Depressive Symptoms Predict Divergent Trajectories of Well-being in U. S. Adults

Depressive Symptoms Predict Divergent Trajectories of Well-being in U. S. Adults | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Analysis Depressive Symptoms Predict Divergent Trajectories of Well-being in U. S. Adults Cassondra Lyman, Anthony Ong, Jonathan Rottenberg This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7123882/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted You are reading this latest preprint version Abstract The extent to which adults sustain psychological well-being across decades—and the role of depression in shaping these trajectories—remains poorly understood. Using data from the 25-year Americans' Changing Lives cohort (N=3,617), we applied growth mixture modeling to identify longitudinal well-being trajectories and assess whether baseline depressive symptoms predict stability versus change. Four distinct patterns emerged: persistent well-being (60% of the sample), declining well-being, gradually increasing well-being, and sharply increasing well-being. While persistent well-being was the mentally healthiest trajectory, many of this group's members still exhibited substantial intra-individual variability in well-being, and some maintained persistently low well-being. Elevated baseline depressive symptoms predicted greater odds of non-stable trajectories, including both decline and marked improvement, independent of baseline well-being level. These findings underscore the dynamic interplay between depressive symptoms and long-term mental health. Results support interventions that go beyond symptom reduction to cultivate the maintenance of psychological well-being across time. Scientific community and society/Social sciences/Psychology Biological sciences/Psychology well-being depressive symptoms longitudinal data analysis growth mixture modeling American’s Changing Lives (ACL) Figures Figure 1 Figure 2 Figure 3 Figure 4 INTRODUCTION Well-being is a multifaced concept referring to emotions, life satisfaction, sense of meaning and purpose, and ability to pursue self-defined goals (Park et al., 2023; Ryff, 2023). Those reporting higher levels of well-being are physically healthier (Keyes & Simoes, 2012) and less likely to develop mental health problems, like depression (Grant et al., 2013). Given the benefits of well-being, both scientists and laypeople are motivated to identify the sources of sustained well-being. Surprisingly little is known about what factors predict the attainment of high levels of well-being over decades. Although depression impairs short-term well-being, little is known about how depression affects long-term trajectories of well-being. To address this question, we leveraged data from the 25-year Americans’ Changing Lives (ACL) study (N=3,617) to (1) identify common well-being trajectories, (2) estimate how often adults sustain well-being across decades in the United States, and (3) test whether depression and other demographic or health factors predict different well-being trajectories (including that of sustained well-being). Defining Well-being According to Keyes’ (2005) dual continuum model of mental health, well-being represents more than the absence of mental illness—it additionally requires the presence of positive emotional, psychological, and social functioning. This holistic view integrates Diener’s (1984) subjective well-being model and Ryff’s (1989) psychological well-being model, and aligns with evidence that human flourishing involves facets including life satisfaction, frequent positive affect, infrequent negative affect, self-acceptance, autonomy, environmental mastery, personal growth, purpose in life, and positive relationships (Disabato et al., 2019; Gallagher et al., 2009; Keyes et al., 2002). A key insight of the dual continuum model is that mental health and mental illness are conceptually and empirically distinct dimensions—not simple opposites (Keyes, 2005; Westerhof & Keyes, 2010). This distinction may explain why depressive symptoms and well-being exhibit more complex relationships over time (e.g., bidirectionality, Huppert, 2009; Joshanloo & Blasco-Belled, 2023; Lamers et al., 2015). Well-being Trajectories A well-being trajectory refers to the pattern of change(s) in a person’s well-being over time. A longstanding debate in affective science concerns the degree of consistency in long-term well-being. On the side of consistency is evidence that well-being does not differ across age-cohorts (Ryff, 1995) and often persists over long follow up intervals (Cintron & Ong, 2024; Ryff et al., 2015). For example, in joint trajectory analyses of longitudinal data spanning 18 years, most (83%) U.S. adults reported persistently high positive affect and low negative affect, while two smaller groups demonstrated either improving or declining affective well-being (Cintron & Ong, 2024). Similarly, analysis of psychological well-being in the same sample’s first two timepoints revealed that most (78-83%) participants’ well-being scores did not shift more than two quartiles. Alternatively, there is evidence of well-being exhibiting meaningful change over time (Cintron & Ong, 2024; Ryff et al., 2015). While early models, such as set-point theory (Lykken & Tellegen, 1996), proposed that well-being is largely stable and returns to a baseline following life events, subsequent research has shown that this is not always the case. For example, major life events—including marriage, disability, and unemployment—can lead to lasting changes in subjective well-being (Lucas, 2007) or to people spending several consecutive years above and then several consecutive years below their long-term average well-being level (Headey & Muffels, 2016). These studies highlight that although well-being often persists over time, it is not necessarily stable. Why Study How Depression Influences Well-being Trajectories? Although many may take comfort in the idea that well-being persists over decades, people with depression likely do not. Clinical depression appears incompatible with well-being cross-sectionally or over short spans (Rottenberg et al., 2019), but people with depression are known to strongly desire the attainment of well-being (Wood & Tarrier, 2010). In fact, many patients with depression prioritize affective balance and restored daily functioning over conventional outcomes, like symptom reduction (Chevance et al., 2020; Demyttenaere et al., 2015; Zimmerman et al., 2006). As people often exhibit preferences for stability (Frijda, 1988; Samuelson & Zeckhauser, 1988; Swann & Read, 1981) and people with persistently high well-being demonstrate more favorable health outcomes over time than those simply reporting high well-being at baseline (Ryff et al., 2015), it stands to reason that people with depression would desire sustained well-being over temporary periods of well-being. However, it remains unclear how often they achieve those desires. Although emerging evidence indicates that around 10% of people with a history of depression subsequently achieve relatively high levels of well-being (Devendorf et al., 2022; Rottenberg et al., 2019), questions remain about whether these gains are maintained over time. Some scholars suggest that they are maintained, proposing that depressive episodes themselves may function as well-being turning points when people make meaningful life change in response to the experience (Nesse, 2000; Ridge & Ziebland, 2006). However, other work suggests that these gains are not maintained, as increased affective reactivity among people with current or remitted depression may contribute to less stable well-being over time (Blysma et al., 2011). Still, the impact of this reactivity on long-term well-being may vary based on the person’s tendency to monitor mood (Thompson et al., 2013), ability to regulate mood (Lischetzke & Eid, 2003), and the valence of events they experience (Blysma et al., 2011). Taken together, these studies suggest that a history of depression may make well-being trajectories less stable, potentially setting the stage for either increases or decreases in long-term well-being. The Current Study Although there is moderately convincing evidence that most people report persistent well-being over long time periods, uncertainties remain. First, although previous work indicates that persistent well-being is common, relatively few studies use growth mixture modeling to examine the full range of well-being trajectories over decades. Second, the nature of persistent well-being has not been well-characterized. It has not been determined whether people who exhibit mean-level stability in well-being invariably maintain low intra-individual variability in well-being or what portion of this group might experience persistently high versus low well-being. Third, the role of depression in predicting people’s long-term well-being trajectories (stability versus change) is unknown. Study Design The present investigation used data from the ACL study, a nationally representative, long-term investigation of health and well-being in the United States. Participants ( N = 3,617) were recruited via stratified, multistage area probability sampling with oversampling of underrepresented groups (e.g., African Americans, adults over age 60) and completed five face-to-face interviews over the span of 25 years. The ACL dataset is well-suited for trajectory analysis due to its longitudinal design and rich psychosocial content. Based on an integrated model of well-being (Keyes et al., 2002; Ryff, 1989), we constructed a battery assessing hedonic (life satisfaction, affect) and eudaimonic dimensions (self-acceptance, mastery, positive relationships) of well-being. We used latent growth mixture models (LGMMs) to identify unobserved subpopulations (i.e., latent classes) typified by similar well-being trajectories and clarify how common sustained well-being is. Afterward, we used multinomial logistic regression analyses to investigate who tends to experience each well-being trajectory and whether depressive symptom severity impacts well-being sustainability. We anticipated that GMM would identify well-being trajectories for both sustained and fluctuating well-being groups, with prior studies on well-being dynamics informing an expectation of 3-6 qualitatively distinct trajectories (e.g., Ryff et al., 2015; Cintron & Ong, 2024). In line with prior findings on sociodemographic predictors of well-being (Fujita & Diener, 2005; Navarro-Carrillo et al., 2020; Ryff, 1995), we hypothesized that men and those of higher SES would be more likely to sustain high well-being. We further expected that elevated baseline depressive symptoms would be associated with well-being trajectory membership and lower likelihood of achieving sustained well-being. To address the concept of well-being turning points, we conducted exploratory analyses of people who made a transition from elevated depressive symptoms to high levels of well-being. ONLINE METHODS Participants Data were extracted from Waves 1, 2, 3, 4, and 5 of the ACL study (1986; 1989; 1994; 2002; and 2011; ACL: https://acl.isr.umich.edu), a nationally representative cohort longitudinal investigation of noninstitutionalized, English-speaking adults (age 25 and up) recruited via stratified, multistage area probability sampling and oversampling African Americans and people 65 or older. At baseline (i.e. Wave 1), respondents participated in a 90-minute face-to-face interview ( N = 3,617). Surviving respondents were re-interviewed in 1989 (N = 2,867, 83% of survivors), 1994 ( N = 2,562 with 164 proxy respondents, 83% of survivors), 2002 ( N = 1,787 with 95 proxy respondents, 74% of survivors), and 2012 ( N = 1,427 with 108 proxy respondents, 81% of survivors); proxy respondents, such as a spouse or nurse, completed questionnaires in collaboration with the surviving respondents when the surviving respondent was experiencing disabilities that would impede their ability to complete the measures (e.g., hearing loss). One thousand and seventy-one respondents (30%) completed all 5 waves, while 1,361 (38%) completed all possible waves prior to their death. Of those who dropped out ( N = 654, 18%), 264 (7%) completed only the first wave, 128 (3%) the first and second waves, 159 (5%) the first three waves, and 103 (3%) all but the fifth wave. The remaining 531 (15%) had intermittent response patterns. Participants were mostly female (62.45%), mostly Caucasian (64.22%), and ranged from age 25 to 95 at baseline ( M = 53.58; SD = 17.63). GMMs and subsequent analyses were conducted using responses from all participants who provided complete or near-complete data at three or more waves (N=2,420). Of these, participants were mostly female (63.68%), mostly Caucasian (68.51%), and ranged from age 25 to 90 at baseline ( M =49.15; SD =15.88). This sample contained 393 people with elevated baseline depressive symptoms (i.e. CESD-11≥8.859; Table 1 ). Nonresponse to three or more waves was associated with age (t=-22.56, p <.001), with nonresponders being significantly older (M=62.56) than responders (M=49.12). Sex and race were associated with nonresponse (χ2=4.51, p =.034; χ2=68.42, p <.001), with post-hoc comparisons revealing that women were less likely to respond to three or more waves than men, and people identifying as Black or Hispanic were less likely to respond to three or more waves than people identifying as other races. Materials Depressive Symptoms At each wave, depressive symptoms were assessed with the brief 11-item version of the Center for Epidemiologic Studies Depression Scale (CES-D-11; Kohout et al., 1993). The CES-D-11 includes items 2, 6, 7, 11, 12, 14, 15, 16, 18, 19, and 20 of the original CES-D (Radloff, 1977), where participants answer items using a 3-point scale ranging from 0 ( Hardly ever ) to 2 (Most of the time ). Responses for items 12 and 16 were reverse-coded. Total scores range from 0 to 22, with higher scores being indicative of greater depression severity. The CES-D-11 demonstrated strong internal consistency (α range: .82 to .85). A cut-off score of 16 on the CES-D is used to identify participants with significant depressive symptoms (Vilagut et al., 2016), and this score can be converted to the CES-D-11 equivalent via O’Hara and colleagues’ (1985) formula: CES-D = 1.866*CES-D-11 + 0.5318. Consequently, a cut-off score of 8.289 was used to identify participants with significant depressive symptoms. Well-being Life Satisfaction. At each wave, overall life satisfaction was assessed using a 5-point scale ranging from 1 ( Completely satisfied ) to 5 (Not at all satisfied ). The item was “Now please think about your life as a whole. How satisfied you are with it?” Frequent Positive Affect. At each wave, respondents answered a face-valid item assessing the frequency of positive affect using a 3-point scale ranging from 1 (H ardly ever ) to 3 ( Most of the time ). The item was, “In the past week [R] was happy.” Infrequent Negative Affect. At each wave respondents answered a face-valid item assessing the frequency of negative affect using a 3-point scale ranging from 1 (H ardly ever ) to 3 ( Most of the time ). The item was, “In the past week [R] was sad.” Self-acceptance. At each wave, one’s ability to accept and be satisfied with themselves was evaluated via an item from the Rosenberg Self- Esteem Scale (Rosenberg, 1965) using a 4-point scale ranging from 1 ( Strongly agree ) to 4 ( Strongly disagree ). The item was "I take a positive attitude toward myself.” Environmental Mastery. At each wave, one’s ability to manage their life and surroundings by effectively handling daily tasks and challenges, making use of available opportunities, and adapting the environment to suit personal needs was assessed via an item from the Pearlin Mastery Scale (Pearlin & Schooler, 1978) using a 4-point scale ranging from 1 ( Strongly Agree ) to 4 ( Strongly Disagree ). The item was “Sometimes I feel that I am being pushed around in life.” This item was reverse-scored. Positive Relationships with Others. At each wave, one’s ability to build and maintain positive relationships with others was assessed via three questions about the quality of relationships with their partner, child(ren), and friends. For each relationship, respondents answered the question, “How much does your [person] make you feel loved and cared for?” using a 5-point scale ranging from 1 ( A great deal ) to 5 ( Not at all ). These items demonstrated acceptable internal consistency (mean inter-item correlation range: .22 to .25) and were used to create a composite score reflecting one’s ability to build and maintain positive relationships with others. Data Analytic Plan All data manipulation and analysis were conducted in RStudio version 2022.12.0. Handling Missing Data To minimize bias in class estimation, well-being outcomes and demographic variables were retained as observed. Given their use as predictors and moderate levels of missingness, CES-D-11 values were imputed using Bayesian methods to support stable estimation in subsequent models. Patterns and amounts of missing data were examined and imputed via Bayesian methods using the mice package (van Buuren & Groothuis-Oudshoorn, 2011). To reduce estimation bias and power issues, Bayesian methods were used to create 40 imputed data sets for pooled analyses (Lee & Harring, 2023). Cleaning Data Prior to analyses, data were examined for multivariate normality using the mvn package (Korkmaz S, 2014) and variables were evaluated for univariate skew and kurtosis (Kim, 2013; West, Finch, & Curran, 1995). No variables exhibited absolute skew ≥ 2 or absolute kurtosis ≥ 7. Testing Well-Being Model Assumptions for Structure and Measurement Invariance To evaluate the structure of the well-being construct prior to modeling trajectories, we conducted a confirmatory factor analysis (CFA) using the lavaan package (Rosseel, 2012). We assessed model fit using the χ 2 goodness of fit test and standard model fit indices (CFI, RMSEA, and SRMR), applying conventional benchmarks for acceptable fit (Hooper, Coughlan, & Mullen, 2008: CFI>0.90; RMSEA<0.10; SRMR<0.08). Items with standardized loadings <0.40 were removed as they insufficiently represented the single underling factor. Model assumptions were tested via the REdaS package (Maier, 2022) and CFA models were conducted and fit via the psych package (Revelle, 2024). No model assumptions were violated, and the CFA model of well-being fit using pairwise deletion to address missing data and robust WLSMV estimators to estimate model parameters was of good fit (Robust CFI=.934; Robust RMSEA=.091, 90% CI: .078-.104; SRMR=.040). The single factor identified was labeled “Well-being” as items loading to it assessed people’s overall life satisfaction, frequency of positive affect, infrequency of negative affect, self-acceptance, environmental mastery, and positive relationships with others ( Figure S1 ). This factor exhibited acceptable internal consistency (α=.65), which was not unexpected given the relatively small number of items and their coverage of distinct facets of well-being. Additionally, we tested longitudinal measurement invariance to ensure the well-being construct could be interpreted equivalently across waves. As χ 2 tests are sensitive to sample size and may detect trivial misfit (Cheung & Rensvold, 2002), we used alternative fit criteria (ΔCFI<.01, ΔRMSEA<.015, ΔSRMR<.030) to evaluate metric invariance (Chen, 2007; Cheung & Rensvold, 2002). As we used ordinal indicators and fixed all intercepts to zero for model identification, establishing metric invariance allowed us to interpret mean-level changes over time as reflecting true changes in the latent construct (Liu et al., 2017). When testing measurement invariance across all five waves, the configural model demonstrated good fit (Robust CFI=.943; Robust RMSEA=.085, CI: .077-.093; SRMR=.037). As expected, the χ 2 test for differences between configural and metric models was significant ( p =.002), and the difference between model fit indices for the configural and metric models (ΔRCFI = -0.004, ΔRRMSEA = -0.011, and ΔSRMR = 0.004) did not surpass prespecified thresholds. Thus, full metric invariance was established and factor scores for the latent factor, Well-being, were extracted and used as the outcome variable in subsequent analyses. Thus, we extracted well-being factor scores for each wave, standardized them, and used these scores as input for subsequent growth mixture modeling. Modeling Well-being Over Time Growth mixture models fit via the lcmm package (Proust-Lima, Philipps, & Liquet, 2015) were used to estimate the prevalence of stable well-being over time and explore which, if any, demographic variables predicted said stability. GMM was used because it is ideal for uncovering unobserved heterogeneity in longitudinal data. All GMMs were fit using step-by-step procedures, including baseline growth model identification, and GMM specification, estimation, selection, and interpretation (Ram & Grimm, 2009). Following usual growth curve modeling procedures, the baseline growth model was identified by fitting a series of models (e.g., no growth, linear, quadratic). A critical step in model specification is class enumeration (i.e., selecting number of latent classes). To ensure robust class enumeration while avoiding overfitting, we performed 5-fold cross validation for each GMM fit (Grimm, Mazza, & Davoudzadeh, 2017). Thus, each series of GMMs was estimated five times: once to each distinct training set containing 80% of the data. To allow for the possibility of rejecting models with additional latent classes and thereby provide statistical evidence that we identified the appropriate number of classes, each series of GMMs was fit to include models with up to one more class than the best fitting model, starting with a 1-class model. The parameters for these models were then used as the basis for a series of GMMs fit to the corresponding hold-out set containing 20% of the data. For each fold, the −2 log-likelihood (-2LL) of each GMM model was saved, and the mean and standard deviation across all five validation attempts were calculated. Three approaches to determining the correct number of latent classes were implemented: 1. The model with the fewest classes whose mean cross-validated -2LL was within one standard error of the best fitting model was selected; 2. The model with the fewest classes whose mean cross-validated -2LL was outside one standard error of the model with one fewer class was selected; 3. The model with the fewest classes whose mean cross-validated -2LL was outside one standard error of the model with one more class was selected. Results from all three methods were used to determine the model of best fit, which was then estimated using all data. We inspected each model to ensure convergence and proper specification of variance-covariance structures (Ram & Grimm, 2009). Specifically, we sought to verify that the model of best fit outperformed other models (better fitting models have lower AIC and SABIC; VLMR-LRT p .80 and class-level entropy >.70 indicate adequate separation between latent classes and confident classification). Mean trajectories for each latent class were plotted, as good fitting models yield classes with expected or logical group differences, but poor fitting or overfit models may yield classes with considerable overlap or unexpected patterns of change (Ram & Grimm, 2009). Following final model selection, we interpreted the meaning of each latent class. To identify the percentage of people belonging to the stable well-being latent class, as well as what range of well-being those people tend to fall in and how stable their well-being was, we extracted the proportion of individuals in each class and used thresholds calculated from baseline data for people who reported minimal or no baseline depressive symptoms to characterize well-being. Scores within the bottom quartile were considered low, within the interquartile range, average, and within the upper quartile, high (Devendorf et al., 2022; Rottenberg et al., 2019). A score change of one or more quartiles between timepoints was interpreted as indicative of intra-individual variability (Ryff et al., 2015). Identifying Predictors of Well-being Trajectories A multinomial logistic regression was fit to assess whether demographic variables (i.e. age, gender, SES) or depressive symptom severity at baseline uniquely predicted latent class membership. The