Abstract
Theories and research on the relationships between values and well-being have predominantly focused on how values may affect well-being, with less attention to the reverse direction—how well-being may shape values. We present three alternative theoretical perspectives on this latter direction: well-being as an activator, reinforcer, and operator of value change. Using diverse measures of well-being and personal values, we conducted three longitudinal studies centered on the same life transition—entering university—across distinct samples from Mainland China (N = 218), Hong Kong (N = 252), and the United Kingdom (N = 196). Results from parallel process latent growth models (LGMs) showed more substantial evidence for well-being predicting later value change than the reverse. Specifically, well-being predicted later value changes in three cases: Initial life satisfaction predicted increases in the openness to change value dimension in Study 1, and initial self-esteem predicted increases in both openness to change and self-enhancement value dimensions in Study 3. Values predicted later well-being changes only once: Initial self-enhancement value dimension predicted increases in positive affect in Study 2. These results illuminate the intricate nature of the relationships between well-being and values, highlighting well-being’s role as a catalyst for value development and the importance of contextual factors during life transitions.
Plain language summary
Have you ever wondered whether your well-being influences what you value in life—or whether your values shape your well-being? We explored this by tracking first year university students in three regions: Mainland China, Hong Kong, and the United Kingdom. After they began their studies, we repeatedly measured each student’s life satisfaction, positive emotions, self-esteem, and the importance they placed on values such as openness to new experiences and personal success. Our findings show that initial well-being more often predicts later changes in values than vice versa. For example, students who started with higher life satisfaction later placed greater value on change, and those with higher self-esteem came to prioritize both openness to change and personal success. In contrast, values predicted subsequent well-being only once: students who initially emphasized personal success reported increased positive emotions over time. These results suggest that supporting students’ well-being early in their university transition can foster values that aid their adjustment to campus life. Moreover, societal contexts appear to shape the interplay between well-being and values during this critical period.
Introduction
Recent research on personality dynamics indicates that, alongside stable individual differences, people undergo meaningful intraindividual changes in thoughts, feelings, motivations, and behaviors across situations and over time (Kuper et al., 2021). To uncover the processes driving this variability, researchers must consider individual traits alongside the temporal and situational factors that lead to context-specific responses and outcomes (Mõttus et al., 2020). Longitudinal studies centered on distinct life events are particularly well suited to reveal these adaptive mechanisms and their consequences (e.g., Bardi et al., 2014; Bleidorn et al., 2016; Luhmann et al., 2012; Scheppingen et al., 2018). However, because a given transition can unfold differently across cultural settings, it is essential to distinguish which processes generalize and which depend on societal context—especially in light of critiques about WEIRD (Western, Educated, Industrialized, Rich, Democratic) samples (Henrich et al., 2010). Therefore, the present research investigates the reciprocal longitudinal links between personal values and well-being via three parallel longitudinal studies of the same life transition—entry into university—conducted in Mainland China, Hong Kong, and the United Kingdom to assess both robustness and context specificity.
As fundamental beliefs that guide human behavior, personal values are generally stable but can shift in response to life changes (e.g., Bardi & Goodwin, 2011; Daniel et al., 2013, 2022; Lönnqvist et al., 2011; Sortheix et al., 2017). Research has demonstrated that group-level values can change meaningfully in response to major societal events or transformations (Daniel et al., 2022; Danis et al., 2011; Sortheix et al., 2017; Verkasalo et al., 2006). At the individual level, values have also been shown to change systematically across different life stages and situational contexts (e.g., Bardi et al., 2009; Milfont et al., 2016; Vecchione, Döring, et al., 2016; Ye & Ng, 2019). Gaining a deeper understanding of how such value changes relate to well-being is crucial for advancing our knowledge of personality processes.
Building on early cross-sectional research linking personal values to well-being (e.g., Bobowik et al., 2011; Joshanloo & Ghaedi, 2009; Sagiv & Schwartz, 2000), scholars have developed theoretical models to explain the relationship between these constructs (Sagiv et al., 2004, 2015; Schwartz & Sortheix, 2018; Sortheix & Schwartz, 2017). However, this theorizing has primarily emphasized pathways from values to well-being—a direction that has received limited and inconsistent support in recent longitudinal studies (e.g., Fetvadjiev & He, 2019; Grosz et al., 2021). In the present research, we propose three theoretical rationales for the reverse direction: how well-being may influence the development and prioritization of values. We also identify key issues in the existing longitudinal literature on value change and test our hypotheses across three longitudinal studies conducted in both Eastern and Western contexts—specifically, in Mainland China, Hong Kong, and the United Kingdom. We begin by outlining the structure and dynamics of personal values, followed by a review of theoretical perspectives on the bidirectional pathways between values and well-being.
The structure of values and their change
Values are broad life goals (e.g., achievement, tradition) that are important to people. They serve as guiding principles that influence a wide range of perceptions, goals, attitudes, and behaviors (Bardi et al., 2008; Boer & Fischer, 2013; Sagiv & Schwartz, 2022). According to Schwartz’s (1992) value theory, basic values are organized along a circular structure, where the distances between pairs of values represent motivational compatibilities or conflicts. Neighboring values are based on compatible motivational goals and are positively associated, while opposite values stem from conflicting motivations and are negatively associated. Furthermore, the circle of values can be divided into four higher-order categories, organized along two dimensions: (1) openness to change vs. conservation and (2) self-enhancement vs. self-transcendence. On the first dimension, values of openness to change (self-direction, stimulation, and often hedonism) motivate individuals to embrace novel ideas and behaviors, whereas values of conservation (security, conformity, and tradition) encourage the maintenance of social order and stability. On the second dimension, values of self-enhancement (achievement, power, and sometimes hedonism) drive individuals to pursue personal success, while values of self-transcendence (benevolence and universalism) motivate the pursuit of the welfare of others, both close and distant others 1 .
Past research has shown that values can change in response to major life events, such as financial crises (Sortheix et al., 2017), terrorist attacks (Verkasalo et al., 2006), education (Hofmann-Towfigh, 2007; Sheldon, 2005), and the COVID-19 pandemic (Daniel et al., 2022). According to Schwartz’s (1992) value theory, human values are rooted in individual needs for survival, group needs for survival and support, and societal needs for coordinated social interactions. Therefore, value changes are adaptive and functional in coping with various challenges. Conversely, a static and rigid value system can be maladaptive and detrimental to human development and survival. To understand the role and process of value change, researchers have sought to relate values to well-being.
