2
Conceptualizing Experienced (or Hedonic) Well-Being
In its narrowest and most traditionally measured form, experienced well-being (ExWB) is the series of momentary affective states that occur through time. In practice, a number of measurement approaches and objectives coexist. These range from the moment-to-moment assessments of affect to instruments that require reflection by respondents about longer time periods, such as how they felt “yesterday.” At the longer intervals, well-being assessments are likely to take on the characteristics of “life evaluation” measures. Experience measures can, in a sense, be viewed as a subspectrum of the overall subjective well-being (SWB) continuum, which at one end involves a point-in-time reference period and is purely hedonic (“How do you feel at this moment?”) and at the other end involves evaluation of a comparatively very long reference period (“Taking all things together, how would you evaluate your life?”). The ExWB portion of the continuum ranges from the momentary assessments of affect (the shortest framing period) to global-day assessments or day reconstructions at the longer end.1 As the reference and recall periods lengthen, a measure takes on more and more characteristics of an evaluative well-being assessment. Specification of the reference period has a strong impact on the results of affect questions and, indeed, on what is being measured.2
________________
1 Week-long reference periods have also been used in ExWB assessments, particularly in health contexts (e.g., a respondent may be asked about pain last week).
2 Another consideration for evaluating the associations among the types of SWB is that there may be a confounding of construct and measurement technique. One feature of how some SWB assessments instruct respondents is to ask them to answer for a particular period of time, say, about the past month, the past week, the past day, or about the current moment.
ExWB is defined by people’s emotional states but may also include sensations such as pain or arousal, ruminations, a sense of purpose or meaning, or other factors. Hedonic well-being typically is used in association with the narrower, emotional (or affect) component of ExWB. For this reason, the term “hedonic well-being” is—in this report—replaced with “experienced well-being” to convey this slightly broader construct.
ExWB is a term with a very close connection with the much older and extensive field of mood and emotions. A reasonable argument can be made that the terms hedonic well-being and emotions are synonymous; and sometimes hedonic well-being is called “emotional well-being” (see, for example, Zou et al., unpublished). The fact that they incorporate similar partitions of positive and negative aspects further confirms their similarities. Emotions can be fleeting states that vary from minute to minute; however, when emotions are aggregated over longer periods of time, they become more stable and reliable measures that may better fulfill the needs of wellbeing researchers. Historically, the “standard” period studied for hedonic well-being or ExWB analysis has been a single day. The initial thinking behind this was that 24 hours was a period that provided some stability and could be assessed without too much concern about recall biases; the panel discusses the implications of these assumptions later, along with the alternatives.
2.1 DISTINCTIVENESS OF EXPERIENCED AND EVALUATIVE WELL-BEING
An important consideration for determining the value of ExWB data and statistics—for research, policy, and general information purposes—is its distinctiveness from measures of evaluative well-being. One might expect people with high levels of overall SWB to report, in most cases, relatively high levels for both its evaluative and experienced dimensions.3 Very high associations of ExWB with evaluative measures would mitigate the case for regularly including both types of measures in data collections. The goal
________________
This is known as the reporting period. The problem arises when a hedonic construct such as happiness, which can fluctuate throughout a day, is assessed with a long reporting period, say, “over the past week.” Long reporting periods are associated with a shift from an immediate recall of emotions during recent experience to respondents’ overall perception of their emotion (Robinson and Clore, 2002). Thus, hedonic SWB measures that use longer reporting periods can start to look more like evaluative well-being measures, creating a confounding effect.
3 A fairly extensive literature exists on the relationship between evaluative well-being and ExWB. As just one example, Zou et al. (unpublished) found life satisfaction and emotional well-being (their ExWB construct) distinct, though with significant overlap when assessed by multiple indicators.
of this section is to go beyond an intuitive impression of the associations among types of SWB by examining the empirical evidence.
ExWB measures are designed to capture emotions as they fluctuate from moment to moment and in response to day-to-day events and activities. They therefore aim to be reactive to a respondent’s immediate focus. For example, for individuals at work, their reported affect is likely related to the immediate task at hand and not to broader issues such as the state of their marriage or their financial circumstances—topics that typically fall more squarely into the evaluative well-being domain. Issues that are only infrequently on a respondent’s mind at any particular time during the course of the day (politics, the state of the economy, etc.) are more likely to surface as a measurable effect on SWB upon reflection—as in evaluative measures—or if the respondent is explicitly prompted to consider them. This suggests a significant difference in what is likely to be captured by—and in turn, what is the purpose of—measures of life satisfaction (reconstructed) versus experienced (momentary) well-being. One example of how this difference plays out occurs in measures that track the day-to-day experiences of the unemployed but do not track the unemployment rate.
Just as evaluative well-being and ExWB are conceptually distinct, at the empirical level positive and negative experiences are also separable and influenced by different factors.4 As detailed below, evidence of this distinctiveness rests not only on correlations and factor analysis but also on multimethod assessments employing measures of SWB beyond self-report surveys. Furthermore, when variables that predict evaluative well-being, positive experience, and negative experience are compared, there are differences in which variables most strongly predict each of these aspects of SWB.
