Introduction

What do people mean when they use terms such as ‘happiness’, ‘a good life’, or ‘satisfaction’? To what extent does the understanding and use of such concepts among the general public resemble or differ from scholarly scientific discourse in the research domains of well-being and quality of life?

The World Health Organization defines quality of life as “an individual’s perception of their position in life in the context of the culture and value systems in which they live, and in relation to their goals, expectations, standards and concerns” (WHOQOL Group 1994). However, unanimously accepted definitions of quality of life and well-being are currently not available within the social sciences (Fernández-Ballesteros 2010; Gasper 2010). Camfield and Skevington (2008) have highlighted the conceptual convergence between quality of life and well-being, particularly with regard to the subjective aspects of these concepts. At the subjective level, both comprise inner psychological states (emotions, cognitive evaluations, beliefs and self-perceptions) as well as perceptions of the external world (Michalos 2008). They represent complex multi-componential constructs encompassing people’s evaluations of different life domains (Veenhoven 2011). This complexity is reflected in the structure of the instruments used to assess well-being and quality of life, which are often composed by subscales investigating each life domain separately (Cummins 2000; Skevington et al. 2004).

The conceptual multiplicity that still dominates well-being research is problematic not only for researchers, but also for policy makers. Over a decade ago Veenhoven (2002) underscored the potential use of subjective indicators of well-being in designing social policy. Identifying people’s wants and needs - a key prerequisite for effective intervention - can be attained only by directly asking citizens to subjectively evaluate their present conditions, and to express their desires and expectations for the future. More recently, the World Health Organization (WHO 2013) has stressed the need for increasing the conceptual clarity of the term ‘well-being’ in order to best serve policy-making processes. Studies of the everyday meanings of this and related terms in local or national contexts might assist in developing sensitive and potentially more precise policy tools for assessing and improving citizens’ quality of life.

Notably, in the scientific literature well-being is rarely investigated per se, when conceptualized as a subjective experience. Many studies are instead focused on three related concepts: happiness, good life, and satisfaction with life, which are defined and operationalized in various and overlapping ways and sometimes used as interchangeable proxies for well-being. Moreover, in both basic and applied research these concepts are predominantly explored through scaled instruments that are surely useful for assessing their levels, but not the meaning that participants attribute to them. Only a small number of studies have investigated how laypersons’ definitions of terms related to well-being map onto the conceptual distinctions developed by philosophers and researchers (Delle Fave et al. 2011, 2016; Oishi et al. 2013). Moreover, there has been little discussion regarding whether or how such mapping could inform conceptual models and intervention strategies.

Based on these considerations, the present study aimed to investigate folk conceptions of happiness, good life and satisfaction in a Norwegian community sample. By analyzing responses to open-ended questions, we sought to assess the extent of differentiation or overlap between these concepts in everyday understanding, their conceptual heterogeneity, and possible socio-demographic predictors of their patterns of use. Our study thus aims at contributing to the conceptual clarification of three specific terms that we consider as related to, but not necessarily synonymous with, the broader notions of well-being and quality of life. The present findings may help refine the empirical approach to the study of well-being, which is often based on loosely defined concepts of uncertain cross-cultural validity.

Conceptual and Theoretical Approaches to Happiness, Good Life and Satisfaction

The current study examined everyday use of the three terms ‘happiness’, ‘good life’ and ‘satisfaction’. By ‘term’, we mean the words themselves. By ‘concept’, we refer to the shared cognitive organization or understanding of phenomena. First, we will briefly describe some ways in which the scientific literature has addressed and used the three terms under examination.

In the literature, the term ‘happiness’ is used in very different ways. It sometimes refers to affective aspects of well-being, in contrast to life satisfaction (Haller and Hadler 2006; Kahneman 1999). However, it is also used as a synonym for life satisfaction, which is conventionally understood as a cognitive evaluation (Veenhoven 2012). Although ‘life satisfaction’ and ‘happiness’ are semantically different terms, they are nevertheless often used interchangeably in the scientific literature (Frey 2008). Furthermore, ‘happiness’ is frequently used as a synonym for subjective well-being (SWB), which is conventionally understood as the combination of both affect (the prevalence of positive emotions relative to negative ones) and life satisfaction (Diener et al. 2010; Kashdan and Steger 2011). Some authors (e.g., Diener 2000) distinguish the colloquial usage of the term ‘happiness’ that covers both the affective and the evaluative aspects of SWB, from the more restricted usage within the SWB research tradition that refers only to affective dimensions. In contrast, other authors employ a much broader concept of happiness, beyond affect and satisfaction. For example, Easterlin (2003) equates welfare with happiness: “I take the terms happiness, utility, well-being, life satisfaction, and welfare to be interchangeable” (p. 11176). In the psychological tradition based on Aristotle’s concept of eudaimonia, some authors refer to ‘eudaimonic happiness’ (Maltby et al. 2005; see also Waterman 2008). The term has been assigned to a wide range of psychological dimensions. The lack of conceptual consensus has urged some authors to suggest that it may be best simply to discard altogether the use of ‘happiness’ as an academic term (Algoe et al. 2011).

A good life (or, sometimes, the good life) is an overarching concept often referred to in philosophical literature (Tiberius 2013). Any account of well-being will arguably draw on prevailing representations of what a good life amounts to. However, there is no consensus about the formal or defining features of the concept. The notion that something (e.g., a life) is good can be understood in terms of prudential, aesthetic, perfectionist or ethical value (Sumner 1996). In the well-being literature, the ‘good’ of a good life appears to be conceived of either with reference to prudential value (good for the person whose life it is) or perfectionist value (good compared to some standard of excellence or virtue). Linguistically, the usage of the term differs from ‘happiness’, because ‘good life’ explicitly refers to life as a whole and does not necessarily denote a psychological state.