Pathways from values to well-being
Based on the associations between values and well-being, researchers have proposed three possible pathways from values to well-being (Sagiv et al., 2004, 2015; Sagiv & Schwartz, 2000; Schwartz & Sortheix, 2018). The first pathway, referred to as the “healthy values” perspective, focuses on the content of values, suggesting that endorsing certain values tends to promote well-being, while others may undermine it. For example, self-determination theory (Ryan & Deci, 2000; Ryan et al., 2008) posits that pursuing intrinsic goals and values—such as self-direction and benevolence—fulfills basic psychological needs, thereby fostering self-motivation and psychological well-being. In contrast, the pursuit of extrinsic goals and values—such as conformity, tradition, and power—can have detrimental effects on well-being. Similarly, Sortheix and Schwartz (2017) associate growth- and self-expansion-oriented values (e.g., self-direction, benevolence, universalism, stimulation) with greater well-being, and anxiety- and self-protection-oriented values (e.g., conformity, tradition, security, power) with lower well-being. With its focus on value content, this first perspective suggests a direct relationship between specific personal values and well-being.
The second pathway, known as the “goal-attainment” perspective, focuses on the process through which values may influence well-being. Unlike the “healthy values” perspective, it does not assume that particular values are inherently linked to positive well-being. Instead, it posits that well-being may result from the successful attainment of personally important goals and values, regardless of their specific content. For example, Oishi et al. (1999) examined how success across different life domains relates to subjective well-being and found that intraindividual changes in well-being were significantly predicted by success in domains that individuals personally valued. Specifically, satisfaction with achievement predicted well-being more strongly among those who highly valued achievement. Similarly, Conte et al. (2023) demonstrated, across three experiments, that biospheric values influenced the intensity of emotional reactions to both positive and negative environmental information, an effect mediated by primary appraisal. Emphasizing the process, this second perspective illustrates an indirect effect of values on well-being via goal attainment.
The third pathway, referred to as the “value congruency” perspective, focuses on the role of context, emphasizing the importance of fit between personal values and the prevailing values in one’s environment for maintaining well-being. Sagiv and colleagues (Sagiv & Schwartz, 2000, 2004, 2015) proposed three mechanisms through which value congruence affects well-being: (a) environmental affordances, whereby individuals experience higher well-being in environments that support the pursuit of their values; (b) social sanctions, wherein individuals may face rejection or disapproval if their values conflict with dominant social norms; and (c) internal conflict, where a mismatch between personal and contextual values undermines psychological harmony. Supporting this, Sortheix and Lonnqvist (2015) found that person–group value congruence was associated with subjective well-being, mediated by the quality of interpersonal relationships. With its emphasis on context, the third perspective highlights how the effects of specific values on well-being are moderated by environmental factors.
While these perspectives have advanced our understanding of the relationships between values and well-being, two key issues warrant further attention. First, much of the existing evidence is based on cross-sectional data and lacks consistent longitudinal support. For example, Vecchione et al. (2016b) found that conservation, self-transcendence, and power values increased during early adulthood. If the theoretical perspectives described above are valid, such shifts in values should be accompanied by changes in well-being. Specifically, an increase in conservation values might directly reduce well-being (according to the “healthy values” perspective), lead individuals to pursue secure goals that enhance well-being (as proposed by the “goal-attainment” perspective), or improve well-being in contexts that emphasize those values (per the “value congruency” perspective). However, empirical evidence supporting these pathways remains limited and inconsistent. For instance, while Hanel et al. (2020) identified significant associations between value congruence and well-being, Du et al. (2021) found no such effect. The second issue is that prior research has largely overlooked the potential influence of well-being on values. Emerging longitudinal evidence suggests that intraindividual changes in well-being, such as fluctuations in life satisfaction, can predict subsequent changes in traits like personality (Hounkpatin et al., 2018; Soto, 2015). Although early work acknowledged the possibility of reverse causality (Sagiv & Schwartz, 2000; Schwartz & Sortheix, 2018), there has been limited theoretical development or empirical investigation of this direction. In the following sections, we outline three theoretical rationales for how well-being may influence value change, each corresponding to one of the previously discussed perspectives.
Pathways from well-being to values
First, the “well-being as an activator” perspective (the reverse of the “healthy values” perspective) posits that individuals who are happy and satisfied with their lives tend to prioritize growth-oriented values, whereas those with lower well-being focus on anxiety-avoidance values. Sagiv and Schwartz (2000) argue that higher well-being confers greater emotional resources, enabling pursuit of autonomy (self-direction), tolerance (universalism), and concern for others (benevolence). In contrast, lower well-being can narrow focus to personal threats, driving a preference for security or power to alleviate anxiety and meet basic survival needs. Multiple theoretical frameworks support this perspective, particularly for openness to change and conservation values. Gray’s behavioral activation and inhibition systems (Gable et al., 2000; Watson et al., 1999) suggest that positive emotions engage the activation system, promoting behaviors and values associated with openness to change. Conversely, negative emotions engage the inhibition system, directing individuals toward conservation values. Similarly, Fredrickson’s broaden-and-build theory (2001, 2005) posits that positive emotions expand cognitive and behavioral repertoires, fostering exploration, creativity, and the adoption of growth-oriented values.
Second, the “well-being as a reinforcer” perspective (the reverse of the “goal-attainment” perspective) posits that people modify their value priorities in response to the outcomes of their pursuits. Selection by consequences theory (Skinner, 1974, 1981) posits that behaviors and traits that yield functional benefits are more likely to be retained and reinforced. Similarly, cybernetic models emphasize that individuals continually adjust their goals, choices, and interpretations in response to feedback (DeYoung, 2015; DeYoung & Tiberius, 2023). When striving for a particular value yields satisfaction, the associated goals and behaviors are reinforced, thereby increasing the subjective importance of that value. Conversely, if pursuit leads to negative experiences, the value may be de-emphasized, and the motivation to pursue it declines. This mechanism aligns with Fischer and Karl’s (2022) value inference model, which argues that emotional reactions shape our judgments about held values. Unlike the “well-being as an activator” perspective, which ties well-being directly to specific values, the reinforcer perspective accommodates diverse shifts in value priorities driven by individuals’ unique experiences of well-being.