The literature consistently indicates that one aspect of SWB can be moved by a manipulation while another aspect of SWB moves much less or sometimes not at all. Longitudinal research (e.g., Lucas et al., 2003; Stutzer and Frey, 2004) indicates that people experiencing important life events such as marriage or childbirth may react more strongly as measured by one construct compared to another, and over time the different measures show differential patterns of adaptation. Bradburn (1969) found that positive and negative affect are not opposite ends of one dimension but are largely independent of one another; a person can be high on one and either high or low on the other. Bradburn’s findings have been replicated many times; for example, Gere and Schimmack (2011) found that, even after controlling for measurement error and bias, positive and negative feelings were distinct. Andrews and Withey (1976) confirmed that life satisfaction is separable
________________
4 Just as, when assessing the economy, more than just gross domestic product is needed to capture its important aspects (growth rates, inflation, employment rates, Gini index, and so on), more than one measure is needed to capture the most important features of SWB.
from positive and negative affect. Lucas et al. (1996), using multimethod measurement (both self-reports and peer reports) and measures over time (a 2-year period), found that evaluative well-being and ExWB were distinct, as were measures of negative and positive experience. Kapteyn et al. (2013), using a specially designed experimental module for the RAND American Life Panel that included measures of evaluative well-being and ExWB, also found life satisfaction and the positive and negative dimensions of ExWB to be distinct, although they found additional factors when different response scales were employed.
Evaluative well-being and ExWB have different causes and correlates as well. Luhmann et al. (2011) found that people react to certain events, such as marriage and childbirth, more strongly in their evaluations of longer-term well-being than in their reports of experienced reactions. However, other events (bereavement, reemployment, and retirement) produced stronger experienced reactions. In examining adaptation to these same events in a meta-analysis of longitudinal studies, the authors found that people adapt more quickly to marriage and childbirth along the ExWB dimension, relative to evaluative well-being, but more slowly to unemployment and reemployment. They found that, for virtually every life event they studied, there was a different pattern for ExWB versus evaluative well-being. For some variables, such as childbirth, they found that ExWB and evaluative well-being could move in opposite directions (Luhmann et al., 2011).
In a review of the evidence, Schimmack (2008) concluded that—even after taking into account measurement error and other factors—life satisfaction, positive experience, and negative experience are to some degree distinct. Thus, people’s SWB cannot be fully or accurately captured without assessing all three.
CONCLUSION 2.1: Although life evaluation, positive experience, and negative experience are not completely separable—they correlate to some extent—there is strong evidence that multiple dimensions of SWB coexist. ExWB is distinctive enough from overall life evaluation to warrant pursuing it as a separate element in surveys; their level of independence demands that they be assessed as distinct dimensions.
As discussed in detail in the next chapter, ExWB can and often is usefully parsed into even narrower groupings. For instance, negative feelings can be divided into anger, sadness and depression, and worry and anxiety. Although there is often a tendency to experience these emotions together, and the same people who frequently experience one of them are also likely to feel the others frequently, these different types of negative feelings can be separated. It may be desirable to measure them separately because they are at times associated with different circumstances. However, time limitations
in surveys may prevent thoroughly assessing each type of affect, or each subcomponent of the various types of SWB. Furthermore, feelings, such as anger and worry, can be parsed even more finely; the question of how fine the measures need to be is to some extent a practical issue depending on time constraints in administering the survey. However, for some policies, particular forms of ExWB, such as depression or anger, might be particularly salient and should be assessed. What must not be lost sight of is the fact that the dimensions of SWB described above have been broadly studied and much is understood about their structure and causes. Substantial evidence indicates that, in a full assessment of SWB, evaluative well-being (or life satisfaction) and both the negative emotion and positive emotion components of ExWB should be measured separately. If survey time allows, finer distinctions can be assessed within each of these constructs, as outlined below.
What unique information, then, do ExWB measures add beyond that which can be gleaned from evaluative well-being surveys, as well as other economic or demographic measures? It would make little sense to measure ExWB (and, in turn, recommend data collection on it to statistical offices) if it did not add important new information, given that evaluative wellbeing can be self-reported using one question easily attachable to existing surveys. The panel’s position, developed above, is that both the stand-alone content of the ExWB metrics and the information that stems from contrasts between them and evaluative and eudaimonic metrics are potentially valuable for statistical purposes and relevant to a range of policy questions.
Evaluative well-being and ExWB may have different implications for policy (Diener, 2011; Graham, 2011; Kahneman et al., 2006). The latter is more directly related to the environment and context of people’s lives. Using data from the Gallup World Poll, Deaton (2012) found, for example, that health state correlates more strongly with ExWB (though it is also important for evaluative well-being) and that marital status and social time are more strongly correlated with ExWB.5 Other aspects of daily behavior, such as the nature of a person’s commute to work and the nature of a person’s social networks, are reflected in positive and negative affective states (separable aspects of ExWB). The quality of people’s daily experiences is also linked to health status and other outcomes via channels such as worry and stress on the one hand and pleasure and enjoyment on the other.
