The Role of Violated Caregiver Preferences in Psychological Well-Being When Older Mothers Need Assistance

Abstract

Purpose: Theory and research suggest that congruence between individuals’ preferences for future care and the patterns of care received will affect well-being. In this article, we explore whether older mothers’ psychological well-being was affected by the children they preferred as future caregivers and provide assistance at a later point when the mothers experience illness or injury. Design and Methods: In this article, we use a combination of quantitative and qualitative data collected from 234 older mothers at two points 7 years apart, beginning when the mothers were 65–75 years of age. Results: Multivariate analyses demonstrated that mothers who received assistance from children whom the mothers did not identify as their preferred future caregivers reported higher depressive symptoms at the second wave; receiving care from children identified as preferred caregivers did not affect well-being. Qualitative data suggested that these patterns occurred because the “alternate” caregivers did not possess the socioemotional attributes of preferred children. Implications: These findings contribute to a growing body of research demonstrating the consequences of violated preferences, particularly when individuals are in need of support in later life.

There has been an increasing interest in adult children’s care for their parents, resulting in a substantial body of knowledge documenting the experiences of these family caregivers. However, this trend has not been accompanied by a focus on the experiences of care recipients (Pruchno, Burant, & Peters, 1997). This pattern has created a substantial gap in what is known about the complex relationship between family caregivers and care recipients.

In this article, we examine the experiences of mothers who suffered a recent illness or injury for which they needed help, focusing on the consistency between mothers’ preferences for a particular adult child caregiver and which children actually assumed that role when the need arose. Theory and research suggest that consistency between preferences for care and actual patterns of care will have a major impact on mothers’ psychological well-being. To shed light on this question, we use a combination of quantitative and qualitative data collected from 234 older mothers at two points 7 years apart, beginning when the mothers were 65–75 years of age.

Parental Favoritism, Caregiving Expectations, and Mothers’ Well-Being

Some studies of older parents have shown that receiving care from adult children has negative effects on psychological well-being of care recipients, even when controlling on declines in health that necessitated the assistance (Lee, Netzer, & Coward, 1995; Silverstein, Chen, & Heller, 1996). Such effects have been attributed primarily to parents’ perceptions that the care was excessive, inadequate, or did not meet the parents’ expectations in some other regard. We propose that an additional factor explaining this counterintuitive pattern may be that the assistance was not provided by children whom the parents would have preferred as their caregivers.

Studies of parental favoritism in adulthood have shown that older mothers typically have clear preferences as to which children would be serving the role as caregiver (Suitor & Pillemer, 2006). Further, these patterns remain remarkably stable across time as mothers move from being completely independent to requiring assistance (Suitor & Pillemer, 2011). Evidence from other stages of the life course (particularly the transition to parenthood) has demonstrated that when preferences and expectations for assistance and support from role partners are violated, individuals often experience psychological distress (Belsky, 1985; MacDermid, Huston, & McHale, 1990; Ruble et al., 1988). We expect that violated preferences will also be consequential for older parents when facing health problems for which they need assistance. The basis for our argument can be drawn from Carstensen's theory (1992) of socioemotional selectivity, which posits that as people age and their time perspective alters, they focus on interpersonal relationships that are most rewarding and increasingly withdraw than those that are not. This may help to explain why mothers prefer adult child caregivers whose socioemotional characteristics increase the likelihood of positive experiences and reduce the risk of negative experiences when in need of assistance (Pillemer & Suitor, 2006; Suitor & Pillemer, 2006).

Given the emphasis of literature about the caregiving role of adult children’s availability and resources (Pavalko, 2011), we might expect that these factors would form the basis of mothers’ identification of preferred caregivers. However, when mothers have been asked to articulate why they would prefer one child over another when facing illness or disability in the future, their responses emphasized socioemotional characteristics of the parent–child dyad (Pillemer & Suitor, 2006; Suitor & Pillemer, 2006). Most salient features appeared to be within-family differences in their values and experiential similarity to their adult children. Specifically, both quantitative and qualitative analyses revealed that mothers strongly preferred children as caregivers with whom they shared a common outlook on life as well as a common set of life experiences. Not surprisingly, their concern with experiential similarity also led them to prefer daughters more than sons (Suitor & Pillemer, 2006).

