, and cannot do. Higher values indicate greater functional difficulty. Depression at baseline (α = .78) is assessed with a subset of eleven items (nine negative and two positive items) from the 20-item Center for Epidemiologic Studies Depression (CES-D) scale (Radloff, 1977). Higher values represent higher depression.

Finally, all analyses control for the duration between the baseline and follow-up interviews. This variable is controlled because although all follow-up interviews were conducted on average at 6 and 18 months following spousal death, the duration between the baseline and the follow-up interviews differs across respondents. This variable is assessed in months. We also tested the interaction between widowhood and the duration between the baseline and follow-up interviews, in order to see if the effect of widowhood differs depending on this duration. Data are not presented but findings are summarized in the results section.

Intergenerational dependence

To examine the second research question, we control for parent’s perception of their dependence on children and their children’s dependence on them at the follow-up waves as factors that potentially mediate the relationship between widowhood and social support. Parents’ dependence on children at Wave 1 (α = .47) and at Wave 2 (α = .49) are evaluated with the following questions: “How much do you depend on your children for (a) emotional support; (b) help or advice with financial and legal matters; and (c) help with errands or other chores?” Children’s dependence on respondent at Wave 1 (α = .47) and at Wave 2 (α = .50) are assessed with the following questions: “How much do your children depend on you for: (a) emotional support; (b) help or advice with financial and legal matters; and (c) help with babysitting or other errands?” Response categories include not at all, a little, some, and a lot. Higher scores reflect higher levels of dependence in both scales.

Two important limitations of this measure should be noted. First, the dependence measures used here reflect older parents’ perceptions of their dependence on children and children’s dependence on them. Thus, they may not capture “objective” levels of instrumental and emotional dependence between generations. Furthermore, parents’ assessment of intergenerational dependence may be different from children’s evaluation of the relationship. Nevertheless, these measures allow us to understand how parent’s subjective appraisal of parent-child dependence influence older adults’ positive and negative social interactions with their children. Second, the reliability coefficients of these scales are low, probably because individual items in the scales capture different aspects of dependence. In order to address this limitation, we further explored how dependence in specific domains of support (e.g., emotional support, financial or legal advice, and instrumental help) influences parent-child ambivalence following spousal loss. These sub-analyses can offer greater insights into the ways that intergenerational dependence in specific domains influences ambivalence within intergenerational relationships. The findings are briefly summarized in the results section.

Analytic plan

We first present t-tests and chi-square results, comparing means (for continuous variables) or proportions (for dichotomous variables) and standard deviations of dependent and independent variables between the widowed and married samples. Next, we present ordinary least squares (OLS) regression models examining the effect of widowhood on parents’ perception of ambivalence in their relationship with children at Waves 1 and 2. In examining the relationship between widowhood and ambivalence, hierarchical regression models are used to further explore the mediating effects of intergenerational dependence. Model 1 includes widowhood and other control variables; Model 2 controls for parents’ perceptions of their dependence on children, whereas Model 3 controls for children’s dependence on them.

Results

Sample characteristics

Table 1 shows descriptive statistics and t-test results comparing the means (or proportions) of the widowed and still-married samples at the 6- (Wave 1) and 18-month follow-up (Wave 2). The mean ambivalence score at Wave 1 was lower among widowed than married persons. However, the differences were not significant at Baseline and at Wave 2.

Table 1.

