Running Head: Subjective Well-Being

Draft: February 28, 2000

Department of Psychology

University of Illinois

603 E. Daniel Street

Champaign, IL 61820

U.S.A.

Email: ediener@s.psych.uiuc.edu

Abstract

Early research on subjective well-being (SWB) focused first on describing and comparing the happiness of people in various categories, especially along demographic dimensions such as age, sex, income, education, and age. In Stage 2 of SWB research, theoretical models are explored, and there is an emphasis on variables such as temperament, goal fulfillment, adaptation, and social comparison. Research in Stage 2 also includes improvement and validation of the SWB measures. Current research at the cutting edge, Stage 3, consists of exploring theoretical accounts of SWB that explicitly recognize the varying psychological processes that affect different measures, exploring theoretical models in terms of multiple measures and longitudinal designs.

New Directions in Subjective Well-Being Research:

The Cutting Edge

Subjective well-being (SWB) represents people=s evaluations of their lives, and includes happiness, pleasant emotions, life satisfaction, and a relative absence of unpleasant moods and emotions. In other words, we gauge a person=s evaluation of his or her life by his cognitive and emotional reactions. Subjective well-being is increasingly important in a democratic world in which we want people to live fulfilling lives as evaluated by themselves, not simply as judged by policy makers, autocrats, or experts. As people in the world come to meet their basic biological needs, they become increasingly concerned with happiness and fulfillment. Thus, it is not surprising that in an international survey of college students we found that life satisfaction and happiness were rated as extremely important (Diener, 2000). For example, on a 7-point importance scale, on which 1 was Anot at all important,@ and 7 was Aextraordinarily important,@ respondents in India on average rated life satisfaction a 5.75 and happiness a 5.97. In comparison, money was rated only 4.81, indicating that the Indian respondents believe that happiness and satisfaction are more important than money.

During the past decades, psychology has been a discipline that treats problems such as depression, anxiety, and other maladies. Psychologists have not been very concerned with the positive B with helping people be happier, more fulfilled, more altruistic, and so forth. Thus, a new direction in which psychologists can expand their activities is in helping people live more rewarding lives. Although the image of the good life is bound to vary somewhat across cultures, it will probably always to some degree include close relationships, responsibilities to one=s community, and enjoyment of one=s life. The positive psychology movement started in the U.S.A. by Martin E. P. Seligman (Seligman and Csikszentihalyi, 2000) is devoted to increasing the scope of psychological research and practice so that it focuses not just on problem alleviation, but also on helping people to enhance their lives, for example by experiencing greater SWB. The idea here is that we can aid people in having happier and more rewarding marriages and families, work and recreation, and friendships, and not just strive to solve problems.

In western nations such as the U.S.A., high levels of positive emotions appear to be related to many good outcomes B for instance, better health, less divorce, less job turnover, greater work productivity, and more responsible behavior at work. We do not know whether these findings will generalize to other less westernized nations such as India. However, it is important for us to find out if they do. If Indians also benefit from positive emotions, it will be the challenge of psychologist practitioners (clinicians, personnel and organizational psychologists, and so forth) to help people and organizations to move in positive directions and lead happier lives. An important element in developing positive psychology within India is to accumulate indigenous research findings. Given the many benefits of a happy life and workforce, it is possible that organizations will be willing to hire psychologists to institute positive interventions.

SWB research has progressed through two stages in the last 50 years, and is now entering the third and most sophisticated phase. By reviewing these stages, we hope to give readers an idea of what research is needed at each level, and how they might, if resources are available, conduct research of the most sophisticated type.

In the earliest studies in this field, researchers obtained simple happiness and satisfaction measures from various groups of respondents and then described the average levels of SWB of these groups. This work was descriptive, and did not shed light on the psychological processes that control SWB. The happiness level of various groups was simply described as an average score on a simple one-item measure, and sometimes hypotheses were advanced after the study for why groups scored as they did.

