Well-Being: Concepts and Measures
It is an article of faith that the supreme criterion by which a government can be judged is the quality of life its citizens experience, including, of course, the duration of life itself. In life expectancy and in material standards, the 20th century was remarkable. Especially in the prosperous and industrialized sectors of the world—the United States, Scandinavia, Japan, and Western Europe—the magnitude of positive change was without precedent. Demographers estimate that increases in longevity during the past 100 years equal or exceed all previous gains from the Bronze Age to the end of the 19th century.
The causes of these gains are many, including public health and sanitation, improved nutrition and drinking water, and the achievements of medicine. Widespread use of vaccination and immunization, the development of antibiotics, pharmacological treatment of chronic diseases, surgical treatment of cardiovascular and orthopedic disease—all these belong in large part to the 20th century.
Even in countries that have been most advantaged in these ways, however, the gains have not been uniform in all life sectors. Gerontologists Matilda and John Riley (1994) have called the latter stages of the 20th century a time of structural lag, in which national policies and institutions have failed to keep pace with gains in longevity and health. The nature of employment and retirement, the place of education in the life course, and the structures for providing medical care have yet to catch up with the facts of octogenarian life expectancy, decades of life after usual retirement, and vigorous capability into genuine old age. Nor is the lag only institutional; public attitudes about aging and about older people—who
they are, what they want, and what they can do—have the mark of the past about them. Lag in all these respects means that we have yet to achieve the quality of life made possible by the added years of life achieved.
Most people would agree that gains in life expectancy and material goods and services, important as they are, are not all we mean when we use terms such as well-being and quality of life. The idea of a more comprehensive set of measures—a complete and continuing index of national well-being—is not new. In the United States in 1929, then President Herbert Hoover created the President's Committee on Social Trends for the purpose of generating the necessary data and thus providing an improved basis for policy decisions. In the nature of such things, the committee's report, Recent Social Trends, appeared much later, in the midst of a great depression and the early years of the Roosevelt presidency. Forty years later, Wilbur Cohen, Secretary of Health, Education, and Welfare, submitted to President Lyndon Johnson an ambitious document with a modest title: Toward a Social Report (Cohen, 1969). This report provided some data and urged the collection of much more in seven main areas: health and illness, social mobility, physical environment, income and property, public order and safety, learning and science and art, and participation and alienation.
As these topics suggest, the 1969 report included some direct measures of well-being—health and illness, for example—and a number of economic and behavioral factors assumed to cause or enable well-being. Throughout the following decade, interest in the measurement of well-being ran high in many countries as part of a larger effort at creating “social indicators.” The social indicators movement, as it was called, appears in retrospect to have been part of a larger international concern with well-being and the role of governments in its achievement. It was a time when such phrases as “welfare state” and “great society” were terms of aspiration rather than derision. In more recent years, interest in such measures has continued in the United States (e.g., as seen in the National Survey of Midlife Development in the United States), but with less visibility and with markedly less support from the major funding agencies.
Efforts at measuring quality of life continue in many other countries as well. For example, the multidisciplinary Berlin Aging Study looks at sources of well-being in very old age (see Baltes and Mayer, 1999, for an extended discussion). Some coordination and exchange of information is being accomplished through the MAPI Research Institute in Lyon, France, which publishes the Quality of Life Newsletter. Coordination and international exchange is also provided by the International Society for Quality of Life Research (ISOQOL). During the year 2000, the first National Quality of Life Symposium was held in China, and the Seventh Annual Sym
posium on Quality of Life Evaluation took place in the United States. Both of these symposia, like most of the research reported in the MAPI newsletter, focused on quality of life among people with specific diseases. The ACROSS Project of the European Union, for example, emphasizes outcome measures in rehabilitation in six European countries.
In 1995, the World Health Organization (WHO) reported the completion of pilot work on a 100-question form (WHOQOL) based on a broader conception of health as “a state of complete physical, mental, and social well-being, not merely the absence of disease.” The aim is to assess “individuals' perceptions of their position in life in the context of the culture and value systems in which they live and in relation to their goals, expectations, standards and concerns” (Orley, 1995). WHOQOL-100 consists of six main sections: physical health, psychological health, level of independence, social relations, environment, and spirituality. Most of the questions within these sections ask for responses in terms of satisfaction or dissatisfaction on a five-point scale ranging from very satisfied to very dissatisfied. Some questions, however, ask about ability, confidence, or adequacy rather than satisfaction. The complete questionnaire, with its response scales, is available on the Internet, as well as from the WHO Division of Mental Health in Geneva, Switzerland.
In addition to these efforts that concentrate on the measurement of subjective well-being, many other survey-based datasets include some indicators of well-being, whether or not they are so labeled. Among them are measures of mortality and morbidity, measures of ability, and direct measures of life quality as experienced. Examples of ability are the ADL (Activities of Daily Living) scale; the IADL (Independent Activities of Daily Living) scale; and items intended to measure capacity to perform more strenuous activities, such as walking a half-mile or more or doing heavy housework. Direct measures of subjective well-being, as we shall see, take several forms. Unfortunately, relationships among the several conceptual levels represented by these various measures are seldom specified or fully examined.
The remainder of this chapter begins by examining various approaches to the measurement of well-being. We then summarize findings from research conducted to date on well-being among the elderly. The chapter ends with the panel's recommendations for future research in this domain.
