Science Briefs

Father Knows Best (Unless I Know Better): Family Relationships in Late Life

Family members are involved in many aspects of elder care, and psychological research has an important role to play in helping families.

By Brian D. Carpenter, PhD

Robert "Bob" E. Welch, 56, of Enfield passed away Saturday (January 29, 2005) at the West Haven Veterans Hospital after a long battle with his family at his bedside.

-- obituary from the Hartford Courant 

The quote above is both a humorous misstatement and a pointed reminder that family relationships are complicated across the lifespan, even at the point of death. As people grow older, they face many decisions regarding their health care, housing, and finances, among other concerns. Often, family members are involved in those decisions, providing input, offering advice, or even making decisions when older adults are no longer able to speak for themselves due to cognitive or physical impairment. Many state laws have codified this kind of participatory role for family members (American Bar Association, 2004), reflecting the belief that family members know each other well, or at least are likely to act in each other's best interests (President's Commission, 1983). Yet recent experience and research suggest that family members may not be the knowledgeable, unbiased informants we expect. Consider, for instance, the recent case of Terri Schiavo, a Florida resident in a persistent vegetative state whose family members were in conflict over the continuation of her medical care (Quill, 2005). This was but one example of family members who had radically different beliefs about how a person would have wanted to live.

The reality is that family members are involved in many aspects of elder care, and psychological research has an important role to play in helping families. In a number of research studies, we have attempted to learn more about the knowledge-or lack of knowledge-of family informants, with the ultimate goal of developing interventions to help families face the challenges that aging will bring to the entire family system.

What do family informants know?

Previous studies have explored informant knowledge in the realm of health care preferences. Roberto (1999), for example, found only low to moderate agreement (kappas <.60) between older adults and family members who were asked to predict preferences regarding hypothetical health care procedures and life-sustaining treatments. When other researchers ( Muncie, Magaziner, Hebel, & Warren, 1997) examined informants' ability to predict opinions about hypothetical medical treatments, informants' estimates were closer to what they would have chosen for themselves, rather than what their family member would have wanted. As Muncie et al. explain, "It appears that the proxy does not stand in the shoes of the charge. Rather, this study suggests that the proxy decides as though the charge were in the proxy's shoes." (p. 932).

In our research we have taken an expanded perspective on preferences by also looking at family members' knowledge about a broad range of preferences, including those for medical care. In a first study (Carpenter, Lee, Ruckdeschel, Van Haitsma, & Feldman, 2005) we examined the psychosocial preferences of frail older adults, asking them about preferences regarding their daily routine, leisure activities, living environment, and other minutiae of daily life. We consider these seemingly minor details to be important to quality of life, perhaps as or even more important than what might appear to be momentous decisions. A brief conversation with most nursing home residents about what time they are required to get up in the morning, how they are bathed, what clothes they can choose to wear, not to mention whether they can have a beer with dinner or a private space for intimacy, makes clear how important these details are.

In our studies we have we settled upon a methodology that, at the risk of endangering our credibility, is not unlike The Newlywed Game, the once popular TV show in which newly married couples were tested on how well they knew each other. In our adaptation, older adults tells us about their preferences and wishes, and then in separate interviews family members predict how their older relatives answered those same questions. Our results suggest that, overall, adult children are good at predicting concrete, observable parent preferences, such as leisure activities (average intraclass correlation = 0.73) but are less accurate when estimating parent preferences for continued personal growth (traveling, attending cultural events; average ICC = 0.59) and autonomy (choosing their own routine and surroundings; average ICC = 0.64). Across the board, adult children tend to underestimate how important these preferences were to their parents.

In this first study we learned many lessons, perhaps the most important of which was that we had overlooked important people in the family. A frequent comment was something like, "My sister would probably know a lot more about these things than me." So in our next study we expanded our sampling technique to include two adult children. And we broadened the preferences we asked about to include not only psychosocial preferences but also financial, housing, and medical preferences. As in the previous study, we have found that, overall, children are only moderately accurate at making predictions about their parent's preferences. To give some examples, many children are mistaken about parent preferences regarding organ donation (underestimate), their willingness to be on mechanical ventilation (overestimate), readiness to accept financial assistance from children (overestimate), and housing (children tend to think their parents would be interested in living with them, while parents are firm in not wanting to live with their children). Once again, family informants seem to have spotty knowledge about older adults.