In 1967 Wilson wrote the first broad review of the area of SWB, and summarized the descriptive research up to that point. He wrote that the happy person is a "young, healthy, well-educated, well-paid, extroverted, optimistic, worry-free, religious, married person with high self-esteem, job morale, modest aspirations, of either sex and of a wide range of intelligence." (p. 294). Similarly, Pavot and Diener (1993) listed the life satisfaction scores of various groups such as nuns, mental patients, prisoners, students, and senior citizens. In addition, Stage 1 research was based on simple satisfaction and happiness measures with unknown validity and reliability. For this reason, Wilson referred to his conclusions in terms of avowed happiness rather than actual subjective well-being.

In terms of current conclusions about differences in demographic groups on SWB, the reader is referred to: Diener and Oishi (2000) and Diener and Biswas-Diener (2000) for income, Diener and Suh (1997) for age, Myers and Diener (1995) for religion, Diener, Suh, Lucas, and Smith (1999) for marriage, and Diener and Lucas (1999) for temperament. Several of Wilson=s conclusions have been overthrown by more recent research; for example, youth and education are no longer seen as prerequisites for SWB. It appears that on average men and women do not differ strongly in levels of SWB, and the same can be said of adults in different age groups. In most studies married people report being happier, and people with higher incomes are also reliably higher in SWB (although this advantage is often small). In the western nations, unemployed people are often quite unhappy. Unfortunately, most of the conclusions of Stage 1 descriptive research come from wealthy and highly westernized societies, and we therefore do not have a firm idea of which of these conclusions will generalize to less economically developed nations such as India.

One of the broad conclusions from much of the Stage 1 research is that demographic variables are not very powerful in explaining the variance in SWB. For example, Campbell, Converse, and Rodgers (1976) discovered that demographic factors (e.g., age, sex, education, marital status) accounted for less than 20% of the variance in SWB. Andrews and Withey (1976) could only predict 8% of the variability in life satisfaction using demographic variables; and Argyle (1999) concluded that demographic variables could account for only 15% of the variance in SWB. Thus, in Diener=s broad 1984 overview of the field, a call was made for more development and testing of theories of well-being, which might shed greater light on who has high SWB, and offer deeper insights than the demographic categories were able to afford.

Stage 2 Research: Theory Building

During the next stage in SWB research, ushered in by Diener=s 1984 call for more psychological and theoretical research in this field, investigators carefully examined several conceptual models for explaining SWB. One theoretical approach emphasized temperament and personality as important underpinnings of whether people are happy. For example, Costa and McCrae (1980) proposed that two major personality traits, extraversion and neuroticism, underlie people=s propensity to react positively or negatively, respectively, to events. Confirming this hypothesis, Lucas, Diener, Grob, Suh, and Shao (in press) found that across cultures there is a tendency for extraverts to report more positive emotions. Similarly, Tellegen, Lykken, Bouchard, Wilcox, Segal, and Rich (1988) concluded that genes account for 40% of the variance in positive emotionality and 55% of the variance in negative emotionality, whereas shared family environment accounts for 22% and 2% of the variance in positive emotionality and negative emotionality, respectively. Thus, a number of investigators have concluded that inborn temperament is a very important influence on people=s long-term level of SWB, although immediate events will move respondents up or down from their baseline. However, Inglehart and Klingemann (2000) demonstrated that there are very large differences between nations in SWB, and therefore heritability is not the only influence on happiness. Similarly, longitudinal studies on events such as unemployment show that some circumstances can have long-lasting effects on SWB.

Another important theoretical idea is Aadaptation@ (see Frederick and Loewenstein, 1999, for a review) -- the idea that over time people habituate to both good and bad events so that these circumstances no longer influence SWB. Brickman and Campbell (1971) first proposed the idea of a Ahedonic treadmill@ B the proposition that people would never be able to remain happy over the long-run because they would always adapt to conditions, both good and bad and thereafter return to hedonic neutrality. In support of this idea, Brickman, Coates and Janoff-Bulman (1978) found that lottery winners were not significantly happier than the comparison group (and in fact were less pleased with small, everyday pleasures), and people with severe disabilities were not as unhappy as was expected. More compelling evidence for adaptation came from Silver (1982), who found that people who were seriously disabled in an accident were at first very upset and unhappy, but over a relatively short time became happier.