Evaluative well-being, while also sometimes influenced by these factors, is more likely to reflect people’s longer-term outlook about their lives as a whole. It may also be related to, and reflected in, longer-term behaviors such as investments in health and education. The World Happiness Report (Helliwell et al., 2012), which focuses primarily on life-evaluation mea-
________________
5 Bradburn (1968) and Bradburn and Orden (1969) also confirm this in their studies of the dimensions of marriage happiness.
sures, documents their closer linkages (relative to ExWB) to life circumstances, which may give them relevance to long-term macro policy making (and perhaps other areas, such as informing service delivery). With respect to the role that children play in most peoples’ lives, the differing assessments that come from time-use ExWB metrics (largely negative) versus life evaluations (largely positive) are a good example of how the former capture effects of the day-to-day environment while the latter capture respondents’ aspirations about their lives as a whole.6
Kahneman and Deaton (2010) found that, in the United States, income correlates more closely with evaluative well-being than with ExWB (they specify “emotional” well-being). The correlation between ExWB and annual income tapers off at roughly $75,000, or roughly the median U.S. income, while the relationship between income and evaluative well-being continues in a linear fashion. After a certain point more income does not seem to make people enjoy their daily lives more (although very low income is clearly linked with suffering and negative moods), but higher levels of income offer people many more choices about how to live and what to do with their lives.
Diener et al. (2010) found that income better predicts life evaluation scores, whereas “psychosocial wealth,” which includes factors such as social support and learning new things, better predicts life satisfaction. Their study of Gallup World Poll data showed that income influences life satisfaction but less so than does experience (affect). Positive feelings, such as enjoying life, were more strongly predicted by psychosocial wealth. Similarly, Graham and Lora (2009) found that the most important variables for the reported life satisfaction of the “poor” (respondents below median income) in Latin America, after having enough food to eat, were having friends and family to rely on in times of need. In contrast, the most important variables for the life satisfaction of the “rich” (respondents above the median) were work and health. It is likely that friends and family are the vital safety nets that make daily life tolerable for the poor, while work and health are what provide respondents with more means to make choices in their lives.
Individuals who focus primarily on daily experiences—due to low expectations, lack of agency, or imposed social norms—may have less incentive to invest in the future. In rapidly growing developing economies, Graham and Pettinato (2002) found lower levels of reported evaluative well-being among respondents with relatively high levels of income mobility compared to very poor rural respondents. It seems that people are better able to adapt to unpleasant certainty and retain relatively high levels of evaluative well-being (and likely higher in ExWB than in evaluative wellbeing) than to live with uncertainty, even when that uncertainty is associated with progress (Graham, 2008, 2011; Graham et al., 2011).
________________
6 See Clark and Senik (in press); see also Dolan (2012) and Graham (2011).
Individuals who have a longer-term focus and are more achievement oriented, meanwhile, may at times sacrifice daily experiences for longer-term objectives and anticipated evaluative well-being in the future. An example is those who choose to migrate to another country to provide their children with opportunities or to participate in social unrest for a broader societal objective. Graham and Markowitz (2011), based on data from Latin America, found lower levels of evaluative well-being among individuals who planned to migrate in the next year—a relatively extreme behavioral choice with future benefit in mind.
Health also correlates differently with different aspects of SWB. Positive affect has been found to predict response to illness (Cohen et al., 2003), with higher levels correlated with lower incidence of cardiovascular disease (Boehm and Kubzansky, 2012). Daily stress and other dimensions of negative affect are positively correlated with illness and with lack of access to health insurance. In contrast, the relation between evaluative well-being and cardiovascular disease, if it exists, is less well known. While ExWB is clearly associated with a reduced likelihood of smoking, the relationship between evaluative well-being and smoking, while still negative, is less consistent (Kahneman and Deaton, 2010).
CONCLUSION 2.2: To a larger degree, evaluative well-being, positive experience, and negative experience have different correlates and (presumably) causes, and can reflect different aspects of life that are relevant to policy. Thus, measuring all in national and specialized surveys is recommended.
To summarize the preceding discussion, sometimes people make sacrifices that lower their ExWB in order to achieve long-run higher evaluative well-being. Conversely, some people may seek greater levels of ExWB and forgo long-run evaluative well-being. This should not be a surprise, as achieving certain overarching objectives, such as advancing a science, completing a doctorate, or performing risky surgery, all of which are likely linked to higher levels of evaluative well-being, are likely to entail an increase in stressful or unpleasant days. Assessing these dimensions separately will shed light on how people view these trade-offs. It will also make evident policies that might affect one type of well-being but not another; assessing evaluative well-being and ExWB as distinct constructs will allow consideration of whether one type of SWB is sometimes bought at the expense of another type.