Mothers also disproportionately preferred those children as caregivers with whom they had the most stable history of supportive exchanges (Pillemer & Suitor, 2006; Suitor & Pillemer, 2006, 2011). Thus, when considering their future caregiving needs, mothers are very deliberate in identifying children whom they believe will most likely understand and share their concerns and respond to their needs based on similarity and history. In other words, consistent with socioemotional selectivity theory, the mothers are selecting those children whom they have the greatest confidence will be a source of reassuring support but not selecting those whom they think will be a source of unmet expectations or conflict.

Based on these theoretical arguments and empirical findings, we hypothesized that receiving care from children whom the mothers identified in advance as their preferred caregivers would be associated with greater psychological well-being, whereas receiving care from children who were not identified would have detrimental consequences.

Methods

The data used were collected as part of the Within-Family Differences Study (WFDS). The design of the WFDS involved selecting a sample composed of mothers 65–75 years of age with at least two living adult children and collecting data from mothers regarding each of their children. (For a more detailed description of the WFDS design, see Suitor & Pillemer, 2006). The first wave of interviews took place with 566 women between 2001 and 2003; the original study was expanded to include a second wave of data collection from 2008 to 2011. At T2, a total of 420 mothers were interviewed regarding 1,514 of their children. In this article, we used data from 234 mothers who reported having needed assistance when they became ill or disabled at some point within two years prior to the T2 interview.

Procedures

Massachusetts city and town lists were used as the source of the original WFDS sample. With the assistance of the Center for Survey Research (CSR) at the University of Massachusetts, Boston, the researchers drew a probability sample of women aged 65–75 with two or more children of the greater Boston area. The T1 sample was composed of 566 mothers, representing 61% of those who were eligible for participation, a rate comparable to that of similar surveys in the past decade (Dixon &Tucker, 2010; Wright & Marsden, 2010).

For the follow-up study, the survey team attempted to contact each mother who participated in the original study. At T2, 420 mothers were interviewed. Of the 146 mothers who participated at only T1, 78 had died between the two waves, 19 were too ill to be interviewed, 33 refused, and 16 could not be reached. Thus, the 420 represent 86% of mothers who were living at T2. Comparisons between the mothers alive at T2 who did and not participate revealed that they differed on only education and subjective health; those who participated were better educated and in better health.

For this article, we used the analytic sample of 234 mothers who reported an injury or illness for which they needed assistance at some point within two years prior to their T2 interview. Mothers were included if they experienced a new health problem for which they needed help within the two previous years or if they had a chronic condition for which they reported that they needed help within the same period. In this article, there were no missing data on the variables of interest. We conducted a lagged endogenous regression analysis using OLS in SPSS Version 19. Using this technique in which we control on mothers’ depressive symptoms at T1 allows us to predict changes in depressive symptoms from T1 to T2.

Table 1 presents the demographic characteristics of the analytic sample. Although the mean number of living children is higher in this subsample than would be found in a national representative sample of women in this age group, this feature is primarily due to the criterion that all participants must have at least two living adult children. The mean number of children of women in the subsample is similar to that found in national representative samples, such as the National Survey of Families and Households (Sweet & Bumpass, 1996), when compared specifically to mothers in the same age group who have two or more children.

Table 1.

Description of the Variables

Characteristics	Mean, SD, % (N = 234)
Marital status (in %)
Married	35.9
Divorced/Separated	14.5
Widowed	48.7
Never married	0.9
Education (%)
Less than high school	17.5
High school graduate	42.8
Some college	13.7
College graduate	26
Number of children (SD)	3.7 (1.6)
Race (%)
White	71.4
Not White	28.6
Age in years (SD)	77.9 (3.2)
CES-D at T1	11.2 (4.1)
CES-D at T2	11.6 (4.6)
Subjective health at T2	2.9 (1.0)
Received care from child/children, no T1 caregiver preference	20.1
Received care from children preferred at T1	42.7
Received care from child not preferred at T1	13.7
Received no care from children	23.5

Note: CES-D: Center for Epidemiological Studies Depression Scale.