regression was specified with well-being trajectory class membership as the outcome variable and age, gender, SES, and depressive symptom severity at baseline as the predictors. Prior to fitting the regression model, we checked for multicollinearity among the independent variables: no predictors needed to be removed due to multicollinearity (VIF>10; Menard, 2010). As is standard, we estimated the variability accounted for by the model (via McFadden’s R 2 ) and the impact of significant predictor variables on the likelihood that a person belonged to the stable well-being trajectory (i.e. relative risk ratio). RESULTS Modeling Well-being Over Time Based on results when fitting 1-class baseline growth models ( Table 2 ), we fit both quadratic and sinusoidal models. Examination of -2LL estimates provided clear support for quadratic over the sinusoidal models but supported either three or six classes depending on the approach used for interpretation ( Figure 1a ). Given this, we fit quadratic GMMs with up to seven classes to the full data ( Table 3 ) and planned to use model fit indices to decide between the 3- and 6-class models. Unfortunately, none of the models fit demonstrated adequate entropy (>.80), even when refit with variance constrained across random intercepts. Having noted large variance in the intercept for each class across models, we plotted the mean trajectories of each latent class grouped by initial well-being level ( Figure S2 ). These plots revealed heterogeneity within latent classes, such that phenomenologically distinct trajectories were grouped together (e.g., each model identified a latent class that included people experiencing increasing and then sustained well-being and people sustaining relatively high well-being). Subsequent inspection of class-level entropies and individual-level posterior probabilities for class assignment revealed that average class confidence was acceptable for two classes (M PP : .80 and .72) but not the other (M PP =.61), and class confidence was ambiguous [1] for 505 (20.9%) participants. This indicated a need for alternative modeling approaches. To address model misfit driven by heterogeneity in baseline well-being, we reduced within-class variability by refitting models using well-being change scores. Across models, information criterion values reduced and entropy values increased, suggesting fit improvement. As before, examination of -2LL estimates provided clear support for quadratic over the sinusoidal models; however, examination of -2LL estimates now also provided clear support for a 4-class model ( Figure 1b ). Given this, we fit quadratic GMMs with up to five classes to the full data ( Table 3 ). Results were consistent with the five-fold cross-validation method for class enumeration: all information criterion indicated that a 4-class model was most appropriate and likelihood ratio tests supported a 4-class model. Variance in the intercept and heterogeneity within latent classes were more reasonable (i.e. lower) and plots of mean trajectories for each latent class yielded classes with logical group differences, even when grouped by initial well-being level ( Figure 2 ). Entropy was also much improved (.636 vs .491) and class-level entropies increased such that all were acceptable (>.70). Further, the number of ambiguously classified participants decreased (484 vs 505), the classification certainty for ambiguously classified participants improved (M MAX(PP) of .58, .58, .56, and .57 vs M MAX(PP) of .52, .52, and .48), and plots of ambiguously classified participants’ trajectories generally supported the assumption that they were correctly classified ( Figure S3 ). Overall, model fit improved across multiple metrics. Consistent with the idea that a persistent well-being group would predominate, class 1 encompassed most participants (N 1 =1,443, 59.63%), and members of this class had good probability of being assigned to it (M=.79, SD=.12). Class members tended to report intermediate levels of initial well-being (M=0.11, SD=.42), and their well-being levels were relatively stable over time. There was significant within-group variation in intercept but not slope, indicating that class members differed in their initial well-being levels but not in how their well-being changed over time. Thus, we interpret Class 1 as encompassing a group of people experiencing persistent well-being. To further characterize well-being levels in this group, we used thresholds calculated from baseline data for people who reported minimal or no baseline depressive symptoms: scores within the bottom quartile were considered low, within the interquartile range, average, and within the upper quartile, high. Interestingly, Class 1 was far from monolithic. In fact, it included people at every level of well-being, with 75% maintaining average or relatively high levels of well-being (N 1.mid =720, 49.90%; N 1.high =361, 25.02%) but also with a quarter maintaining relatively low levels of well-being (N 1.low =362, 25.09%). Furthermore, although class members sustained their well-being over time (stable average of well-being), being in class 1 did not preclude intra-individual variability. It was actually the minority of Class 1 members (n=413, 28.62%) who demonstrated perfect intra-individual consistency (i.e. no quartile change in well-being scores across the 5 timepoints). Class 4 was the next largest group, encompassing just under a third of participants (N 4 =736, 30.41%), and class members had similarly good probability of being assigned to it (M=.78, SD=.12). Class members tended to report low initial well-being (M=-0.26, SD=.40), and exhibited gradual increases in well-being and slowing of improvement over time. Significant within-group variation in intercept and little variation in slope indicated that although participants had different baseline well-being levels, their well-being generally followed a similar slow growth trajectory. Non-significant covariance between intercept and slope indicated that initial well-being level was not associated with different rates of improvement. Thus, Class 4 encompasses a group of people experiencing gradually increasing well-being. Almost all members (94.97%) of Class 4 began with low or average levels of well-being (N 4.low =440, 59.78%; N 4.mid =259, 35.19%), but a small minority had relatively high initial well-being (N 4.high =37, 5.03%). Of those with low initial well-being, most (n=265, 60.23%) achieved average levels of well-being and some (n=112, 25.45%) achieved relatively high levels of well-being. Around half (n=221, 58.62%) sustained the highest level of well-being they achieved, with 52 (12.27%) sustaining relatively high levels of well-being. The remainder experienced regression toward their baseline well-being level, with 100 (37.74%) regressing to average well-being levels and 14 (5.28%) regressing back to relatively low well-being levels. Of those with average initial well-being, 205 (79.15%) achieved relatively high levels of well-being, and around half (n=109, 53.17%) sustained that new, relatively high level of well-being. Others (n=73, 35.61%) regressed toward their baseline well-being level, and some (n=23, 11.22%) experienced significant declines in well-being such that their last documented level of well-being was relatively low. In comparison, far fewer participants (N 2 =183, 7.56%) belonged to Class 2. Still, class members demonstrated good probability of being assigned to it (M=.79, SD=.16). Class members tended to report relatively high initial well-being (M=0.37, SD=.15), and demonstrated steep declines in well-being over time. Significant within-group variation in intercept, minimal variation in slope, and nonsignificant covariance between intercept and slope indicated that while participants in this class had different baseline well-being levels, they experienced similar rates of decline. Overall, Class 2 encompasses a group of people experiencing declining well-being. Almost all people (90.71%) in Class 2 began with high or average levels of well-being (N 2.high =91, 49.73%; N 2.mid =75, 40.98%), but some had relatively low baseline well-being (N 2.low =17, 9.29%). Of those who began with high well-being, 18 (19.78%) went on to experience more typical (or average) levels of well-being and 73 (80.22%) went on to experience relatively low well-being. Six (6.59%) managed to recover their initial, higher level of well-being, 20 (21.98%) experienced improvements in well-being after their initial decline, and 65 (69.75%) maintained their new, lower level of well-being. Of those who began with average well-being, all went on to experience relatively low levels of well-being and 56 (74.67%) maintained that new, relatively low level of well-being. That said, 16 (21.33%) recovered their initial level of well-being and 3 (4%) went on to experience relatively high levels of well-being. Finally, the smallest group was Class 3 (N 3 =58, 2.4%). Class members demonstrated high probability of being assigned to it (M=.82, SD=.17) and low probability of being assigned to other classes. Like members of Class 4, members of Class 3 tended to report low initial well-being (M=-0.78, SD=.43). However, their average initial well-being scores were much lower (-0.78 vs -0.26), and they demonstrated more marked increases in well-being than those in Class 4. They were also more likely to regress toward their initial well-being level than participants in Class 4. As in other classes, participants experienced similarly strong upward trajectories in well-being regardless of differences in baseline well-being. Overall, Class 3 encompasses a small group of people experiencing marked increases in well-being. Almost all people in this class (n=57, 98.28%) began with low levels of well-being: of those, a third (n=19, 33.33%) achieved average levels of well-being, and most (n=36, 63.16%) achieved relatively high levels of well-being. Half (n=28, 49.12%) sustained the highest level of well-being they achieved, with 18 sustaining relatively high levels of well-being. Identifying Predictors of Well-being Trajectories As baseline depressive symptoms and well-being were highly correlated ( r =-0.76), we included baseline well-being level in analyses ( Appendix SA includes results without this adjustment [2] ). Additionally, because 20% of participants (n=484) were classified with uncertainty, we ran models with full and reduced datasets: results were consistent across both. Sex and age at baseline were not significant predictors of well-being trajectory class, but baseline SES, depressive symptom severity, and well-being level were ( Table 4; Figure 3 ). Higher SES at baseline was associated with greater odds of gradual well-being improvement and lower odds of declining well-being. Importantly, baseline depressive symptoms were predictive of well-being class membership. When accounting for baseline well-being level, more severe baseline depressive symptoms were associated with increased odds of being in either the declining well-being or the markedly increasing well-being classes: a one-unit increase in baseline depressive symptom severity was associated with a 1.10-fold increase in the odds of declining well-being class membership and a 1.20-fold increase in the odds of markedly increasing well-being class membership. Higher baseline well-being was associated with increased odds of being in the declining well-being class and reduced the odds of being in the increases in well-being class. Each quartile increase in baseline well-being was associated with a 3.47-fold increase in the odds of being in the declining well-being class, and a respective 98% and 69% reduction in the odds of being in the marked or gradually increasing well-being classes. Overall, these findings indicate that baseline well-being is the strongest predictor of well-being change direction , whereas baseline depressive symptom severity is a strong predictor of well-being change likelihood . People with lower baseline well-being were more likely to experience improvements in well-being, and people with elevated baseline depressive symptoms were more likely to experience change in well-being. Thus, people with both low well-being and elevated depressive symptoms at baseline