The bottom-line question of this section is “what dimensions of experience factor significantly into people’s SWB and should therefore be prioritized when designing surveys?” This consideration is crucial in practical terms as organizations such as the UK Office for National Statistics (ONS) begin constructing and fielding SWB modules; it is also a major focus of the recently completed OECD Guidelines (2013) project. For the ExWB component of SWB, the most obvious analytic decision for survey design is how to allocate questions between negative and positive affect, but there are other (sometimes more specific) emotions and sensations as well. Going forward, statistical agencies will also be asked to consider additional lines of demarcation that may not fit neatly into the positive and negative emotion categories, some of which the panel discusses later in this section.
2.2.1 Negative and Positive Experiences—Selecting Content for Surveys
Empirical SWB research (see Diener et al., 1999; Kahneman, 1999) strongly supports the separation of positive and negative emotional states. The two dimensions also have different correlates in the general population, which may carry policy implications. A number of researchers—Tellegen et al. (1994, cited in Watson and Clark, 1999) and Diener et al. (1995, cited in Watson and Clark, 1999)—have shown very low raw correlations between positive and negative affect, but higher, though still moderate, relationships after controlling for various random and systematic errors.
CONCLUSION 2.3: Both positive and negative emotions must be accounted for in ExWB measurement, as research shows that they do not simply move in an inverse way. For example, an activity may produce both negative and positive feelings in a person, or certain individuals may be predisposed to experience both positives and negatives more strongly. Therefore, assessments of ExWB should include both positive and negative dimensions in order for meaningful inferences to be drawn.
Additionally, indicators of negative emotion are distinct from one other. Evidence suggests that dimensions of negative affect—sadness, worry, stress, anger, frustration, etc.—tend to be more differentiated than those on the positive side, which tend to move more in unison, carries implications for data collection. Research on how different adjective terms cluster (e.g., Kapteyn et al., 2013) show that negative emotion measures generally seem to have lower intercorrelations than do positive ones and may be subject to more variability as a function of specific adjectives used in survey questions. Positive measures appear to be more robust in this
sense. Within the clusters, however, the construct measures appear to be fairly robust with regard to the selection of particular adjectives from the cluster.
The multidimensional character of negative emotion suggests a need for more questions about it (relative to positive emotions) on surveys intending to cover the full range of ExWB. The ONS approach, for example, reflects the positive–negative dichotomy, as the two dimensions are separated in the question structure. However, for now, the ONS surveys still lack multiple dimensions/questions on the negative side.
RECOMMENDATION 2.1: When more than two ExWB questions can be accommodated on a survey, it is important to include additional ones that differentiate among negative emotions because—relative to the positive side—they are more complex and do not track in parallel (as the positive emotion questions tend to).
While “happiness” has received a great deal of attention in the media, and the positive dimensions of SWB are actively researched in the literature on evaluative well-being, a number or researchers have emphasized measurement of negative emotions (suffering),7 and this choice of focus is to some degree a policy choice. Kahneman’s view8 on the positive–negative balance is that:
the focus on happiness is misguided and … in part is an accident of language. We measure length and not shortness, we measure depth and not shallowness, and we only see in dimensions that are marked on the one side we are thinking of. We should be measuring suffering. And we should act as a society to reduce suffering…. I am much less concerned about happiness and [in favor of] reducing human suffering.
At this point, there is not enough empirical evidence about these two dimensions of ExWB to know which is more policy responsive. Clearly, as Kahneman argues, reducing negative experience, particularly prolonged
________________
7 A major exception to the “focus on happiness” is found in the field of mental health, where far more attention has been directed to the negative side of the emotional balance than to the positive side. Western cultural biases, embodied in the psychiatric conceptions of mental health, have led to a concentration on efforts aimed at reducing negative affect and suffering. However, Sheldon Cohen’s research (available: http://www.psy.cmu.edu/~scohen/AmerPsycholpaper.pdf [October 2013]) has examined how positive emotions/attributes lead to higher resistance and identifies specific mechanisms through which different types of social constructs influence physical health (including social stress and immune function issues)—revealing a potentially powerful effect.
8 From a talk, available: http://stevensonfinancialmarketing.wordpress.com/2013/04/11/8345 [October 2013].
suffering, is often a rightful policy objective, even if the exact policy levers have not been identified. Knowing more about the relationships between determinants and negative experience is important contextual information.9 If data can reveal the links, it can be left to researchers to discover if policy could be creatively used to have an impact. A targeted policy to assist the poor may focus on negative experience, but it may be linked to positive affect as well.
This line of reasoning suggests the value of framing measurement in terms of experience, which can reasonably include pain and other sensations that factor into suffering but may be omitted by a narrower hedonic approach. In other words, measuring “experience” seems essential for addressing issues of long-term suffering in various populations. Relatedly, the metric for characterizing emotions and suffering—that is, ExWB—could be based on the duration of the day (or other time period) spent in that state. Of course, it is not clear that all methods for capturing ExWB are capable of yielding such temporal metrics; it probably requires momentary assessment or reconstruction of a day to achieve duration-weighted indices.