Measures

Depressive Symptoms.—To measure depressive symptoms, we employed the seven-item version of the Center for Epidemiological Studies Depression (CES-D) scale (Ross & Mirowsky, 1988). The CES-D scale asks respondents how often in the past week they have felt a certain way. It should be noted that the CES-D was not intended for use as a diagnostic tool. Rather, it provides a valid and reliable means for ordering individuals on the basis of the frequency and severity of their symptoms. The CES-D scale’s reliability and validity for use in community surveys have been clearly established (Radloff, 1977). The items comprising the scale are as follows: (a) Everything I did was an effort, (b) I had trouble getting to sleep or staying asleep, (c) I felt lonely, (d) I felt sad, (e) I could not get going, (f) I felt I could not shake off the blues, and (g) I had trouble keeping my mind on what I was doing. In this sample, the scale for mothers at T2 ranged from 7–28, with a mean of 11.64 (SD = 4.58) and an Alpha coefficient .83. The scale at T1 (which we included as a control) ranged from 7–22, with a mean of 11.23 (SD = 4.07) and a Cronbach’s Alpha coefficient of .80.

Congruence between Mothers’ Caregiver Preferences and Outcomes

Creating a measure of congruence between mothers’ preferences and outcomes regarding caregivers required comparing which children they preferred at T1 as their future caregivers with those children, if any, who actually provided them care when they were ill or injured in the 2-year period prior to the T2 interview.

We began by creating measures of mothers’ caregiving preferences at T1 and actual provision of assistance at T2. To assess mothers’ future caregiver preferences at T1, each mother was asked to select which child she would prefer help from if she (the mother) became ill or disabled. Approximately, three quarters of the mothers named a particular child at T1 whom they would prefer help from if they became ill or disabled. Based on these responses, each child was coded as having or not having been chosen as their mother’s preferred future caregiver. At T2, each mother was asked whether any of their children had provided help for their recent health problem(s), and if so, which child or children had provided help; each child was then classified as having or not having provided assistance for those problems.

The congruence measure was created by comparing the mothers’ stated caregiver preferences at T1 to which children actually provided care for a health problem within the two years prior to T2. We began by developing a 4-category variable of congruence between mothers’ caregiver preference and actual caregiving provided by children for the recent health event. The categories include (a) received no care from children, regardless of T1 preferences (n = 55), (b) received care from children preferred at T1, either alone or in combination with children not preferred (n = 100), (c) received care from children but had expressed no caregiving preferences at T1 (n = 47), and (d) received care only from children not preferred at T1 (n = 32). Finally, we created a set of dummy variables reflecting these four categories. Having received no care from adult children was assigned as the referent category in the regression analysis.

Control Variables

Race was measured by asking the mothers to select from a card listing several races and ethnicities (e.g., White, Black or African-American, Hispanic or Latina, Native American, Asian). The mothers could choose more than one race or ethnicity. For this article, the analytic sample included 167 mothers who identified themselves as White, 60 as Black, 3 as Hispanic, 3 as Native American, and 1 as Asian. Based on the literature on later-life families, which has shown greater filial responsibility in Black, Asian, and Hispanic than White families, we coded race as White = 1 and Not White = 0.

Marital status was coded as married = 1 and not married = 0. Age at T2 was the age mothers provided at T1 plus 7 (the number of years between interviews). We measured family size by asking mothers for the names of each of their children at T1, and asking them to confirm this at T2.

We included subjective health as a control, because it has been found to be a strong predictor of depressive symptoms (Geerlings, Beekman, Deeg, & van Tilburg, 2000); poor = 1 and excellent = 5.

The descriptive statistics of the variables included in the analysis are presented in Table 1.