The concept of satisfaction usually refers to the fulfilment of needs (e.g., Ryan and Deci 2000), desires (e.g., Davis 1981) or expectations (e.g., Sumner 1996), with the latter involving a comparison to some set of standards or ideals. However, theoretical discussion of this concept has been relatively neglected in the literature (Aspinal et al. 2003). In most studies, satisfaction is presumed to correspond to a specific evaluative judgment, of either life as a whole or specific life domains (Pavot and Diener 2008). In policy settings and large scale surveys, it is often assumed that perceived life satisfaction is a good proxy for overall well-being. Although satisfaction with life as a whole is sometimes construed as an aggregate of satisfaction across various domains, there is a lack of consensus concerning which specific domains should be included in such measures (Charlemagne-Badal et al. 2015). Satisfaction with life or with specific domains is often considered the cognitive component of SWB (Pavot and Diener 2008). However, evaluative judgments rely on not only cognition, but also affective experience. How affect and cognition jointly contribute to such evaluations is by no means a resolved issue (Ajzen 2005).

The Investigation of Everyday Understandings of Well-being

The above-mentioned conceptualizations of happiness, good life and satisfaction are drawn from the scholarly literature, and are largely theoretically or philosophically derived. In contrast, folk psychological understandings of these concepts are rarely investigated, although this endeavor should be a primary concern for psychology (Bruner 1990; Smedslund 2009). It is through folk psychological understandings that people find meaning, seek goals, motivate themselves for action and judge the extent to which they find their lives worthwhile. Indeed, folk psychological understandings are key constituents of culture. Reflecting on the well-being of oneself or others inescapably involves the use of culturally transmitted symbols and representation systems (Baumeister et al. 2013). Folk conceptions of well-being are not only descriptive but also potentially normative in the sense that they represent the ideals toward which individuals strive (Scollon and King 2011).

Most empirical studies conducted on everyday conceptions of well-being focus on antecedents, experiential dimensions, or outcomes rather than on the meaning that people – including academics – attribute to well-being by making use of various terms. In other words, the concepts of well-being are far less empirically studied than the phenomena related to well-being. Nonetheless, some researchers have explored the conceptual understanding or terminological usage of well-being across countries, either taking the general concept of well-being as their point of departure (Ryff 1989; Westerhof et al. 2001; Sastre 1999; McMahan and Estes 2012) or specifically investigating the concepts of happiness, good life or satisfaction, to which we now turn.

Studies of Happiness Conceptions

A number of studies have investigated folk conceptions of happiness by following rather different approaches. Some of these studies (Kim et al. 2007; Galati et al. 2006; Lu and Gilmour 2004; Pflug 2009; Delle Fave et al. 2011) have made use of open-ended questions, thus refraining from imposing researcher-defined conceptualizations onto participants. Several of these studies (Galati et al. 2006; Kim et al. 2007) have primarily investigated factors that are perceived as causing or leading to happiness. It can be argued that they did not address what kind of entity or experience happiness is perceived to be. Somewhat in contrast, another study explored the semantic-associative network of happiness, specifically investigating co-occurring emotional terms in 12 million web blogs (Mogilner et al. 2011). The findings showed that younger people primarily associated happiness with excitement, whereas older people connected happiness with feeling peaceful. Consequently, being ‘happy’ could correspond to different emotions across the life span. A recent study analyzed dictionary entries for happiness in 30 nations (Oishi et al. 2013). Stating that “the linguistic analysis of the term happiness is critical to advance psychological theory and the scientific understanding of well-being” (p. 559), the authors observed that efforts to measure happiness in different cultures in fact may be measuring different concepts.

Of direct relevance to the present work are previous studies that have examined everyday terminological understanding by employing fully open-ended question formats, e.g. “what is happiness for you?” (Delle Fave et al. 2011; Lu and Gilmour 2004; Pflug 2009; Uchida and Kitayama 2009). As participants themselves offer free descriptions of their understandings of happiness, this approach allows for the collection of a large corpus of information that can be subjected to both qualitative and quantitative analysis. Importantly, this approach enables participants to freely define happiness through either antecedents, substantive conceptual components, or consequences. Some studies have emphasized cultural differences as well as similarities in conceptualizations (Delle Fave et al. 2011; Oishi et al. 2013; see also a review by Uchida et al. 2004; Uchida and Kitayama 2009). A general finding emerging from these studies is that relationships appear to be almost universally cited components of happiness. Findings from a wide range of western countries indicated that the family environment was clearly the most prominent contextual component (Delle Fave et al. 2011). Furthermore, the findings in that study and those in a more recent one, conducted with the same procedure in twelve nations across five continents (Delle Fave et al. 2016), highlighted the prominence of inner balance and harmony as psychological ingredients of happiness across cultures. A comparative study conducted among US and Japanese participants highlighted that, although both groups included hedonic descriptions of happiness, Americans tended to associate hedonic experience with personal achievement, whereas Japanese were more likely to associate it with social harmony (Uchida and Kitayama 2009). Such studies facilitate a more nuanced view of the degree of universality vs. culture-specificity of conceptual components of happiness.

Our study also draws on the conceptual referent hypothesis (Rojas 2005), which highlights the importance of conceptual heterogeneity. This approach states that the meaning of a term, i.e., its conceptual referent, is not fixed, but will differ among persons or groups. A study based on this framework, conducted with a stratified sample of Mexican adults (Rojas 2005), showed that socio-demographic variables such as income, education and age were associated with the propensity to embrace certain conceptual referents of the term ‘happiness’. The eight referents investigated included stoicism, virtue, enjoyment, carpe diem, satisfaction, utopia, tranquility and fulfillment. However, they were not drawn from the participants’ descriptions, but from a researcher-generated analysis of the philosophical literature. This study additionally investigated whether embracing certain conceptual referents was associated with higher levels of self-reported happiness. No such pattern was found, although the carpe diem and utopian understandings of happiness were related to lower levels of perceived happiness.