RECOMMENDATION 2.2 (Research): A scale of suffering that has a duration dimension would be a useful measurement construct and should be developed. Such a measure might capture and distinguish between things like minutes of pain or stress versus ongoing poverty, hunger, etc. Suffering is not the absence of happiness or the presence of only negative experience, and the scale should reflect this in a way that suggests relevant classes of policies. Little work has been done on a scale of suffering, so the research will have to begin at the conceptual level. This research should examine the information content of alternative descriptive adjectives, some of which have perhaps not yet been used in the literature on SWB, but which could round out the set.
________________
9 While not an emotion scale and thus not a measure of SWB (though likely a predictor of it), a scale of negative life events was developed for use in the General Social Survey (GSS) as a component of the GSS index of societal well-being. This approach, by registering exposure to the negative circumstances and events experienced by people (e.g., hospitalization, death of a family member, eviction, crime victimization), was designed to report “objective experiences that disrupt or threaten to disrupt an individual’s usual activities, causing a substantial readjustment in that person’s behavior” (Thoits, 1983). As described by Smith (2005), this approach has been used extensively not only to account for differing levels of reported well-being among individuals or groups but also for understanding and predicting individual illness (both psychological and physiological); in so doing, it provides “factual data for the formulation of public policies to deal with these problems” (p. 18).
A focus on suffering may also resonate with the public in a way that discussions of happiness do not. When people are asked what is more important for programs and policy, reducing suffering (which requires monitoring negative affect) or increasing happiness (which requires tracking positive affect), the available evidence suggests that the majority prioritize the former. Dolan and Metcalfe (2011) surveyed people to ask whether government policy should seek to (1) improve happiness or (2) reduce misery, and there was more support for the second option. Such findings have important implications for how happiness (and misery) are discussed in the popular press and public debate. Thus, while it is not obvious that a good measure for suffering has yet been developed, it may be politically more acceptable to aim policies at reducing negative affect as opposed to increasing positive emotions. In fact, monitoring (and reduction of) suffering and stress has been the more common objective of government policy.
RECOMMENDATION 2.3: Given the importance of both positive and negative experience, the one-dimensional term “happiness” should not be used to label most ExWB measures. Another limitation of the term is that it is often also used as a descriptor in evaluative measures, which creates another likely source of confusion. Instead, including a term signifying misery or suffering in addition to positive emotions would be more balanced.
For the fullest possible descriptive accuracy, having two words (one for positive and one for negative experiences) incorporated under the ExWB dimension of SWB has an advantage, even though it is well understood by researchers in this field that “hedonic” refers to both positive and negative experiences (again, more broadly defined than emotions). While it may or may not be intuitive that there are both increases and decreases in “well-being,” it is clear that SWB measurement is about much more than “happiness.” This general point applies to measures of evaluative wellbeing as well, where an overarching term such as “happiness” can easily mask the great depth of findings in well-being research. It seems clear that labeling (word choice), especially in the popular press, does influence the public debate. Linguistic biases need to be addressed, both in survey construction and in presentation of information. Certainly, labeling measures (or measurement programs) as “well-being/suffering” would dilute the relentless focus on the positive.
There are alternative ways to characterize the positive and negative sides of ExWB—most notably as unidimensional or bidimensional. “Balance” metrics have also been used that combine positive and negative poles. Bidimensional (and possibly multidimensional) approaches offer a more relevant concept—relative to unidimensional measures such as
happiness—because these measures give a richer picture of experience and the possible environmental factors that might influence affective experience. For example, people who report suffering directly may be different from those reporting low levels of satisfaction. Insofar as the various dimensions of emotions are driven by different factors, measuring them separately offers more insight into situations where polices potentially could improve SWB. Balance measures also allow those, such as policy makers, who want to concentrate on reducing negative experience to still have access to information underlying that end of the spectrum. There are issues in how to define the balance measure: Is it just an average? Is it enough to know that a respondent had x minutes of positive affect and y minutes of suffering during that day? Or do we need hours, events, or intensities? A counterargument in favor of the value of unidimensional measures is that they allow people to scale or integrate for themselves how the different measures of emotions or different aspects of their lives should be weighted.
A balance concept could encourage investigating actions that might increase positive emotion (and possibly thereby increase positive aspects of SWB, and possibly health), as well as actions that might reduce suffering. Such a concept need not presume that suffering is the opposite of happiness, because it is possible to be moderately happy while experiencing a moderate degree of suffering as well as a moderate degree of positive emotion. It is not yet clear exactly which balance metrics would be appropriate and therefore should be considered for national statistics.
Beyond and possibly intertwined with evaluative well-being and ExWB are additional types of psychological well-being or SWB that may be of potential interest to policy makers, leaders, and citizens. A number of alternative or supplemental forms of psychological well-being have been placed under the rubric of eudaimonic well-being: these include optimism; quality of social relationships; meaning and purpose in life; mastery, skills, and achievement; freedom to make decisions regarding one’s own life; engagement, interest, and flow; and self-worth. Eudaimonic well-being comes into play if one assumes that people commonly strive for more than just “happiness” and one believes a worthwhile societal goal is to encourage citizens to pursue meaning and purpose in their lives, to give and receive social support, and to have skills and self-esteem.10
________________
10 The literature on “noncognitive skills” has addressed the role of some of these individual personality traits and abilities in various outcomes, such as labor market or educational success. Heckman et al. (2006) list as noncognitive skills various social skills, time preferences, motivation, and the ability to work with others.