However, a number of modifications have had to be made to the hedonic treadmill idea as new data were collected to test it. For example, based on the temperament data, it is clear that not all people return to the same baseline (hedonic neutrality in the original theory), and that in fact most people return to a slightly happy baseline (Diener and C. Diener, 1996). Therefore, the idea of adaptation has been combined with temperament in the Adynamic equilibrium@ model of Headey and Wearing (1992), in which it is predicted that life events will make people happy or unhappy, depending on whether the events are good or bad, but that over time these individuals will return to a baseline that is determined by their temperaments. The dynamic equilibrium model has not been fully supported by data, in that people do not fully adapt to some circumstances such as unemployment (Clark, Diener, & Georgellis, 2000) and living in extremely poor conditions (Diener, Diener, & Diener, 1995). Nevertheless, the dynamic equilibrium model appears to have a degree of validity B SWB is influenced by personality and people do to some degree adapt to both good and bad events over time.

Another line of theorizing about SWB stressed the importance of goals and values. The idea was that people have different goals and desires, and therefore what makes them happy will differ. If people make progress toward their particular goals, and act in accordance with their values, they are likely to be happy, according to goal theory. For example, Emmons (1986) found that people have more positive affect if they succeed at their particular goals, have greater negative affect if their goals are in conflict, and have greater life satisfaction if they possess important goals. Similarly, Oishi, Diener, Suh, and Lucas (1999) found that what made people happy depended on their values. For students who highly valued achievement, getting good grades was predictive of their satisfaction, whereas for those who valued conformity, family harmony was more important to their life satisfaction. Similarly, people with different personalities are most satisfied with their lives if they include activities that are concordant with their temperament. Oishi, Schimmack, & Diener (2000) found that sensation-seekers were more influenced by hedonism B their satisfaction was more dependent on the degree of physical pleasures they experienced than was the satisfaction of low sensation-seeking respondents.

Also in support of the goal approach, Diener and Fujita (1995) found that the resources that are most related to a person=s SWB are those resources that help with his or her particular goals. For example, intelligence might be related to happiness for someone aspiring to be a physician, but not for someone aspiring to be a laborer. Kasser and Ryan (1996), however, suggested that certain goals will be more beneficial to happiness than other desires, and therefore they contend that not all goals are equally helpful in obtaining happiness. For example, we found substantial evidence that an individual placing too much value on making money is toxic to happiness within the United States. A recent review by Diener, Suh, Lucas, and Smith (1999) concluded that the happy person is blessed with a positive temperament, tends to look on the bright side of things and does not ruminate excessively about bad events, is living in an economically developed society, has social confidants (intimate friends), and possesses adequate resources for making progress toward valued goals. Both similarities and differences from Wilson=s earlier conclusions can be seen here, but note that progress toward goals was mentioned in the account of Diener et al.

Other theorists proposed additional conceptual models for understanding SWB. For example, in Csikszentmihalyi=s (1997) theory of flow, engagement in interesting activities is seen as a key to a happy life. In this theory the positive interest value of an activity depends on a balance between challenge and skill. We believe that whether the activity is congruent with the person=s values is also a necessary element for obtaining long-lasting pleasure from the activity. According to Csikszentmihalyi=s model, a person should be happy if he or she is usually involved in interesting activities that present challenges that the person can meet.

Another conceptual model is based on social comparison, the idea that people will be happy if they are better off than those around them, and will be unhappy if they are worse off than people in their comparison group. However, Diener and Fujita (1997) concluded that this model is oversimplified, and that people sometimes pay little attention to social comparisons. Further, sometimes people draw courage from upward comparisons and feel empathy and sadness when they make downward comparisons. Thus, being around others who are better or worse off does not necessarily make us feel bad or good, respectively. Furthermore, other standards such as goals may be more salient to people much of the time, and the effects of social comparison might be most powerful when they influence a person=s goals.