THE MEASUREMENT OF WELL-BEING
Kahneman et al. (1999) propose five conceptual levels as relevant for research on well-being. In descending order, from molar to molecular, they are as follows:
External (“objective”) conditions (e.g., income, neighborhood, housing)
Subjective well-being (e.g., self-reports of satisfaction/dissatisfaction)
Persistent mood level (e.g., optimism/pessimism)
Immediate pleasures/pains, transient emotional states (e.g., joy, anger)
Biochemical, neural bases of behavior
The conceptual distinction that Kahneman and colleagues propose between objective (external) conditions and subjective well-being is familiar and important, although it is a methodological as well as a conceptual distinction. External conditions, such as neighborhood and housing, can be assessed by self-report (subjective) as well as by independent (“objective”) observation. The distinction between the physiological or biochemical level and the others is also clear, although population-based research on well-being has yet to incorporate the biochemical level, and biochemical researchers show little interest in linking their work to more molar and subjective indicators of well-being. This linkage, nevertheless, has great scientific potential. Perhaps least obvious and most controversial is the distinction between persistent mood level and immediate pleasures or pains. We deal with this issue later in the section on stocks and flows in well-being analysis.
Kahneman et al. are critical of research on well-being in two main respects: the fact that most researchers concentrate on a single level without examining its relationship to the others, and that most rely on self-reported data without examining their biases. Kahneman et al. point out that measures of reported well-being were relatively stable during decades when income and other economic indicators were increasing substantially, citing this as but one example of the inappropriately modest relationships between objective conditions and subjective well-being. Their explanation for this lack of relationship emphasizes “construal processes”—biases in self-reported well-being that include the effects of temporary moods or associations, differing frames of reference and comparisons with others, and social acceptability bias (the wish to present oneself in favorable terms). They also discuss the phenomenon of the hedonic treadmill, proposed by Brickman and Campbell (1971) some 30 years ago in discussing theories of adaptation, which the panel believes is a more important explanation.
The Hedonic Treadmill
Diener (2000:36) summarizes the Brickman-Campbell theory as follows:
In a classic 1971 article, Brickman and Campbell suggest that all people labor on a ‘hedonic treadmill.' As they rise in their accomplishments and possessions, their expectations also rise. Soon they habituate to the new level, and it no longer makes them happy. On the negative side, people are unhappy when they first encounter misfortune, but they soon adapt and it no longer makes them unhappy. On the basis of this reasoning, Brickman and Campbell proposed that people are destined to hedonic neutrality in the long run.
In the decades since its introduction, research on the hedonic treadmill has supported the concept in the main but added two qualifications: first, the process of hedonic adaptation may be toward a moderately positive rather than an absolutely neutral level of life satisfaction; second, after an experience of particular happiness or unhappiness, the tendency may be to return not to a universal level of satisfaction, but to a set point that reflects the temperament and personality of the individual. Recent analyses support both of these qualifications.
More than a million people responding to surveys in 45 countries have answered questions about their overall happiness or unhappiness with their lives. On a 10-point scale in which 10 represents maximum happiness, 1 maximum unhappiness, and 5 neutrality, the median response was slightly over 7 and the mean response not much lower (see Figure 7-1).
Explanations for the stability of well-being and happiness during periods of substantial improvement in objective circumstances emphasize adaptation processes and the distinction between transient moods and longer-term affective states. According to proponents of this distinction, an increase in wages, for example, creates a short-term elevation in mood but not a long-term alteration in level of well-being or happiness. Similarly, being turned down for a raise in pay creates an immediate reaction of unhappiness, often mixed with anger and resentment, but is not likely to alter significantly the individual's characteristic level of happiness or well-being. This adaptational dynamic explains the stability of happiness data in the United States over a period of 40 years during which real income more than doubled (see Figure 7-2).
While this and related evidence supports the hypothesis of the hedonic treadmill, other data clearly demonstrate the concept's limits. Cross-national comparisons (see Figure 7-3) indicate that countries high in gross national product (GNP) per capita also tend to be high in subjective well-being. Countries lowest in reported well-being were those in eastern Europe, in which many people were suffering both low income and recent severe reductions in income. Unaccustomed poverty adds to the fact of objective deprivation a sense of loss and the unavoidable comparison with an earlier and more prosperous time. In complementary
fashion, countries in which GNP has been growing rapidly, even though they are still poor by international comparison, score relatively high in subjective well-being. In China, for example, self-reported well-being is almost as high as in Japan, although the two countries are almost as far apart in GNP per capita as the scale allows.
Comparative data on self-reported happiness, a concept perhaps more unambiguously “hedonic” than well-being, show a similar pattern. Veenhoven (1991) compared self-reported happiness and income (GNP) per capita for seven parts of the world. The results indicate a clear relationship between happiness and per capita GNP, with some tendency toward curvilinearity. Africa, while low in reported happiness, is not as low as the level of GNP would predict. And Latin America scores almost
as high in reported happiness as western Europe, which has more than double the Latin American GNP. The implication is that GNP is driving this relationship, but that cultural factors also play a significant role (see Figure 7-4).
Existing Research: The State of Play
A substantial amount of research has been conducted on the various types of well-being measures. Historically, much of this research has been concerned with two of the five categories proposed by Kahneman and colleagues.
First are the measures of external objective conditions, such as personal income growth, neighborhood characteristics, housing status, lon
gevity, health, and disability. These measures were originally designed to supplement the conventional measures of societal progress reflected by developments in the economic sphere—growth rate of GNP, level of investment and saving, distribution of income, level of consumption, and the like. To the extent that these supplemental objective conditions have an impact on well-being that is not reflected in the standard economic variables, they provide a richer and broader picture of societal well-being.
The second category of well-being measures on which research has focused—subjective self-reports of satisfaction/dissatisfaction—experienced a rapid development in the 1970s and 1980s. This research encom
passes much of what is usually characterized as the subjective well-being literature. Studies in this area include not only the global satisfaction measures associated with various aspects of people's lives—such as their income; their neighborhood; and their relationships with their children, their spouse, and other family members—but also research focused on human development and mental health. These latter measures, discussed below, focus on assessment of how people feel about themselves as indicated by scales of self-assessment, personal growth, and relationships with others.