In both these studies it is important to note that we found a broad range of knowledge among adult children: some children are quite knowledgeable while others are no better than chance at estimating what their parent would want. (A hyperbolic way of putting it is that for some parents, their child is as knowledgeable about them as any random child plucked from the street.) In trying to identify correlates of childrens' knowledge, we have found a few promising candidates. Parents who are more physically frail tend to have children who are more knowledgeable, suggesting the possibility that frailty may prompt the family to begin to investigate parent preferences. In contrast, older parents who are depressed seem to have children who are less knowledgeable, perhaps indicating that these parents have withdrawn from their children, leaving children less aware. And family members who experience more caregiving burden are less knowledgeable, which may reflect the psychological hurdle of making a prediction through the scrim of one's own distress. No other demographic or contextual variables have been consistently predictive of childrens' knowledge-not gender (of parent or child), age, geographical proximity, etc.

More recently we have begun to explore family-level attributes that also may be informative. We approach our research from a family systems perspective, believing that longstanding patterns of interaction within the family may be related to contemporary family functioning. (And yes, we finally learned our lesson and now try to include as many family members as possible.) In particular, structural family theory (Minuchin, 1974) suggests that families may have maladaptive patterns of behaving with one another that impede the family's ability to respond to developmental challenges, such as parent care (see Carpenter, 2001). A number of other research groups, for instance, have found that family dynamics help explain how families respond to caregiving challenges (Lieberman & Fisher, 1999; Mitrani, Feaster, McCabe, Czaja, & Szapocznik, 2005). We have added an observational component to our protocols, asking families to complete simple problem solving and interaction tasks, which we videotape and code for a range of communication and affective features. These data are still under analysis, but our aim is to determine if aspects of family interactions, such as interruptions, overspeaking, disengagement, and criticality, may predict childrens' knowledge. That kind of data might, in turn, have implications for family intervention. Here I should add that few assessment tools, observational or self-report, have been validated with older families, meaning that much more basic research is needed to create reliable and valid assessment instruments that can be used in family geropsychology.

From surveys to services

During our conversations with research families-and even amongst ourselves-it became clear that families just do not talk about preferences and wishes. As an explanation, adult children tell us they are not sure how to bring up a topic that seems morbid. Older parents tell us they try to bring the topic up with their children but they are rebuffed. When Cicirelli (1992) asked adult children and their parents if they had talked about future caregiving preferences, 42% of the parents said they had never discussed their preferences with their children. Meanwhile, only 20% of the children agreed, suggesting that adult children may think they have discussed parent preferences adequately even when parents still have more to talk about. Instead of talking about preferences directly, adult children tend to rely on implicit assumptions about what a relative would want (Pecchioni, 2001). And so, decisions about elder care are based on speculation rather than conversation. As clinical scientists with a commitment to applied research, we plan to use what we are learning to develop family education programs that improve communication about parent preferences.

In a recent feasibility study, we tested an intervention with a small sample (n = 24) of families. We broadened our recruitment once again, inviting all adult children in the family to participate, and we developed a brief psychoeducational program. The program involves meeting with each family, teaching basic communication skills and strategies for collaborative decision making, and describing the kinds of decisions families commonly face as parents age. The core of the intervention, however, is a guided discussion through a personalized workbook that we create for each family. In the workbook we list parent preferences side by side with the assumptions each child made about parent preferences. In this way everyone in the family sees what children know and don't know about their parent. We lead families through a discussion of points of agreement and disagreement, and the family keeps the workbooks for future reference. We hope this kind of tool will make conversations less threatening and will provide the structure and detail families need to consider all the decisions they may some day encounter. A formal evaluation of the psychoeducational program is in the planning phases.