Descriptive Statistics, Changing Lives of Older Couples Study, 1987–93

	Wave 1: 6-month follow-up			Wave 2: 18-month follow-up
	Widowed	Married		Widowed	Married
	(N=188)	(N=78)		(N=160)	(N=178)
	Mean (SD)	Mean (SD)	T-test or X2	Mean (SD)	Mean (SD)	T-test or X2
Dependent variable
Ambivalence, W1 or W2	− 0.05 (1.13)	0.38 (1.22)	2.69**	0.13 (1.34)	0.28 (1.21)	1.10
Independent variables
Baseline measures of ambivalence	.58 (1.29)	.62 (1.28)	0.22	0.52 (1.21)	0.50 (1.19)	− 0.16
Pre-loss marital context
Marital quality, BL	−0.10 (1.11)	0.01 (0.95)	0.78	−0.19 (1.17)	−0.09 (0.95)	0.87
Spouse’s health, BL (1=poor health)	0.59	0.30	18.97***	0.60	0.35	25.53***
Sociodemographic and health characteristics
Sex (1=female)	0.71	0.74	0.36	0.70	0.70	0.01
Race (1=white)	0.87	0.85	0.32	0.88	0.84	1.05
Age, BL	70.24 (6.95)	68.98 (6.03)	− 1.49	70.12 (6.87)	70.37 (6.79)	0.34
Education (in years)	11.35 (2.85)	11.67 (2.88)	0.85	11.42 (2.70)	12 (3.05)	1.85
Depression, BL	0.14 (1.05)	0.02 (1.11)	− 0.83	0.13 (1.05)	−0.09 (0.87)	− 2.09*
Functional difficulty, BL	−0.16 (0.75)	−0.10 (0.92)	0.55	−0.14 (0.80)	0.12 (1.10)	2.48*
Months between BL and W1 (W2) Interviews	36.88 (18.33)	59.64 (8.54)	13.81***	42.17 (14.00)	46.79 (14.60)	2.97**
Perceived intergenerational dependence
Parents’ dependence on children, W1 or W2	0.62 (1.19)	0.02 (0.92)	− 4.43***	0.32 (1.14)	0.07 (1.05)	− 2.08*
Children’s dependence on parents, W1 or W2	−0.42 (0.89)	−0.15 (0.89)	2.23*	−0.32 (0.92)	−0.19 (0.97)	1.30

Note: Two-tailed t-tests were used to assess significant differences between means. Chi-square tests were used to assess significant differences in proportions. N’s are weighted. BL refers to baseline; W1 refers to Wave 1 (6-month follow-up); W2 refers to Wave 2 (18-month follow-up).

^*p < .05;

^**p < .01;

^***p < .001 (2-tailed)

Among pre-loss marital context variables, a significant difference in spousal health was noted between widowed and married persons. The proportion of widowed respondents reporting poor health of spouse was about twice that of married respondents. The results also showed significant differences in baseline depression and functional difficulty between Wave 2 samples of widowed and married individuals. The duration between baseline and the follow-up interviews was shorter for the widowed sample than for the married sample at Wave 1, because data collection of the married controls started later than that of the widowed sample, due to a temporary cut in funding. The average duration between the baseline and Wave 2 interview was greater among married persons, due to the controls being interviewed after their widowed counterparts. Widowed persons showed greater dependence on children at both waves, while they reported lower levels of children’s dependence than their married counterparts at Wave 1.

Effect of widowhood on parent-child ambivalence

Table 2 shows the regression models predicting the effect of widowhood on parent’s perception of ambivalence in their relationship with children. Model 1 shows the results of the regression models that do not include indicators of intergenerational dependence as predictors. At Wave 1, widowed persons experienced lower levels of ambivalence than married persons, controlling for baseline level of ambivalence. However, there was no significant difference between widowed and still-married persons at Wave 2.

Table 2.

OLS Regression Predicting the Effect of Widowhood on Change in Parents’ Ambivalence Toward Children

	Wave 1: 6-month follow-up			Wave 2: 18-month follow-up
	Model 1	Model 2	Model 3	Model 1	Model 2	Model 3
Independent variables
Widow (1=widowed)	− 0.44**	−0.45**	− 0.34*	− 0.22	− 0.21	− 0.15
Ambivalence, BL	0.42***	0.42***	0.37***	0.51***	0.51***	0.36***
Pre-loss context of widowhood
Marital quality, BL	− 0.17**	−0.17**	− 0.18**	− 0.10	− 0.10	− 0.10
Spouse’s health, BL (1=poor health)	0.08	0.08	0.12	0.04	0.04	0.09
Sociodemographic and health characteristics
Sex (1=female)	− 0.03	− 0.03	− 0.07	− 0.36**	− 0.34*	− 0.32*
Race (1=white)	− 0.14	− 0.15	− 0.16	− 0.27	− 0.27	− 0.31
Age, BL	0.02	0.02	0.03**	0.00	0.00	0.01
Education (in years)	0.01	0.01	0.01	− 0.03	− 0.03	− 0.03
Depression, BL	0.15*	0.15*	0.14*	− 0.01	− 0.01	0.04
Functional difficulty, BL	− 0.14	−0.15	− 0.12	− 0.03	0.03	0.00
Months between BL and W1 interviews	0.00	0.00	0.00	− 0.01	0.01	0.00
Perceived intergenerational dependence
Parents’ dependence on children		0.01			− 0.00
Children’s dependence on parents			0.28***			0.44***
(Constant)	−1.34	−1.32	− 1.82	1.10	1.03	0.22
Adjusted R2	0.35	0.34	0.38	0.29	0.29	0.37
N	267	267	267	338	338	338

Note: BL refers to baseline; W1 refers to Wave 1 (six-month follow-up).