Third, the “well-being as an operator” perspective (the reverse of the “value congruency” perspective) emphasizes that well-being actively drives efforts to align personal values and prevailing values in one's environment. The traditional value congruency account posits that misalignment undermines well-being, yet treats this relationship as static. In contrast, Bandura's (2001) notion of human agency suggests that individuals both shape and are shaped by their environments. When value incongruence produces dissatisfaction, people do not passively accept it; instead, their level of well-being can spur corrective actions. First, they may alter their circumstances or seek out new contexts that better match their values—whether by voting (e.g., Caprara et al., 2017), changing jobs (Pelled & Xin, 1999), migrating (Tartakovsky et al., 2017), or pursuing different educational paths (Bardi et al., 2014). Second, they may achieve a sense of congruence by reframing their perception of the environment rather than changing their own values. For instance, Hanel, Wolfradt et al. (2018) found that people tend to perceive community and societal values in ways that favor their own priorities, creating a subjective fit without altering their value structure.
These three theoretical perspectives highlight the importance of examining how well-being may influence values as much as how values shape well-being. Addressing these bidirectional pathways requires longitudinal evidence. In the sections that follow, we review the key challenges in recent longitudinal research and demonstrate how the present study addresses them.
The current research
As reviewed above, the evidence on the interplay between values and well-being remains limited, as most prior research has relied on cross-sectional data (Bobowik et al., 2011; Joshanloo & Ghaedi, 2009; Sagiv & Schwartz, 2000). To date, only a few longitudinal studies have been conducted (Fetvadjiev & He, 2019; Grosz et al., 2021), and their findings have been mixed. Several key methodological issues require attention in this area.
First, to better understand how values and well-being influence each other over time, it is beneficial to focus on a specific life transition within a relatively short time frame. This approach enables a clearer understanding of the psychological impact of the transition. Both Fetvadjiev and He (2019) and Grosz et al. (2021) examined changes over more extended periods (5–7 years) without focusing on a particular life transition, unlike Bardi et al. (2014), who centered their research around a defined event. A broad temporal scope risks conflating the effects of multiple life events, each of which may lead to distinct changes in values and well-being, thereby complicating interpretation. Prior research has shown that major life transitions—such as childbirth (Scheppingen et al., 2018), retirement (Schwaba & Bleidorn, 2019), and community relocation (Kling et al., 2003)—can lead to substantial psychological changes. Therefore, isolating a specific life transition is crucial for identifying coherent patterns of change in values and well-being, as well as understanding their mutual dynamics. In the current research, we focused on the transition to university, a significant life event that has been shown to prompt value changes (Schwartz, 2007). University education often encourages values associated with openness to change (as opposed to conservation) and self-enhancement (as opposed to self-transcendence), as students are exposed to new ideas and experiences (Paulsen, 2013), and their achievements are directly tied to positive outcomes, such as academic success and future career opportunities. By concentrating on this transition into a value-rich environment, the present research offers a focused lens through which to examine and interpret changes in personal values and well-being, as well as the dynamic interplay between them.
Second, the age groups of the study samples can also confound results. According to Schwartz (2005), personal values may change in response to the aging process (e.g., physical strength, cognitive capacity). For instance, conservation and self-transcendence values tend to increase among older adults, while younger adults typically emphasize openness to change and self-enhancement values (Milfont et al., 2016). Additionally, previous research suggests that values cluster more closely among age groups that rate them as more important (Borg et al., 2017; Gollan & Witte, 2014). More recently, Leijen et al. (2022) distinguished between generational cohort effects (i.e., societal influences associated with the era of birth) and intra-individual change (i.e., maturation or aging processes), demonstrating that these sources contributed differently to the development of values. However, the recent longitudinal studies on values and well-being (Fetvadjiev & He, 2019; Grosz et al., 2021) included participants from diverse age groups and cohorts, potentially introducing confounding effects to the findings. To address these issues, the current research focuses on a specific age group to achieve a clearer value structure (Fontaine et al., 2008) and minimize the confounding effects of various life events and conditions (e.g., employment, marriage, child-rearing) on well-being (see Lucas, 2007; Lucas et al., 2003, 2004, for details on the set-point model).
Third, to understand the relationships between values and well-being, it is essential to gather evidence from different societal contexts to establish generalizability (Fontaine et al., 2008; Roccas & Sagiv, 2010; Sortheix & Schwartz, 2017). However, the recent longitudinal studies on the relationships between values and well-being (Fetvadjiev & He, 2019; Grosz et al., 2021) were conducted in single Western countries (the Netherlands and Germany, respectively). Given the increased emphasis on the reproducibility of scientific findings, it is worthwhile testing the same model across different samples, particularly in non-WEIRD contexts (Henrich et al., 2010). Significant cultural differences have been identified in value instantiation (Hanel, Maio, et al., 2018), the use of values as self-guides (Cheung et al., 2016), and the associations of values with other variables such as attitudes (Boer & Fischer, 2013) and behaviors (Roccas & Sagiv, 2010). To establish the commonality of the relationships between values and well-being, the current research includes three different samples from Mainland China, Hong Kong, and the United Kingdom.
Lastly, a significant concern with the two most relevant longitudinal studies (Fetvadjiev & He, 2019; Grosz et al., 2021) is that their findings were based on raw, rather than ipsatized, scores of values. According to Schwartz’s (1992) value theory, scores should be ipsatized to reflect the relative importance of values and control for individual differences in scale use (see Rudnev, 2021, for a review). Correcting for scale use tendencies is particularly important in research on values and well-being, as such tendencies have been linked to optimism (Borg & Bardi, 2016), which closely relates to well-being (e.g., Robins et al., 2001). Partly due to this reason, the longitudinal relations between values and well-being in Fetvadjiev and He (2019) were all in the same direction, even for competing values that are expected to work in opposite ways (See the longitudinal evidence in Bardi et al., 2009). The conflicting results make it challenging to derive coherent theoretical interpretations.