Studies on Good Life Conceptions

Only a few empirical studies have focused specifically on the notion of a (or the) good life. In one study, convergence was found between folk understandings and the SWB tradition, in that happiness and meaningfulness were considered important, whereas monetary wealth was not (King and Napa 1998). In another study, folk concepts of a good life were investigated from the two perspectives of the desired life for oneself and a person’s life being considered morally good. Participants considered people who exert effort in their lives as leading morally better lives than those who do not. The role of effort as part of a desired good life depended in part on whether effort was seen as energy depletion or engagement (Scollon and King 2004).

Most investigations of good life notions also highlighted the importance of relationships. In a study conducted in China, India, Japan and Canada (Tafarodi et al. 2012), university students were asked by which criteria they would decide whether their lives had been worthy, according to their personal beliefs, at the imagined future age of 85. Across cultures, the most prominent category cited was social relationships, followed by practical or material achievement, emotional life, and virtues. A study of beliefs about a good life among mainland Chinese and Canadians of East Asian, South Asian and West European descent (Bonn and Tafarodi 2013) highlighted the primary importance across cultural or ethnic groups of close and enduring relationships. Similarly, students and adult community members in the United States reported fulfillment in relationships as a key aspect of a good life, while work fulfilment was considered less integral (Twenge and King 2005). Other studies (e.g., Scollon and Wirtz 2014) discussed how material wealth appears as part of good life descriptions in most cultures, although to a varying extent, and possibly fulfilling different psychological functions in collectivistic versus individualistic cultures (see also Kim et al. 2007).

Studies on Satisfaction Conceptions

As noted above, there is a lack of consensus regarding the conceptual meaning of satisfaction. This state of affairs appears to have spurred less interest compared with the debate around the happiness concept. The few empirical studies on the conceptual understanding or usage of the term ‘satisfaction’ concern particular roles and life domains, such as patient satisfaction (Collins and Nicolson 2002).

Concerning life satisfaction, a study conducted in Norway and Greenland explored how participants understood the conceptual meaning of the five items composing the Satisfaction With Life Scale (Vittersø et al. 2005). Based on item response modeling, the study highlighted divergences between the two countries as well as across latent classes of participants within each country. However, this study addressed participants’ understanding of researcher-defined operationalizations of life satisfaction, rather than the everyday term. More recently, a study contrasting everyday understandings of ‘satisfaction’ and ‘harmony’ showed that satisfaction associations tapped particularly into general aspects of well-being (contentment and pleasure), fulfilment and gratification, satisfaction of external needs, and self-enhancing and achievement-oriented themes (Kjell et al. 2016).

To summarize, existing studies of conceptions of happiness suggest that people’s definitions include both the psychological sphere (mental states and experiences) and the external life domains. The psychological components comprise not only hedonic or affect-oriented elements, but also harmony and eudaimonic dimensions related to personal growth and self-actualization. Among the external life domains, family and social relationships stand out in almost all studies. The extent to which the definitional components are endorsed varies across cultures and demographic groups (particularly with regard to age). Conceptions of a good life have been less studied, but the existing evidence indicates that the relationship realm is central for this concept also. Empirical studies of how people define satisfaction are rare.

Aims of the Present Study

Considering the lack of consensus regarding the definitions of happiness, (life) satisfaction and a good life, the present study primarily aims at identifying the everyday meaning potential of the terms ‘happiness’, ‘good life’ and ‘satisfaction’ in a Norwegian context, since professional language and conceptual frameworks ultimately develop from everyday meaning systems. We expected that the conceptual structure of these three terms would largely reflect previous findings. For each of the terms, we therefore expected the usage to include both external life domains and psychological states or experiences, the latter encompassing hedonic and eudaimonic components. Among the life domains, we expected relationship themes to be particularly prevalent, in addition to work, leisure and material aspects of life.

Furthermore, we set forth hypotheses about the extent of conceptual differentiation or overlap between ‘happiness’ and the two other terms. Based on the psychological literature, one might expect the concept of good life to be substantively similar to happiness, as the terms are sometimes used interchangeably. However, from the philosophical perspectives briefly outlined above, one might instead expect the everyday concept of happiness to be understood first and foremost as referring to the inner world (endorsing mental states or experiences), and the concept of a good life to contain more contextual features. Furthermore, a good life could be expected to represent a broader concept than happiness, arguably referring to welfare or well-being in the broadest sense, encompassing external life domains as well as internal psychological states (cf. Sumner 1996; Brülde 2007). Accordingly, we expected happiness to more likely be categorized as a psychological concept, compared to good life. Additionally, we hypothesized that among the external domains, standard of living would be more frequently mentioned as part of good life than of happiness, because of the possible material connotations of the term ‘good life’, and the semantic similarity of ‘standard of living’ and ‘good life’ in the Norwegian language. As satisfaction can be primarily seen as a mental state (Sumner 1996), we expected this term to be more likely categorized as a psychological concept, compared to happiness. Furthermore, we expected satisfaction to be more likely categorized as hedonic rather than eudaimonic, which would be consistent with common practice in psychological literature (Huta and Waterman 2014).