The Ryff Multidimensional Scales of Well-Being (Ryff and Keyes, 1995) is an example of a widely used, predominantly eudaimonic scale; it consists of six dimensions of wellness (autonomy, environmental mastery, personal growth, positive relations with others, purpose in life, self-acceptance). However, the underlying latent structure and factorial validity of this model remains highly contentious; specifically, there is evidence of a high correlation (lack of distinctiveness) among four of the six dimensions (Springer et al., 2006).11
Eudaimonic well-being is broadly related to the opportunities that people perceive they have to exercise choice and to pursue fulfilling lives. While distinct,12 eudaimonic well-being may also figure into assessments of both evaluative well-being and ExWB. For example, perceived meaning attached to one’s job or taking care of one’s child may play a role in a person’s self-reported well-being, or it may be a factor in predicting whether a person will continue to engage in an activity that scores poorly in a momentary assessment. While eudaimonic well-being is surely important and worth measuring, the field has much less experience with metrics for this type of well-being, and further research and testing are necessary before recommending its inclusion in large-scale surveys.
The evidence for the independence of subtypes (or dimensions) of eudaimonic well-being from each other and from SWB is more limited than it is for evaluative well-being and for the dimensions of ExWB. The new OECD guidelines on measuring SWB include a separate measure of eudaimonic well-being. Literature cited in that volume suggests that eudaimonia correlates less closely with the other SWB measures than do measures of positive or negative affect or of life evaluation. Gallup World Poll data for the OECD countries show the highest correlation between positive and negative affect (–0.39) and the lowest between purpose (the Gallup organization’s measure of eudaimonia) and negative affect (–0.09).
________________
11 Diener et al. (unpublished) have recently begun developing a “Comprehensive Psychological Well-being Scale,” which includes a eudaimonic (meaning and purpose) component. In reference to the debates about Ryff’s factor structure (which does not show six clearly differentiable factors), they are conducting a factor analysis of the new scale to demonstrate the correlation and separability among self-assessment components, which fall into four categories: Relationships (perceived support, social capital, trust, respect, loneliness, and belonging); Mastery/Engagement (flow, engagement, interest; using one’s skills; learning new things; control of one’s life; and achievement, accomplishment, and goal progress assessments); Meaning and Purpose in Life; and Subjective Well-being (optimism, life satisfaction, positive feelings, and negative feelings).
12 The distinctiveness of SWB components may be present even at the cellular level. Fredrickson et al. (2013) investigated “molecular mechanisms underlying the prospective health advantages associated with psychological well-being” and found that “hedonic and eudaimonic well-being engage distinct gene regulatory programs despite their similar effects on total well-being and depressive symptoms” (p. 1).
Life satisfaction has a correlation of approximately 0.23 with positive affect and –0.23 with negative affect; its correlation with purpose is 0.13 (OECD, 2013, pp. 33-34).
The purpose (or lack of it) dimension of eudaimonic well-being seems particularly important, as it is associated with much of what we do. Purposefulness (or worthwhileness) can be an important driver of behavior and is experienced in much the same way as emotion. And of course, an ExWB measure might capture some purpose dimensions. In thinking about the full dimensionality of SWB, the concept of “worthwhileness” or “meaningfulness” has been given considerable attention in the literature and has apparently been deemed central to it by ONS, which includes a question on eudaimonic well-being in its SWB module. This dimension may be important for understanding (or predicting) why and when people engage in various activities during the day or in life more generally.
For example, a parent may be less unhappy changing a child’s diaper because he finds taking care of his child a worthwhile activity. Or reading the same story over and over to one’s children may not always bring a great deal of pleasure, but it is purposeful (or worthwhile, meaningful, fulfilling, rewarding). And the reader (parent) feels that purpose at the time. Activity-based data suggest that time spent with children is relatively more rewarding than pleasurable and time spent watching television is relatively more pleasurable than rewarding—but both are drivers of behavior (Kahneman and Krueger, 2006). A rich conception of the flow of feelings places both pleasure and purpose on experiential footings.
Calling purpose a feeling suggests that it is an emotion that can be placed on a comparable footing to more recognized emotions like joy, anxiety, anger, etc. To most psychologists, feelings are emotions and so any feelings of purposefulness (or purposelessness) would simply add to (or subtract from) the overall “goodness” of an emotional experience. Feeling that something is purposeful (or purposeless) adds to (or subtracts from) the overall “goodness” of the sentiments associated with an experience. This is somewhat related to Fred Feldman’s attitudinal hedonism (Feldman, 2004).