Survey Approaches to Well-Being
Surveys of well-being use one or more of three definitions of well-being: (1) satisfaction with life, (2) health and ability/disability, and (3)
composite indexes of positive functioning. This section provides examples of each of these approaches, along with an indication of their relative strengths and weaknesses.
Well-Being as Satisfaction with Life
“Now I want to ask you about your life as a whole. How satisfied are you with your life as a whole these days?” This question, from the 1976 national survey of the quality of American life (Campbell et al., 1976), is typical of those asked in many subsequent surveys. The scaling of responses, however, varies.
Campbell and colleagues used a seven-point scale, ranging from completely satisfied to completely dissatisfied. Andrews and Withey (1976) experimented with a number of different ways of scaling the global satisfaction question. They assumed that a person's assessment of life quality involves both a cognitive evaluation and some degree of positive or negative feeling. They attempted to capture both elements by presenting survey respondents with a seven-point scale anchored at one end by the word “delighted” and at the other by the word “terrible.” Their D-T scale, as it was called, has not been widely used in spite of the considerable methodological evidence in its favor. It was used, however, by Headey and Wearing (1991) in their longitudinal study of well-being in Victoria, the most densely populated state in Australia. Their index of life satisfaction was based on six items, all scaled on nine points ranging from delighted to terrible. The question about “life as a whole” was asked twice, once at the beginning and once at the end of the interview. Other questions addressed “sense of meaning or purpose,” “what you are accomplishing,” “how exciting is your life,” and “the extent to which you are succeeding and getting ahead.”
Current survey usage favors direct questions about life satisfaction and a return to the familiar five-point scale. Typical is the study of Americans Changing Lives (House, 1998). The question is worded as follows: “Now please think of your life as a whole. How satisfied are you with it? Are you completely satisfied, very satisfied, somewhat satisfied, not very satisfied, or not at all satisfied?” The wording used reflects an attempt to spread the distribution and avoid the tendency for responses to cluster toward the satisfied end of the scale.
As Kahneman et al. (1999) point out and as Campbell (1981) had noted earlier, the use of such questions assumes that the varied experiences of a person's life somehow combine to produce a global state of well-being or its lack, that a person's sense and level of well-being are relatively stable rather than a labile response to transient events, that people are able to describe their state of well-being to an interviewer, and
that they can be induced to do so. These assumptions have been challenged on two grounds. The first of these is the counterintuitive stability of well-being responses across cohorts and age groups, even during decades of substantial gains in material standard of living. Second is the sensitivity of subjective well-being responses to construal processes, that is, the cognitive and emotional factors that shape each respondent's assessment of his or her own life circumstances (Schwarz and Strack, 1999). The panel believes the latter criticism is the more valid.
Within the broad category of construal processes, Schwarz and Strack present evidence for five sources of bias in self-reported global assessments of subjective well-being. These include an assimilation effect, when some happy incident comes to mind and heightens the feeling of life satisfaction; and a contrast effect, when a happy event of the past is somehow used as a standard of comparison for present feelings and thus reduces the overall sense of well-being. Similar elevations or reductions in answering the well-being question may reflect transient moods, evoked by events of the moment. In addition, one's response about sense of well-being may be altered by silent comparison with others, a frame of reference that in most surveys is neither specified by the interviewer nor volunteered by the respondent. Finally, responses about well-being are subject to the bias of social acceptability, a tendency that varies among individuals and is itself measurable but seldom measured.
These methodological problems are serious and deserve more attention than they have received. The other criticism posed by Schwarz and Strack, which has to do with the stability of well-being levels across cohorts and age groups, can be countered more readily. The fact that reported levels of well-being did not show increases commensurate with gains in material indexes (e.g., income, automobile ownership) during the decades before 1980 does not necessarily indicate a problem in measurement. There are at least two plausible explanations. The first is the phenomenon of the hedonic treadmill described earlier—the tendency of expectations to rise (or fall) as life's objective circumstances and resources change. The second explanatory factor, a complement rather than an alternative to the hedonic treadmill, is that people are bringing other, nonmaterial criteria to bear upon the assessment of well-being. Allardt (1976) had proposed earlier that well-being depends on the satisfaction of three main categories of human needs—having, relating (loving), and being. The first of these, having, is gratified by the possession of material things, and income is an appropriate measure. The second, relating, can be met only by affectionate, supportive relationships with spouse or partner, family, and friends. Finally, the need for being or self-actualization can be met only by the actions of the individual, through the acquisition and exercise of valued traits and abilities. If one takes this view, the fact
that well-being has not closely tracked increases in income, far from being a sign of measurement inadequacy, is an indication that well-being requires more than the satisfaction of material needs and thus is not measured sufficiently by income and other aspects of material satisfaction. Moreover, as we have seen, responses to global questions about life satisfaction and happiness, while not strongly related to objective economic data within countries, do show significant relationships in the expected direction when comparisons among countries are made.
Well-Being as Health
Virtually all studies that attempt to assess well-being include one or more questions about health. Self-reported health is an independent predictor of longevity, even after the major risk factors for early mortality have been statistically controlled (see Chapter 6). 1 A direct question about overall health, as asked in the SF-36 and its shorter version, the SF-12, is typical (Ware and Sherbourne, 1992). These widely used questionnaires, before entering into a more detailed inventory of health problems, begin by asking: “In general, would you say that your health is (Excellent, Very Good, Good, Fair, Poor)?” Campbell et al. (1976), in an attempt to avoid the biases of brief minor illnesses, asked: “Of course, most people get sick now and then, but overall, how satisfied are you with your health?”
When surveys of life quality have looked more intensively at health, the questions have usually addressed health-determined limitations on activities. The SF-12 questionnaire is typical. It begins with a question about overall health, on a five-point scale ranging from excellent to poor. This is followed by questions about health as limiting the performance of specific activities, about emotional problems as interfering with daily activities, and about mood and energy level.