had the highest odds of experiencing marked improvements in well-being. In this sample, 57% of people with both low well-being and elevated depressive symptoms at baseline experienced increases in well-being. Understanding Transitions from Depression Symptoms to Well-being Three hundred and ninety-six people exhibited elevated depressive symptoms at baseline (i.e. CESD-11≥8.289). While baseline depressive symptom severity was a robust predictor of less stable well-being over time, many people with elevated baseline depressive symptoms (n=169, 42.67%) exhibited persistent well-being. Of those in the persistent well-being class, most began with low well-being (n=146, 85%) and experienced stable persistent low well-being (n=98, 57%), but others experienced high intra-individual variability despite demonstrating persistent well-being (n=70, 41%), and five reported one or more instances of relatively high well-being. Fifty-six other people with elevated baseline depressive symptoms also experienced at least one instance of relatively high well-being over the five assessment points. In total, 15.4% of people with elevated baseline depressive symptoms went on to report at least one instance of elevated well-being and 2.5% sustained the elevated level of well-being they achieved ( Figure 4 ). DISCUSSION How consistent is human well-being over decades? Supportive of perspectives that view consistency as normative, persistent well-being was the most common trajectory in this sample of U.S. adults, with around 60% of the sample demonstrating little change from baseline well-being 25 years later. However, results also lend credence to perspectives that highlight intra-individual change in well-being, as only 23% of people maintained their baseline well-being level at all five timepoints and we identified three other meaningful classes: gradually increasing well-being, declining well-being, and markedly increasing well-being. Thus—although persistent well-being is common—it does not necessarily indicate stable well-being. When well-being is assessed at enough time points and with short enough lags, meaningful intra-individual change (i.e. instability) in well-being is common. While persistent well-being was generally the healthiest class, membership in this class did not guarantee high levels of well-being. A quarter of people in the persistent well-being class reported consistently low well-being. Although often treated as part of a psychologically “healthy” reference group, this subgroup’s existence highlights the need to distinguish persistent or stable well-being that might reflect resilience and that which might reflect enduring difficulty (i.e. languishing). That over half of those experiencing persistently low well-being did not report elevated baseline depressive symptoms, supports the dual continua model and peril of equating the absence of well-being with the presence of psychopathology (Keyes, 2005; Westerhof & Keyes, 2010). Demographic features had a muted role in predicting well-being trajectories. Despite robust prior association between SES and well-being (Navarro-Carrillo et al., 2020; Tan et al., 2020), higher SES was only modestly associated with reduced odds of declining well-being. Thus, although elevated SES may increase access to helpful resources and decrease exposure to chronic stressors, these changes seem to protect well-being over time rather than enhance it. In contrast, baseline well-being level and depressive symptom severity were both strong predictors of well-being trajectory. Indeed, both low baseline well-being and elevated baseline depressive symptoms were associated with subsequent improvements in well-being; however, those with elevated baseline depressive symptoms were also less likely to sustain well-being improvements. These differences between the gradually and markedly improving classes further indicate that depressive symptoms may predict less stable well-being trajectories. Why depressive symptom severity is associated with less stable well-being trajectories is difficult to explain from our research design, but prior evidence on the heterogeneity of depression (Goldberg, 2011; Lynall & McIntosh, 2023) may help. People who experience depression with more cognitive symptoms may have higher odds of experiencing declining well-being (Liao et al., 2022). Likewise, depressed people prone to heightened affective variability may have higher odds of experiencing early well-being gains followed by reversal (Bylsma et al., 2011). In contrast, people whose symptoms began in response to addressable life stressors, may be more likely to experience well-being turning points (Monroe et al., 2019). Other factors—such as treatment-seeking, effective emotion regulation, and engagement in health-related behaviors (e.g., regular exercise)—may also drive well-being change direction among people with elevated baseline depressive symptoms. Understanding which of these factors is most decisive may provide leverage for initiating improvements. Regardless, the fact that both depressive symptom severity and well-being were independent significant predictors of well-being trajectory supports recent calls for depression treatment to go beyond symptom reduction by incorporating activities that help clients recognize, cultivate, and sustain well-being (Fava et al., 2017; 1998). Of course, this study’s results should be considered in the context of several limitations. This was a secondary analysis of archival data, which constrained the measurement instruments available. Nevertheless, our well-being battery included all facets of Diener’s (1984) model of well-being and several key facets of Ryff’s (1989) model of well-being and demonstrated acceptable reliability. Second, while the ACL data allowed us to model five timepoints over 25 years, the time between waves increased in a near exponential pattern, with the longest time lag being 10 years. This complicates claims that similar well-being scores across later waves indicate well-being stability , as the increasing interval between waves decreases the certainty that a person’s well-being did not change between assessments. Although the intervals between waves in the ACL are not unusual, it would be valuable if future studies examined well-being trajectories over smaller, more uniform intervals. Finally, the ACL used a cutoff from a validated measure of self-reported depressive symptom severity rather than clinical diagnostic interviews to identify those with significant depression. Although self-report measures are more efficient for large-scale data collection, they may be influenced by individual differences in insight and recall bias and may not differentiate transient distress from clinically significant depression. Despite these limitations, our analyses provide novel insights on typical well-being trajectories and how these trajectories may be altered by depression. We were able to interpret all classes identified in terms of well-being thresholds created using a well-accepted population-based approach and provided initial insight into how often people with depression sustain subsequently achieved high well-being. Finally, while generalizability is always uncertain, it should be underscored that the ACL was a nationally representative longitudinal cohort study of noninstitutionalized, English-speaking adults recruited via stratified, multistage area probability sampling with an oversampling African Americans and people 65 or older. References Bylsma, L. M., et al. (2011). "Emotional reactivity to daily events in major and minor depression." Journal of Abnormal Psychology 120(1): 155. Chen, F. F. (2007). Sensitivity of Goodness of Fit Indexes to Lack of Measurement Invariance. Structural Equation Modeling: A Multidisciplinary Journal, 14 (3), 464-504. https://doi.org/10.1080/10705510701301834 Cheung, G. W., & Rensvold, R. B. (2002). Evaluating Goodness-of-Fit Indexes for Testing Measurement Invariance. Structural Equation Modeling: A Multidisciplinary Journal, 9 (2), 233-255. https://doi.org/10.1207/s15328007sem0902_5 Chevance, A., Ravaud, P., Tomlinson, A., Le Berre, C., Teufer, B., Touboul, S., Fried, E. I., Gartlehner, G., Cipriani, A., & Tran, V. T. (2020, 2020/08/01/). Identifying outcomes for depression that matter to patients, informal caregivers, and health-care professionals: qualitative content analysis of a large international online survey. The Lancet Psychiatry, 7 (8), 692-702. https://doi.org/https://doi.org/10.1016/S2215-0366(20)30191-7 Cintron, D. W., & Ong, A. D. (2024). Trajectories of affective well-being and survival in middle-aged and older adults. Emotion, 24 (5), 1149-1156. https://doi.org/10.1037/emo0001341 Demyttenaere, K., Donneau, A.-F., Albert, A., Ansseau, M., Constant, E., & Van Heeringen, K. (2015). What is important in being cured from depression? Discordance between physicians and patients (1). Journal of Affective Disorders, 174 , 390-396. https://doi.org/10.1016/j.jad.2014.12.004 Devendorf, A. R., Rum, R., Kashdan, T. B., & Rottenberg, J. (2022). Optimal Well-Being After Psychopathology: Prevalence and Correlates. Clinical Psychological Science, 10 (5), 981-996. https://doi.org/10.1177/21677026221078872 Diener, E. (1984). Subjective Well-being. Psychological Bulletin, 95 (3), 542-575. Disabato, D., Goodman, F., & Kashdan, T. B. (2019). A hierarchical framework for the measurement of well-being. Eid, M., & Diener, E. (2004). Global Judgments of Subjective Well-Being: Situational Variability and Long-Term Stability. Social Indicators Research, 65 (3), 245-277. https://doi.org/10.1023/b:soci.0000003801.89195.bc Fava, G. A., Cosci, F., Guidi, J., & Tomba, E. (2017). Well-being therapy in depression: New insights into the role of psychological well-being in the clinical process. Depression and Anxiety, 34 (9), 801-808. https://doi.org/10.1002/da.22629 Fava, G. A., Rafanelli, C., Cazzaro, M., Conti, S., & Grandi, S. (1998). Well-being therapy. A novel psychotherapeutic approach for residual symptoms of affective disorders. Psychological Medicine, 28 (2), 475-480. https://doi.org/10.1017/s0033291797006363 Frijda, N. H. (1988). The laws of emotion. American Psychologist, 43 , 349-358. https://doi.org/10.1037/0003-066X.43.5.349 Fujita, F., & Diener, E. (2005). Life Satisfaction Set Point: Stability and Change. Journal of Personality and Social Psychology, 88 , 158-164. https://doi.org/10.1037/0022-3514.88.1.158 Gallagher, M. W., Lopez, S. J., & Preacher, K. J. (2009, Aug). The hierarchical structure of well-being. Journal of Personality, 77 (4), 1025-1050. https://doi.org/10.1111/j.1467-6494.2009.00573.x Goldberg, D. (2011). The heterogeneity of “major depression”. World Psychiatry, 10 (3), 226-228. https://doi.org/10.1002/j.2051-5545.2011.tb00061.x Grant, F., Guille, C., & Sen, S. (2013). Well-Being and the Risk of Depression under Stress. PloS One, 8 (7), e67395. https://doi.org/10.1371/journal.pone.0067395 Headey, B., & Muffels, R. (2016, 2016/11/01). Towards a Theory of Medium Term Life Satisfaction: Two-Way Causation Partly Explains Persistent Satisfaction or Dissatisfaction. Social Indicators Research, 129 (2), 937-960. https://doi.org/10.1007/s11205-015-1146-8 Hooper, D., Coughlan, J., & Mullen, M. (2008, 11/30). Structural Equation Modeling: Guidelines for Determining Model Fit. The Electronic Journal of Business Research Methods, 6 . Huppert, F. A. (2009). Psychological Well-being: Evidence Regarding its Causes and Consequences. Applied Psychology: Health and Well-Being, 1 (2), 137-164. https://doi.org/10.1111/j.1758-0854.2009.01008.x Joshanloo, M. (2022). Longitudinal Relations Between Depressive Symptoms and Life Satisfaction Over 15 Years. Applied Research in Quality of Life, 17 (5), 3115-3130. https://doi.org/10.1007/s11482-022-10055-x Joshanloo, M., & Blasco-Belled, A. (2023, Jan 29). Reciprocal Associations between Depressive Symptoms, Life Satisfaction, and Eudaimonic Well-Being in Older Adults over a 16-Year Period. International Journal of Environmental Research and Public Health, 20 (3). https://doi.org/10.3390/ijerph20032374 Keyes, C., & Annas, J. (2009). Feeling good and functioning well: Distinctive concepts in ancient philosophy and contemporary science. The Journal of Positive Psychology, 4 (3), 197-201. Keyes, C. L. M. (2005). Mental Illness and/or Mental Health? Investigating Axioms of the Complete State Model of Health [doi:10.1037/0022-006X.73.3.539]. American Psychological Association. Keyes, C. L. M., Dhingra, S. S., & Simoes, E. J. (2010). Change in Level of Positive Mental Health as a Predictor of Future Risk of Mental Illness. American Journal of Public Health, 100 (12), 2366-2371. https://doi.org/10.2105/ajph.2010.192245 Keyes, C. L. M., Shmotkin, D., & Ryff, C. D. (2002). Optimizing well-being: The empirical encounter of two traditions. Journal of Personality and Social Psychology, 82 (6), 1007-1022. https://doi.org/10.1037/0022-3514.82.6.1007 Keyes, C. L. M., & Simoes, E. J. (2012). To Flourish or Not: Positive Mental Health and All-Cause Mortality. American Journal of Public Health, 102 (11), 2164-2172. https://doi.org/10.2105/ajph.2012.300918 Kim, H.