To control for the effects of scale use, an effective approach is to analyze the two value dimension scores (i.e., openness to change vs. conservation and self-enhancement vs. self-transcendence), which has become increasingly common in value research (Grigoryan & Schwartz, 2020; Lönnqvist et al., 2018; Seppala et al., 2012; Tormos et al., 2017). Focusing on these two dimensions offers several advantages (see also Verkasalo et al., 2009). First, compared to basic values, the two dimensions provide a more parsimonious and robust representation of the value structure, enhancing theory building and the generalizability and applicability of findings. Second, this analysis benefits from higher reliability, as each dimension is assessed by more items than the basic values. Third, the issue of multicollinearity caused by ipsatization can be eliminated, allowing for simultaneous analysis of both value dimensions. Fourth, the dimension scores can effectively control for scale use by subtracting one higher-order value from its opposite value. Unlike ipsatization, which uses the personal mean across all values, this approach eliminates the influence of value scores on the other dimension that are theoretically orthogonal and unrelated. This is particularly crucial when examining value change, as a life transition may induce changes in one dimension but not the other. If ipsatization is used, any changes to the value system would be redistributed across all values through mean correction, complicating the study of specific value changes. Consequently, the current research focuses on the two value dimensions to examine patterns of change in values and well-being, as well as their relationship.
In the current research, three longitudinal studies were conducted among university students in three different societal contexts. Study 1 took place in an Eastern culture (Mainland China) and aimed to understand how values relate to cognitive evaluations of well-being (i.e., life satisfaction). Study 2 was conducted in a cosmopolitan city (Hong Kong), where Eastern and Western cultures intersect, and focused on how values relate to both cognitive (i.e., life satisfaction) and affective components (i.e., positive and negative affects) of well-being. Study 3 took place in a Western culture (United Kingdom), testing how values relate to life satisfaction and self-esteem, the latter being a crucial element of well-being in Western cultures (see DeNeve & Cooper, 1998, for a review). Additionally, self-esteem is linked to personal values (Lönnqvist et al., 2009) and contributes to various life outcomes (Luyckx et al., 2013; Marshall et al., 2014; Orth et al., 2008, 2012; Trzesniewski et al., 2006).
Summary of key study characteristics and findings.
Notes. PVQ = Portrait Values Questionnaire; ESS = European Social Survey; SVS = Schwartz Value Survey; SWLS = Satisfaction With Life Scale; PANAS = Positive and Negative Affect Schedule; RSES = Rosenberg Self-Esteem Scale; OTC = openness to change; SE = self-enhancement; PA = positive affect; LS = life satisfaction; n.s. = non-significant.
An Illustrative Diagram for Parallel Process Latent Growth Modeling. Note. Autocovariances between the measurement errors of the value dimension and WB within each time point are omitted for clarity. WB = Well-being.
While we did not commit a priori to a specific directional prediction, the LGM framework enabled us to test both pathways empirically. If the traditional perspective is supported, in which values predict later well-being, then initial values should predict changes in well-being. Conversely, if the emerging perspective is supported, in which well-being predicts value development, then initial well-being should predict changes in values. Support for both pathways would suggest a reciprocal relationship, where the initial levels of values and well-being each predict changes in the other. Given the life context studied—transitioning into university—we expected well-being to be positively associated with openness to change (vs. conservation), a value dimension particularly salient among young adults in higher education (see review above). By contrast, the association between well-being and self-enhancement (vs. self-transcendence) was to be examined, as it could be either negative, as predicted by the “healthy values” and “well-being as an activator” perspectives, or positive, as suggested by the value congruence and “well-being as an operator” perspectives.
Study 1
Study 1 tested the model among a sample of university students in Mainland China, focusing on the longitudinal relationships between values and life satisfaction (LS). According to the hypotheses, it is expected that the initial levels of openness to change (vs. conservation, denoted as OTC) and self-enhancement (vs. self-transcendence, denoted as SE) would predict the slope of life satisfaction (i.e., OTC and SE intercepts → LS slope), if the previous pathways from values to well-being are supported. Conversely, if well-being predicts values, the initial levels of life satisfaction would predict changes in OTC and SE (LS intercept → OTC and SE slopes).
Method
Transparency and openness
We report on how we determined our sample size, all data exclusions, and all measures in Studies 1–3, following the Journal Article Reporting Standard for reporting psychology research in scientific journals. The materials, data, and analysis code are available at https://osf.io/vdbse/. Data were analyzed using SPSS 25 and LISREL 8.8. The studies were approved by the Institutional Review Boards in the universities of the first and second authors.
Participants
For latent growth modeling, a sample size of 200 or more is typically recommended to detect small to medium effects (Byrne et al., 2008; Fan & Fan, 2005; Zhang & Wang, 2009). In the current study, a total of 218 undergraduate students from a university in Mainland China participated in all waves of data collection. Among the participants, 94 were male, 122 were female, and 2 respondents did not indicate their gender. Their ages at Time 1 ranged from 15 to 29 years (M = 18.61, SD = 1.12). The participants came from various disciplines, including science, engineering, and social sciences, among others.
Attrition analysis
To assess potential bias due to attrition, we examined the differences between respondents who participated in all waves of data collection and those who participated only at Time 1, focusing on demographic characteristics and value dimensions at Time 1. No significant differences were observed in demographic variables. A multivariate analysis of variance (MANOVA) on the two value dimensions indicated a significant difference between the two groups, Wilks’ Λ = .98, F (2, 375) = 4.38, p = .013, multivariate η2 = .02. Follow-up ANOVAs revealed that respondents who participated in all waves of data collection (M = −.99. SD = .94) had lower self-enhancement compared with those who participated at Time 1 only (M = −.73. SD = .99), F (2, 375) = 6.63, p = .010, η2 = .02. Although the effect size was small, caution should be exercised in interpreting results regarding this value dimension.
Procedure
At the beginning of the first semester, all first-year undergraduates from a university in Mainland China were invited to participate in the study. Data were collected using a paper-and-pencil questionnaire, which included measures of personal values, life satisfaction, and demographic characteristics. Participants completed the questionnaire in a lecture theater at three time points, with an interval of six months.