Yet another model of SWB is the idea that a person will be happy if she is improving on various dimensions, but will be unhappy if she is declining. For example, according to this approach a person would be satisfied with her grades if they were going up, but would be dissatisfied if they were lower than in the past. Thus, even a person with excellent grades might be dissatisfied if they were somewhat lower than in the past. However, Diener and Biswas-Diener (2000) concluded that this theory is not reliably supported in the case of income. People whose incomes are increasing are not invariably more satisfied, although there is a greater tendency for people whose incomes are decreasing to be dissatisfied. Probably whether increases or decreases cause satisfaction or dissatisfaction depends on a person=s goals. If he or she decides that money is no longer important, then a decrease in income may have no effect.

In Evaluation Theory (2000), Diener and Lucas proposed that a person=s SWB can be influenced by several judgment standards. Which standards are most relevant will depend in part on a person=s temperament, culture, and values. Highly relevant standards are likely to be chronically salient to the person, and therefore to influence his or her SWB much of the time. However, situational variables can intervene to make a particular standard, for instance social comparison, salient at a particular moment. Thus, even standards that are not normally salient to the person can influence his or her judgments of happiness at particular moments in time, depending on whether the situation draws attention to them. For instances, we may not normally think of our physical mobility when we judge our life satisfaction. But Schwarz and Strack (1999) found that when a person in a wheelchair was in view during the satisfaction survey, people were more satisfied with their lives than when a person with physical disability was not present. Thus, the situation of seeing a person with a disability made salient a health well-being factor that often might not be salient in the young subjects= judgments of their satisfaction. Other standards may be more chronically salient and not depend so heavily on the situation. For instance, according to Diener and Lucas, goals are usually related to people=s SWB because goals represent a standard that is usually very salient to the person (because it is what he or she is working and planning to obtain) and therefore is likely to be a standard that is used much of the time.

One of the areas of recent progress in SWB is understanding the cultural dimensions that moderate the variables that influence SWB. Much of the cultural research has contrasted individualistic societies (in which the individual and his or her unique attributes are seen as paramount) with collectivistic societies, (in which people are viewed as embedded in a social matrix that defines who they are) (Oishi, 2000). For example, Diener and M. Diener (1995) found that self-esteem was a much stronger predictor of life satisfaction for women in the U.S.A. than it was for women in India. Similarly, Suh (1999) found that the importance of congruence, acting consistently across situations and in accord with the self, was a better predictor of life satisfaction in the U.S.A. than in Korea. Suh, Diener, Oishi, and Triandis (1998) discovered that individualists use their Aaffect balance,@ the degree to which they experience pleasant emotions more than unpleasant ones, as important information in judging their life satisfaction, whereas collectivists tend to give heavier weight to norms saying how good it is to be satisfied. Finally, Oishi, Diener, Lucas, and Suh (1999) found that satisfaction with the self, freedom, and recreation were more important in predicting the life satisfaction of individualists than of collectivists. Thus, it appears that people who are socialized within an individualistic culture are more likely to think of their own worth, their own feelings, and their own autonomy when judging their life satisfaction. In contrast, people in a collectivistic culture are more likely to consider whether it is appropriate for them to be satisfied, and might consider the well-being of their family more when deciding if they are satisfied.

Individualism versus collectivism is not the sole societal distinction that has received research attention in relation to satisfaction. Diener and M. Diener (1995) found that financial satisfaction was a stronger predictor of life satisfaction in poorer nations than in wealthier nations. We have also found that in Latin countries versus Pacific Rim Asian nations, both of which are collectivistic, there is a distinct difference in norms about pleasant and unpleasant emotions. In the Pacific Rim Confucian cultures, positive and negative emotions are seen as almost equal in desirability, whereas in Latin cultures (e.g., Spain, Puerto Rico, and South America) the positive emotions are viewed as being very desirable, and the unpleasant emotions are perceived as being very undesirable. Thus, it is not surprising that Latin nations report higher levels of happiness, controlling for wealth. It appears that in some cultures there is a positivity disposition B people are socialized to look on the bright side of things B which can lead them to weight good things in their lives and give relatively little weight to bad things when making life satisfaction judgments.