-Y. (2013). Statistical notes for clinical researchers: assessing normal distribution (2) using skewness and kurtosis. Restorative Dentistry & Endodontics, 38 (1), 52. https://doi.org/10.5395/rde.2013.38.1.52 Kohout, F. J., Berkman, L. F., Evans, D. A., & Cornoni-Huntley, J. (1993, May). Two shorter forms of the CES-D (Center for Epidemiological Studies Depression) depression symptoms index. Journal of Aging and Health, 5 (2), 179-193. https://doi.org/10.1177/089826439300500202 Koivumaa-Honkanen, H., Kaprio, J., Honkanen, R., Viinam⍰Ki, H., & Koskenvuo, M. (2004). Life satisfaction and depression in a 15-year follow-up of healthy adults. Social Psychiatry and Psychiatric Epidemiology, 39 (12), 994-999. https://doi.org/10.1007/s00127-004-0833-6 Korkmaz S, G. D., Zararsiz G. (2014). MVN: An R Package for Assessing Multivariate Normality. , 6 (2), 151–162. https://journal.r-project.org/archive/2014-2/korkmaz-goksuluk-zararsiz.pdf Kyriazos, T. A. (2018). Applied Psychometrics: The 3-Faced Construct Validation Method, a Routine for Evaluating a Factor Structure. Psychology (Savannah, Ga.), 09 (08), 2044-2072. https://doi.org/10.4236/psych.2018.98117 Lamers, S. M. A., Westerhof, G. J., Glas, C. A. W., & Bohlmeijer, E. T. (2015). The bidirectional relation between positive mental health and psychopathology in a longitudinal representative panel study. The Journal of Positive Psychology, 10 (6), 553-560. https://doi.org/10.1080/17439760.2015.1015156 Lee, D. Y., & Harring, J. R. (2023). Handling Missing Data in Growth Mixture Models. Journal of Educational and Behavioral Statistics , 107699862211491. https://doi.org/10.3102/10769986221149140 Liao, Y., Zhang, H., Guo, L., Fan, B., Wang, W., Teopiz, K. M., Lui, L. M. W., Lee, Y., Li, L., Han, X., Lu, C., & McIntyre, R. S. (2022, 2022/10/15/). Impact of cognitive-affective and somatic symptoms in subthreshold depression transition in adults: Evidence from Depression Cohort in China (DCC). Journal of Affective Disorders, 315 , 274-281. https://doi.org/https://doi.org/10.1016/j.jad.2022.08.009 Lischetzke, T. and M. Eid (2003). "Is Attention to Feelings Beneficial or Detrimental to Affective Well-Being? Mood Regulation as a Moderator Variable." Emotion 3(4): 361-377. Liu, Y., Millsap, R. E., West, S. G., Tein, J. Y., Tanaka, R., & Grimm, K. J. (2017, Sep). Testing measurement invariance in longitudinal data with ordered-categorical measures. Psychological Methods, 22 (3), 486-506. https://doi.org/10.1037/met0000075 Lucas, R. E. (2007). Adaptation and the Set-Point Model of Subjective Well-Being:Does Happiness Change After Major Life Events? Current Directions in Psychological Science, 16 (2), 75-79. https://doi.org/10.1111/j.1467-8721.2007.00479.x Lykken, D., & Tellegen, A. (1996). Happiness Is a Stochastic Phenomenon. Psychological Science, 7 (3), 186-189. https://doi.org/10.1111/j.1467-9280.1996.tb00355.x Lynall, M.-E., & McIntosh, A. M. (2023). The Heterogeneity of Depression. American Journal of Psychiatry, 180 (10), 703-704. https://doi.org/10.1176/appi.ajp.20230574 Maier, M. (2022). Companion Package to the Book “R: Einführung durch angewandte Statistik”. . In (Version 0.9.4) https://CRAN.R-project.org/package=REdaS Menard, S. (2010). Logistic Regression: From Introductory to Advanced Concepts and Applications https://doi.org/10.4135/9781483348964 Monroe, S. M., Anderson, S. F., & Harkness, K. L. (2019). Life stress and major depression: The mysteries of recurrences. Psychological Review, 126 (6), 791-816. https://doi.org/10.1037/rev0000157 National Academies of Sciences, E., & Medicine. (2022). Ontologies in the Behavioral Sciences: Accelerating Research and the Spread of Knowledge . The National Academies Press. https://doi.org/doi:10.17226/26464 Navarro-Carrillo, G., Alonso-Ferres, M., Moya, M., & Valor-Segura, I. (2020, 2020-June-10). Socioeconomic Status and Psychological Well-Being: Revisiting the Role of Subjective Socioeconomic Status [Original Research]. Frontiers in Psychology, 11 . https://doi.org/10.3389/fpsyg.2020.01303 Nesse, R. M. (2000). Is Depression an Adaptation? Archives of General Psychiatry, 57 (1), 14. https://doi.org/10.1001/archpsyc.57.1.14 O'Hara, M. W., Kohout, F. J., & Wallace, R. B. (1985). Depression among the Rural Elderly A Study of Prevalence and Correlates. The Journal of nervous and mental disease, 173 (10), 582-589. https://journals.lww.com/jonmd/fulltext/1985/10000/depression_among_the_rural_elderly_a_study_of.2.aspx Park, C. L., Kubzansky, L. D., Chafouleas, S. M., Davidson, R. J., Keltner, D., Parsafar, P., Conwell, Y., Martin, M. Y., Hanmer, J., & Wang, K. H. (2023). Emotional Well-Being: What It Is and Why It Matters. Affective Science, 4(1), 10-20. https://doi.org/10.1007/s42761-022-00163-0 Panaite, V., Rottenberg, J., & Bylsma, L. M. (2020). Daily Affective Dynamics Predict Depression Symptom Trajectories Among Adults with Major and Minor Depression. Affective Science, 1 (3), 186-198. https://doi.org/10.1007/s42761-020-00014-w Pearlin, L. I., & Schooler, C. (1978). The Structure of Coping. Journal of Health and Social Behavior, 19 (1), 2-21. https://doi.org/10.2307/2136319 Pinquart, M., & Sörensen, S. (2000). Influences of socioeconomic status, social network, and competence on subjective well-being in later life: a meta-analysis. Psychology and Aging, 15 (2), 187. Proust-Lima, C., Philipps, V., & Liquet, B. (2015). Estimation of extended mixed models using latent classes and latent processes: the R package lcmm. arXiv preprint arXiv:1503.00890 . Radloff, L. S. (1977). The CES-D Scale. Applied Psychological Measurement, 1 (3), 385-401. https://doi.org/10.1177/014662167700100306 Ram, N., & Grimm, K. J. (2009). Methods and Measures: Growth mixture modeling: A method for identifying differences in longitudinal change among unobserved groups. International Journal of Behavioral Development, 33 (6), 565-576. https://doi.org/10.1177/0165025409343765 Revelle, W. (2024). psych: Procedures for Psychological, Psychometric, and Personality Research. In R package (Version 2.4.1) https://cran.r-project.org/package=psych Ridge, D., & Ziebland, S. (2006). “The Old Me Could Never Have Done That”: How People Give Meaning to Recovery Following Depression. Qualitative Health Research, 16 (8), 1038-1053. https://doi.org/10.1177/1049732306292132 Rosenberg, M. (1965). Society and the Adolescent Self-image . Princeton University Press. Rosseel, Y. (2012, 05/24). lavaan: An R Package for Structural Equation Modeling. Journal of Statistical Software, 48 (2), 1 - 36. https://doi.org/10.18637/jss.v048.i02 Rottenberg, J., Devendorf, A. R., Panaite, V., Disabato, D. J., & Kashdan, T. B. (2019). Optimal Well-Being After Major Depression. Clinical Psychological Science, 7 (3), 621-627. https://doi.org/10.1177/2167702618812708 Ryan, R. M., & Deci, E. L. (2001). On happiness and human potentials: A review of research on hedonic and eudaimonic well-being. Annual Review of Psychology, 52 , 141. Ryff, C. D. (1989). Happiness is everything, or is it? Explorations on the meaning of psychological well-being. Journal of Personality and Social Psychology, 57 (6), 1069. Ryff, C. D. (1995). Psychological Well-Being in Adult Life. Current Directions in Psychological Science, 4 (4), 99-104. http://www.jstor.org/stable/20182342 Ryff, C. D. (2023). Flotsam, Jetsam, and Forward-Moving Vessels on the Sea of Well-Being. Affective Science, 4(1), 49-51. https://doi.org/10.1007/s42761-022-00162-1 Ryff, C. D., & Keyes, C. L. M. (1995). The structure of psychological well-being revisited. Journal of Personality and Social Psychology, 69 (4), 719. Ryff, C. D., Radler, B. T., & Friedman, E. M. (2015). Persistent psychological well-being predicts improved self-rated health over 9–10 years: Longitudinal evidence from MIDUS. Health Psychology Open, 2 (2), 205510291560158. https://doi.org/10.1177/2055102915601582 Samuelson, W., & Zeckhauser, R. (1988, 1988/03/01). Status quo bias in decision making. Journal of Risk and Uncertainty, 1 (1), 7-59. https://doi.org/10.1007/BF00055564 Swann, W. B., & Read, S. J. (1981, 1981/07/01/). Self-verification processes: How we sustain our self-conceptions. Journal of Experimental Social Psychology, 17 (4), 351-372. https://doi.org/https://doi.org/10.1016/0022-1031(81)90043-3 Tan, J. J. X., Kraus, M. W., Carpenter, N. C., & Adler, N. E. (2020). The association between objective and subjective socioeconomic status and subjective well-being: A meta-analytic review. Psychological Bulletin, 146 (11), 970-1020. https://doi.org/10.1037/bul0000258 Thompson, R. J., et al. (2013). "The Role of Attention to Emotion in Recovery from Major Depressive Disorder." Depression Research and Treatment 2013: 1-6. https://dx.doi.org/10.1155/2013/540726 van Buuren, S., & Groothuis-Oudshoorn, K. (2011). mice: Multivariate Imputation by Chained Equations in R. Journal of Statistical Software, 45 (3). https://www.jstatsoft.org/v45/i03/ Vilagut, G., Forero, C. G., Barbaglia, G., & Alonso, J. (2016). Screening for Depression in the General Population with the Center for Epidemiologic Studies Depression (CES-D): A Systematic Review with Meta-Analysis. PloS One, 11 (5), e0155431. https://doi.org/10.1371/journal.pone.0155431 West, S. G., Finch, J. F., & Curran, P. J. (1995). Structural equation models with nonnormal variables: Problems and remedies. Westerhof, G. J., & Keyes, C. L. M. (2010). Mental Illness and Mental Health: The Two Continua Model Across the Lifespan. Journal of Adult Development, 17 (2), 110-119. https://doi.org/10.1007/s10804-009-9082-y Wood, A. M., & Joseph, S. (2010, 2010/05/01/). The absence of positive psychological (eudemonic) well-being as a risk factor for depression: A ten year cohort study. Journal of Affective Disorders, 122 (3), 213-217. https://doi.org/https://doi.org/10.1016/j.jad.2009.06.032 Wood, A. M., & Tarrier, N. (2010, 2010/11/01/). Positive Clinical Psychology: A new vision and strategy for integrated research and practice. Clinical Psychology Review, 30 (7), 819-829. https://doi.org/https://doi.org/10.1016/j.cpr.2010.06.003 Zimmerman, M., McGlinchey, J. B., Posternak, M. A., Friedman, M., Attiullah, N., & Boerescu, D. (2006). How Should Remission From Depression Be Defined? The Depressed Patient’s Perspective. American Journal of Psychiatry, 163 (1), 148-150. https://doi.org/10.1176/appi.ajp.163.1.148 Declarations As defined at 45 CFR 46.102, the analysis of de-identified, publicly available data does not constitute human subjects research and thus does not require IRB review. Footnotes 1 2 The predictive effect of baseline depressive symptoms differed across models. Without baseline well-being level included, more severe baseline depressive symptoms were associated with decreased odds of experiencing declining well-being and increased odds of experiencing either gradually or markedly increasing well-being. Tables Table 1. Final Sample Characteristics at Baseline ( N=2,420 ) Characteristic N (%) Sex (male) 879 (36.32%) Race Caucasian 1,658 (68.51%) African American 693 (28.64%) Other 69 (2.85%) SES Low SES ≥11 years education and family income $20,000 ≤12 years of education and family income $20,000 795 (32.85%) High SES ≥16 years education and family income >$20,000 340 (14.05%) Age ( M, SD ) 49.15 (15.88) Baseline Depressive Symptom Severity Likely Not Depressed CES-D-11≤8.289 (n, % ; M, SD) 2020 (83.47%); 3.16 (2.47) Likely Depressed CES-D-11>8.289 (n, % ; M, SD) 396 (16.36%); 11.37 (2.51) Well-being Low Well-being WBLV < -.16 876 (36.20%) Lower-Typical Well-being -.16≤WBLV<.13 517 (21.36%) Upper-Typical Well-being .13≤WBLV<.39 538 (22.23%) Optimal Well-being WBLV≥.39 489 (20.21%) Note. CES-D=depressive symptom severity score. WBLV=well-being latent variable score. Thresholds for well-being classes were determined by identifying the 25 th , 50 th , and 75 th quartiles of well-being among people who were likely not depressed at baseline. Table 2. Fit Statistics for Baseline Growth Model and Growth Mixture Models (N=2420) 1-Class Linear 1-Class Exponential 1-Class Logarithmic 1-Class Sinusoidal 1-Class Quadratic Number of Parameters 6 7 10 10 7 LL -12,666 -19,307 -12,658 -12,654 -12,661 AIC 25,344 38,627 25,335 25,327* 25,336 BIC 25,379 38,668 25,393 25,386 25,376* CAIC 25,344 38,628 25,336 25,328* 25,336 SABIC 25,360 38,646 25,362 25,354* 25,354* Entropy 1.00 1.00 1.00 1.00 1.00 Note. (C)AIC = (Corrected) Akaike Information Criterion. (SA)BIC = (Sample-Adjusted) Bayes Information Criterion. VLMR LRT = Voung-Lo-Mendell-Rubin Likelihood Ratio Test. *model of best fit for each information criterion Table 3. Fit Statistics for Growth Mixture Models using Well-being Change Scores (N=2420) 1-Class 2-Class 3-Class 4-Class 5-Class Sample Size A N c=1 2,420 1,499 1,751 1,443 1,356 N c=2 921 336 736 515 N c=3 333 183 318 N c=4 58 188 N c=5 43 Fit Statistics Number of Parameters 7 12 17 22 27 LL -4,332 -4,063 -3,948 -3,906 -3,901 AIC 8,678 8,149 7,929 7,856 7,855 BIC 8,716 8,219 8,028 7,983 8,011 CAIC 8,718 8,219 8,028 7,983 8,011 SABIC 8,696 8,181 7,934 7,913 7,925 Entropy 1.00 0.480 0.612 0.636 0.551 Adjusted VLMR LRT 516.28** 220.51** 79.86** 10.49 Note. A) Estimated counts for the latent classes based on the posterior probabilities. LL = Log-likelihood. (C)AIC = (Corrected) Akaike Information Criterion. (SA)BIC = (Sample-Adjusted) Bayes Information Criterion. VLMR LRT = Voung-Lo-Mendell-Rubin Likelihood Ratio Test; values are for tests of n vs n-1 classes. All information criteria are rounded to the nearest whole number. Bold font indicates the model of best fit for each information criterion. ** p <.001, * p <.05 Table 4. Multinomial Logistic Regression Model Full Data (N=2420) Reduced Data (N=1936) Term β St. Err Exp(β) 95% CI for Exp(β) β St. Err Exp(β) 95% CI for Exp(β) Class 2 Int. -4.005** 0.543 -4.776** 0.663 Age -0.017* 0.005 0.98 0.97, 0.99 -0.016* 0.006 0.98 0.97, 1.00 Sex Female 0.189 0.174 1.21 0.86, 1.70 0.272 0.212 1.31 0.87, 1.99 SES LowerMC -0.432* 0.215 0.65 0.43, 0.99 -0.249 0.262 0.78 0.47, 1.30 SES UpperMC -0.684* 0.229 0.50 0.32, 0.79 -0.517 0.276 0.60 0.35, 1.03 SES Upper -0.784* 0.301 0.46 0.25, 0.82 -0.753* 0.379 0.47 0.22, 0.99 CES-D -0.095* 0.031 1.10 1.03, 1.17 -0.105* 0.037 1.11 1.03, 1.20 WBLV 1.244* 0.155 3.47 2.56, 4.71 1.257** 0.187 3.52 2.44, 5.07 Class 3 Int. -0.521** 1.334 1.134** 0.420 Age 0.015 0.010 1.02 1.00, 1.03 0.008 0.011 1.01 0.99, 1.03 Sex Female 0.007 0.321 1.01 0.54, 1.89 0.126 0.366 1.13 0.55, 2.32 SES LowerMC 0.282 0.365 1.33 0.65, 2.71 0.059 0.405 1.06 0.48, 2.35 SES UpperMC 0.258 0.389 1.29 0.60, 2.78 -0.044 0.425 0.96 0.42, 2.20 SES Upper -0.369 0.671 0.69 0.19, 2.58 -0.718 0.800 0.49 0.10. 2.34 CES-D 0.183** 0.037 1.20 1.12, 1.29 0.192** 0.040 1.21 1.12, 1.31 WBLV -3.838** 1.023 0.02 0.00, 0.16 -1.527** 0.420 0.00 0.00, 0.00 Class 4 Int. 0.933** 0.311 -1.433** 0.356 Age 0.003 0.003 1.00 1.00, 1.01 -0.0009 0.004 1.00 0.99, 1.01 Sex Female 0.064 0.104 1.07 0.87, 1.31 0.131 0.120 1.14 0.90, 1.44 SES LowerMC 0.129 0.139 1.14 0.87, 1.49 0.418 0.161 1.15 .084, 1.58 SES UpperMC 0.181 0.142 1.20 0.91, 1.58 0.052 0.164 1.05 0.76, 1.45 SES Upper 0.308 0.174 1.36 0.97, 1.91 0.263 0.201 1.30 0.88, 1.93 CES-D 0.018 0.016 1.02 0.99, 1.05 0.010 0.018 1.01 0.97, 1.05 WBLV -1.187** 0.099 0.31 0.25, 0.37 -1.438** 0.115 0.24 0.19, 0.30 Note . MC=Middle Class. CES-D=depressive symptom severity score. WBLV=Well-being latent variable score. * p <.05, ** p <.001 Additional Declarations There is NO Competing Interest. Supplementary Files WBTrajectoriesSuppl5.26.25.docx Supplemental Figures and Analyses Cite Share Download PDF Status: Under Review Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7123882","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Analysis","associatedPublications":[],"authors":[{"id":489393810,"identity":"6bc70d24-f034-4984-a8cd-23e9bcabd2ab","order_by":0,"name":"Cassondra Lyman","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAz0lEQVRIiWNgGAWjYDACdubjPz/+sYFwHhQAiQOEtDCzJUhLNqRBOAkGRGnhMZDgbThMghb+ZgYDA8kd5+X5pc8efADUIsd3IwG/FonDDAkJhWduG87sy0s2AGoxliSkxYCZ4cABCbbbCQZneMwkgFoSNxDWwtjYwMN2DqTF/AdQSz0RWpiZGXjbDoBtAXk/wYCwX9jYmCXOJBvO7OExBjpMwnDmmQf4tfC3939j/FBhJ8/Pw2P44UOFjTzfcQK2YNhKmvJRMApGwSgYBdgBABa3PsCfxL2xAAAAAElFTkSuQmCC","orcid":"https://orcid.org/0000-0003-0329-3438","institution":"Cornell University","correspondingAuthor":true,"prefix":"","firstName":"Cassondra","middleName":"","lastName":"Lyman","suffix":""},{"id":489393811,"identity":"6cd5b97c-c260-48ef-a553-97948698a5a1","order_by":1,"name":"Anthony Ong","email":"","orcid":"","institution":"Cornell University","correspondingAuthor":false,"prefix":"","firstName":"Anthony","middleName":"","lastName":"Ong","suffix":""},{"id":489393812,"identity":"6b1e2928-2940-4087-8968-e06d89a5b218","order_by":2,"name":"Jonathan Rottenberg","email":"","orcid":"https://orcid.org/0000-0001-6128-4359","institution":"Cornell University","correspondingAuthor":false,"prefix":"","firstName":"Jonathan","middleName":"","lastName":"Rottenberg","suffix":""}],"badges":[],"createdAt":"2025-07-14 18:45:45","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7123882/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7123882/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":88009090,"identity":"604a377f-31bb-4a44-8430-2b22a9b0fc49","added_by":"auto","created_at":"2025-07-31 11:36:43","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":85294,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eMean Cross-Validated -2LL by Class\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e\u003cstrong\u003eNote.\u003c/strong\u003e\u003c/em\u003eIn Figure 1a, the 3-class quadratic model was the model with the most classes whose mean cross-validated -2LL was outside one standard error of the model with one fewer class and the model with the fewest classes whose mean cross-validated -2LL was within one standard error of the model with one more class, but the 6-class quadratic model was the model with the fewest classes whose mean cross-validated -2LL was within one standard error of the best fitting model.\u003c/p\u003e","description":"","filename":"1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7123882/v1/aa6fc003fd004878d0e0e6b8.jpg"},{"id":88009091,"identity":"afc1d0bb-0ffb-4c5d-9045-f13282cd4b50","added_by":"auto","created_at":"2025-07-31 11:36:43","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":186435,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eFour-Class Change Score Model: Average Trajectories of Latent Class Grouped by Initial Well-being Level\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e\u003cstrong\u003eNote\u003c/strong\u003e\u003c/em\u003e. Red line demarks 25\u003csup\u003eth\u003c/sup\u003e percentile. Green line demarks 75\u003csup\u003eth\u003c/sup\u003e percentile. Black lines plot the groups’ average trajectories.\u003c/p\u003e","description":"","filename":"2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7123882/v1/cfd4d203acd4407695624f58.jpg"},{"id":88009942,"identity":"6ff275a9-4b10-41e3-b3ef-8bb7d0716b13","added_by":"auto","created_at":"2025-07-31 11:44:43","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":139156,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003ePredicted Probabilities of Well-being Trajectory Class\u003c/em\u003e\u003c/p\u003e","description":"","filename":"3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7123882/v1/67ab08f3faf9bc9fca567de1.jpg"},{"id":88009092,"identity":"97ea6724-8f7f-4201-b6e2-6e7802530fc5","added_by":"auto","created_at":"2025-07-31 11:36:43","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":181544,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003eWell-being Trajectories of People with Elevated Baseline Depressive Symptoms that Experienced Optimal Well-being\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e\u003cstrong\u003eNote\u003c/strong\u003e\u003c/em\u003e. Red line demarks 25\u003csup\u003eth\u003c/sup\u003e percentile. Green line demarks 75\u003csup\u003eth\u003c/sup\u003e percentile. Black lines plot the groups’ average trajectories\u003c/p\u003e","description":"","filename":"4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7123882/v1/4126431cb5a85d7ee35cb616.jpg"},{"id":88010254,"identity":"a6d180b9-4f37-4299-b361-69df2ae83def","added_by":"auto","created_at":"2025-07-31 11:52:44","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1861967,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7123882/v1/60b254b9-ee93-420f-a79a-2c1a9c61eefa.pdf"},{"id":88009943,"identity":"0e7ecd72-1c9b-4aa8-8822-84e14a7a3fbb","added_by":"auto","created_at":"2025-07-31 11:44:43","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":786888,"visible":true,"origin":"","legend":"Supplemental Figures and Analyses","description":"","filename":"WBTrajectoriesSuppl5.26.25.docx","url":"https://assets-eu.researchsquare.com/files/rs-7123882/v1/be40eb3930b567deed82aacf.docx"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e Competing Interest.","formattedTitle":"Depressive Symptoms Predict Divergent Trajectories of Well-being in U. S. Adults","fulltext":[{"header":"INTRODUCTION","content":"\u003cp\u003eWell-being is a multifaced concept referring to emotions, life satisfaction, sense of meaning and purpose, and ability to pursue self-defined goals (Park et al., 2023; Ryff, 2023). Those reporting higher levels of well-being are physically healthier (Keyes \u0026amp; Simoes, 2012) and less likely to develop mental health problems, like depression (Grant et al., 2013). Given the benefits of well-being, both scientists and laypeople are motivated to identify the sources of sustained well-being. Surprisingly little is known about what factors predict the attainment of high levels of well-being over decades. Although depression impairs short-term well-being, little is known about how depression affects long-term trajectories of well-being. To address this question, we leveraged data from the 25-year Americans\u0026rsquo; Changing Lives (ACL) study (N=3,617) to (1) identify common well-being trajectories, (2) estimate how often adults sustain well-being across decades in the United States, and (3) test whether depression and other demographic or health factors predict different well-being trajectories (including that of sustained well-being).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDefining Well-being\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAccording to Keyes\u0026rsquo; (2005) dual continuum model of mental health, well-being represents more than the absence of mental illness\u0026mdash;it additionally requires the presence of positive emotional, psychological, and social functioning. This holistic view integrates Diener\u0026rsquo;s (1984) subjective well-being model and Ryff\u0026rsquo;s (1989) psychological well-being model, and aligns with evidence that human flourishing involves facets including life satisfaction, frequent positive affect, infrequent negative affect, self-acceptance, autonomy, environmental mastery, personal growth, purpose in life, and positive relationships (Disabato et al., 2019; Gallagher et al., 2009; Keyes et al., 2002). A key insight of the dual continuum model is that mental health and mental illness are conceptually and empirically distinct dimensions\u0026mdash;not simple opposites (Keyes, 2005; Westerhof \u0026amp; Keyes, 2010). This distinction may explain why depressive symptoms and well-being exhibit more complex relationships over time (e.g., bidirectionality, Huppert, 2009; Joshanloo \u0026amp; Blasco-Belled, 2023; Lamers et al., 2015).