The Survey on Flourishing: Measuring Subjective Well-Being in an Adolescent Sample
Flourishing is a state of well-being that occurs when individuals have the psychological, social, and physical resources they need to meet psychological, social, or physical challenges [1] It goes beyond just focusing on the absence of pathology, also examining positive outcomes such as emotional regulation, supportive relationships, meaning and purpose, and life satisfaction. Currently, much of the research on flourishing has been conducted using adult populations rather than youth [2]. However, in recent years there has been a push to better understand adolescent well-being.
Well-being research shows that flourishing is associated with several positive life outcomes and circumstances, including supportive social networks and relationships, positive work life, higher levels of physical and mental health, and improved school performance [3–9]. Given the prevalence of psychosocial stressors present during adolescence and the nature of adolescence as a critical period for social and emotional growth, it becomes more important to accurately understand and measure adolescent flourishing.
Improving adolescent flourishing also has broader societal implications. Historically, developmental science, psychology, education, and other fields have underestimated adolescents, tending to focus on the problems they face (e.g., learning difficulties, mental illness, low motivation, substance use, etc.) rather than the strengths they possess [10–11]. However, positive youth development research and other similar areas of research identify adolescents as having unique resources which they can use to meaningfully contribute to their community [10, 12]. Although working to improve adolescent well-being and enabling them to use their strengths to contribute meaningfully to society is an important task, it may be difficult to accomplish if we cannot measure it. Thus, the purpose of the present study is to examine the psychometric properties of the Survey on Flourishing (SURF), a measure of subjective well-being, using a nationally representative adolescent sample from the United States.
What is Subjective Well-Being?
Subjective well-being is a broad, multifaceted construct, and has historically been difficult to define. For many decades well-being was determined to be the absence of physical or mental malfunction. However, more recent research indicates that well-being is not just the absence of problems, but it includes assets, strengths, values, and other positive characteristics [13–14]. Diener, in a classic paper on subjective well-being, defined subjective well-being as a combination of positive emotion and life satisfaction [15]. Currently, definitions of subjective well-being most commonly include two components: emotional well-being, which includes the presence of positive emotion and life satisfaction, and positive functioning, which includes social functioning (e.g., social integration and contribution) and psychological functioning (e.g., autonomy and personal growth) [13].
These components of well-being also apply to adolescents. Researchers have identified certain developmental tasks which may indicate whether a child is doing well. Some of these tasks which are critical in adolescence include academic achievement, forming close peer relationships, learning to follow rules, participating in extracurricular activities, and forming a sense of self-identity [16]. These tasks which are critical to healthy adolescent development generally align with the social, emotional, and psychological components included in subjective well-being.
Taken together, current research suggests that subjective well-being is more than just the absence of pathology, it is subjectively experienced, and it includes emotional, social, and psychological well-being components [17]. This definition also applies across developmental periods, although how they might manifest may differ. This definition of subjective well-being has been used to create various models of well-being which we discuss below.
Models of Subjective Well-being
One recent model that has been suggested is Martin Seligman’s five-factor PERMA model. The PERMA model conceptualizes subjective well-being through the domains of positive emotion, engagement, supportive relationships, meaning, and achievement [18–19]. This model uses these five domains to capture emotional well-being (via the positive emotion domain), social functioning (via the supportive relationships domain), and psychological functioning (via the engagement, meaning, and achievement domains). Although this model has mainly been applied to adults, the EPOCH (engagement, perseverance, optimism, connectedness, happiness) is a model which adapted the PERMA to better apply to adolescents.
A second commonly used model is the Ryff model of psychological well-being [20]. This model focuses specifically on the psychological well-being component of subjective well-being. It describes psychological well-being as encompassing six dimensions: autonomy, environmental mastery, personal growth, positive relations with others, purpose in life, and self-acceptance.
Third, Keyes’ model of social well-being is a frequently used model for describing social well-being [14]. According to this model, social well-being comprises five domains: social integration, social contribution, social coherence, social actualization, and social acceptance. Both the Ryff model and the Keyes model clearly describe components of subjective well-being, and they are brought together in the Mental Health Continuum (MHC), which is a measure of subjective well-being which we discuss in the following section [21].
Measures of Adolescent Well-Being
Using the conceptualization of subjective well-being and the models described above, researchers have developed measures to assess subjective well-being. However, these measures have mainly relied on adult populations to determine their utility. However, there are still some measures which have either been developed specifically for adolescents, or which have been shown to be effective when used with adolescents. In this section we describe the most commonly used measures of adolescent well-being.
A recent literature review identified seven measures of subjective well-being which are available for use with adolescents, and which contain items which measure both the emotional well-being and positive functioning components of subjective well-being [22]. These measures included the MHC-short form, the Ryff scales of psychological well-being, the EPOCH (engagement, perseverance, optimism, connectedness, happiness) measure of adolescent subjective well-being, the Child and Adolescent Wellness Scale (CAWS), the Social and Emotional Health Survey (SEHS), the Warwick-Edinburgh Mental Well-Being Scale (WEMWBS), and the World Health Organization-Five well-being index (WHO-5) [21, 23–27].