The second aim was to investigate the degree of conceptual heterogeneity of each of the three terms. Theory suggests that well-being terms, particularly ‘happiness’, can refer to several quite distinct phenomena. One of the most cited distinctions in the literature is between happiness as a psychological state, and happiness as a more general condition of flourishing or well-being (Haybron 2008). It is therefore of interest to apply structural equation modeling to identify distinct groups or classes of participants who might hold similar understandings of each term. In line with the reviewed literature, we expected to find at least two main classes of understandings of happiness, one of which would be more psychologically oriented, and the other more contextually oriented. Although theoretical and empirical explications of good life and satisfaction are less developed, we tentatively expected a similar distinction between psychological and contextual response classes for these two terms.

Finally, as demographic and cultural variations were detected in the empirical literature with regard to conceptual definitions, the third aim was to examine socio-demographic predictors of these suggested classes.

Method

Participants

The sample comprised 500 Norwegian adults. The mean age was 44.7 (SD = 15.2), with a range of 17–81, and 58 % of the sample was female while 42 % was male. Among these participants, the majority (76 %) were working, 60 % reported education beyond secondary school, and 65 % reported living with a partner. Structural differences between happiness and good life concepts were investigated among 230 participants (subsample 1), while the differences between happiness and satisfaction concepts were investigated among the remaining 270 participants (subsample 2). There were no significant differences in the demographic features between the two subsamples.

Participants were recruited through face-to-face interaction. Most were contacted in public spaces such as stations, shopping centers, or car license renewal offices, whereas others were contacted at selected workplaces. Participants were not offered monetary or other incentives.

Data protection approval was obtained from the Norwegian Social Service Data Services for both paper and electronic data. Informed consent was obtained from the participants. Any information that potentially could identify a participant was deleted, and questionnaires were thus handled anonymously in the process of coding and analysis.

Materials

Participants were asked to respond to the following question: “What is happiness for you? Take your time and provide your definition”. This question was part of the Eudaimonic and Hedonic Happiness Investigation instrument (EHHI, Delle Fave et al. 2011), which was administered to participants within a wider international survey (Delle Fave et al. 2016). The questionnaire pack submitted to the Norwegian participants also included the question “What is a good life for you?” (subsample 1) or “What is satisfaction for you?” (subsample 2). All participants also provided demographic information.

The three open-ended questions on happiness, good life and satisfaction were designed to encourage subjective description in two senses. First, they invited people to present their own ideas. Second, the phrasing for you suggested that the focus was on the participant’s own happiness, good life, and satisfaction. We deliberately used the indefinite form “a good life” rather than “the good life”, as the latter has connotations to luxury and affluence in the Norwegian language. In line with the most commonly used translations in local well-being research, the terms used in Norwegian were lykke (happiness), tilfredshet (satisfaction) and et godt liv (a good life).

Procedure

Data were collected as an extension of the second wave of the EHHI (see Delle Fave et al. 2011, 2013a, b, 2016). The EHHI made use of open-ended questions administered to a sample stratified by age, gender and education level for the 30–60 age range (n = 216) in each participating country. In Norway, the sample was extended in order to include a larger number of participants belonging to a broader age range. The current study relies on this extended sample and is therefore partially stratified. The paper-and-pencil questionnaire included a web address to an online version, which was used by 5.5 % of the participants. Recruiting people with high school education only (especially men, and elderly women) was considerably harder than expected. For the purpose of the present study, the sample was restricted to participants who provided definitions of both ‘happiness’ and either ‘satisfaction’ or ‘good life’ (N = 500).

The open-ended questions were classified using the standardized coding system developed by the EHHI (Delle Fave et al. 2011), based in part on pilot studies and in part on the World Health Organization Quality of Life Assessment (WHOQOL 1998). Codes allow quantification of qualitative answers by partitioning each answer into semantic units referring to different aspects, which were coded separately. As some participants provided multifaceted and elaborate descriptions of happiness, good life and satisfaction, a maximum of nine codes were retained for each description. Although the coding strategy aimed at preserving fine-grained answer units, a general distinction was made between answers that were domain-related and answers referring to psychological dimensions of the three concepts. The life domains comprised work, family, standard of living, interpersonal relations, health, leisure, spirituality/religion, society and community issues, and education. An additional domain, labeled as “life in general” included answers that referred to life as a whole (e.g., to have a good quality of life”) or to unspecified situations (e.g., “a good day” or “a positive environment”). The answers referring to psychological dimensions comprised both hedonic definitions (related to pleasure, satisfaction, and the absence of negative feelings) and eudaimonic definitions (referring to growth, purpose, mastery, autonomy, self-actualization, meaning, harmony, awareness, optimism, and inner states of well-being). For the purpose of the present study, codes were transposed into a dichotomous variable for each domain or dimension as follows: A “0” indicated that it was not mentioned by the participant, and a “1” indicated that it was mentioned at least once.

From the Norwegian sample, 58 responses were selected for an inter-rater reliability analysis (Neuendorf 2002) of the coding scheme. Percent agreement and kappa coefficients were calculated for each of the life domain and psychological variables (mentioned vs. non-mentioned). The overall psychological (internal) set of hedonic and eudaimonic definitions showed an excellent kappa value of .85. All specific life domains (external) showed good kappa values (over .70). The life in general domain showed a kappa level of .43. This value falls into the moderate range as defined by Altman (1991) and Landis and Koch (1977), and the category was retained for analysis. The raters’ pattern of classification of two psychological dimensions - optimism and inner states of well-being - showed low kappa values, suggesting that these particular analytic terms do not clearly differentiate between hedonic and eudaimonic aspects of well-being and might contain elements of both. These two dimensions were thus excluded when categorizing items as belonging to either the hedonic or eudaimonic group of definitions. However, as optimism and inner states of well-being clearly refer to psychological dimensions, they were retained when categorizing items as psychological versus belonging to the life domains. The life domain of education was discarded from further analysis based on the low number of answers.

