My research compares how people make decisions with normative models of the best or most rational method for making decisions. I focus primarily, but not exclusively, on decision processes that are important in the areas of health and medicine. One of my long-standing interests concerns the role of risk and time delay in decision making, and how these play out in preventive health behavior. More recently, I have become interested in decision processes in surrogate decision making. In addition, I am also engaged in a number of other research projects that explore fundamental cognitive processes underlying decision making.
Surrogate decision making. How is making a decision on
behalf of someone else similar to or different from making a decision for
oneself? Medical decisions are frequently made on behalf of someone
else. For example, a family member might decide on end of life medical
treatment for a loved one after consulting with that person’s advance directive
(or living will). With support from an NSF grant, I have recently
been pursuing several research directions in surrogate decision making.
One research question concerns factors that influence the accuracy
with which one person (the surrogate) can predict the preferences of another
person (the principal), and the circumstances under which written communication
from the principal (e.g, an advance directive) improves the surrogate’s
accuracy. In one such study, the parents of Rutgers students wrote
advance directives. The students then had to predict their parents’
preferences in a number of medical scenarios with or without the benefit
of seeing the parent’s advance directive. Preliminary results indicate
that having the advance directive does not increase accuracy except in
the situation where the principal and surrogate are strangers (rather than
parent and child) and the two parties score high on, respectively, ability
to write and interpret advance directives.
Another research question asks whether end of life preferences are subject to framing effects. As her honors thesis, Laura Kressel collaborated with me in a project that compared different formats of advance directive questions. For example, some participants were presented with a list of life sustaining treatments and asked which treatments they would want withdrawn in particular situations, while other participants were asked which treatments they would want provided in those situations. Normatively, the two questions should elicit the same preferences, since a treatment that one wants withdrawn is a treatment that one does not want provided. The advance directive format did make a difference, however, with the “withdraw” version leading to a stronger preference to receive life sustaining treatment than the “provide” version (Kressel & Chapman, in press). Three experiments with undergraduates have demonstrating this framing effect, and a replication with geriatric patients is now underway.
Deciding on behalf of someone else is in some ways parallel to deciding on behalf of one’s future self. Thus, another research question investigates parallels between affective forecasting (predicting one’s own future preferences) and predicting the preferences of another person. In one study with graduate student Jeffrey Vietri, students predicted how often they would use gifts they received at the holidays. Four months later they reported how often they had actually used them. These personal forecasts will be compared to forecasts made by other people who do not know the students. We predict that the stranger surrogates will predict the students’ gift-use behavior more accurately than the students themselves because the surrogates take an outside view that avoids certain judgment biases.
Time and risk in decision making. A major research focus for me in the past decade has been the role of time and risk in decision making. That is, what decision processes are employed in intertemporal choice (decisions between outcomes that occur at different points in time) and risky choice (decisions about uncertain outcomes)? This area has theoretical and practical importance because many decisions have consequences that are uncertain and delayed in time (e.g., investment, health promotion, education). My research has explored several questions, such as why the content domain of decisions (e.g., health vs. money) influences time preferences and why biases or inconsistencies in risk preferences and time preferences occur.
A number of studies examined biases in intertemporal choice. For example, decision makers’ willingness to wait for delayed outcomes is influenced by the length of the delay, the magnitude of the outcomes, and whether the outcomes are gains or losses (Chapman & Elstein, 1995; Chapman, 1996 JEP; Chapman & Winquist, 1998; see Chapman, 1998 PLM and Chapman, 2003 for reviews). All of these biases contradict normative economic theory which says that delayed outcomes should be discounted at a constant rate. My research uncovered an additional way in which decision makers deviate from normative theory; specifically, they use discount rates for monetary outcomes that differ from the rates applied to health outcomes (Chapman, 1996 JEP; Chapman & Elstein, 1995). Effects of decision domain are important to identify because much of previous decision making research has examined only one domain, with the assumption that findings would generalize to other domains. Several studies explored why decision makers apply different discount rates to health and money decisions. One possible account, which was not supported by the data, involved differential familiarity with health and money (Chapman, Nelson, & Hier, 1999). Another account, which was supported by the data, has to do with whether health and money are perceived as tradable (Chapman, 2002).