Some age-related tendencies are common among the numerous surveys that have compared self-reports of life quality by different age groups. Overall satisfaction with life is not greatly different among age groups; indeed there is some evidence for slight increases in satisfaction (but not happiness) in older age groups. If this apparent tendency is confirmed by long-term longitudinal studies, one might conclude that it reflects a process of adaptation, of people's coming to terms with their lives.
Within the more specific life domains, only one—satisfaction with health—declines significantly with age. Given the increasing morbidity and comorbidity that often come with increasing age, reductions in health
1It should be noted, however, that prevalence figures derived from this measure may vary considerably, both within and across societies.
satisfaction are evidence of people's actual experience and of the validity of self-reporting.
Composite Indexes of Well-Being
Various indexes of overall well-being or positive functioning have been used in different surveys. Some are simple summations or averages of domain-specific satisfaction scores. Others use weighting procedures in which responses to the direct question about overall life satisfaction are dominant, or in which satisfaction responses are combined with measures of happiness or positive and negative affect.
Critics of these approaches point out that they have in common an emphasis on gratification and a neglect of other aspects of what might be considered a “good life.” Adler (1987:52) argues that Western philosophy includes two distinct conceptions of happiness. One is gratification of an individual's desires, whatever they may be; the other is “the ancient ethical conception of happiness as a whole life well-lived because it is enriched by the cumulative possession of all the goods that a morally virtuous human being ought to desire.” The key word in this long definition is “ought”; an external standard of values is thus imposed on an individual's experience of happiness. If his or her gratification derives from sources that would not be desired by “a morally virtuous human being,” one is instructed not to count this as happiness, or at least to distinguish between two kinds of happiness on the basis of their sources, moral and amoral.
This argument has value, but the panel does not accept Adler's conclusion. The value comes from the reminder that research on satisfaction, gratification, happiness, and other such hedonic responses becomes much more meaningful when we also investigate their sources—the events, conditions, and relationships that evoke them. But the panel believes that to incorporate this kind of source information into the definition of happiness, as Adler does, is counterproductive.
A different and more constructive criticism is found in the work of Ryff (1989; 1995; see also Singer and Ryff, 1999). She argues that three rather separate lines of research and theory have been concerned with definitions of well-being: theories of human development; of clinical psychology; and of mental health, especially positive mental health. From these sources, she proposes six defining components of well-being: self-acceptance, purpose in life, environmental mastery, autonomy, personal growth, and positive relationships with others.
In the MacArthur studies of midlife (MIDUS, 2000), each of these six components was measured by three items. These items were presented to respondents as statements, and agreement or disagreement was measured
on a seven-point scale ranging from “strongly agree” to “strongly disagree.” Sample items from the scale are shown below, each identified by the dimension it is intended to measure:
“In many ways, I feel disappointed about my achievements in life.” (self-acceptance)
“Some people wander aimlessly through life, but I am not one of them.” (purpose in life)
“The demands of everyday life often get me down.” (environmental mastery)
“Maintaining close relationships has been difficult and frustrating for me.” (personal relationships with others)
“For me, life has been a continuous process of learning, changing, and growth.” (personal growth)
“I tend to be influenced by people with strong opinions.” (autonomy)
In a national sample, cross-sectional comparisons of three broad age groups (young adult, midlife, and older adult) show differing patterns for the six components of well-being. Environmental mastery and autonomy increase between young adulthood and midlife. Purpose in life and personal growth show decreases with age, especially between midlife and older age. Positive relationships with others increase steadily and significantly from the young to the oldest age group. Self-acceptance shows no significant differences with age (Ryff, 1995).
All of the survey approaches reviewed thus far measure well-being either as a present state or a retrospective judgment over an extended time period. Self-reports of this kind are valuable, but they do not reflect the variability of health, happiness, and well-being over time that all people experience in varying degrees. The moment-to-moment reports that Kahneman and colleagues obtained from people undergoing stressful procedures demonstrate short-term variability, and studies of timeuse samples (discussed later in this chapter) provide further insight into the variability experienced by individuals under the conditions of everyday life. The latter variations and the accompanying differences among individuals in their magnitude and duration imply the need for an additional criterion of well-being. This criterion might be termed resilience, and would measure the ability of an individual to deal effectively with stressful life events. Relevant conceptual dimensions would include the magnitude of deflection from the individual's characteristic set point of well-being, the duration of that reduction, and the rapidity and completeness of his or her return to the set point. These issues are of special importance for the well-being of older men and women because of their
increased exposure to bereavement and illness (comorbidity). The concept of resilience is discussed in greater detail in a later section of this chapter.
Work and Well-Being
To understand the meaning of a person's overall satisfaction with life requires breaking down the broad concept of life satisfaction and looking separately at the major life sectors—marriage, family, work, income, housing, neighborhood, community, and others. Of these, work has a special significance. It is a source of income, which in turn determines housing, neighborhood, and the many other aspects of life that are in some degree monetarized. But a person's employment also demands a significant part of his or her time and energy. For most people it is a source of friendships, and for many it provides a means of utilizing valued skills and abilities. For all these reasons, work (employment) ranks high among the determinants of overall life satisfaction. Only family life, income, and all nonwork activities in combination rank higher, and work is the major determinant of income, which in turn determines access to many nonwork activities.