Measuring SWB. In his 1984 Psychological Bulletin review article, Diener emphasized the need for stronger measures of SWB, and this type of advance has also characterized Stage 2 research. For one thing, multi-item measures of SWB were developed because they afford greater reliability than single-item measures. For example, Watson, Clark, and Tellegen (1988) developed the PANAS to measure positive and negative emotions, and we developed the five-item Satisfaction with Life Scale in our laboratory (see Pavot and Diener, 1993, for a review).

The modern self-report scales possess adequate psychometric properties. For example, they show convergent validity with daily mood reports and informant reports (Sandvik, Diener, and Seidlitz, 1993), with spouse reports (Costa and McCrae, 1988), and with recall for positive versus negative life events (e.g., Seidlitz, Wyer, and Diener, 1997). People who score high on global life satisfaction are less likely to attempt suicide (Moum, 1996) and to become depressed in the future (e.g., Lewinsohn, Redner, and Seeley, 1990). The scales show, as predicted, moderate stability and appropriate sensitivity to changing life circumstances (Eid and Diener, 1998; Headey and Wearing, 1992). Furthermore, response artifacts often do not represent serious problems with the SWB scales (e.g., Diener, Lucas, Grob Suh, and Shao, in press). Finally, SWB measures of various concepts such as life satisfaction and negative affect show discriminant validity from each other, as well as from related constructs such as optimism and self-esteem (Lucas, Diener, and Suh,1996).

In addition to the multi-item survey questionnaires, other types of measures have been used to assess subjective well-being. However, these other approaches have mostly been used to validate the self-report measures, not as direct assessment devices. For example, Sandvik, Diener, and Seidlitz (1993) found that self-report scales correlate with smiling, ratings of an interview made by an expert, and other types of measures. In most studies, however, researchers have not used multiple types of measures to compare groups or to test theories. There are a few exceptions. Balatsky and Diener (1993) used both a self-report satisfaction measure and memory for good versus bad life events to assess the SWB of Americans and Russians, and found that the latter group scored very low on both types of measures. Another study that used multiple types of measures to assess subjective well-being was conducted by Oettingen and Seligman (1990). These researchers found that West Germans compared to East Germans had higher SWB as measured by the amount of smiling in the two areas of the city, as well as by the positivity of newspaper accounts of events. Thus, the investigators found that two extremely different measures led to the same conclusions, thus enhancing our confidence in their conclusions.

So far the use of multiple measures of SWB are all too rare. One advantage of multiple measures is that when they converge they will give much greater certainty to the results. For example, many response artifacts such as number use and impression management can be ruled out when some of the other types of SWB measures are included in the study. Even when various types of measures diverge, however, they can help us understand the underlying processes affecting SWB.

An example of Stage 2 research is a project we are conducting in Orissa and West Bengal. We are collecting from our respondents several types of SWB scores. Not only do we have the traditional measures such as the Satisfaction with Life Scale, but we also have experience sampling in which we ask people to record their moods at random moments five times a day for one week. We also have memory measures in which people are asked to recall in a short timed period as many good events as they can from their past week, and then as many bad events as they can in a separate timed period. In addition, we later ask people to recall how they felt during the experience sampling week, in order to determine whether they recall being happier or unhappier than they reported on-line at the random moments. Our study also asks respondents how often they feel various emotions that were not taken from western lists, but which were generated in India. Finally, we also have included informant reports in which we ask friends and family how happy the target subject is. Unfortunately, we do not have any physiological measures such as salivary cortisol included in the project. The goal of the research is to determine how the various measures converge, and see whether the conclusions we derive from the research will be similar across the assessment approaches.