Measures
Personal values
The Portrait Values Questionnaire (PVQ), developed by Schwartz et al. (2001), was used to measure personal values. Participants were asked to compare themselves to a person described by 40 items measuring 10 basic human values, which were further combined into four higher-order values: openness to change (self-direction, stimulation, and hedonism), conservation (security, conformity, and tradition), self-enhancement (achievement and power), and self-transcendence (benevolence and universalism). Each item was assessed on a 6-point Likert scale ranging from 1 (not like me at all) to 6 (very much like me). The measures of openness to change (Cronbach’s αs = .79 to .84), conservation (Cronbach’s αs = .70 to .77), self-enhancement (Cronbach’s αs = .81 to .81), and self-transcendence (Cronbach’s αs = .78 to .84) exhibited good internal reliability across the three measurement occasions.
Theoretically, these four values represent two bipolar dimensions, each consisting of two opposing values on the same continuum. Consistently, openness to change was negatively correlated with conservation (r = −.69 to −.65, p’s < .001), while self-enhancement was negatively correlated with self-transcendence (r = −.63 to −.48, p’s < .001) across the three time points. Following previous studies (e.g., Lipponen et al., 2008; Seppala et al., 2012; Tormos et al., 2017), the two value dimensions were derived by subtracting conservation from openness to change and self-transcendence from self-enhancement.
Life satisfaction
The cognitive component of subjective well-being was assessed using the Satisfaction With Life Scale (SWLS), developed by Diener et al. (1985). This unidimensional measure of global life satisfaction consists of five items (e.g., “I am satisfied with my life”). All items utilized a 7-point Likert response format, ranging from 1 (strongly disagree) to 7 (strongly agree). This scale demonstrated satisfactory internal consistency reliability across the three time points (Cronbach’s α = .76 to .82).
Data analysis
Missing data were imputed using the expectation maximization (EM) algorithm. Three separate sets of univariate LGMs were analyzed for the two value dimensions and life satisfaction. In each univariate LGM, factor loadings on the intercept factor were fixed at 1, representing the starting point of the growth trajectory. The loadings on the slope factor were fixed at 0, 1, and 2 to reflect a linear growth trend across equal intervals. This widely used approach enhances model parsimony and interpretability. If the model results in a poor fit, modifications—such as relaxing constraints for free estimation—would be considered.
Subsequently, parallel process LGMs were applied to examine the relationships between the initial levels and change slopes of values and life satisfaction. As illustrated in Figure 1, the slope of life satisfaction was regressed on the intercept of values, and the slope of values was regressed on the intercept of life satisfaction. Furthermore, following the recommendations of Wright et al. (2016), each slope factor was regressed on the intercept of the same construct to control for the “regression to the mean” effect (Campbell & Kenny, 1999; Finkel, 1995). Additionally, autocovariances were included between the measurement errors of values and life satisfaction within each time point. Two sets of parallel process LGMs were conducted to analyze the two value dimensions separately.
In terms of model estimation, the selection of the estimation method was based on the multivariate normal distribution of the data. Maximum likelihood (ML) estimation was applied if the data adhered to multivariate normality. If this assumption were violated, robust maximum likelihood (RML) would be employed, and Satorra-Bentler (SB) scaled chi-square statistics (Satorra & Bentler, 2001) would be computed. The goodness of fit for the models was assessed using a combination of several fit indices, including the root mean square error of approximation (RMSEA; Steiger, 1990), standardized root mean square residual (SRMR; Bentler, 1995), comparative fit index (CFI; Bentler, 1990), and non-normed fit index (NNFI; Tucker & Lewis, 1973). Acceptable model fit was indicated by RMSEA ≤ .10, SRMR ≤ .08, CFI ≥ .90, and NNFI ≥ .90 (Kline, 2011).
Results
Descriptive statistics
Descriptive statistics in study 1.
Univariate LGMs
Univariate latent growth models of value dimensions and life satisfaction in study 1.
ap < .10. *p < .05. **p < .01. ***p < .001.
The means of the intercepts indicated that students initially endorsed openness to change more than conservation, and self-transcendence more than self-enhancement, consistent with international findings (Schwartz & Bardi, 2001). Participants also reported being reasonably satisfied with their lives. The slope for the self-enhancement dimension was positive and significant (p = .006), suggesting an increase in self-enhancement (vs. self-transcendence) over time. In contrast, the openness to change dimension and life satisfaction tended to be stable, as the slopes for both constructs were not significantly different from zero.
Additionally, the intercepts of the two value dimensions were negatively related to their corresponding slope factors (p’s < .05), indicating a “regression to the mean” effect (i.e., individuals with higher initial levels tend to decrease subsequently, and vice versa). The intercept and slope of life satisfaction were not significantly related.
Parallel process LGMs
Parallel process latent growth models of value dimensions and life satisfaction in study 1.
Note. LS = life satisfaction.
*p < .05. **p < .01. ***p < .001.
Study 2
Study 2 tested the same model among a group of university students in Hong Kong, where Eastern and Western cultures intersect. This study focused on the longitudinal relationships between values and both cognitive and affective components of well-being, namely life satisfaction (LS), positive affect (PA), and negative affect (NA). It was expected that the initial levels of openness to change (vs. conservation) and self-enhancement (vs. self-transcendence) would predict the slope of well-being (i.e., OTC and SE intercepts → LS, PA, and NA slopes) if previous perspectives for pathways from values to well-being are supported; or vice versa (i.e., LS, PA, and NA intercepts → OTC and SE slopes) if the new perspectives for pathways from well-being to values are supported.
Method
Participants
A sample of 252 Hong Kong Chinese undergraduate students participated in all waves of data collection, comprising 65 males and 187 females. They represented various colleges, including business, liberal arts and social sciences, science and engineering, creative media, and energy and environment. Participants’ ages at Time 1 ranged from 17 to 22 years (M = 18.60, SD = .77).
Attrition analysis
We investigated differences between respondents who participated in all waves of data collection and those who participated only at Time 1, focusing on demographic characteristics and value dimensions at Time 1. There was a significantly lower proportion of males among respondents who participated in all waves (25.8%) compared to those who participated only at Time 1 (43.5%), χ2 (1) = 7.57, p = .006. We examined the potential effect of gender on the two value dimensions using a MANOVA, which showed no significant gender differences, Wilks’ Λ = .99, F (2, 311) = 2.32, p = .100, multivariate η2 = .01, suggesting that the attrition of male participants did not strongly impact the results. There were no significant differences in other demographic variables. A MANOVA on the two value dimensions also revealed no significant differences between the two groups, Wilks’ Λ = .99, F (2, 311) = 1.58, p = .207, multivariate η2 = .01. Thus, there were no indications of bias due to attrition.