Thousands of studies have been conducted on work-related subjects, including worker motivation, job stress and its effects, satisfaction and dissatisfaction with job content, working conditions, supervision, rewards, and many others. Despite differences in terminology and theoretical orientation, there is substantial consensus on the job characteristics that determine work satisfaction. They include the following:
Task content, especially variety and the opportunity to use valued skills
Autonomy, especially regarding the methods and pace of work
Supervision, especially supportive behavior by supervisors
Resources, including information, as well as appropriate tools and equipment
Coworker relations and the opportunity for informal contact and conversation
Working conditions, including physical surroundings, noise level, and security
Rewards, wages, and fringe benefits
Promotion and prospects for advancement for outstanding performance
To understand the place of work in people's lives, all these aspects should be encompassed by the inquiry. At a minimum, studies of well-
being should include two questions about a person's job—his or her satisfaction with the work, and his or her rating of the job on a scale from excellent or highly desirable to poor or very undesirable. Evidence is that the adaptation process figures prominently in people's responses on satisfaction. In thinking about their own satisfaction or dissatisfaction with their jobs, individuals take into account their education and experience, their age, and the state of the labor market (Kahn, 1981; Warr, 1999). These processes are not apparent in their responses to the short-answer, fixed-alternative questions by which satisfaction with work is typically measured. They may be revealed, however, when interviews are less structured and responses are recorded verbatim. For example, consider the following excerpt from an interview with a blue-collar worker whose first response to the question of satisfaction with work was: “I got a pretty good job.” The interview then asked: “What makes it such a good job? The answer was instructive: “Don't get me wrong. I didn't say it is a good job. It's an OK job, about as good as a guy like me can expect. The foreman leaves me alone and it pays well. But I would never call it a good job. It doesn't amount to much, but it's not bad” (Strauss, 1972:55).
Some recent innovative but controversial work on well-being explores a set of concepts involving emotional states—both persistent moods and immediate pleasures and pains. Research on emotional states as immediate pleasures and pains is the focus of a substantial effort undertaken recently by psychologists (Kahneman et al., 1999; Seligman and Czikszentmihaly, 2000). This work starts with the proposition that conventional measures of well-being derived from retrospective self-reports of satisfaction or dissatisfaction have serious biases that cannot easily be controlled. Proponents of this view assert that a more valid assessment of subjective well-being requires a model in which the immediate pleasures and pains associated with people's circumstances are monitored continuously. In this literature, the instantaneous record of pleasures and pains is called direct utility; the integral of direct utility over an episode is called total utility or objective well-being. Retrospective recollection of the global satisfaction or dissatisfaction associated with the immediate pleasures or pains of a previous time is called remembered utility, and is essentially equivalent to subjective well-being as described earlier. The basic idea behind this model is that a subject's instantaneous report of the pleasures or pains associated with a particular activity or episode is the least biased of any of the subjective well-being measures, and therefore should constitute the core data for scientific understanding of human happiness and its correlates.
Research along these lines has required experimental subjects to report successive levels of pleasure or pain during brief episodes of an acute or dramatic nature. In the most widely cited case, respondents were asked to report every 60 seconds, on a scale ranging from no pain at all to excruciating pain, the level of pain associated with a colonoscopy, an unpleasant and painful medical diagnostic procedure that lasts between 10 and 20 minutes. But if one is thinking of using instantaneous pleasure and pain as a way to measure utility, the method of measurement should be applicable to routine activities. That is, the measures should be obtained for a sampling of activities that represent what individuals typically do during the course of a day—talking to a spouse, lunching with a friend, walking through the park, commuting to work, using a word processor, and so on. Use of the instantaneous pleasure/pain response as a measure of well-being thus turns out to be operationally equivalent to asking respondents about the level of pleasure and pain associated with a random sampling of the activities on which they spent time during some recent period of days or weeks.
The notion that instantaneously recorded pleasure and pain is a valid representation of objective well-being (because it is less vulnerable to the biases associated with self-reports of global recollections of satisfaction with various aspects of life) depends critically on the premise that meaningful distinctions in levels of pleasure and pain can be measured not only for dramatic events such as colonoscopies, but also for the events that together constitute everyday life. Moreover, the notion that the summing of pleasures and pains yields a good index of well-being depends on the premise that the only source of utility from an activity is the integral of the instantaneous pleasures/pains associated with that activity.
But what if the activity is seen as an unpleasant experience that produces a desirable outcome later in life? That is, suppose the activity is cleaning the house. Most subjects regard this as a dull activity. But having spent time on producing a clean house, they will derive more enjoyment from entertaining friends than if the house were untidy. Thus the utility from the activity of cleaning house is derived in part from the instantaneous pleasures/pains associated with that activity, but may also encompass the future utility obtained when the clean house is used for social entertainment. In economic terminology, one might think of cleaning the house as an investment in a future benefit derived from using a clean and orderly house for social entertainment. The same is true for many monotonous and physically demanding jobs, whose rewards and satisfactions are often extrinsic rather than intrinsic to the activity itself.
Analysis of the instantaneous pleasures or pains associated with episodes as a way to measure well-being has a close counterpart in studies of time use, which sometimes measure the satisfaction derived from activi-
ties as well as the hours devoted to them. While almost all countries have some information on the use of time among their populations, in two cases (Sweden and the United States), information about the level of enjoyment associated with various activities has been collected along with the time-allocation data (see Juster and Stafford, 1986; Klevmarken, 1999). Such data are available for the United States in 1975-1976 and 1981-1982 and for Sweden in 1984-1991. In these studies, data on satisfaction with activities are obtained from responses to a question of the following type, asked for each of a set of daily activities: “Think about a scale from 10 to 0. If you enjoy doing an activity a great deal, rank it as a 10; if you dislike doing it a great deal, rank it as a zero; if you don't care about it one way or the other, rank it in the middle as a 5. For example, if you like it some you might rank it 6 or 7. Keep in mind that we're interested in whether you like doing something, not whether you think it's important to do. How much do you enjoy (activity 1, 2, . . .) ?”
Although the enjoyment scale is not necessarily identical to the instantaneous utility associated with the pleasures or pains of an activity, there is clearly a close correspondence. The most important difference is probably that the enjoyment scale data are a recollection based on a (typical) past event, while the instantaneous pleasure/pain data are for an event currently under way.