Stage 3 research can include several elements: longitudinal designs to better determine causal pathways, measures of underlying processes, experimental manipulations, and multiple measures. Research by Robert Emmons at the University of California at Davis exemplifies the new use of experimentation in this field. Not only can formal experiments shed light on causal influences and underlying processes, but they can also give us possible interventions that will increase SWB. Emmons assigns people randomly to experimental and control groups. In the experimental group, participants are asked to each day think of the things that they are grateful for, and he finds that this group=s SWB is heightened compared to those who think about more neutral topics. He is now pursuing this initial finding with additional studies. Intervention studies are still very rare, but we must do more of them in the decade to come. In the paragraphs to follow, we describe the measurement of SWB in the most sophisticated stage of research.

In Stage 2 research the desirability of having multiple measures is that the findings will be more certain because they replicate across types of measures. Because each method of measurement is likely to have different strengths and weaknesses, results that replicate across several measures are seen as more definitive. For instance, if we find that wealthier people are only slightly happier than poor people, using global self-report measures of happiness, a number of questions arise. Perhaps poor people only say that they are happy, but are not really happy. Perhaps wealthy people are happier, but use the scale differently and calibrate their answers to reflect the amount of happiness they believe they should experience. However, if we obtain additional types of measures B for instance, physiological measures (e.g., cortisol, reflecting stress), experience sampling measures (reflecting the percentage of random moments when people report being happy), memory measures (how many positive versus negative events the respondent can quickly recall in a timed period), ratings of smiling (how frequently the person smiles when describing her or his life), and surveys of Ainformants@ (family and friends report on whether the person is happy) B we will have greater faith in the conclusions. If we reached identical conclusions about the SWB of wealthy and poor people using these various measures, we would have much more confidence in the conclusions than if only global self-report surveys were used. Thus, the use of multiple types of measures is very desirable when it is feasible to include them.

At Stage 3 researchers take the use of multiple types of measures one step farther, trying to infer the psychological processes influencing SWB by the patterning of scores on the measures. No longer are investigators content to hope for convergence between the SWB assessment devices, but here they seek to examine differences between scores on the measures for the possible light these may shed on conceptual variables. The basic idea is that to scientifically understand a phenomena, we must have a theory that incorporates the measures. Take temperature as an example. There were crude measures of temperature even before a satisfactory theory of temperature existed. However, once temperature was seen as the internal energy level of atoms and molecules in an object, and heat was seen as the passage of energy from one object to another, it could be conceived how various processes influenced this, as well as how various types of measurement devices reflect it. For example, we can better understand various thermal measures based on the expansion of materials when they are heated if we understand how heat moves between objects, how the measurement medium expands with heat, and so forth. Furthermore, divergences between various types of temperature measures can be understood within the broad framework of the theory. Similarly, we need a theory of SWB that includes measurement as an integral aspect. The first beginnings in this direction are evident in research where divergence between measures can be given a theoretical explanation.

In our laboratory we have used satisfaction with narrower versus broader variables to assess SWB in different ways. These two types of measures are thought to reflect, respectively, a larger bottom-up influence or a larger top-down (personality) influence. For instance, we ask respondents about satisfaction with their education, but also with satisfaction with components of their education such as their textbooks, lectures, grades, and professors. Rather than just examine the correlates of satisfaction with these various domains, we also examine the factors that predict one type of measure more strongly than the other. In this case, life satisfaction is predicted by the broader domain measures than by the narrower domain measures. In addition, the difference between narrower and broader domains predicts life satisfaction, indicating that the global domains are influenced more by a positivity disposition, just as is life satisfaction (Diener, Napa-Scollon, Oishi, Dzokoto, and Suh, 2000). Thus, not only can we determine with which variables both broad and narrow domain satisfactions correlate, but we can also gain insights into the processes leading to them by examining the differences between them.

We have analyzed different types of satisfaction measures in yet another way that gives us theoretical insights. We examined the relation between the best life domain and life satisfaction, and between the worst domain and life satisfaction (Diener, Oishi, Lucas, and Suh, 2000). In this case, the two types of measures are global life satisfaction and domain satisfaction ratings. We found that happy people were more likely to weight good domains in judging their life satisfaction, and to give relatively less weight to their worst domain. In contrast, unhappy people were more likely to give greater weight to their worst domain. Again, the value of analyzing two types of SWB measures in reference to one another is evident.