Statistical Analysis

To determine whether the participants referred to structurally different conceptual properties when defining ‘good life’ or ‘satisfaction’ as compared to when they defined ‘happiness’, the McNemar test for differences in dependent proportions was used. Latent class analysis was applied to examine conceptual heterogeneity of each of the three terms using the statistical software Mplus version 6.1 (Geiser 2013; Muthén and Muthén 2010). Latent class analysis allows the identification of latent homogenous subgroups of observations, based on observed dichotomous response variables (Goodman 2002). If more than one latent group is detected, the sample is characterized by latent heterogeneity (Eid et al. 2003). Latent class analysis is particularly useful for situations in which relatively distinct entities or clusters (e.g., conceptually separate word meanings) might underlie the observed data. In other words, latent class analysis can detect the extent to which particular conceptual meanings typically appear together in the participants’ descriptions. To determine the number of classes that best fit the data, the sample-size adjusted Bayesian information criterion (SS-BIC), as well as the bootstrap likelihood ratio test (BLRT), were used. For each added class, lower SS-BIC values reflect an improved fit, whereas a non-significant p value for the BLRT indicates that the model does not fit the data significantly better than a more parsimonious model with one less class. Although SS-BIC and the BLRT have been identified as the best performing indices (Nylund et al. 2007), we also included the Vuong-Lo-Mendell-Rubin likelihood ratio test (VLMR) and the Lo-Mendell-Rubin adjusted likelihood ratio test (LMRT) as these are commonly reported in studies utilizing latent class analysis. Model estimation (Geiser 2013) was based on running 1000 random sets of start values and selecting the 100 sets with the largest log-likelihood for second stage optimizations, to avoid inaccurate parameter estimates. Initial stage iterations were set to 100.

For examining socio-demographic predictors of the suggested latent classes, a pseudo-class draw approach was used (see e.g., Bray et al. 2015; Clark and Muthén 2009; Muthén and Muthén 2010). The main advantage of this approach is that it takes the uncertainty in the latent class assignment into account when relating latent classes to external variables. Based upon differences in the individuals’ probability of class assignment, individuals are randomly assigned to one of the latent classes multiple times (20 draws was the default). Thus, individuals are allowed to change to neighboring classes at a rate specified by their posterior probabilities. The socio-demographic variables were defined as “auxiliary” variables when the latent class analysis was performed (Muthén and Muthén 2010). This means that the socio-demographic variables were not included when the latent classes were estimated, but instead they were subsequently included as covariates in a multinomial logistic regression model with latent class membership as the dependent variable. This multinomial logistic analysis was repeated 20 times (once for each draw), and the results presented are based upon the combined number of draws.

Results

Structural Properties of Concepts

As illustrated in Table 1, significant variation of conceptual meaning was found across the three terms. In the comparison of ‘happiness’ and ‘good life’ (subsample 1), standard of living was significantly more likely to be mentioned as part of good life definitions. The same finding was detected for leisure, community and society, and to a lesser degree, health. The difference for the unspecific domain “life in general” was non-significant. A large majority of participants reported at least one external life domain in their descriptions of both ‘happiness’ and ‘good life’. Over half of the participants referred to at least one inner, psychological dimension in describing both ‘happiness’ and ‘good life’.

Table 1 Domain endorsementa for each of the three concepts
Full size table

In the comparison between happiness and satisfaction (subsample 2), a large majority of participants included at least one external life domain in the description of ‘happiness’, while a significantly smaller proportion did so for ‘satisfaction’. Around two-thirds of the participants referred to at least one psychological dimension in their descriptions of ‘happiness’, and a significantly higher percentage did so when referring to ‘satisfaction’. Furthermore, hedonic definitions were significantly more often mentioned for ‘satisfaction’ compared to ‘happiness’. As Table 1 shows, the relational domains (family and interpersonal relationships) were significantly less endorsed within satisfaction descriptions than within happiness descriptions.

On an individual level, we also examined the extent to which participants retained or changed their descriptions, with regard to the proportion of psychological answers, external domain answers and mixed answers. Whereas 43 % of the participants changed the balance of these components when describing good life compared to describing happiness, 53.3 % did so when describing satisfaction compared to happiness. The proportions were significantly different (χ 2 (1) = 5.264, p = .022). People with partners were significantly more likely (p = .045) to change the balance when defining satisfaction compared to happiness. No other significant associations between socio-demographic variables and the tendency to change the balance of descriptions were detected.

Latent Class Analysis

Happiness

As shown in Table 2, a comparison of the SS-BIC and BLRT results for the various solutions indicated that a model containing three classes provided the best fit when examining the conceptual heterogeneity of the term ‘happiness’. In contrast, the VLMR and the adjusted LMRT values favored a two-class model. However, as these two tests are considered less reliable than the SS-BIC and the BLRT (Nylund et al. 2007), and due to the near-significance of the three-class solution for the VLMR and the LMRT, we retained the three-class model.

Table 2 Model fit and classification quality statistics for each of the three concepts
Full size table

Based on estimated class size, 24.7 % of the participants belonged to Class 1. This class was characterized by a highFootnote 1 probability of mentioning both hedonic and eudaimonic definitions, an intermediate probability of mentioning interpersonal relationships and life in general, and a low probability of mentioning any other life domains. Class 2, comprising 44.6 % of the participants was characterized by a very high probability of mentioning family, high probabilities of mentioning the interpersonal and health domains, intermediate probabilities of reporting most other specific life domains and eudaimonic definitions, and low probabilities of including hedonic definitions and life in general. The remaining 30.8 % of the participants were characterized by a high probability of mentioning interpersonal relationships and eudaimonic definitions, intermediate probability of family, standard of living, health, leisure and hedonic definitions, and low probabilities of the other life domains (see Fig. 1).