Interestingly enough, questions about the satisfaction derived from specific activities produce results that are apparently inconsistent with conventional economic views of utility maximization. Economists tend to think of work as an activity producing a major extrinsic reward (income) but having, especially at the margin, negative intrinsic rewards. That is, economists think of people as working up to the point at which the income produces utility that is just equal to the disutility attached to the work itself—otherwise people would do more work (if they could) than they do. Data on satisfaction with activities, according to this model, ought to show that work ranks poorly on the scale of intrinsic satisfaction or enjoyment. But in fact work is the second or third most highly rated activity, just behind playing with children and socializing with friends, and it outranks all other leisure activities except for these two. The activity associated with the least intrinsic benefits or satisfaction is clearly housework.
Stocks and Flows in Well-Being Analysis
Before examining the state of knowledge relative to the well-being measures discussed above, we may note another analytically useful distinction among the measures: whether they can be thought of as representing the degree of satisfaction with stable features of the social and
economic landscape, or as behaviors that produce well-being directly and may also do so indirectly by changing the environment. The first could be labeled a stock, the second a flow. The conceptual framework here tends to reflect the way economists think about the world, while much of the well-being literature reflects the perceptual schemes of psychologists.
The basic idea of stocks is that they represent a state of being that reflects the cumulative impact of past history on current state. With regard to financial variables, for example, total assets would represent the stock since they show the cumulative impact of lifetime saving behavior, along with any return to that saving. And if assets represent the stock, the change in assets, or current-period saving plus or minus capital gains, represents the flow.
A similar framework can be applied to noneconomic sources of well-being. For example, the current state of one's relationship with a spouse, children, or friends can be thought of as reflecting a past history of interaction that has produced a characteristic level of satisfaction with those relationships. Thus the stock of associations can be thought of as an assessment of the overall degree of satisfaction with the relationship, while the flow can be thought of as the transient emotional states (joy, anger, and the like) evoked by some specific event, interaction, or activity. The cumulative effect of such events, of course, is to change the stock. For example, discussing a family problem with one's spouse is an activity that almost certainly generates immediate affect or flow (e.g., anger, pleasure, relief, guilt) but also may, and often will, have a persisting impact on satisfaction with the marriage (stock) and perhaps other family relationships as well.
As this example suggests, a concern for measuring well-being leads to an interest in activities that affect well-being temporarily or enduringly—that is, affect both the flow and the stock. A useful way to think about the relationship between satisfaction and activities is to recognize that all human activities can be thought of as producing two kinds of satisfaction or reward. One is intrinsic satisfaction, reflecting the degree of enjoyment associated with performing the activity. The other is the extrinsic satisfaction, associated with a product or outcome of the activity. Activities typically generate both intrinsic and extrinsic satisfaction or dissatisfaction. For example, cleaning the house can be thought of as yielding the extrinsic rewards associated with a clean and orderly house, which enhance the satisfaction associated with living in the house, plus any intrinsic satisfaction (or minus any intrinsic dissatisfaction) associated with the activity.
The panel regards conceptualizing satisfaction along stock and flow dimensions as a useful way of linking the concepts of activity and satisfaction. For the most part, the measures of well-being reported in the
existing literature are stocks rather than flows; respondents are asked to sum up their satisfaction with life as a whole or with certain of its domains—neighborhood, family relationships, housing, and so on. Similarly, aspects of well-being associated with human development and mental health (e.g., self-acceptance, purpose in life, environmental mastery) can be conceptualized usefully as stocks. In some unspecified way, however, these summations are the result of the continuous flow of life experiences and the activities that create those experiences or expose the individual to them.
Moreover, the activities in which people engage during the course of the day, week, month, or year have immediate effects on flows of satisfaction or dissatisfaction (intrinsic rewards or their lack), short-term effects through their products (extrinsic satisfaction or dissatisfaction), and potential effects on the persisting level of life satisfaction or well-being (stocks). To continue an earlier example, cleaning one's house not only generates intrinsic satisfaction or dissatisfaction with the activity itself and immediate extrinsic satisfaction with the result, both of which are flows, but may also add to one's self-esteem, which is a stock.
Satisfaction effects are measurable as both stocks and flows, at least in principle. Conventional measurements of satisfaction as stocks are the standard fare of well-being analysis. Yet the immediate satisfaction associated with activities has also been measured, although less often, by eliciting subjective judgments from respondents about the level of enjoyment derived from a specific list of activities or the instantaneous pleasure/pain associated with the activities. Data on satisfaction with activities tend to be used in analyses of time use or time allocation, in which the objective is to understand the choice of activities in terms of both satisfaction with the activities themselves (the flows) and satisfaction associated with the enhancements of the stock that result from the activities.
Among the important information needed to provide an adequate basis for assessing changes in life satisfaction in older people is data on the way older people use their time, combined with data on the satisfaction obtained from those activities. A good deal of methodological work has explored the reliability and bias associated with various types of time-use measures (Juster and Stafford, 1986). It has been found that the most unbiased measures of time use consist of a 24-hour diary of events, with the optimum survey design involving the use of multiple time diaries reflecting different days of the week and seasons of the year. The major problem with using time diaries to document changes in activity patterns as people age is that the diaries are extremely expensive relative to other
survey methods. Measuring activities during a 24-hour period takes between 20 and 25 minutes of survey time, and if multiple diary days are needed to provide a good representation of activities for different days of the week, that amount of time must be multiplied by a factor of at least two and probably more.
Alternative methods of measuring patterns of time use include “stylized” questions. For example, “During the past week, about how much time did you spend doing x, y, z, where x, y, z?” represents activities such as watching television, doing the dishes, working in the garden, working for pay, and reading. The evidence indicates that these types of time-use measures will typically yield underestimates, and for some activities, very serious underestimates. It is also clear from the methodological literature that social desirability biases are substantial for some of the stylized measures of time use. For example, parents are likely to seriously overreport the amount of time they spend reading to their children when asked about that activity in a stylized mode, and similarly to underreport the amount of time their children spend watching television.