Another difference between health and money time preferences was explored in a set of experiments on preferences for sequences (Chapman, 1996 OBHDP, Chapman, 2000 JBDM). Interestingly, decision makers sometimes show different sequence preferences for health and money such that, assuming the total amount of health or money was held constant, they prefer monetary income that improves over time but overall health that declines. That is, they want most of the health up front but most of the money at the end. Some additional work investigated time preferences for outcomes for future generations (Chapman, 2001) and found that decision makers treat inter- and intra-generational trade-offs quite similarly.
More recent research, with former graduate student Bethany Weber, examined mechanisms underlying some biases in risky decision making such as the peanuts effect, or the tendency for decision makers to be more willing to take risks for small stakes (Weber & Chapman, 2005b). Our findings indicate that the peanuts effect occurs because decision makers experience disappointment when they miss out on large outcomes but not when they miss out on small outcome. Another related stream of research examined parallels between risky decision making and intertemporal choice. For example, the special status given to immediate outcomes is in some sense equivalent to the special status given to certain outcomes in that adding delay “undoes” the special preference given to certainty and adding uncertainty removes the special preference given to immediacy (Weber & Chapman, 2005a). In addition, certain biases in intertemporal choice parallel other biases in risky choice in the sense that individuals who show a large bias in one domain are likely to show a large bias in the other domain as well (Chapman & Weber, in press).
Preventive health behavior. Many examples of “imprudent” health decision making appear to involve intertemporal or risky choice. For example, a failure to engage in preventive health behavior such as exercising, eating a healthy diet, brushing teeth, or taking medication may be interpreted as insufficient weight given to delayed outcomes or inattention to risks. Each of these actions involves a cost or inconvenience that starts immediately, and an uncertain benefit that accrues only after a delay (Chapman, 2005).
In one series of studies I examined the relationship between time preference (willingness to wait for delayed rewards) and adherence to preventive health behaviors, including taking hypertension medication among elderly participants, adherence with cholesterol-lowering medication among patients from a clinical practice, and getting vaccinated against influenza. These studies found no or only a small relationship between time preferences and health behavior (Chapman, Brewer, Coups et al., 2001; Chapman & Coups, 1999 MDM; 1999 PM). These studies did, however, find other psychological predictors of adherence (Brewer, Chapman, Brownlee, & E. Leventhal, 2002; Chapman & Coups, 1999 MDM; 1999 PM).
I conducted a 5-year longitudinal study on decision making about the flu shot, funded by a Rutgers Busch Biomedical grant and AHRQ. This study provided results on time preference and health behavior (Chapman, Brewer, Coups et al., 2001). Former graduate student Elliot Coups and I (Coups & Chapman, 2002) also examined how lay people integrate information to assess the relationship getting the flu shot and contracting the flu. Another analysis examined the role of emotions in health decisions, finding that the emotions of worry and regret mediated the relationship between perceived risk of getting the flu and the decision to get vaccinated (Chapman & Coups, in press). Another analysis explored reasons why healthcare workers are more likely to get vaccinated than are non-clinicians (Capolongo, DiBonaventura, & Chapman, in press). An additional analysis explored moderators of the relationship between intending to get vaccinated and actually getting vaccinated (DiBonaventura & Chapman, provisionally accepted). People who have the same vaccination intention from year to year are more likely to behave as they intended. Finally, some other analyses explore the role of social norms and barriers in vaccination decisions (DiBonaventura & Chapman, under review; in preparation).
A related intervention study examined helmet use among bicyclists. In collaboration with Fern Goodhart, Director of Health Education, and graduate student Eric Hekler, I measured attitudes toward helmets and behavior at two time points. In between the two assessments, participants were randomly assigned to one of several intervention conditions such as persuasive messages or a free bike helmet. The interventions, however, had no effect on helmet use.
Additional Research Projects. In addition to the study of time preferences, I am engaged in several research projects on decision making that are described here very briefly.