Fig. 1
figure 1

Descriptions of happiness: Latent class analysis. Conditional probabilities of endorsing external domains and psychological dimensions

Full size image

A further argument for retaining a three-class solution is that the classes were evaluated as theoretically meaningful, supporting the expected separation between psychologically oriented vs. contextually oriented understandings. Class 1, specifically characterized by reference to psychological dimensions, was labeled as “mostly inner states” (MIS). In clear contrast, Class 2 focused on contextually oriented domains, especially family, and was named “family and life domains” (FLD). Class 3 was similar to Class 1 in that psychological dimensions were predominant, but life domains were also mentioned to a greater extent than in Class 1. Moreover, in contrast to Class 2, interpersonal relationships rather than family were particularly endorsed. We thus labeled Class 3 as “inner states and friends” (ISF).

The entropy level was rather low, thus raising questions about the overall classification quality of the model. The average latent class assignment probability diagonal values (Geiser 2013) for most likely latent class membership were deemed as good (.84) for the contextually oriented Class 2, but less acceptable for the psychologically oriented Classes 1 and 3 (.74 and .77, respectively). The most likely latent class membership for an individual is based on the posterior probabilities of belonging to a certain latent class. In particular, the low entropy appears to be caused by members of Class 1 showing a high probability of belonging to Class 3 instead.

Good life

For the term ‘good life’, a similar latent class analysis was conducted with SS-BIC and BLRT as the main criteria for determining the number of classes (see Table 2). Here, the information criteria diverged in that SS-BIC indicated a three-class solution and the BLRT pointed toward a two-class solution. Due to reported performance limitations of the SS-BIC at lower sample sizes (Nylund et al. 2007), we decided to rely mainly on the BLRT. Accordingly, a two-class model for ‘good life’ was favored. VLMR and LMRT were inconclusive, as none of these values indicated a clear cutoff point for determining the number of classes.

Based on estimated class size, 74.9 % of participants belonged to Class 1, and 25.1 % to Class 2. The interpretation of the profiles for the two classes supported the expected separation between contextual and psychological understandings of ‘good life’. As shown in Fig. 2, Class 1 was characterized by a high probability of mentioning contextual domains such as family, standard of living and interpersonal relationships, and intermediate mentioning of work, leisure, health as well as hedonic and eudaimonic definitions. The considerably smaller Class 2 included more psychologically oriented understandings. In particular, this class was characterized by a high probability of providing eudaimonic definitions, an intermediate probability of mentioning life in general and hedonic definitions, and a low probability of mentioning any specific life domain. We thus labeled the larger Class 1 “family and life domains” (FLD) and the smaller Class 2 “mostly inner states” (MIS), respectively. For classification purposes, entropy was deemed as satisfactory, and the average latent class probability diagonals (.95 and .85 for Classes 1 and 2) as good.

Fig. 2
figure 2

Descriptions of good life: Latent class analysis. Conditional probabilities of endorsing external domains and psychological dimensions

Full size image

Satisfaction

With regard to ‘satisfaction’, SS-BIC comparisons as well as the BLRT tests indicated the best fit through a two-class model (see Table 2). This solution was also supported by the VLMR and LMRT values. Regarding classification quality, the average latent class probability diagonals showed good values (.91 and .87). The classes were of similar size. Based on estimated class proportions, 45.4 % of the participants belonged to Class 1, and 54.6 % to Class 2. As reported in Fig. 3, contextual understandings of satisfaction were predominant in Class 1. This class was characterized by a high endorsement of family, as well as intermediate endorsement of hedonic and eudaimonic definitions and all other external domains except community and society. Class 2 was characterized by a more psychological orientation, showing a low endorsement of all specific life domains, intermediate endorsement of life in general, and high endorsement of hedonic and eudaimonic definitions. In other words, the two classes differed concerning the description of satisfaction, either as a mixed psychological/contextual term with a particular emphasis on family, or as a term almost entirely referring to life in general and to psychological dimensions. We thus named Class 1 “Family, life domains and inner states” (FLI) and Class 2 “mostly inner states” (MIS). Although the entropy level was only borderline acceptable, the fit indexes precluded running an alternative three-class solution. The average latent class probability diagonals (.91 and .87 for Classes 1 and 2 respectively) were deemed as good.

Fig. 3
figure 3

Descriptions of satisfaction: Latent class analysis. Conditional probabilities of endorsing external domains and psychological dimensions

Full size image

Socio-Demographic Predictors

After determining the latent classes, we proceeded to examine how gender, age, education and living with a partner predicted class membership. Table 3 presents the proportions of participants belonging to the various socio-demographic groups, by the assigned latent class.

Table 3 Proportions of participants belonging to demographic groups, by latent class
Full size table

Happiness

As shown in Table 4, multinomial logistic regression analyses with pseudo-class draws indicated that the latent class assignment for ‘happiness’ was significantly associated with age group and living with a partner. Using Class 1 (MIS) as the reference category, Class 2 (FLD) included a significantly higher proportion of participants in the middle-aged and older age groups, as well as people living with partners. Additional analysis further revealed that class 2 (FLD) included a significantly higher proportion of both middle-aged (b = 0.70, p < .05) and older participants (b = 1.04, p < .01) compared to class 3 (ISF). There was no significant association between class assignment and gender or education.