Procedure
At the beginning of the first semester, first-year undergraduates from a university in Hong Kong were invited to participate via an internal online platform. A unique and personalized link to the online survey was sent to interested students. The survey included measures of personal values, life satisfaction, positive and negative affect, as well as demographic characteristics. Participants completed the survey across four measurement occasions, with an interval of six months.
Measures
Personal values
Personal values were measured using the 21-item European Social Survey (ESS) version of the Portrait Values Questionnaire (PVQ; Schwartz et al., 2001). Similar to Study 1, the four higher-order values (openness to change, conservation, self-enhancement, and self-transcendence) were calculated. Despite using fewer items than in the original PVQ, the internal consistency reliability for the measures of openness to change (Cronbach’s α = .70 to .78), conservation (Cronbach’s α = .65 to .72), self-enhancement (Cronbach’s α = .66 to .72), and self-transcendence (Cronbach’s α = .70 to .78) was generally acceptable across the four measurement occasions, comparable to typical reliabilities reported in previous research (Schwartz, 2003; Ye et al., 2015).
As in Study 1 and 2 value dimensions were computed by subtracting conservation from openness to change and self-transcendence from self-enhancement. Openness to change was negatively correlated with conservation (r = −.80 to −.69, p’s < .001), and self-enhancement was negatively correlated with self-transcendence (r = −.52 to −.32, p’s < .001) across the four time points, indicating that the use of these two value dimensions was empirically justified.
Life satisfaction
The Satisfaction With Life Scale (SWLS) was used to assess life satisfaction, as in Study 1. This scale demonstrated satisfactory internal consistency reliability across the four time points (Cronbach’s α = .84 to .90).
Positive and negative affect
The Positive and Negative Affect Schedule (PANAS; Watson et al., 1988) was employed to measure positive and negative affect. This scale consists of 10 items for positive affect (e.g., interested, excited, strong) and 10 items for negative affect (e.g., afraid, scared, nervous). Respondents rated their affective experiences on a 5-point scale (ranging from not at all to always). This instrument showed good reliability across the four time points (Cronbach’s α = .85 to .89 for PA and .88 to .91 for NA).
Data analysis
The statistical procedures used were essentially the same as in Study 1, except that one additional time point was included in the LGMs.
Results
Descriptive statistics
Descriptive statistics in study 2.
Univariate LGMs
Univariate latent growth models of value dimensions and subjective well-being in study 2.
*p < .05. **p < .01. ***p < .001.
Parallel process LGMs
Parallel process latent growth models of value dimensions and life satisfaction in study 2.
Note. LS = life satisfaction.
*p < .05. **p < .01. ***p < .001.
Parallel process latent growth models of value dimensions and positive affect in study 2.
Note. PA = positive affect.
ap < .10. *p < .05. **p < .01. ***p < .001.
Parallel process latent growth models of value dimensions and negative affect in study 2.
Note. NA = negative affect.
*p < .05. **p < .01. ***p < .001.
Study 3
Study 3 tested the model in a United Kingdom sample—an individualistic cultural context—focusing on longitudinal links between values and two well-being indicators: life satisfaction (LS) and self-esteem (SES). As in Studies 1 and 2, it was expected that the initial levels of openness to change (vs. conservation) and self-enhancement (vs. self-transcendence) would predict the slope of life satisfaction and self-esteem (i.e., OTC and SE intercepts → LS and SES slopes), if previous perspectives for pathways from values to well-being are supported; or vice versa (i.e., LS and SES intercepts → OTC and SE slopes) if the new perspectives for pathways from well-being to values are supported.
Method
Participants
A sample of 196 undergraduate students in the United Kingdom participated in three waves of data collection, comprising 56 males and 138 females, with one participant not indicating their gender (for additional details, see Bardi et al., 2014, Study 2). Participants were from the psychology (n = 131) and business (n = 65) departments. The ages of participants at Time 1 ranged from 18 to 43 years (M = 18.57, SD = 1.98).
Attrition analysis
We compared respondents who participated in all waves of data collection with those who participated only at Time 1. There were no significant differences in age, gender, or ethnicity between the two groups. A MANOVA on the value scores revealed no significant differences in the overall pattern of value importance between respondents who participated in all waves and those who participated only at Time 1, hence there was no support for bias due to attrition.
Procedure
Participants were approached at the end of a lecture at the beginning of their first, second, and third years (i.e., at 12-month intervals) and completed a pencil-and-paper questionnaire for course credit or the equivalent of $7. Their university email addresses, which did not reveal their names, were used to link data across the three time points. We also emailed a link to an online version of the questionnaire to those not attending the Year 2 and/or Year 3 lectures. The value questionnaire was presented first, followed by the well-being questionnaires.
Measures
Personal values
Participants completed the Schwartz Value Survey (SVS; Schwartz et al., 2000). This questionnaire includes a list of value items followed by definitions, such as “AUTHORITY (the right to lead or command),” measuring power values. Each of the 10 values is assessed using three to eight items. Participants rated the importance of each value item as a guiding principle in their lives on a 9-point scale ranging from −1 (opposed to my values) to 0 (not important) to 7 (of supreme importance). The asymmetry of the scale reflects the distinctions among values that people naturally make (Schwartz & Bardi, 2001), and the use of negative values is quite rare (see more detail in Bardi et al., 2009), minimizing issues with scale asymmetry. The measures of openness to change (Cronbach’s α = .71 to .77), conservation (Cronbach’s α = .73 to .81), self-enhancement (Cronbach’s α = .74 to .75), and self-transcendence (Cronbach’s α = .78 to .83) demonstrated satisfactory internal consistency reliability across the three time points.
As in Studies 1 and 2, two value dimensions were computed by subtracting conservation from openness to change and by subtracting self-transcendence from self-enhancement. Openness to change was negatively correlated with conservation (r = −.72 to −.64, p’s < .001), while self-enhancement was negatively correlated with self-transcendence (r = −.70 to −.67, p’s < .001) across the three occasions, supporting the use of these two value dimensions.
Life satisfaction
The Satisfaction With Life Scale (SWLS) was used to assess life satisfaction, as in Study 1. This scale demonstrated satisfactory internal consistency reliability across the three time points (Cronbach’s α = .77 to .84).