Another way that we have examined divergences between types of measures of SWB is in analyzing the discrepancy between on-line reports of emotions at the moment (experience sampling in which respondents record their immediate moods at random moments when an alarm sounds), and retrospective recall of those moods. Again, we are trying to gain leverage in understanding SWB by comparing the two types of measures, not simply use two measures to achieve convergent validation of our results. We find that European-American students and Asian-American students are on average similar in their moods when their affect is sampled at random moments over time. When the two groups of participants are asked to later recall their moods, however, the European-Americans report being happier than they were, and happier than the Asian-Americans recall being. The explanation seems to be that because the European-Americans value positive emotions more than do the Asian-Americans, they are more likely to have biased recall in a positive direction.

Of course we need greater confirmation of this interpretation. But note that the discrepancy between the types of measures, global and on-line, leads to a theoretical interpretation of whether two ethnic groups in the U.S.A. differ in happiness. If on-line experience is used as the definition of happiness, the two groups appear to be similar. If retrospective recalled happiness is used, however, then European Americans would be considered to be happier. Because Asian-Americans show less positivity disposition, they tend to recall their moods more accurately than European Americans. Thus, discrepancies in the measures give us insights into the types of SWB experienced by the two groups.

A similar set of findings are reported by Robinson (Robinson and Johnson, 1997; Robinson, Johnson, and Shields, 1998) on the question of whether women are more emotional than men. Robinson and his colleagues found that in terms of on-line reports, women and men showed approximately the same levels of emotions. However, in retrospective recall (especially when that recall was delayed for some period of time), women reported higher levels of emotion. The biased recall of men and women are probably due to different emotion norms for the two sexes in the U.S.A., with respondents recalling their emotions in a way concordant with sex role stereotypes.

Yet another example of discrepancies between various SWB measures was reported by Shedler, Maymann, and Manis (1993), studies in which some subjects showed very different physiological stress responses than was indicated by their verbal reports. In other words, when exposed to stress inducing stimuli, some subjects said that they were unaffected and yet at the same time showed strong physiological reactions. Perhaps these individuals were Arepressors@ who did not admit their emotional reactions to themselves. The point here is not that one type of measures is wrong and that the other is right, although that is a simple interpretation of the results. Instead, we see the difference between the two types of measures as being particularly powerful in throwing light on the processes occurring. It appears that some subjects evaluated the stressful stimuli as indeed stressful, and these subjects showed physiological reactions to this. At the same time, these reactions did not reach full conscious awareness and labeling for some of the participants. By comparing the two measures, we come to understand that some individuals might deny negative emotions and therefore report higher SWB, even though in some emotion channels they manifest substantial negative reactions.

In one study we obtained experience-sampling reports on emotional intensity from subjects over a period of time, and also obtained a number of reports from informants (who judged how intense they believed the emotions of the target respondents to be). Interestingly, discrepancies between the two types of measures predicted visits to the health center. Subjects who said their emotions were less intense than their friends reported were more likely to visit the health center, as were people who said their emotions were more intense than their friends and family thought. It appeared that both people who suppressed the outward display of their emotions, as well as those who inhibited their emotions from fully entering conscious awareness, were more likely to experience health difficulties (or at least to seek medical care). Obviously these very preliminary results require replication, but they do once again point to the interesting conclusions that can be drawn when SWB measures diverge.

In terms of life satisfaction reports, it appears that people selectively choose information that is salient to them at the moment, and that is congruent with their identity. In contrast, narrower and more concrete reports (e.g., what is your mood at this moment, how satisfied are you with your bicycle) are more likely to be grounded in the recall of actual experience. Thus, there are likely to be discrepancies between global life satisfaction measures and scales measuring satisfaction with more concrete things. These discrepancies are likely to reflect factors such as the person=s self-concept, their idea of what is normative or desirable, and the participants= chronically available concepts. Thus, discrepancies between the two types of measures can help us understand the person=s SWB. Rather than ask which of the two types of measures tells whether the person is truly happy, we can define happiness in a scientific way that includes both types of responses.