Table 4 Bivariate multinomial regression using pseudo-class draws. Parameter estimates with standard errors in parentheses
Full size table

Good Life

Latent class assignment for ‘good life’ was significantly associated with gender, age group, and having a partner (see Table 4). Women, people who were middle-aged and older, and people with a partner were significantly more represented in Class 1 (FLD) compared to Class 2 (MIS), which was the reference category. Similar to the happiness concept, we found no significant association between class assignment and education.

Satisfaction

For the term ‘satisfaction’, latent class assignment was significantly associated with age group (Table 4). A significantly higher percentage of older participants (aged 56 and above) were assigned to Class 1 (FLI), compared to Class 2 (MIS). No significant associations were found between class assignment and gender, education or having a partner.

Finally, for each of the three terms, analyses were re-estimated as multivariate multinomial regression analyses, yielding only trivial changes in parameter estimates. The only difference compared to bivariate analyses was detected for ‘happiness’, for which the association between latent class and living with a partner became non-significant (p = .16).

Discussion

The findings of the present study highlight that Norwegian folk conceptualizations of happiness, satisfaction and a good life are not limited to affective or evaluative internal dimensions. As expected, a common feature of the three concepts is that they were described through a broader range of both psychological and contextual components. Furthermore, the extent to which the different components were endorsed varied across the concepts. While approximately 90 % of the participants included life domains in their definitions of happiness and good life, only approximately 70 % referred to them to define satisfaction. On the other hand, over 75 % of the participants mentioned psychological dimensions in satisfaction definitions, whereas the corresponding proportion was lower for happiness definitions, although still endorsed by a majority of the participants. Relational components (family and interpersonal relationships) were the most frequent topics among the external life domains for all three concepts.

Thus, on the conceptual level, the analysis highlighted important similarities as well as clear differences between the understandings of happiness, good life and satisfaction. Good life appears to be the most broadly encompassing concept. While in scholarly literature this term is sometimes used to denote well-being in a general sense, the present study indicates that its domain characteristics in everyday language are relatively similar to those associated with happiness. Contrary to expectations, the overall proportions of internal (psychological) vs. external (life domain) components were not significantly different for happiness and good life. Similarly, the proportions of hedonic and eudaimonic definitions were not significantly different between the two concepts. In this sense, happiness and good life share a similar conceptual structure, at least among Norwegians. Furthermore, on the individual level, the majority of participants did not change the balance of psychological, external domain or mixed descriptions when defining good life compared to defining happiness. However, despite these similarities, happiness and good life were not synonymously understood across participants, as the concept-specific distributions of external life domains differed significantly. As hypothesized, the major difference concerned standard of living, which was given greater importance in good life conceptions. Thus, compared to happiness, the concept a good life shows stronger connotations towards living conditions and material aspects of life, and is also more inclusive of health, leisure and community and society issues.

Consistent with our expectations, the everyday concept of satisfaction contained more psychological connotations than the concept of happiness. Most life domains were significantly less endorsed in satisfaction descriptions, and this was particularly evident for the relational domains. Importantly, and consistent with expectations, satisfaction was understood as a significantly more hedonic concept than happiness in a Norwegian context. Furthermore, a majority of participants changed the balance of psychological, external domain or mixed definitions when turning to satisfaction, after defining happiness. These findings suggest that treating happiness and satisfaction as largely synonymous concepts prevents from capturing their connotative differences in everyday language.

For each of the three concepts, latent class analysis was used to classify participants into homogenous subgroups based on the observed pattern of responses. The analysis showed that happiness was best described with a three-class solution, and the other two concepts with two-class solutions. The three-class solution of happiness should be interpreted with some care, because of classification uncertainty. Thus, it is premature to conclude that the happiness concept includes greater variability than the other two concepts, although it is a possibility. In any case, the latent patterns for each concept were quite similar. Within all three concepts, one class primarily referred to external life domains, and the other class (two in the case of happiness) prominently included psychological dimensions. This distinction corroborated theoretically grounded expectations. For happiness, however, an additional Class 3 emerged as a hybrid class including interpersonal relationships in addition to psychological dimensions. Among the external domains, relationships (either family or interpersonal connections) were predominant. For each class, we found significant socio-demographic predictors, despite problems of low entropy for two of the concepts. For all three concepts, younger participants showed a greater propensity for endorsing more psychological definitions than older participants, who were instead more likely to endorse external life domains.

Many authors consider happiness to be a largely psychological concept, as contrasted to the more general concept of well-being (cf. Haybron 2008; Tiberius 2013). Therefore, it is somewhat surprising that a large class of happiness descriptions (FLD) endorsed hedonic or eudaimonic aspects with a lower probability than any of the classes identified for good life or satisfaction. This highly contextualized, life domain-endorsing pattern of descriptions raises some important questions. First and foremost, it suggests that laypeople, at least in Norway, do not necessarily have a largely psychological concept in mind when they talk about happiness. This interpretation is consistent with findings obtained in an international study conducted with the EHHI, highlighting that participants belonging to the Germanic linguistic group emphasized domain-related definitions of happiness to a greater extent than participants belonging to the Romance linguistic group (Delle Fave et al. 2013b) as well as in a more recent study including a Norwegian sample (Delle Fave et al. 2016). As an alternative interpretation, it might be argued that people confound happiness (as an inner state) with its antecedents (as found in external life domains). However, if this is the case, why should people place such large emphasis on the life domains? In other words, why do the objects of affective or cognitive appraisal show such a strong presence in many people’s descriptions? This finding raises the intriguing possibility that subjective experiences of happiness, good life or satisfaction are not experientially independent of the life domains that produced them. For example, happiness as experienced in leisure might be qualitatively different from happiness experienced in relationships.