One of the attractive features of assessing changes in life satisfaction over time by analyzing time diary data is that it is a relatively simple matter to attach satisfaction measures to the activities. As noted earlier, both U.S. and Swedish time-use data provide measures of the satisfaction associated with different activities. The data on satisfaction with specific activities—called process benefits—turn out to be quite illuminating about the sources of life satisfaction and associated measures of well-being.
FINDINGS ON WELL-BEING AMONG THE ELDERLY
Subjective Well-Being and Aging
As noted earlier, most studies find that overall subjective well-being, measured as satisfaction, is about the same for different age groups, or even slightly higher among older people than among middle-aged and younger adults. Satisfaction with health, however, is an exception; consistent with data on morbidity and comorbidity, satisfaction with health is lower among older people. The finding of overall stability in life satisfaction, while based almost entirely on cross-sectional research, is consistent with theories of adaptation, which argue essentially that people come to terms with their lives as they grow older. When well-being is measured on a scale of happiness and unhappiness, however, older people tend to give fewer responses of either extreme happiness or unhappiness. This apparent dampening of affect or emotion with increasing age has been interpreted as one aspect of a more general “slowing down” and reduction in intensity of response (Campbell, 1981).
Age-related changes in patterns of activity are in sharp contrast with the relative stability of satisfaction responses, and the most conspicuous of such changes in activity is retirement from paid employment. In the United States, national surveys show that hours of paid work decline after age 50 and drop sharply after age 60; hours of unpaid productive activities, on the other hand, begin to decline later in life and persist in a significant amount beyond age 75. Volunteering does not, however, show a significant increase after retirement, even though many people report expectations of becoming more active as volunteers after retirement; participation in voluntary organizations varies as a function of health and degree of connectedness with others. Time spent with other family members shows a substantial increase as people leave paid work and as work-related social interactions tend to diminish over time. Finally, leisure activities, such as watching television, traveling, and visiting, are all likely to rise on average.
The ability to recover from negative (stressful, threatening, damaging) life events is an important attribute at any age, but acquires additional importance as people age. Even older people who describe themselves as very happy or as high on a scale of well-being are unlikely to have been in that state at all times. A few may have avoided, by design or fortuitously, the stressful life events and challenges that threaten well-being. Some may have been born with the kind of genetic endowment that confers relative immunity from the infirmities and chronic diseases common in old age. Still others, in old age, may be reaping the benefits of a lifetime of health-promoting habits of diet and exercise. All these factors, especially in combination, may enable people to go through later life in a state of good physical health, sustained cognitive function, and unvarying sense of well-being. The more common experience of aging, however, involves some number of accidents and illnesses, losses and bereavements, crises of adult children, involuntary retirement, moves from a longtime home, and the like. The high incidence of such events in later life raises the question of how successfully people deal with them, that is, how such events affect well-being in both the short term and beyond.
For research in this area, the panel proposes the concept of resilience noted earlier, defined as the ability to recover quickly and completely from such misfortunes and challenges. Holmes and Rahe (1967) developed a broad inventory of such negative events in a questionnaire form that has been widely used. Cross-sectional comparisons of people reporting different levels of stressful life events do not always show consistent
differences in well-being, nor do such comparisons reveal the speed and completeness of recovery. Only bereavement over the loss of a child or spouse has been studied more intensively and with attention to the magnitude and duration of negative impact (see, for example, Wortman et al., 1993). Findings of this research show frequent prolonged grief effects in such cases. Wortman et al. (1993) found that, on a measure of life satisfaction, widows and widowers scored significantly lower than a control group for 10 years, and that they scored higher on depression for almost 20 years. Of the potential moderators of adaptation to negative events, social support has been most studied and has been shown to have a significant buffering effect. Advance notice or warning of an impending negative event also moderates its effects (Berenbaum et al., 1999).
Resilience can be inferred from cross-sectional comparisons and retrospective accounts. Stronger, although more difficult, research designs would require longitudinal study of specific event sequences that would include assessment of well-being before the stressing challenge is encountered and subsequent monitoring to observe the initial decremental effect, the time required to regain stability of well-being, and the level of well-being regained.
7-1. Cross-national research on aging should include an overall measure of subjective well-being (life satisfaction), with comparable questions for major life domains. Cross-national comparisons should also include the experience of major life events, especially the stressful events that are increasingly likely in old age, such as illness, bereavement, retirement (complete or partial), and changes in activity patterns. The range of methods for assessing well-being has increased greatly in recent years. Scales of satisfaction-dissatisfaction and happiness-unhappiness still predominate, although the work of some researchers (e.g., Ryff and colleagues) goes beyond that approach. The debate between objective and subjective definitions of well-being persists. That debate, we believe, is mistaken: neither objective conditions nor subjective responses are sufficient; both are needed to understand the quality of life. The environment, neighborhood, and community in which people live; the work they do; the income they receive—all are essential determinants of the life they lead. But these objective measures are not sufficient to reveal whether people find their lives satisfying or dissatisfying, fulfilling or frustrating. Conversely, to know only subjective responses, with their inevitable processes of adaptation and compromise
and their limitations in perspective, is inadequate for both science and public policy.
7-2. Standardized and validated measures of subjective well-being (physical, cognitive, happiness) should be developed for continuing use in ongoing national surveys. National policies differ in the benefits and services provided to older men and women, and cultures differ in the opportunities and demands presented to people as they age. These differences do much to define the quality of life experienced by older people. Cross-national data are available on GNP and other economic indicators. But the ultimate criterion by which nations should be compared and by which governments should be judged is the quality of life they provide to their citizens and residents. Data on well-being are an essential element in assessing quality of life.