One question that has dogged us throughout our research is whether what we are learning about families in our idealized lab- and home-based protocols will generalize to real-life situations, a concern raised by others (see Roberto, 1999). Decisions about where a parent should live, whether to start (or end) a medical treatment, and how to disperse a deceased parent's assets are made under time constraints and in the presence of powerful emotions. Even if we manage to improve knowledge about parent preferences, there is no guarantee family informants will act on that knowledge. For many complicated reasons, surrogate decision makers override or ignore parent preferences. The practical, legal, and ethical issues in this area deserve their own empirical attention.


Family members play an important role in the support and care of aging parents. Demographic patterns suggest that the vast majority of people will at some point be involved in parent care. Psychologists can play an important role in helping families navigate the challenges of aging. We bring to the table an understanding of lifespan development, cognitive aspects of decision making, and interpersonal dynamics, and we have skills in developing and evaluating intervention programs. In our own research, we have the modest aim of getting families to talk about some of the aging issues they might face before a crisis arises. These conversations are not easy for families to have, but they might help avert problems and conflicts in their futures.


American Bar Association. (July 1, 2004). Surrogate consent in the absence of an advance directive. Downloaded from on 7/8/05.

Carpenter, B. D. (2001). Attachment bonds between adult daughters and their older mothers: Associations with contemporary caregiving. Journals of Gerontology, 56B, P257-P267.

Carpenter, B. D., Lee, M., Ruckdeschel, K., Van Haitsma, K., & Feldman, P. H. (2005). Adult children as informants about parent's psychosocial preferences. Manuscript under review.

Cicirelli, V. G. (1992). Family caregiving: Autonomous and paternalistic decision making. Newbury Park, CA: Sage.

Lieberman, M. A., & Fisher, L. (1999). The effects of family conflict resolution and decision making on the provision of help for an elder with Alzheimer's disease. The Gerontologist, 39, 159-166.

Minuchin, S. (1974). Families and family therapy. Cambridge, MA: Harvard University Press.

Mitrani, V. B., Feaster, D. J., McCabe, B. E., Czaja, S. J., & Szapocznik, J. (2005). Adapting the Structural Family Systems Rating to assess the patterns of interaction in families of dementia caregivers. The Gerontologist, 45, 445-455.

Muncie, H. L., Magaziner, J., Hebel, R., & Warren, J. W. (1997). Proxies' decisions about clinical research participation for their charges. Journal of the American Geriatrics Society, 45, 929-933.

Pecchioni, L. L. (2001). Implicit decision-making in family caregiving. Journal of Social & Personal Relationships, 18, 219-237

President's Commission for the Study of Ethical Problems in Medicine and Biomedical and Behavioral Research. (1983). Final Report on Studies of the Ethical and Legal Problems in Medicine and Biomedical and Behavioral Research. Washington, DC: U.S. Government Printing Office.

Quill, T. E. (2005). Terri Schiavo - A tragedy compounded. The New England Journal of Medicine, 352, 1680-1633.

Roberto, K. A. (1999). Making critical health care decisions for older adults: Consensus among family members. Family Relations, 48, 167-175.


This research program has benefited from the generous support of the Brookdale Foundation, the Fan F. Fox and Leslie R. Samuels Foundation, the American Psychological Foundation, the National Institute on Aging ( NIA P50 AG05681-18), the Administration on Aging (90AM2612), and the Washington University Center for Aging.

About the Author

Brian D. Carpenter earned his PhD in Clinical Psychology from Case Western Reserve University in 1997, followed by postdoctoral fellowships at the Philadelphia Geriatric Center and the University of Pennsylvania. He is currently an Assistant Professor at Washington University in St. Louis. His research uses a family systems perspective to examine how families address aging, with a particular interest in intergenerational preparedness. Other interests include family response to diagnostic disclosure in Alzheimer's disease and family involvement in end-of-life care. In 2002 he was awarded a Brookdale Foundation National Fellowship, and he is a member of both APA's Division 20 and Division 12/Section II (clinical geropsychology).