Self-esteem
The Rosenberg Self-Esteem Scale (RSES, Rosenberg, 1965) was used to assess global self-esteem. This scale consists of 10 items, with a sample item being “On the whole, I am satisfied with myself.” All items were rated on a 7-point Likert scale ranging from 1 (disagree strongly) to 7 (agree strongly). The internal consistency reliability for the RSES was good across the three time points (Cronbach’s α = .84 to .86).
Data analysis
A similar statistical procedure to Study 1 was adopted.
Results
Descriptive statistics
Descriptive statistics in study 3.
Univariate LGMs
Univariate latent growth models of value dimensions and subjective well-being in study 3.
ap < .10. *p < .05. **p < .01. ***p < .001.
Parallel process LGMs
Parallel process latent growth models of value dimensions and life satisfaction in study 3.
Note. LS = life satisfaction.
ap < .10. *p < .05. **p < .01. ***p < .001.
Parallel process latent growth models of value dimensions and self-esteem in study 3.
Note. SES = self-esteem.
ap < .10. *p < .05. **p < .01. ***p < .001.
General discussion
Our research examined the longitudinal interplay between personal values and subjective well-being during the transition to university—a period of significant personal growth shaped by changing educational contexts—through three studies conducted in Mainland China (Study 1), Hong Kong (Study 2), and the United Kingdom (Study 3), using diverse well-being measures (life satisfaction, positive/negative affect, and self-esteem). Our findings reveal intricate relationships, with well-being exerting a more pronounced effect on changes in values: in Mainland China, higher baseline life satisfaction predicted increased openness-to-change values; in the UK, higher initial self-esteem forecasted growth in both openness-to-change and self-enhancement values; and, conversely, in Hong Kong, stronger initial self-enhancement values predicted rises in positive affect.
These findings challenge traditional value-theory frameworks—which primarily position values as antecedents of well-being—by demonstrating the significant role of well-being in predicting later value endorsement and by highlighting the context-dependent nature of these dynamics across diverse sociocultural settings. In the sections that follow, we discuss the theoretical implications of the pathways, explore the variability of value–well-being dynamics, and address the study’s limitations, as well as provide directions for future research.
Longitudinal pathways between values and well-being
The interplay between personal values and subjective well-being (SWB) is a dynamic process, where each may shape the other over time.
Pathways from values to well-being
In Study 2, we observed that among Hong Kong university students, those who initially emphasized self-enhancement values (e.g., achievement, power) over self-transcendence values (e.g., benevolence, universalism) experienced greater increases in positive affect over time. This finding is noteworthy, given that participants across all three studies, on average, placed greater importance on self-transcendence than self-enhancement values. It challenges the “healthy values” perspective, which holds that self-transcendence values are inherently more conducive to well-being than self-enhancement values (Sagiv & Schwartz, 2000). Instead, it supports the value congruence perspective, which emphasizes the importance of alignment between personal values and sociocultural context (Sagiv et al., 2004). In Hong Kong’s fast-paced, competitive environment, values such as achievement and power are highly emphasized (Ng et al., 2016; Ng & Ye, 2016). Students who endorse these values may therefore experience greater positive affect due to the perceived fit between their value orientation and their surrounding societal norms.
This evidence suggests that the impact of values on well-being may not be universal but contingent on sociocultural congruence. By demonstrating that self-enhancement values can enhance well-being in specific contexts, our findings challenge the assumption of fixed categorizations of “healthy” and “unhealthy” values. This advances theoretical understanding by highlighting the need for models that account for contextual variability, moving beyond universal claims about value–well-being relationships.
Pathways from well-being to values
Compared to the effect of values on well-being, our studies revealed a more pronounced effect of well-being on subsequent value change. Specifically, in Mainland China, higher baseline life satisfaction predicted increases in openness to change (vs. conservation) values. In the United Kingdom, higher initial self-esteem predicted growth in both openness to change (vs. conservation) and self-enhancement (vs. self-transcendence) values. These findings illuminate the potential roles of well-being in shaping value priorities, as interpreted through two theoretical lenses: the “well-being as an activator” perspective and the “well-being as an operator” perspective.
The “well-being as an activator” perspective posits that higher well-being encourages individuals to pursue personal growth and novel experiences, fostering values such as openness to change and self-transcendence. The “well-being as an operator” perspective, by contrast, suggests that greater well-being facilitates the adoption of values prevalent or rewarded within one’s social context. The observed effects on openness to change in both Mainland China and the United Kingdom support both perspectives: higher well-being appears to motivate individuals to seek out new experiences (as an activator) while also aligning their values with those emphasized in their environments, such as university settings that often prioritize openness to change (as an operator).
However, the effect of well-being on self-enhancement values in the United Kingdom aligns solely with the operator perspective. Here, higher self-esteem predicted increased self-enhancement (e.g., achievement, power), suggesting that individuals with greater well-being are more likely to embrace values that are valued or reinforced in their social settings. This finding challenges the activator perspective, which would predict that higher well-being activates growth-oriented values like self-transcendence (e.g., universalism, benevolence), rather than anxiety-avoidant values like self-enhancement. The divergence indicates that well-being’s influence on values may depend on contextual factors, such as the cultural or social salience of specific values.
The variability of value–well-being dynamics
While we obtained some findings of longitudinal relations between well-being and personal values, the fact that some hypothesized relationships failed to reach significance demonstrates that these dynamics are neither uniform nor universal but are moderated by societal factors and underlying mechanisms. This variability highlights the complexity of value–well-being dynamics and cautions against broad generalizations.
The role of societal factors
These variations highlight the pivotal role of societal factors in modulating this relationship. Among university students in Mainland China, higher initial life satisfaction significantly predicted increases in openness to change values but not self-enhancement values. This pattern may stem from the shift from China’s rigid, exam-driven high school system (Lai et al., 2016) to a university environment where academic achievement is no longer the sole measure of success (Davey et al., 2007). Instead, broader goals—such as social engagement and interpersonal relationships—gain greater importance (Zhang et al., 2016), potentially diminishing the relevance of self-enhancement, a value linked with self-interest measured by traditional proverbs (Ye et al., 2018).