Qualitative Data

Combining quantitative and qualitative data has become increasingly common in research on families across the past decade (Carr, 2005; Creswell & Plano-Clark, 2010; Neal, Hammer, & Morgan, 2006; Plano-Clark, Huddleston-Casas, Churchill, Green, & Garrett, 2008; Wenger & Burholt, 2001). Although quantitative analyses can identify patterns of relationships among variables, such analyses are less fruitful for pursuing the processes underlying statistical relationships (Morgan, 2007; Neal et al., 2006; Umberson, 2003; Umberson & Montez, 2010). The aims of this article are particularly well suited for such an approach. Specifically, our goal is to test hypotheses regarding the association between violation of mothers’ caregiving preferences using the quantitative data and to attempt to explain why this association might occur using the qualitative data.

At both waves of data collection, the interviews were conducted in person and were audio-taped in almost all cases. A research team of nine students transcribed the interviews, coded the open-ended items, and prepared detailed case summaries of each family. Codes were developed for the open-ended items as data preparation continued, rather than being established, prior to the coding process. In contrast to having coders working independently and calculating kappas based on coders' consistency, we used a consensus approach based upon the group interactive analysis component of Borkan’s “immersion/crystallization” method for analyzing qualitative data (Borkan, 1999). Each week, one of the principal investigators (PIs) surveyed all the open-ended coding that had been completed during the previous week. Approximately 90% of the coders’ original decisions were in agreement with those of the PI; any coding that was not in agreement with the PI’s assessment was discussed by the entire group at weekly team meetings until consensus could be reached.

For this article, we have focused primarily on the mothers’ responses to questions regarding which children they identified as their preferred future caregivers at T1 and T2, as well as their responses to a series of open-ended questions about which children they preferred in other relational contexts (confiding, help in an emergency, socializing), which children the mothers felt were the most similar to themselves, and the ways in which they felt that their children differed from one another. We also examined the statements of mothers who had experienced conflict with or had been disappointed in their children regarding care during the recent illness or injury. The lead author and a coauthor, who was also a member of the research team that coded and transcribed the interview material, used this set of responses to identify themes in the mothers’ discussions of their experiences as care recipients and the basis for preferring particular children as caregivers. They also selected examples to illustrate these processes. All names used in the qualitative section are pseudonyms.

Results

Multivariate Analyses

Table 2 presents the predictors of mothers' depressive symptoms. This analysis indicates that neither receiving care from children whom the mothers preferred at T1 nor receiving care when mothers had no preference at T1 predicted changes in mothers′ depressive symptoms. In contrast, having received care from only children whom the mothers did not prefer predicted changes in depressive symptoms between T1 and T2. In fact, receiving care from a child not preferred at T1 was as strong a predictor of mothers’ depressive symptoms at T2 as subjective health and one of only three significant predictors.

Table 2.

OLS Analysis Predicting Depressive Symptoms

	b (SE)	β
Married	.02 (.58)	0.01
Education	.10 (.15)	0.04
Family size at T2	−.01 (.15)	−.01
White	.60 (.58)	0.06
Age at T2	.04 (.08)	0.03
CES-D at T1	.66** (.06)	0.58
Subjective health at T2	−.61* (.25)	−.14
Received care from child/children, no T1 caregiver preference	−.66 (.76)	−.06
Received care from child/children preferred at T1	.43 (.66)	0.05
Received care from child not preferred at T1	1.69* (.86)	0.13
Model statistics
Adjusted R 2	0.37
Df	10
N	234

Note: Having received no care from adult children was assigned as the referent category. ^* p < .05, ^** p < .01.

To assess whether the absence of effects of receiving care from the preferred child could be accounted for by pooling mothers who received care from only the preferred child and those who received care from the combination of preferred and not preferred children, we conducted a second analysis. In this analysis, we disaggregated the “preferred only” and the “mixed preferred and not preferred” cases. The second analysis in which these two variables were entered separately along with “care only from child not preferred” also revealed no effects of receiving care from preferred children.

Thus, receiving care only from children who had not been identified as preferred caregivers at T1 was associated with an increase in mothers’ depressive symptoms between T1 and T2, whereas receiving care from preferred children, whether alone or as part of a team of caregivers, did not predict mothers’ psychological well-being. Further, given the referent category, the findings suggest that mothers’ depressive symptoms at T2 were higher when they received care from not preferred children than when they received no care from offspring.