For measurement, it may not make much difference whether one thinks of purpose as contributing directly to good and bad emotions or as sitting alongside but separate from them, as a distinct sentiment. What matters is that the adjectives for purpose (fulfillment, etc.) are distinct from those used for pleasure (fun, etc.) and that a range of good feelings (good emotions, good sentiments) contributes to overall well-being. Hedonic emotions and purpose are both felt experiences and ideally both would be measured. If anything, the purpose dimension is a simpler construct than other emotions in that it is largely nonaroused and so either good (purposeful, worthwhile, meaningful, fulfilling) or bad (pointless, worthless, meaningless, unfulfilling). One would not need to measure both if pleasurable experiences, for
example, were highly correlated with purposeful ones. The evidence on this issue is scarce, but data from the Day Reconstruction Method suggest that, while some activities are high in both purpose and pleasure (e.g., exercising) or low in both (e.g., commuting), others are high in pleasure and low in purpose, such as watching television, and others are low in pleasure and high in purpose, such as volunteering for unpleasant tasks (White and Dolan, 2009).
CONCLUSION 2.4: An important part of people’s experiences may be overlooked if concepts associated with purpose and purposelessness are not included alongside hedonic ones like pleasure and pain in measures of ExWB. Crucially, central drivers of behavior may also go missing. People do many things because they are deemed purposeful or worthwhile, even if they are not especially pleasurable (e.g., reading the same story over and over again to a child, visiting a sick friend, volunteering); they also do many things that are pleasant even if they are not viewed as having much long-term meaning in the imagined future.
In terms of relevance to policy, there appear to be differences in the way ExWB and evaluative well-being relate to time spent volunteering when purpose is accounted for. Greenfield and Marks (2004) found that among an older population group, volunteering was associated with more positive ExWB but not with significantly less negative affect. The extent to which volunteering makes people happier is unclear, as is the extent to which happier people tend to engage in more volunteerism. However, the latter seems to be part of the story, as the measurable association is reduced considerably when fixed effects are controlled (Meier and Stutzer, 2006). This has potentially important implications for those in government trying to “sell” the idea of helping others; it may have more traction if it is presented as a way of increasing happiness (or decreasing unhappiness) through purpose. In any case, more can be said about how well life is going if purpose is accounted for, as well as pleasure.
Not accounting for purpose alongside pleasure is potentially a threat to the legitimacy of ExWB measures. One of the attractions to policy makers of the constructs of evaluative and eudaimonic well-being is that they allow for consideration of nonhedonic sources of happiness and suffering. But beyond the reflective exercise of asking respondents to consider their lives overall, additional insights can be gained by assessing degree of purpose, or lack thereof, as it is revealed during such life experiences as looking after friends or family (or not having those connections), pursuing goals (or not having them), or dedicating oneself to work (or finding work pointless).
RECOMMENDATION 2.4: Where possible, adjectives of purpose can—and should—be added to experiential assessment methods and techniques, although more needs to be learned about them. Cognitive testing and other psychometric work is needed to find out what members of the public make of these and other possible descriptions. Also needed is quantitative analysis of the correlations between candidate descriptions.
2.2.3 Other Candidate Emotions and Sensations for Measures of ExWB
In thinking about exactly which adjectives best capture the ExWB constructs of interest and that warrant measurement, it is important to consider those that may not sort neatly as positive or negative emotions. Again, thinking in terms of experiences and not just emotions allows for inclusion of more of these factors. Whether or not an ExWB measure should include factors beyond the realm of emotions depends on the research or policy question at hand. For example, sensations such as physical pain, numbness, heat, or cold could be part of the conceptualization of ExWB at the momentary level of measurement—particularly if the context is people’s health or housing conditions. Certainly, people experiencing pain will on average report higher levels of negative well-being, all else being equal (Krueger and Stone, 2008). The 2012 Health and Retirement Study is a good example of a survey module that asks about negative emotions and physical pain, as is the 2010 version of the American Time Use Survey. The National Health Interview Survey and the National Health and Nutrition Examination Survey, major data collection programs of the National Center for Health Statistics of the Centers for Disease Control and Prevention, are other good candidates.
RECOMMENDATION 2.5: Pain may be an important dimension of ExWB given that it affects people’s ability to engage in day-to-day activities. Therefore, while still experimental at this stage of research, pain questions should be included in ExWB questionnaires, particularly in domains such as health or housing where this information is particularly germane to research and policy questions.
Ultimately, for a given question, how one characterizes “the momentary” will dictate the value of additional experience considerations such as feelings of pain, spiritual elevation, flow, love, etc. The Positive Affect Negative Affect Scale, often called “PANASX,” is a popular emotion scale for which several other adjectives have been identified and tested, such as those related to hostility, guilt, fear, joviality, serenity, shyness, self-assurance, fatigue, surprise, and attentiveness. Anger is particularly complex; Bradburn (1969) and Harmon-Jones (2004) found that anger appears to be largely un-
related to global measures of either positive or negative affect. In principle, surveys can ask about the presence of these states at the momentary level of analyses, but this has not, in the panel’s experience, usually been done.