^*p < .05;

^**p < .01;

^***p < .001

Because the long duration between baseline and the follow-up interviews may confound the relationship between widowhood and ambivalence, in supplementary analyses (not shown but available from the first author), we examined how the effect of widowhood would differ depending on the time elapsed between baseline and follow-up interviews by including an interaction term (widowhood x duration). A significant interaction effect was observed at Wave 2 only. Specifically, the effect of widowhood was most pronounced when the gap between the two interviews was minimal. However, as the duration between the two interviews increased, the differences in levels of ambivalence between widowed and married sample decreased.

The baseline measure of ambivalence was a strong and significant predictor of ambivalence at both follow-up waves, suggesting that the nature of social relationships remains constant to a large extent, despite a stressful life event. The analyses showed that various contextual factors also play important roles in predicting ambivalence in parent-child relationships. At Wave 1, those who had better marital quality at baseline and those who were less depressed reported lower levels of ambivalence. At Wave 2, women reported lower ambivalence than men.

In order to examine to what extent these changes in ambivalence following widowhood are due to positive or negative components of ambivalence, in supplementary analyses, we further examined how widowhood differentially affects these components of ambivalence (data are not presented). At the 6-month follow-up, widowhood did not significantly affect positive interactions with children, yet it was significantly associated with lower levels of negative social interactions. Thus, the diminished ambivalence that we found in the main analysis (Table 2) is likely to be more attributable to decreased negative social interactions between parents and children than to increased positive support. At the 18-month follow-up, however, widowhood did not significantly influence either positive or negative components of ambivalence.

Mediating effects of intergenerational dependence

In examining the impact of widowhood on parent-child ambivalence, we further analyzed the potential mediating effects of parents’ perceptions of their dependence on children (Model 2) and children’s dependence on them (Model 3).

Looking at the main effects of dependence, contrary to our proposed hypothesis that older adults’ greater dependence on children would predict greater intergenerational ambivalence, parents’ perception of their dependence on children had no significant impact at both Waves 1 and 2. However, children’s dependence on parents was an important predictor of intergenerational ambivalence, consistent with previous literature that found significant association between children’s dependence and tension in mother-daughter relationships (Fingerman, 1996).

In order to further explore which aspects of dependence produce significant effects, we conducted additional analyses (data are not presented) examining how intergenerational dependence in specific domains (e.g., emotional, financial, and instrumental) influences changes in ambivalent feelings. However, none of the three kinds of parents’ dependence on children predicted changes in intergenerational ambivalence at both waves. Instead, at Wave 1, children’s dependence on parents for instrumental support predicted an increase in intergenerational ambivalence. Similarly at Wave 2, children’s dependence on parents for financial and legal advice, as well as for instrumental support, predicted an increase in intergenerational ambivalence. Taken together, these findings suggest that, adult children’s dependence produce mixed emotions for parents while the parents’ dependence on adult children does not.

Discussion

Using ambivalence theory, this paper examined the extent to which widowhood affects changes in older adults’ ambivalence toward their children. In this section, we summarize major findings, describe limitations, and discuss their implications for research and practice.

First, inconsistent with ambivalence theory, these analyses revealed that widowhood is not a source of increased ambivalence, but of decreased ambivalence during the early stages of bereavement. At least two reasons may account for why the impact of widowhood on ambivalence differs from the proposed hypothesis. One reason may be that widowhood is a life transition that affects social networks as a whole, rather than the widowed person as an individual. Children share their widowed parents sense of loss and feelings of grief (Umberson & Chen, 1994), and thus, providing support to the surviving parent during the time of adjustment may be seen as a way of coping with the loss together rather than as a burden. Furthermore, adult children may be aware of the surviving spouse’s intense distress during bereavement, and thus, try to minimize conflict, which may contribute to lower levels of ambivalence among widowed parents than among still-married parents. Our component analysis results supported this speculation, showing that widowhood is associated with lower negative interactions (i.e., demands and criticism) especially in the short term.