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eWell-being Trajectories\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA well-being trajectory refers to the pattern of change(s) in a person\u0026rsquo;s well-being over time. A longstanding debate in affective science concerns the degree of consistency in long-term well-being. On the side of consistency is evidence that well-being does not differ across age-cohorts (Ryff, 1995) and often persists over long follow up intervals (Cintron \u0026amp; Ong, 2024; Ryff et al., 2015). For example, in joint trajectory analyses of longitudinal data spanning 18 years, most (83%) U.S. adults reported persistently high positive affect and low negative affect, while two smaller groups demonstrated either improving or declining affective well-being (Cintron \u0026amp; Ong, 2024). Similarly, analysis of psychological well-being in the same sample\u0026rsquo;s first two timepoints revealed that most (78-83%) participants\u0026rsquo; well-being scores did not shift more than two quartiles.\u003c/p\u003e\n\u003cp\u003eAlternatively, there is evidence of well-being exhibiting meaningful change over time (Cintron \u0026amp; Ong, 2024; Ryff et al., 2015). While early models, such as set-point theory (Lykken \u0026amp; Tellegen, 1996), proposed that well-being is largely stable and returns to a baseline following life events, subsequent research has shown that this is not always the case. For example, major life events\u0026mdash;including marriage, disability, and unemployment\u0026mdash;can lead to lasting changes in subjective well-being (Lucas, 2007) or to people spending several consecutive years above and then several consecutive years below their long-term average well-being level (Headey \u0026amp; Muffels, 2016). These studies highlight that although well-being often persists over time, it is not necessarily stable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eWhy Study How Depression Influences Well-being Trajectories?\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAlthough many may take comfort in the idea that well-being persists over decades, people with depression likely do not. Clinical depression appears incompatible with well-being cross-sectionally or over short spans (Rottenberg et al., 2019), but people with depression are known to strongly desire the attainment of well-being (Wood \u0026amp; Tarrier, 2010). In fact, many patients with depression prioritize affective balance and restored daily functioning over conventional outcomes, like symptom reduction (Chevance et al., 2020; Demyttenaere et al., 2015; Zimmerman et al., 2006). As people often exhibit preferences for stability (Frijda, 1988; Samuelson \u0026amp; Zeckhauser, 1988; Swann \u0026amp; Read, 1981) and people with persistently high well-being demonstrate more favorable health outcomes over time than those simply reporting high well-being at baseline (Ryff et al., 2015), it stands to reason that people with depression would desire sustained well-being over temporary periods of well-being. However, it remains unclear how often they achieve those desires.\u003c/p\u003e\n\u003cp\u003eAlthough emerging evidence indicates that around 10% of people with a history of depression subsequently achieve relatively high levels of well-being (Devendorf et al., 2022; Rottenberg et al., 2019), questions remain about whether these gains are maintained over time. Some scholars suggest that they are maintained, proposing that depressive episodes themselves may function as well-being turning points when people make meaningful life change in response to the experience (Nesse, 2000; Ridge \u0026amp; Ziebland, 2006). However, other work suggests that these gains are not maintained, as increased affective reactivity among people with current or remitted depression may contribute to less stable well-being over time (Blysma et al., 2011). Still, the impact of this reactivity on long-term well-being may vary based on the person\u0026rsquo;s tendency to monitor mood (Thompson et al., 2013), ability to regulate mood (Lischetzke \u0026amp; Eid, 2003), and the valence of events they experience (Blysma et al., 2011). Taken together, these studies suggest that a history of depression may make well-being trajectories less stable, potentially setting the stage for either increases or decreases in long-term well-being.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eThe Current Study\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAlthough there is moderately convincing evidence that most people report persistent well-being over long time periods, uncertainties remain. First, although previous work indicates that persistent well-being is common, relatively few studies use growth mixture modeling to examine the full range of well-being trajectories over decades. Second, the nature of persistent well-being has not been well-characterized. It has not been determined whether people who exhibit mean-level stability in well-being invariably maintain low intra-individual variability in well-being or what portion of this group might experience persistently high versus low well-being. Third, the role of depression in predicting people\u0026rsquo;s long-term well-being trajectories (stability versus change) is unknown.\u003c/p\u003e\n\u003ch3\u003e\u003cstrong\u003eStudy Design \u003c/strong\u003e\u003c/h3\u003e\n\u003cp\u003eThe present investigation used data from the ACL study, a nationally representative, long-term investigation of health and well-being in the United States. Participants (\u003cem\u003eN \u003c/em\u003e= 3,617) were recruited via stratified, multistage area probability sampling with oversampling of underrepresented groups (e.g., African Americans, adults over age 60) and completed five face-to-face interviews over the span of 25 years. The ACL dataset is well-suited for trajectory analysis due to its longitudinal design and rich psychosocial content. Based on an integrated model of well-being (Keyes et al., 2002; Ryff, 1989), we constructed a battery assessing hedonic (life satisfaction, affect) and eudaimonic dimensions (self-acceptance, mastery, positive relationships) of well-being. We used latent growth mixture models (LGMMs) to identify unobserved subpopulations (i.e., latent classes) typified by similar well-being trajectories and clarify how common sustained well-being is. Afterward, we used multinomial logistic regression analyses to investigate who tends to experience each well-being trajectory and whether depressive symptom severity impacts well-being sustainability.\u003c/p\u003e\n\u003cp\u003eWe anticipated that GMM would identify well-being trajectories for both sustained and fluctuating well-being groups, with prior studies on well-being dynamics informing an expectation of 3-6 qualitatively distinct trajectories (e.g., Ryff et al., 2015; Cintron \u0026amp; Ong, 2024). In line with prior findings on sociodemographic predictors of well-being (Fujita \u0026amp; Diener, 2005; Navarro-Carrillo et al., 2020; Ryff, 1995), we hypothesized that men and those of higher SES would be more likely to sustain high well-being. We further expected that elevated baseline depressive symptoms would be associated with well-being trajectory membership and lower likelihood of achieving sustained well-being. To address the concept of well-being turning points, we conducted exploratory analyses of people who made a transition from elevated depressive symptoms to high levels of well-being.\u003c/p\u003e"},{"header":"ONLINE METHODS","content":"\u003ch2\u003e\u003cstrong\u003eParticipants\u003c/strong\u003e\u003c/h2\u003e\n\u003cp\u003eData were extracted from Waves 1, 2, 3, 4, and 5 of the ACL study (1986; 1989; 1994; 2002; and 2011; ACL: https://acl.isr.umich.edu), a nationally representative cohort longitudinal investigation of noninstitutionalized, English-speaking adults (age 25 and up) recruited via stratified, multistage area probability sampling and oversampling African Americans and people 65 or older. At baseline (i.e. Wave 1), respondents participated in a 90-minute face-to-face interview (\u003cem\u003eN\u0026nbsp;\u003c/em\u003e= 3,617). Surviving respondents were re-interviewed in 1989 (N = 2,867, 83% of survivors), 1994 (\u003cem\u003eN\u003c/em\u003e = 2,562 with 164 proxy respondents, 83% of survivors), 2002 (\u003cem\u003eN\u003c/em\u003e = 1,787 with 95 proxy respondents, 74% of survivors), and 2012 (\u003cem\u003eN\u003c/em\u003e = 1,427 with 108 proxy respondents, 81% of survivors); proxy respondents, such as a spouse or nurse, completed questionnaires in collaboration with the surviving respondents when the surviving respondent was experiencing disabilities that would impede their ability to complete the measures (e.g., hearing loss). One thousand and seventy-one respondents (30%) completed all 5 waves, while 1,361 (38%) completed all possible waves prior to their death. Of those who dropped out (\u003cem\u003eN\u003c/em\u003e = 654, 18%), 264 (7%) completed only the first wave, 128 (3%) the first and second waves, 159 (5%) the first three waves, and 103 (3%) all but the fifth wave. The remaining 531 (15%) had intermittent response patterns. Participants were mostly female (62.45%), mostly Caucasian (64.22%), and ranged from age 25 to 95 at baseline (\u003cem\u003eM\u003c/em\u003e= 53.58; \u003cem\u003eSD\u003c/em\u003e= 17.63).\u003c/p\u003e\n\u003cp\u003eGMMs and subsequent analyses were conducted using responses from all participants who provided complete or near-complete data at three or more waves (N=2,420). Of these, participants were mostly female (63.68%), mostly Caucasian (68.51%), and ranged from age 25 to 90 at baseline (\u003cem\u003eM\u003c/em\u003e=49.15; \u003cem\u003eSD\u003c/em\u003e=15.88). This sample contained 393 people with elevated baseline depressive symptoms (i.e. CESD-11\u0026ge;8.859; \u003cstrong\u003eTable 1\u003c/strong\u003e). Nonresponse to three or more waves was associated with age (t=-22.56, \u003cem\u003ep\u003c/em\u003e\u0026lt;.001), with nonresponders being significantly older (M=62.56) than responders (M=49.12). Sex and race were associated with nonresponse (\u0026chi;2=4.51, \u003cem\u003ep\u003c/em\u003e=.034; \u0026chi;2=68.42, \u003cem\u003ep\u003c/em\u003e\u0026lt;.001), with post-hoc comparisons revealing that women were less likely to respond to three or more waves than men, and people identifying as Black or Hispanic were less likely to respond to three or more waves than people identifying as other races.\u003c/p\u003e\n\u003ch2\u003e\u003cstrong\u003eMaterials\u003c/strong\u003e\u003c/h2\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eDepressive Symptoms\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAt each wave, depressive symptoms were assessed with the brief 11-item version of the Center for Epidemiologic Studies Depression Scale (CES-D-11; Kohout et al., 1993). The CES-D-11 includes items 2, 6, 7, 11, 12, 14, 15, 16, 18, 19, and 20 of the original CES-D (Radloff, 1977), where participants answer items using a 3-point scale ranging from 0 (\u003cem\u003eHardly ever\u003c/em\u003e) to 2 \u003cem\u003e(Most of the time\u003c/em\u003e). Responses for items 12 and 16 were reverse-coded. Total scores range from 0 to 22, with higher scores being indicative of greater depression severity. The CES-D-11 demonstrated strong internal consistency (\u0026alpha; range: .82 to .85). A cut-off score of 16 on the CES-D is used to identify participants with significant depressive symptoms (Vilagut et al., 2016), and this score can be converted to the CES-D-11 equivalent via O\u0026rsquo;Hara and colleagues\u0026rsquo; (1985) formula: \u003cstrong\u003e\u003cem\u003eCES-D = 1.866*CES-D-11 + 0.5318.\u003c/em\u003e\u003c/strong\u003e Consequently, a cut-off score of 8.289 was used to identify participants with significant depressive symptoms.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eWell-being\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eLife Satisfaction.\u003c/strong\u003e At each wave, overall life satisfaction was assessed using a 5-point scale ranging from 1 (\u003cem\u003eCompletely satisfied\u003c/em\u003e) to 5 \u003cem\u003e(Not at all satisfied\u003c/em\u003e). The item was \u0026ldquo;Now please think about your life as a whole. How satisfied you are with it?\u0026rdquo;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFrequent Positive Affect.