In Hong Kong, initial self-enhancement values predicted increases in positive affect among university students. However, openness to change values showed no predictive effect on well-being. In contrast to students in Mainland China and the UK, many Hong Kong university students continue living with their families (Forrest & Xian, 2018; Li, 2014), and the city’s compact geography provides fewer opportunities for new lifestyles or cultural experiences. This relative stability may limit the development of openness to change values.
In the UK, higher initial self-esteem predicted increases in both openness to change and self-enhancement values among university students. However, no significant effects were found for life satisfaction, highlighting the central role of self-esteem in non-collectivistic cultures—a pattern consistent with previous research (Luyckx et al., 2013; Marshall et al., 2014; Orth et al., 2008, 2012; Trzesniewski et al., 2006), including findings on the longitudinal effects of self-esteem on self-direction values (Daniel et al., 2021). These contextual differences underscore the importance of considering societal factors when examining the relationship between values and well-being.
The role of underlying mechanisms
The variations in results may also reflect the complexity of the underlying mechanisms that govern the interplay between values and well-being. First, well-being is shaped by a wide range of influences beyond the pursuit of personal values. Research has identified numerous contributing factors, including physical health, social relationships, financial stability, and work performance (see review in Diener et al., 2018). Although the pursuit of personally meaningful values can enhance well-being, these alternative pathways offer individuals multiple avenues for maintaining psychological health. As a result, the specific influence of values on well-being may be diluted, potentially contributing to the null findings observed in our study.
Second, value pursuit is inherently dynamic and context-sensitive. When an individual’s values are misaligned with their immediate environment—for example, a university student who endorses conservation values in a setting that emphasizes openness—they may find alternative contexts to express those values, such as religious communities or other like-minded social groups (see Bardi et al., 2014). Such adaptability suggests that the relationship between initial value priorities and well-being is not necessarily linear or direct; rather, individuals may modify their environments or coping strategies to maintain well-being, even when their core values are not immediately fulfilled. Together, these insights underscore the importance of considering broader ecological and psychological processes when examining the relationship between values and well-being.
Limitations and future research
Several limitations of the current research should be acknowledged. First, the findings were based on a single life transition—entering university. While this focus helps pinpoint specific changes in values and well-being, the generalizability of these findings to other life transitions—such as immigration, marriage, childbirth, or retirement—requires further investigation. The pathways between well-being and values may vary qualitatively across different transitions. For example, individuals with higher well-being may experience increased openness to change during the transition to university but may shift toward more conservative values when entering parenthood. Additionally, the observed changes in values and well-being in this research were generally small, as indicated by the slope estimates, thereby limiting the ability to detect significant pathways. More impactful and enduring transitions, such as childbirth (Scheppingen et al., 2018) or retirement (Blekesaune & Hansen, 2021; Gruszczyńska et al., 2019), could reveal stronger dynamics between well-being and values.
Second, the findings were based on general well-being measures. Future research should test the models using domain-specific well-being measures (e.g., job satisfaction, relationship satisfaction) to investigate how they relate to value change. For instance, past research has shown that congruence between personal and organizational values contributes to work-related outcomes (Arieli et al., 2020). It remains unclear, however, whether employees with higher job satisfaction are more likely to adopt values that align with their organizations. Similarly, spousal similarity in values predicts marital satisfaction (Gaunt, 2006; Leikas et al., 2018; Ye et al., 2023). Our findings suggest that it would be worthwhile to examine whether marital satisfaction fosters value convergence over time, as satisfied couples may be more inclined to adjust their values for greater alignment.
Third, the statistical power of our parallel process LGMs may have been insufficient to detect smaller effects. Although our studies were powered to detect medium-sized effects—evidenced by significant paths such as β = 0.35 in Study 1, β = 0.42 in Study 2, and β = 0.33 in Study 3—smaller effects (e.g., β < 0.30) may not have reached significance due to limited sample sizes 4 . Notably, these smaller effects, which align with pathways from values to well-being, might become significant in larger samples. Conversely, the lack of replication for the significant results across different contexts (e.g., Western vs. Eastern) and the fact that most hypothesized pathways did not reach significance suggest the preliminary nature of our findings. This pattern raises the possibility of Type I errors (false positives), especially when testing multiple potential pathways in the analyses, as the family-wise error rate increases with the number of comparisons. Future research should replicate these findings with larger, more diverse samples to verify both significant and non-significant results and to further clarify the pathways between well-being and values.
Fourth, this research relied on self-report data, which may not fully capture the processes underlying the interplay between values and well-being. Future studies should incorporate a broader range of methodologies—such as interviews (Grosz et al., 2021), experiments (Hanel et al., 2017; Wolf et al., 2021), and experience sampling methods (Hanel et al., 2023; Hektner et al., 2007; Schimmack & Diener, 2003; Scollon et al., 2003)—to gain deeper insight into the dynamics of value–well-being relationships and to more effectively test theoretical frameworks that emphasize underlying processes (e.g., the “goal-attainment” perspective and the “well-being as a reinforcer” perspective).
Finally, our research design varied across the three studies in terms of well-being measures. While each study included life satisfaction, other indicators differed. This approach allows us to capture a broader range of components—such as emotional states and self-worth—which can enrich our findings by revealing nuanced relationships. However, it also presents challenges for direct comparisons and complicates the interpretation of differing results across studies. Future research should standardize well-being measures across samples to improve comparability and bolster the replicability of findings.
Conclusion
This research introduced three theoretical frameworks for how well-being may drive changes in personal values and empirically tested the bidirectional pathways across three parallel longitudinal studies focused on the transition to university. Although findings varied by context and measure, the overall pattern showed that well-being more often predicted subsequent shifts in values than values predicted later changes in well-being. By establishing well-being not only as an outcome but also as a precursor to value change, our results call for reframing value–well-being models to recognize well-being as an active driver in the evolution of personal values.
Footnotes
Declaration of conflicting interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study is supported by a grant from the Research Grants Council of the Hong Kong Special Administrative Region, China (Project No. CityU 11410314), awarded to the first author. Funding for Study 3 was provided to the second author by the Nuffield Foundation (Grant SGS/33555).
Open science statement
Disclosure of prior uses of data
Ethical approval
All procedures in studies involving human participants were performed in accordance with the ethical standards of the institution’s Human Research Ethics Committee.
Informed consent
Informed consent was obtained from all participants included in the study.