Our second, and perhaps greatest contribution, is that we investigated the mechanisms through which widowhood affects ambivalence in parent-child relationships. Consistent with our hypothesis, we found that children’s greater dependence on their parents is associated with increased ambivalence. This finding concurs with previous literature, which has suggested that children’s failure to achieve adult status and independence increases parents’ feelings of ambivalence and intergenerational tension (Fingerman, 1996; Pillemer & Suitor, 2002). However, contrary to our hypothesis, the effect of parent’s dependence on their children did not have any significant effect. Though unexpected, this finding may be due to our outcome measure that captured only parents’ perceptions of intergenerational ambivalence rather than both parents’ and children’s perceptions; thus, this measure may not be sufficiently sensitive to parental dependence on children.

Our third major finding is that there is a temporal change in the impact of widowhood on ambivalence. Consistent with our hypothesis, during the early stages of bereavement, widows were far less likely than their married counterparts to experience ambivalence whereas at a later stage of bereavement (i.e., 18-month follow-up), the difference in the effect of widowhood on ambivalence was minimal. This finding may suggest that parent-child relationships are highly adaptive. During early bereavement, widowed persons may experience lower levels of ambivalence because their adult children accommodate their needs through decreased negative social interactions. However, as widowed parents adjust to the acute stress of widowhood, parent-child relationships may return to long-standing patterns of positive and negative exchanges of support, resulting in similar degrees of intergenerational ambivalence when comparing widowed and still-married parents.

Limitations and future directions

By examining the interplay between widowhood, intergenerational dependence, and ambivalence in parent-child relationship, this study offers important insights into understanding changes in social relationships following spousal loss. However, several limitations remain to be addressed by future studies.

First, the current study used parents’ separate evaluation of positive and negative social interactions with their children to construct the measure of ambivalence. Although previous studies have used similar measures to capture the ambivalent nature of intergenerational relationships (e.g., Willson et al., 2003), it should be noted that these measures are limited in assessing the full spectrum of ambivalent feelings. For example, the measure of negative social interactions captures the degree to which others are demanding or critical of the respondents, but it does not inform us about the degree to which respondents have negative feelings toward those individuals. With the indirect measures utilized in this study, we are also limited in evaluating the extent to which respondents have mixed feelings toward their children. Measures of ambivalence are still in the process of refinement and development (Lettke & Klein, 2004), and different predictors influence interpersonal closeness, stress, and ambivalence (Pillemer & Suitor, 2002); researchers should take into account the implications of different measures of ambivalence in interpreting and comparing different study findings. Future studies also should examine the impact of widowhood using direct measures of ambivalence, such as the degree to which one has mixed feelings toward others (Pillemer & Suitor, 2002).

Second, the measures of intergenerational dependence as well as positive and negative social interactions used in this study reflect only widowed parents’ perceptions. Although parent’s widowhood is a shared experience of loss within a family, because aging parents and their adult children are at different stages in adult development (Fingerman, 1996), bereaved persons and their children may hold different views or expectations of each other. Bereaved persons may expect increased positive support from children, and consider their dependence on children unavoidable. On the other hand, children may view their bereaved parents’ increased need for support as burdensome. They may also expect their parents to overcome grief in a short time period. Examining the perceptions of adult children vis-á-vis dependence and comparing them with that of their widowed parents would provide important insights into understanding the dynamics of intergenerational ambivalence during stressful life transitions.

It should be also noted that the questions on children’s sit is not clear whether respondents think about all of their children, the child with whom they’re closest, or the child with whom they are most distant when answering the question items.

Finally, widowhood is only one example of a stressful life transition in old age. Because the death of a spouse is a shared bereavement experience for family members, children may be more willing to provide support. Future studies should examine how other stressful life transitions that are more person-centered, such as retirement, affect intergenerational ambivalence over time.

In conclusion, this study contributes to the intergenerational relationship literature by applying ambivalence theory to examine changes in social relationships with adult children following late-life widowhood. The study expands the application of ambivalence theory in several important ways. First, it extends the scope of empirical research in this area by examining the effect of late-life transition, as opposed to transitions that occur from adolescence to adulthood. Furthermore, it examines how parents’ evaluation of intergenerational relationship changes in response to their own transition, as opposed to their adult children’s transitions. In addition, the findings shed light on the mechanisms by which dependence contributes to intergenerational ambivalence. Future studies should build upon this research by examining how adult children’s view of intergenerational relationships change as their parents experience life transitions in old age.