The Ryff scales of psychological well-being constitute a 20-item measure which is based upon the Ryff model of psychological well-being discussed above. While Rose and colleagues reported that this scale includes both emotional well-being and psychological functioning components [22], the creators of this study only aimed to capture the psychological well-being aspect of subjective well-being [28]. It measures psychological well-being along the six domains of autonomy, environmental mastery, personal growth, positive relations with others, purpose in life, and self-acceptance. Items from this measure were adapted to create the MHC, which is a more comprehensive measure of subjective-well-being.
The Mental Health Continuum—Long Form is a 35-item measure which draws from the Ryff and Keyes models discussed above [21, 29]. The MHC-short form is an adapted version of this measure which includes 14 items— three which assess emotional well-being, five which assess social well-being, and six which assess psychological well-being. In a recent study summarizing research on the latent profile analysis of the MHC, a bi-factor model (like that found for the PERMA Profiler) appeared to demonstrate the best fit with observed data, although the general subjective well-being factor accounted for a substantially greater amount of variance than either of the other latent variables [30]. A three-factor model (with factors representing MHCM’s three main foundations) also demonstrated good fit, although slightly less than the bi-factor model [30].
The EPOCH is a 20-item self-report measure which was developed for adolescents and adapted from the PERMA model described above [31]. The EPOCH measures engagement, perseverance, optimism, connectedness, and happiness. The researchers aimed to develop a measure using domains which influence the PERMA domains in adulthood. They determined that the EPOCH demonstrated adequate psychometric properties, although more research is needed to determine the extent of its utility.
The EPOCH’s counterpart is the PERMA-profiler, which is a 23-item measure designed for adults that is also based on Seligman’s PERMA model [18]. A recent study examining this measure’s effectiveness suggested that this model was an accurate and valid well-being measurement tool. They also used confirmatory factor analysis (CFA) to determine that a bi-factor model of subjective well-being best fit the data. The model contained one general factor which accounted for much of the shared variance between all the items, and five secondary factors that represented the PERMA domains [32]. They determined that the general factor explained most of the variance in scores, while the secondary factors explained additional (although relatively weak) amounts of variance.
The Warwick-Edinburgh Mental Well-Being Scale (WEMWBS) is also a widely used subjective well-being measurement tool [33]. Items in this measure were developed to reflect the domains of positive affect, psychological functioning (autonomy, competence, self-acceptance, personal growth) and interpersonal relationships [34]. Similar to the MHCM and the PERMA Profiler, research examining the WEMWBS’ performance with adolescents suggested that a bi-factor model best fit the observed data, with a broad factor representing general subjective well-being, and two relatively weak secondary factors representing psychological and social well-being [35].
The CAWS is a 100-item measure that assesses child well-being through ten domains: empathy, connectedness, self-efficacy, initiative, adaptability, social competence, conscientiousness, optimism, emotional regulation, and mindfulness [24]. This scale measures domains which are not generally included in adult measures of well-being, including social competence and emotional regulation. While the original validation study noted that this measure demonstrated good reliability, they did not examine the factor structure of the measure.
The SEHS is a 36-item measure designed to assess subjective well-being along the domains of belief in self, belief in others, emotional competence, and engaged living [26]. This model is based largely on social-emotional learning theory, which suggests that developing social and emotional competencies are among the most important tasks that will help adolescents live meaningful lives [36]. This model also drew from positive psychology research on the importance of supportive relationships on promoting well-being, although it was not explicitly grounded in the models of subjective well-being described above.
In summary, researchers have developed many models of subjective well-being for adolescents. While these measures differ somewhat in their underlying theories and content, researchers generally agree that subjective well-being is a subjectively experienced, multifaceted construct. There is strong overlap among these measures which suggest subjective well-being consists of social (e.g., connection, supportive relationships, etc.), psychological (purpose, achievement, etc.), and emotional (e.g., life satisfaction, positive emotion, etc.) components. Statistical approaches used to examine some of the most commonly referenced models suggest that using a bi-factor approach may be an effective method for measuring subjective well-being. Specifically, after accounting for general subjective well-being or positive emotion, other components may explain additional (although relatively small) differences in people’s levels of subjective well-being. It is important to note, however, that the literature did not provide a consensus on how many secondary factors exist or what exactly they represent. In the studies above, the researchers specified the factors based on their theoretical model (e.g., the PERMA-profiler used five secondary factors while the MHC identified three). However, these factors generally seemed to represent aspects of social, psychological, and emotional well-being.