Another candidate for consideration is a crosscutting dimension of affect known as “activation-deactivation” (Larson and Diener, 1992). There is considerable evidence that the range of emotions can be usefully characterized as a two-dimensional space, with high and low arousal as one of the dimensions and positive and negative emotion as the second (see the “circumplex” model of affect, Watson et al., 1988). Arousal is especially relevant when measuring affect in populations that are ethnically and/or age diverse. We discuss this in more detail below, in section 4.1 on cultural effects.
While there may not be enough evidence to include questions about sensations and other factors (beyond emotions) influencing people’s experiences in broad surveys, in the same way that questions on evaluative well-being or “feelings yesterday” have been added, there are cases where such factors are clearly relevant and should be included; for example, in assessing pain and mobility in surveys of the elderly (Health and Retirement Study) or in measuring arousal in cross-country comparisons. Descriptors for pain and anger are among the most prominent adjectives of interest beyond the “hedonic” that may not always fit into a positive/negative emotion construct. Other factors, not discussed here and not well understood, may also influence people’s moment-to-moment experiences; Box 2-1 describes one example: what respondents happened to be thinking about, which may take the form of intrusive or fleeting thoughts.
BOX 2-1
Pop-Ups: The Likely Importance of Intrusive Thoughts
One of the main benefits of assessments of ExWB is that they overcome some of the focusing-effect problems associated with global assessments of life satisfaction. The focusing effect is used to explain the higher effect that income has on life satisfaction as compared to the Day Reconstruction Method (DRM): Kahneman et al. (2006, p. 1908) conclude that “the belief that high income is associated with good mood is widespread but illusory.”
Yet the DRM’s attempt to capture experienced utility may create a focusing effect of its own by asking respondents to report their feelings when thinking about the activities in their lives. This formulation neglects the way in which people’s attention drifts between current activities and concerns about other things. It therefore may be useful to consider and measure the impact on experienced utility of important pop-ups: things that pop into people’s heads as they go about their daily activities but which are not captured especially well by routine assessments of affect.
One of the great advantages of ExWB measures is that they seek to capture the flow of experienced utility over time. Experienced utility is largely influenced by where attention is directed, sometimes voluntarily (such as when an author is focusing on writing a sentence) and at other times involuntarily (such as when the author’s children just popped into his head). Measures of emotional well-being do a good job of picking up feelings in general but often miss important pop-ups.
Just like purposefulness, these pop-ups (also called intrusive thoughts or mind-wanderings) potentially drive a lot of behavior. People can be expected to give a lot of weight to those things that grab their attention, even if they do so only fleetingly. Most of the research to date has been conducted on clinical populations, but there is some evidence that general mind wanderings are frequent, occurring in up to 30 percent of randomly sampled moments during an average day (Smallwood and Schooler, 2006). Generally negative intrusive thoughts have a negative association with well-being (Watkins, 2008). The relationship is not straightforward, however, because the suppression of intrusive thoughts can make things even worse (Borton et al., 2005). There is also some suggestion that even unwanted thoughts may still play an important adaptive role in problem solving and learning (Baars, 2010). On the other hand, it has been suggested that even positive intrusive thoughts can be a source of unhappiness; our ability
to think about something other than what we are doing at the present moment comes at an emotional cost (Killingsworth and Gilbert, 2010).
The presence of intrusive thoughts about health helps to explain the difference between experienced utility and decision utility in the valuation of health states (Dolan, 2011). In particular, pop-ups about health could explain why people are often willing to make large sacrifices in life expectancy in order to alleviate conditions for which there is a considerable degree of hedonic adaptation (Smith et al., 2006). Recent research has also shown how intrusive thoughts can be used to change behavior. By using therapies that focus on shifting attention, researchers were able to reduce intrusive thoughts about smoking and see positive results on patients’ ability to stop smoking (May et al., 2011).
Policy makers will be interested in the consequences, as well as the causes, of intrusive thoughts. In a health context, for example, negative thoughts such as worry are associated with increased cortisol levels and increased heart rate, and they may actually cause increased heart disease and fatigue and slower recovery from surgery (Watkins, 2008).
Of course, negative thoughts might also be good for health through their effects on health-promoting behaviors. There is evidence, for example, that increased worry about breast cancer is associated with a greater probability of undertaking screening (Hay et al., 2006). Among other things, new data will enable policy makers to better determine when to reduce negative thoughts, by how much, and for whom.
One of the attractions to policy makers of “happiness” as represented by evaluative SWB measures is that it allows people to consider the importance of a range of things, including intrusive thoughts. But it does so in rather artificial and abstract ways by asking respondents to consider their life overall. It is much better to pick up the effects of pop-ups where they really show up—in the experiences of life.
Every survey will focus attention in one way or another. It seems unlikely that pop-ups can be picked up effectively by asking people to focus attention on them. To be more specific, a future DRM-type study could certainly ask respondents about thoughts and feelings before asking what the respondent is doing about them and who they are with. It may be that asking people about their main activity before asking them about their mood draws their attention away from what they were thinking about. Such a study would allow researchers to explore the importance of focusing effects as they relate to current activity and to general mood.
This page intentionally left blank.