7-3. Cross-national research should include documenting, at appropriate intervals, the changes in time use and lifestyle that characterize the adult life course, especially the transition from full-time employment to retirement. Complementary to data on subjective well-being (i.e., how people feel) is the question of what they do. Especially important are data on the various forms of productive activity (paid and unpaid, formal and informal) and the satisfaction or dissatisfaction these activities generate.
7-4. Data on self-perceived health—people's sense of their own health on a scale from excellent to poor—should be obtained for all countries. Long life is a prime value in all countries, and it is therefore a criterion for assessing the relative success of nations. Moreover, countries differ greatly in life expectancy, and these differences are not entirely explained by economic factors. Research shows that an individual's rating of his or her own health is a strong predictor of mortality or survival. Such ratings are easily obtained, and they can be considered as leading indicators of things to come. They are therefore of great potential value to policy makers.
7-5. In those countries that have the necessary technical and fiscal resources, data should be collected that link specific activities (work and nonwork) to the pleasure or pain they generate. Most measures of well-being are linked to life events and activities by means of recollection; past events are then analyzed to explain present well-being or its lack. Such retrospective efforts, however carefully undertaken, are vulnerable to the biases and limitations of individual memory. Much of this vulnerability can be avoided if people are asked to rate the pleasure or pain and/or satisfaction or dissatisfaction associated with specific activities at the time those activities are being performed. The methods used for such research, already available through studies of time use in a few countries, need both wider utilization and further development.
1987 We Hold These Truths . New York : Macmillan .
1976 Dimensions of welfare in a comparative Scandinavian study. Acta Sociologica 19: 227-239 .
1976 Social Indicators of Well-Being . New York : Plenum ., and
1999 The Berlin Aging Study. Aging from 70 to 100 . Cambridge : Cambridge University Press ., and
1999 Disturbances in emotion. In Well-Being: The Foundations of Hedonic Psychology , D. Kahneman et al., eds. New York : Russell Sage ., , and
1971 Hedonic relativism and planning the good society. In Adaptation-Level Theory: A Symposium , M.H. Apley, ed., pp. 287-302 . New York : Academic Press ., and
1981 The Sense of Well-Being in America . New York : McGraw Hill .
1976 The Quality of American Life . New York : Russell Sage ., , and
1969 Toward a Social Report . Ann Arbor : University of Michigan Press .
2000 Subjective well-being: The science of happiness and a proposal for a national index. American Psychologist 55(1): 34-43 .
1991 Subjective well-being: A stocks and flows framework. In Subjective Well-Being: An Interdisciplinary Perspective , F. Strack, M. Argyle, and N. Schwarz, eds. Oxford : Pergamon Press ., and
1967 The social readjustment scale. Journal of Psychosomatic Research 11: 213-218 ., and
1998 Age, work, and well-being: Toward a broader view. In Impact of Work on Older Adults , K.W. Schaie and C. Schooler, eds. New York : Springer .
1997 Culture Shift in Advanced Industrial Society . Princeton, NJ : Princeton University Press .
Juster, F.T., and F. Stafford, eds. 1986 Time, Goods, and Well-Being . Ann Arbor, MI : Institute for Social Research .
1981 Work and Health . New York : Wiley .
Kahneman, D., E. Diener, and N. Schwarz, eds. 1999 Well-Being: The Foundations of Hedonic Psychology . New York : Russell Sage .
1999 Measuring investment in young children with time diaries. In Wealth, Work, and Health: Innovations in Measurement in the Social Sciences , J.P. Smith and R.J. Willis, eds., pp. 34-63 . Ann Arbor, MI : University of Michigan Press .
2000 Available: http://www.icpsr.umich.edu/cgi/ab.prl?file=2760 [February 28, 2001].
1996 The pursuit of happiness: New research uncovers some anti-intuitive insights into how many people are happy—and why. Scientific American 274(5): 70-72 ., and
1995 The WHOQOL Measure: Production of the WHOQOL-100 Field Trial Form. Quality of Life Newsletter 12: 3 (April) .
1994 Generational relations: A future perspective. In Aging and Generational Relations over the Life Course: A Historical and Cross-Cultural Perspective , T.K. Hareven, ed. Berlin : De Gruyter ., and
1989 Happiness is everything—or is it? Explorations on the meaning of psychological well-being. Journal of Personality and Social Psychology 57: 1069-1081 .
1995 Psychological well-being in adult life. Current Directions in Psychological Science 4: 99-104 .
1999 Reports of subjective well-being: Judgmental processes and their methodological implications. In Well-Being: The Foundations of Hedonic Psychology , D. Kahneman et al., eds., pp. 61-84 . New York : Russell Sage ., and
2000 Positive psychology: An introduction. American Psychologist 55(1): 5-14 ., and
1999 Hierarchies of life histories and associated health risks. Annals of the New York Academy of Sciences 896: 96-115 ., and
1972 Is there a blue-collar revolt against work? In Work and the Quality of Life , J. O'Toole, ed. Cambridge : Massachusetts Institute of Technology Press .
1991 Questions on happiness: Classical topics, modern answers, blind spots. In Subjective Well-Being: An Interdisciplinary Perspective , F. Strack, M. Argyle, and N. Schwarz, eds., p. 13 . Oxford : Pergamon Press .
1992 The MoS 36-item, short form health survey (SF-36): Conceptual framework and item selection. Medical Care 30: 473-483 ., and
1999 Well-being and the work place. In Well-Being: The Foundations of Hedonic Psychology , D. Kahneman et al., eds. New York : Russell Sage .
1993 The meaning of loss and adjustment to bereavement. In Handbook of Bereavement: Theory, Research, and Interventions , M. Stroebe, W. Stroebe, and R. Hansson, eds., pp. 349-366 . Cambridge : Cambridge University Press ., , and