It is worth noting that T1 CES-D explained a great deal of the variance in T2 CES-D (32%), as would be expected. In contrast, mothers’ violated caregiving expectations accounted for slightly less than 2% of the variance in T2 CES-D scores. However, to put this result in perspective, the same percentage of variance in T2 CES-D scores was explained by subjective health—a factor that is considered to be strongly predictive of psychological well-being (Beekman, Kriegsman, Deeg, & van Tilburg, 1995; Geerlings, Beekman, Deeg, & van Tilburg, 2000). Another way to interpret the effect of violated preferences is that mothers whose preferences were violated reported CES-D scores nearly 2 points higher than did mothers who received no care from their children. Thus, given the large number of factors that affect individuals’ psychological well-being, this is a notable finding.

Finally, we investigated whether mothers with a large number of children would be less likely to have their caregiving preferences violated because they would have more “alternate” preferred caregivers. When we investigated this issue in a separate analysis, however, there were no differences by family size. Further, in that analysis, we also considered other characteristics of mothers that might differentiate between those whose caregiving preferences were and were not violated, including age, race, marital status, educational attainment, and psychological well-being at T1. We found no characteristics that differentiated between these groups of mothers (tables not shown).

Using Qualitative Data to Explain Why Violated Expectations Matter

Next, we turned our attention to the qualitative data to shed light on the question of why receiving care from children the mothers had not identified at T1 as preferred caregivers led to higher depressive symptoms at T2. To explore this issue, we examined the complete transcripts of each of the 32 mothers who were provided care by children whom they had not specified as preferred future caregivers at T1. In three-quarters of the cases in which mothers’ caregiving preferences had been violated, their discussions made clear the issue of primary concern appears to have been differences in the socioemotional attributes of the children they preferred and those who later provided them with care.

The mothers drew clear contrasts between the children they had identified at T1 as their preferred future caregivers and those who provided care at a later point. In particular, the mothers’ statements highlighted marked differences between the personalities of the preferred and not preferred children, and the overall quality of the relationship between the mother and the preferred and not preferred children. Mothers clearly preferred those children as caregivers whom they considered to be their most empathetic, loving, and supportive children.

In some cases, the mothers shared experiences that had reinforced their beliefs that some of their children would be “better” caregivers than others, based on differences in their personalities. For example, at T1, Joan, a 73-year-old widowed mother of six made explicit comparisons between the daughter she “preferred” as her caregiver, who lived further away and the daughter whom she thought would become her caregiver because she lived nearby:

My oldest daughter, Nancy, we can talk about anything. I think she understands me the best. [My younger daughter] Linda’s good, but I wouldn’t tell Linda everything that I would tell Nancy. [Linda] would get annoyed with me. Although she’s my rock here because she’s the one that’s here. But again—see it’s hard because if Nancy were here, she’d be the one, you see. [Nancy is] a very loving, caring person. Very [loving] … and she would be more understanding.

Between the first and second interviews, Joan experienced two serious falls, both requiring that she be on bed rest for an extended period. Following the first accident, Nancy came and stayed with her. However, for the more recent incident, Nancy was unable to come, and Linda assumed the role of caregiver. When asked at T2 which of her children she would prefer as her caregiver when she became ill or disabled again and needed assistance, she responded without any hesitation that she would prefer Nancy. When asked why, she laughed and explained:

Because it’s happened! I had a terrible fall [more than two years ago] and Nancy came and took care of me for five weeks. Whereas later, I had another fall [more recently] and my daughter Linda had me go into a [residential] rehab.

Interestingly, Joan said that she would call Linda if there was a different type of crisis, both because she “would have to” because Linda was the most proximate and because Linda was especially “strong” and “wouldn’t fall apart.” However, Joan made it clear that for problems in which understanding or empathy was essential, she continued to prefer Nancy. Her reasoning at T2 reflected the same differences that she had described at T1:

[Nancy’s] more understanding … I could tell Nancy just about anything, whereas, Linda is not as emotional. Like Nancy [would] say, “Oh Mom, that’s alright. I’m here for you.” [In contrast] Linda would say, “Mom, get over it.” … that’s just how she is.