Furthermore, and notably, demographic groups showed different preferences for domain inclusion. For all three terms, older participants were more likely to include external life domains in their descriptions. A possible implication of this finding is that experiences of well-being are contingent on the developmental stage of the individual. However, whether increasing age is associated with “turning inward” or “turning outward” is an unresolved issue in the research literature (Dörner et al. 2005). Alternatively, in light of recent historical trends toward increased individualism, younger people may be more likely to describe well-being in terms of functioning of the self (Westerhof et al. 2001; Nafstad et al. 2007).

In addition to illustrating variation in conceptual understanding across terms and participant groups, these findings raise a more fundamental point: Concepts are by necessity at the core of research practices, including exploration, hypothesis generation, analysis, and dissemination. Although sophisticated methods are applied for measurement and analysis, well-being related constructs carry wide meaning potentials and surplus meaning (Rommetveit 2003), and there is no final consensus about what they supposedly refer to. If common understandings are historically and culturally contingent, and differ across socio-demographic groups, it may be futile to seek an ultimately valid or final operationalization for each concept. Instead, more work should be directed to conceptual analysis, including longitudinal and cross-cultural studies.

Some scholars have argued that the understanding of psychological terms cannot be dichotomously categorized as “lay” and “expert”, because scientific conceptualizations feed into everyday ones, and vice versa, in complex ways (Shaw 2002; Wagner 2007). For instance, both everyday and medical-scientific conceptualizations of depression have changed congruently across time and cultures (Ehrenberg 2010; Healy 2004), and the propensity of individuals to endorse biomedical rather than contextual or “fate” understandings of depression has been shown to vary across socio-demographic groups (Lawrence et al. 2006). Such potentially shifting patterns of concept usage and development pose challenges to quantitative research in particular, because the use of quantitative methods requires some degree of construct stability across people, cultures and time to ensure consistent observation and analysis. Further work is therefore required to gain a deeper understanding of the dynamic, bidirectional and culturally contingent influences between everyday and professional language of well-being. Such conceptual endeavors are of particular relevance for applied settings, such as in social planning and healthcare, involving joint efforts by practitioners and service users.

In this regard, the findings of the present study point to the importance of participant co-authorship (Hermans and Kempen 1993; Bruner 1990) in well-being research, in view of two potentially useful outcomes. First, functional communication of research findings may benefit from more elaborate understanding of how concepts related to well-being are understood and used in everyday language within a given culture. Second, and more fundamentally, everyday language may contain folk psychological knowledge that can inform science, including conceptual and methodological development (Bruner 1990; Billig 2011). The present analysis illustrates how concepts such as happiness, good life, and satisfaction do not necessarily correspond to unambiguous, clearly delineated objects. They are tools to grasp complex psychological phenomena, relational functioning and people’s interaction with the social and material world. Sometimes, conceptual tools are too large and crude for the finely grained research functions we expect them to have. Therefore, for the sake of scientific precision, we agree with Algoe et al. (2011) that the science of well-being may benefit from unpacking wide, elusive terms into their assumed constituent parts. With regard to the development of research instruments, recommendations following from the present study include careful choice of survey wording in addition to including alternative items to capture various nuances of meaning.

Limitations and Future Directions

We have consistently used the English terms ‘happiness’, ‘satisfaction’, and ‘good life’ throughout this paper. This linguistic presentation might mask important concerns regarding translation. Although we adhere to common translational practice of well-being research in Norway, we cannot take for granted that the Norwegian words lykke, godt liv, and tilfredshet precisely mirror their English counterparts (Oishi et al. 2013; Wierzbicka 2009). Strictly speaking, the inferences we draw are valid only in a Norwegian language setting, and further research is required to investigate whether similar conclusions can be drawn in other cultural contexts.

It should also be noted that we requested participants to define ‘satisfaction’ as a general term, rather than ‘satisfaction with life’. The context of the questionnaire, in which the word ‘life’ appears in several instances, arguably invites participants to understand the term in the context of overall life. Future research might shed light on how the understanding of ‘satisfaction’ in general relates to the more specific understanding of ‘satisfaction with life’ as well as to satisfaction in specific domains.

The current study required participants first to define ‘happiness’, and thereafter to define ‘good life’ or ‘satisfaction’. Therefore, participants may have been primed by their own happiness conceptualizations in the description of the other terms. Ideally, we should have extended the study to include alternating orders of questions. However, as considerable definitional differences emerged between ‘happiness’ and the other two terms, furthermore pointing in different directions for ‘good life’ and ‘satisfaction’, we believe that our findings are robust enough to warrant the conclusions put forth. Further research may probe deeper into how order effects may influence subjective definitions.

Indicators such as the SS-BIC and BLRT have been found to perform well in identifying the number of latent classes (Nylund et al. 2007). It is nevertheless important to note that, as class assignment is based upon probability, some individuals will endorse the core features of the class more closely than others assigned to the same class, and some individuals could share some features with other classes. The classes identified should therefore be considered a heuristic rather than a “true” representation of the heterogeneity across the individuals studied (Lanza and Rhoades 2013; Geiser 2013).

Among the demographic findings, age group was as a consistently important predictor of conceptual understanding. As the present study is cross-sectional, we cannot disentangle developmental from age-cohort effects. Studies based on longitudinal designs are required to better explore these issues (Westerhof et al. 2001; McMahan and Estes 2012).

Social scientists as well as policymakers and international organizations have emphasized the need for increasing conceptual clarity in well-being research. The present study has highlighted that, at least in the Norwegian context, the terms ‘happiness’, ‘good life’ and ‘satisfaction’ do not refer to the same thing, and may mean different things to different people. This evidence suggests that investigating everyday understandings of well-being terms can provide valuable insights both for scientific conceptual development, and for purposes of dissemination.