Conclusions

There are now a number of different ways to measure SWB, and using more than one method can substantially strengthen a study. In one study we found that Russians scored lower both on global self-reports of SWB, and also on a memory measure of SWB (how much quicker the respondent recalled good events versus bad events). In a recent study we found that people reporting high life satisfaction are able to respond more quickly in a computer task to the pairing of self terms with positive traits than to the pairing of self terms with negative traits. In yet another study in our laboratory, Allman (1990) found that people with disabilities that confined them to a wheelchair scored as high as a control group on several types of SWB measures, thus indicating that this group was not simply avowing happiness to make a positive impression on the researcher. In each of the above cases, multiple measures have successfully been used to improve the validity of conclusions by assessing SWB in several ways. However, we are also arguing that when the multiple measures do not converge, it may give special conceptual insights into psychological processes that affected the measures. Of course this sanguine conclusion must be tempered by the warning that sometimes when different SWB measures diverge, we will be unable to create a compelling reason as to why this occurred. Although we cannot guarantee that in every case the use of multiple measures will eventuate in theoretical insights, in many cases when we have used them it has proved to be beneficial. Using multiple measures might be difficult or impossible in some very large-scale surveys, although even here measures of broad versus narrow domain satisfactions, and memory for good versus bad events, can probably often be included. In smaller studies, we might also include momentary experience sampling of moods and emotions, reports by family and friends, and interviews to obtain additional SWB scores.

Cutting edge research in SWB now requires a number of additional qualities. We more frequently need to use longitudinal designs in which respondents are assessed at a number of points over time. Because most previous research on SWB has been cross-sectional, it is difficult to establish the causal direction of correlations. For instance, when married people report being happier in the U.S.A., is this because marriage leads to happiness or because happy people are more likely to get married and stay married? If we conduct longitudinal research in which people are sampled over time, we are likely to gain insights into the direction of causality between marriage and happiness. For example, we might find that married people and unmarried people differed in happiness long before marriage age, and this might help us to conclude that there is a tendency for happy people to more often marry. Thus, in addition to using multiple measures, more sophisticated levels of research will usually be based on longitudinal designs. We also need to include measures of processes such as goal success in our SWB research, which will help us better to understand the SWB scores. We also more often need to include experimental manipulations in SWB studies. Finally, we need to include measures that are sensitive to the cultural context, not only measures that were created in the west. For example, in our studies in India we include measures of specific emotions generated by respondents in India, and we ask for participants= reactions to emotion episodes that are created in that society. These new methodological improvements are expensive in both time and money, but the expense will be worthwhile in terms of greater scientific understanding.

Although sophisticated Stage 2 and Stage 3 research is needed in India, in this cultural context additional Stage 1 research is also required that describes the basic levels of SWB of various groups. For example, we are currently collecting data on very poor people in Calcutta, on sex workers in Calcutta, and on farm families in West Bengal. Because of the paucity of SWB data from India, studies describing mean levels of happiness of various groups can still make a contribution to our understanding. However, more sophisticated levels of research in economically developing nations are also very desirable at this point. One research priority should be to determine whether happy people and positive emotions have salutary benefits in the Indian context B for health, longevity, marital stability, work productivity, and so forth.

We need much more SWB data from India and similar nations, but we also need more findings based on sophisticated measures and research designs. In the not-too-distant future, India is likely to lead the world in population, and it seems very desirable that we understand the subjective well-being of this large segment of humanity. Although research in western nations has often dominated psychology over the past decades, in the future we are confident that indigenous research from emerging nations such as India will bring new and exciting discoveries about subjective well-being.

References

Footnotes

We would like to extend our gratitude to the helpful collaborators in India who have made our research there possible: Ahalya Hejmadi; Ramanauj Majumdar, Indian Institute of Management, Calcutta; F. M. Sahoo and B. N. Puhan, Utkal University, Bhubaneswar, Orissa; and Sriya Majumdar, University of Calcutta.