Rarely were mothers’ caregiving preferences violated because the preferred children deliberately chose to disregard the mothers’ wishes. Most often, the preferred caregivers were unavailable because they had moved away for their own or their spouses’ employment, or had medical problems themselves, making it difficult to provide care to their mothers in a health emergency. Nevertheless, other children were often unable to fill the caregiving void satisfactorily from the mother’s perspective, primarily because the “alternate caregivers” did not share the socioemotional attributes of the children whom the mothers had identified at T1 as their preferred caregivers.

Wanda, a 75-year-old divorced mother of four sons, identified Charles as her preferred caregiver because “He has more patience.” However, between the T1 and T2 interviews, Charles needed to move across the country and Brian, who remained nearby, assumed the role of caregiver for his mother when she developed a chronic illness. In the T1 interview, Wanda made few statements that differentiated among her children. However, at the T2 interview, she drew many distinctions among them. The clearest contrasts were between Charles and Brian. When asked to discuss the differences among her children, she described Charles as being very affectionate and concerned about her well-being:

He’s very concerned about me. Always calling me, telling me he loves me, and he comes, you know. Very affectionate.

She also described Brian as being concerned but with a much more instrumental focus:

He’s helpful, and, he checks on me every day, and night sometimes. He wants to know everything I’m doing.

She also described him as being knowledgeable, but her wording suggested irritation regarding his attitude:

Because he thinks he knows it all. He’ll try to tell you, to the best of his knowledge, what he thinks about it.

Early in the T2 interview, Wanda was asked who she thought would become her caregiver, and who she preferred as her caregiver when she needed help again in the future. She responded that she would prefer her youngest son, James, but explained that he would be unavailable because he was suffering from a life-threatening illness. Thus, she expected that Brian would be her future caregiver because he was most proximate and more available, but she was not entirely happy with this arrangement:

[Brian] would help, but it’s a different attitude [from James]. Is that clear or what? [Brian], he’d be reluctant—he’ll help, but he’ll, you know, give a little sigh or you know.

However, later in the interview, her disappointment with Brian as a caregiver became more obvious when she described her recent experience with him in this role:

Sometimes I’d want him to do something for me and he’d always say, “Well, I’m not the only one, you know.” When I would get mad with him [over this] then he would call [his brothers] and ask them “How much do you help Mom?” [He would] try to put it on them too ...

She found Brian’s attitude particularly frustrating because he lived nearby and was not employed at the time, leading her to feel that he was the one on whom she should be able to depend the most. Further, unlike James, he was healthy as well as less encumbered with other responsibilities. Thus, Wanda did not receive care from either Charles, whom she had preferred at T1, or James, whom she preferred at T2, but instead received care from a son who was emotionally cool and willing to directly express his irritation at having to provide her care.

One point that became apparent throughout the cases in which mothers’ preferences had been violated was that even high quality care from other children was unlikely to compensate for the absence of the preferred caregiver. Loretta, a mother of five, provided an illustration of this point. Loretta was 66 and married when she participated at T1. She became widowed and experienced a serious decline in her health over the next several years, requiring by T2 that she have assistance with several activities of daily living. At T1, Loretta described her daughter, Rhonda, as the child most similar to her as well as the one she would prefer as both her confidant and her caregiver. As Loretta described her, “Rhonda is much more mature than my other children—she is my sweetheart” in contrast to another daughter who “is a yeller and a screamer.”

However, two years prior to the T2 interview, Rhonda’s partner died suddenly, leaving her with seven children, three of whom were still young and living at home. Rhonda had not completed high school and was struggling to support herself and her children. As Loretta explained, the situation “took out a lot on” Rhonda. As a result, looking to Rhonda for assistance with activities of daily living was not a consideration when Loretta’s health declined substantially. Instead, each of Loretta’s other children provided support on a regular basis, yet Loretta said that she never felt that any of them lived up to her expectations for care. Further, despite the fact that each of her other children was also coping with his or her own health problems, Rhonda, who contributed no care, was the only child with whom Loretta did not express disappointment regarding her care.