Limitations to Current Measures
Although the development of measures of adolescent subjective well-being represents significant progress, there are some notable limitations that impact their utility. First, measure content must be considered. There is a consensus among well-being researchers that subjective well-being is a broad, multifaceted construct. While three generally agreed on domains include social, emotional, and psychological well-being, there are many important domains which have not yet been tapped into, such as gratitude, transcendence, or mindfulness. Other measures of subjective well-being focus on a particular facet of well-being. Although this approach may be intentional, these measures may be too narrow to capture certain elements important to broader adolescent subjective well-being. Similarly, Seligman suggests that no single measure can capture the breadth and depth of well-being [31]. Thus, although these measures may provide valuable information, they may be most useful when used in conjunction with measures that examine alternate facets of well-being. The depth and breadth of subjective well-being suggests that there is a need for additional measures that expand on the content of current measures.
Second, the generalizability of these measures depends on the sample which was used to examine their psychometric properties. When examining these measure’s validation studies, we observed that only the EPOCH obtained a nationally representative sample from within the United States [27]. Additionally, most of these validation studies relied on samples obtained from outside of the United States. Although this should be noted as a strength for those using these measures with clients from the areas where the measure was validated, it would be inappropriate to expect these measures to perform equally across cultural groups. Thus, because the purpose of this study is to examine the psychometric properties of the SURF with data from a representative sample from the United States, these validation studies suggest there may be a need for additional measures which are supported for use within this population.
One additional concern regarding the samples used in these validation studies relates to the size of the sample. Although a general rule of thumb regarding sample size states that researchers should collect ten responses per test item, several adolescent subjective well-being measures fall short of this recommendation [37–38]. When developing a measure, having a too-small sample size may increase measurement error and lead to inaccurate or biased measurement.
Finally, there are other relatively smaller limitations that may affect these measure’s utility. First, some measures, such as the 100–150 item CAWS, are extensively lengthy [24]. Shorter measures may be more time-sensitive while still demonstrating good psychometric properties with little item overlap. Second, accessibility determines the extent to which the measures can be used for many practical purposes. Out of these measures, the PGI is not available in the public domain, and the CAWS, SEHS, WEMWBS, and SEHS are free to use with developer permission. The EPOCH, MHC-SF, Ryff scales, and the WHO-5 are free with developer acknowledgement [22]. Although paid measures may be effective, free measures, such as the EPOCH and the SURF may display similar effectiveness and allow for more widespread use.
In summary, although the development of measures of adolescent subjective well-being is a step forward, these measures have some limitations such as content domain, sample population, sample size, accessibility, and length which impact their utility as adolescent subjective-well-being outcome measures. Because adolescent well-being is becoming a greater priority in society, it is important that accurate measurement tools are available to help individuals understand and improve it. In this paper, we examine the psychometric properties of the Survey on Flourishing adolescent version (SURF) as a novel measure of subjective well-being that is based on a nationally representative US sample, accessible, and quick to administer. We also examine its reliability, validity, and factor structure.
The Survey on Flourishing
This study aims to use current research on adolescent well-being to examine the psychometric properties of the Survey on Flourishing (SURF) in a nationally representative adolescent sample within the United States. The original SURF questionnaire was designed to obtain a measurement of subjective well-being by including items reflecting both positive functioning and emotional well-being. The SURF was shown to have good reliability and validity when used in an adult population. We expect that the SURF will demonstrate similar psychometric properties and structure when used in an adolescent population. Thus, this study aimed to examine the utility of the SURF through examining its internal consistency, factor structure, and convergent and discriminant validity.
Regarding the reliability of the SURF, we expected that the SURF would demonstrate good internal consistency by having a Cronbach’s alpha score (average inter-item correlation) between .80 and .90. Having a Cronbach’s alpha statistic in this range means that the test displays strong internal consistency, which is one facet of reliability.
Regarding the factor structure of the SURF, we expected the items to load onto a single general factor of adolescent subjective well-being (see Methods). This suggests the SURF measures a unitary construct, which would align with previous research on the measure.
Regarding, the SURF’s validity, we expected the SURF to show a strong positive correlation (r > .70) with similar measures of well-being such as the PANAS Positive Affect subscale (PANAS-Pos) and the Satisfaction With Life Scale (SWLS), while showing a weak negative correlation (r < -0.5) with discriminant measures such as the PANAS Negative Affect Subscale (PANAS-Neg). These predictions were based on a previous study which showed that the SURF demonstrated similar psychometric properties when used with adults and adolescents. Good convergence with the PANAS Positive Affect subscale and the SWLS would suggest that the SURF is measuring a similar construct, whereas a low correlation with discriminant measures would suggest that the SURF is not measuring constructs that are different from adolescent subjective well-being. Taken together, these measures provide evidence that the SURF is measuring what it purports to measure.
Of note, we planned to examine the test-retest reliability and to calculate the SURF test-retest reliability and Reliable Change Index (RCI), although due to invalid second phase data provided to us by the data collection site, we were unable to conduct these analyses. We discuss this further in the discussion section.