In summary, the qualitative data both confirmed the patterns shown by the multivariate analyses and revealed that the effects of the violation of the mothers’ preferences emanated primarily from the inability to secure care from those children whose socioemotional characteristics made them their mothers’ first choice when facing illness or disability.

Discussion

The question we posed in this article was “Do mothers’ preferences for caregivers play a role in their psychological well-being when facing illness or disability at a later point in their lives?” The results we have presented suggest that this is the case. However, what appears to be most important is not whether mothers received care from the children they identified as their preferred caregivers, but instead, whether they received care only from children whom they did not identify as their preferred caregivers. In fact, mothers’ depressive symptoms were not predicted by receiving care either from the preferred caregivers alone or from the preferred caregivers in concert with their siblings.

These patterns mirror those found in the broader literature on support and well-being. One of the most consistent findings of this line of work is that negative interpersonal events are more salient than positive events. Research has shown that support that does meet individuals’ expectations is associated with increased stress and lower well-being (Baumeister et al., 2001; Rook, 1984, 2001; Schuster, Kessler, & Aseltine, 1990). We suggest that parental caregiving is a context in which unmet expectations are especially likely to reduce well-being given that the presence of harmonious and supportive relationships plays a salient role in health and recovery from illness in the later years (Birditt, Antonucci, & Tighe, 2012; Gallant, Spitze, & Prohaska, 2007; Cornwell & Waite, 2009).

Further, qualitative analyses suggested that mothers’ distress was the result of the inability to receive care from children whom they felt possessed highly desirable socioemotional characteristics. The mothers’ comments revealed that in selecting their preferred future caregivers, they valued socioemotional characteristics of the children and the parent–child dyad that ensured predictability and harmonious relations—a pattern consistent with principles of socioemotional selectivity theory (Carstensen, 1992). Specifically, the mothers chose those children whom they perceived to be the most empathetic and similar to themselves from among their offspring, consistent with other research (Pillemer & Suitor, 2006; Suitor et al., 2006). Further, their perceptions by T2 were that the children they had preferred earlier were still the most empathetic and similar to them, even after receiving care from other children. Thus, mothers’ preferences were strong, enduring, and very specific, making the violation of these preferences particularly problematic.

This study suggests several directions for future research. First, it is possible that unmet expectations for specific caregivers could be particularly distressing when the children’s own problems preclude providing care to their mothers, as was the case for several of the mothers in our study. However, given that only 32 of the mothers reported not being provided care by their preferred children, there is not a sufficient number of cases to address this issue. Second, the present analysis is based on the reports only from the perspective of the mothers. Given that mothers’ and children’s perspectives often vary (Giarrusso, Feng, & Bengtson, 2004), including regarding parental differentiation (Suitor, et al., 2006), it is important to study caregiving expectations from the perspectives of both parties. Third, it is interesting to note that the mothers’ preference for children focused specifically on positive expressive attitudes and behaviors, with little regard for situational factors—a finding that is consistent with the literature on fundamental attribution errors (Follett & Hess, 2002) Scholars might find fundamental attribution error a fruitful framework when studying parental differentiation.

These findings have important implications for both theory and practice. First, one of the central issues in social gerontology is determining the conditions under which caregiving is the most beneficial and the least stressful for both the caregiver and care recipient. The patterns we have presented indicate that adhering to individuals’ caregiving preferences—especially avoiding children whom the parents do not want as their caregivers—plays a salient role in positive outcomes from the perspective of care recipients. Second, theoretical literature on sense of control has called for studies exploring the consequences of maintaining or losing sense of control in clearly defined contexts (Krause, 2003); this research meets this criterion by focusing on a clearly defined, highly salient, and increasingly common context. It is important that future research expands the study of met and violated expectations in the context of caregiving, including examining whether the importance of this factor varies by parents’ gender and by the point in the life course when individuals face serious illness and disability.