\u0026nbsp;\u003c/strong\u003eAt each wave, respondents answered a face-valid item assessing the frequency of positive affect using a 3-point scale ranging from 1 (H\u003cem\u003eardly ever\u003c/em\u003e) to 3 (\u003cem\u003eMost of the time\u003c/em\u003e). The item was, \u0026ldquo;In the past week [R] was happy.\u0026rdquo;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eInfrequent Negative Affect.\u0026nbsp;\u003c/strong\u003eAt each wave respondents answered a face-valid item assessing the frequency of negative affect using a 3-point scale ranging from 1 (H\u003cem\u003eardly ever\u003c/em\u003e) to 3 (\u003cem\u003eMost of the time\u003c/em\u003e). The item was, \u0026ldquo;In the past week [R] was sad.\u0026rdquo;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u0026nbsp;Self-acceptance.\u003c/strong\u003e At each wave, one\u0026rsquo;s ability to accept and be satisfied with themselves was evaluated via an item from the Rosenberg Self- Esteem Scale (Rosenberg, 1965) using a 4-point scale ranging from 1 (\u003cem\u003eStrongly agree\u003c/em\u003e) to 4 (\u003cem\u003eStrongly disagree\u003c/em\u003e). The item was \u0026quot;I take a positive attitude toward myself.\u0026rdquo;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u0026nbsp;Environmental Mastery.\u003c/strong\u003e At each wave, one\u0026rsquo;s ability to manage their life and surroundings by effectively handling daily tasks and challenges, making use of available opportunities, and adapting the environment to suit personal needs was assessed via an item from the Pearlin Mastery Scale (Pearlin \u0026amp; Schooler, 1978) using a 4-point scale ranging from 1 (\u003cem\u003eStrongly Agree\u003c/em\u003e) to 4 (\u003cem\u003eStrongly Disagree\u003c/em\u003e). The item was \u0026ldquo;Sometimes I feel that I am being pushed around in life.\u0026rdquo; This item was reverse-scored.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u0026nbsp;Positive Relationships with Others.\u003c/strong\u003e At each wave, one\u0026rsquo;s ability to build and maintain positive relationships with others was assessed via three questions about the quality of relationships with their partner, child(ren), and friends. For each relationship, respondents answered the question, \u0026ldquo;How much does your [person] make you feel loved and cared for?\u0026rdquo; using a 5-point scale ranging from 1 (\u003cem\u003eA great deal\u003c/em\u003e) to 5 (\u003cem\u003eNot at all\u003c/em\u003e). These items demonstrated acceptable internal consistency (mean inter-item correlation range: .22 to .25) and were used to create a composite score reflecting one\u0026rsquo;s ability to build and maintain positive relationships with others.\u003c/p\u003e\n\u003ch2\u003e\u003cstrong\u003eData Analytic Plan\u003c/strong\u003e\u003c/h2\u003e\n\u003cp\u003eAll data manipulation and analysis were conducted in RStudio version 2022.12.0.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eHandling Missing Data\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo minimize bias in class estimation, well-being outcomes and demographic variables were retained as observed. Given their use as predictors and moderate levels of missingness, CES-D-11 values were imputed using Bayesian methods to support stable estimation in subsequent models. Patterns and amounts of missing data were examined and imputed via Bayesian methods using the \u003cem\u003emice\u003c/em\u003e package (van Buuren \u0026amp; Groothuis-Oudshoorn, 2011). To reduce estimation bias and power issues, Bayesian methods were used to create 40 imputed data sets for pooled analyses (Lee \u0026amp; Harring, 2023).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eCleaning Data\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003ePrior to analyses, data were examined for multivariate normality using the \u003cem\u003emvn\u0026nbsp;\u003c/em\u003epackage (Korkmaz S, 2014) and variables were evaluated for univariate skew and kurtosis (Kim, 2013; West, Finch, \u0026amp; Curran, 1995). No variables exhibited absolute skew \u0026ge; 2 or absolute kurtosis \u0026ge; 7.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eTesting Well-Being Model Assumptions for Structure and Measurement Invariance\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo evaluate the structure of the well-being construct prior to modeling trajectories, we conducted a confirmatory factor analysis (CFA) using the \u003cem\u003elavaan\u003c/em\u003e package (Rosseel, 2012). We assessed model fit using the \u0026chi;\u003csup\u003e2\u003c/sup\u003e goodness of fit test and standard model fit indices (CFI, RMSEA, and SRMR), applying conventional benchmarks for acceptable fit (Hooper, Coughlan, \u0026amp; Mullen, 2008: CFI\u0026gt;0.90; RMSEA\u0026lt;0.10; SRMR\u0026lt;0.08). Items with standardized loadings \u0026lt;0.40 were removed as they insufficiently represented the single underling factor. Model assumptions were tested via the \u003cem\u003eREdaS\u003c/em\u003e package (Maier, 2022) and CFA models were conducted and fit via the \u003cem\u003epsych\u003c/em\u003e package (Revelle, 2024). No model assumptions were violated, and the CFA model of well-being fit using pairwise deletion to address missing data and robust WLSMV estimators to estimate model parameters was of good fit (Robust CFI=.934; Robust RMSEA=.091, 90% CI: .078-.104; SRMR=.040). The single factor identified was labeled \u0026ldquo;Well-being\u0026rdquo; as items loading to it assessed people\u0026rsquo;s overall life satisfaction, frequency of positive affect, infrequency of negative affect, self-acceptance, environmental mastery, and positive relationships with others (\u003cstrong\u003eFigure S1\u003c/strong\u003e). This factor exhibited acceptable internal consistency (\u0026alpha;=.65), which was not unexpected given the relatively small number of items and their coverage of distinct facets of well-being.\u003c/p\u003e\n\u003cp\u003eAdditionally, we tested longitudinal measurement invariance to ensure the well-being construct could be interpreted equivalently across waves. As \u0026chi;\u003csup\u003e2\u003c/sup\u003e tests are sensitive to sample size and may detect trivial misfit (Cheung \u0026amp; Rensvold, 2002), we used alternative fit criteria (\u0026Delta;CFI\u0026lt;.01, \u0026Delta;RMSEA\u0026lt;.015, \u0026Delta;SRMR\u0026lt;.030) to evaluate metric invariance (Chen, 2007; Cheung \u0026amp; Rensvold, 2002). As we used ordinal indicators and fixed all intercepts to zero for model identification, establishing metric invariance allowed us to interpret mean-level changes over time as reflecting true changes in the latent construct (Liu et al., 2017). When testing measurement invariance across all five waves, the configural model demonstrated good fit (Robust CFI=.943; Robust RMSEA=.085, CI: .077-.093; SRMR=.037). As expected, the \u0026chi;\u003csup\u003e2\u003c/sup\u003e test for differences between configural and metric models was significant (\u003cem\u003ep\u003c/em\u003e=.002), and the difference between model fit indices for the configural and metric models (\u0026Delta;RCFI = -0.004, \u0026Delta;RRMSEA = -0.011, and \u0026Delta;SRMR = 0.004) did not surpass prespecified thresholds. Thus, full metric invariance was established and factor scores for the latent factor, Well-being, were extracted and used as the outcome variable in subsequent analyses. Thus, we extracted well-being factor scores for each wave, standardized them, and used these scores as input for subsequent growth mixture modeling.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eModeling Well-being Over Time\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eGrowth mixture models fit via the \u003cem\u003elcmm\u0026nbsp;\u003c/em\u003epackage (Proust-Lima, Philipps, \u0026amp; Liquet, 2015) were used to estimate the prevalence of stable well-being over time and explore which, if any, demographic variables predicted said stability. GMM was used because it is ideal for uncovering unobserved heterogeneity in longitudinal data.\u003c/p\u003e\n\u003cp\u003eAll GMMs were fit using step-by-step procedures, including baseline growth model identification, and GMM specification, estimation, selection, and interpretation (Ram \u0026amp; Grimm, 2009). Following usual growth curve modeling procedures, the baseline growth model was identified by fitting a series of models (e.g., no growth, linear, quadratic). A critical step in model specification is class enumeration (i.e., selecting number of latent classes). To ensure robust class enumeration while avoiding overfitting, we performed 5-fold cross validation for each GMM fit (Grimm, Mazza, \u0026amp; Davoudzadeh, 2017). Thus, each series of GMMs was estimated five times: once to each distinct training set containing 80% of the data. To allow for the possibility of rejecting models with additional latent classes and thereby provide statistical evidence that we identified the appropriate number of classes, each series of GMMs was fit to include models with up to one more class than the best fitting model, starting with a 1-class model. The parameters for these models were then used as the basis for a series of GMMs fit to the corresponding hold-out set containing 20% of the data. For each fold, the \u0026minus;2 log-likelihood (-2LL) of each GMM model was saved, and the mean and standard deviation across all five validation attempts were calculated. Three approaches to determining the correct number of latent classes were implemented: 1. The model with the fewest classes whose mean cross-validated -2LL was within one standard error of the best fitting model was selected; 2. The model with the fewest classes whose mean cross-validated -2LL was outside one standard error of the model with one fewer class was selected; 3. The model with the fewest classes whose mean cross-validated -2LL was outside one standard error of the model with one more class was selected. Results from all three methods were used to determine the model of best fit, which was then estimated using all data.\u003c/p\u003e\n\u003cp\u003eWe inspected each model to ensure convergence and proper specification of variance-covariance structures (Ram \u0026amp; Grimm, 2009). Specifically, we sought to verify that the model of best fit outperformed other models (better fitting models have lower AIC and SABIC; VLMR-LRT \u003cem\u003ep\u003c/em\u003e\u0026lt;.05 indicates the \u003cem\u003ek\u003c/em\u003e class model is better fitting than the \u003cem\u003ek-1\u003c/em\u003e class model), and reasonable classification confidence (entropy \u0026gt;.80 and class-level entropy \u0026gt;.70 indicate adequate separation between latent classes and confident classification). Mean trajectories for each latent class were plotted, as good fitting models yield classes with expected or logical group differences, but poor fitting or overfit models may yield classes with considerable overlap or unexpected patterns of change (Ram \u0026amp; Grimm, 2009). Following final model selection, we interpreted the meaning of each latent class. To identify the percentage of people belonging to the stable well-being latent class, as well as what range of well-being those people tend to fall in and how stable their well-being was, we extracted the proportion of individuals in each class and used thresholds calculated from baseline data for people who reported minimal or no baseline depressive symptoms to characterize well-being. Scores within the bottom quartile were considered low, within the interquartile range, average, and within the upper quartile, high (Devendorf et al., 2022; Rottenberg et al., 2019). A score change of one or more quartiles between timepoints was interpreted as indicative of intra-individual variability (Ryff et al., 2015).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eIdentifying Predictors of Well-being Trajectories\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA multinomial logistic regression was fit to assess whether demographic variables (i.e. age, gender, SES) or depressive symptom severity at baseline uniquely predicted latent class membership. The regression was specified with well-being trajectory class membership as the outcome variable and age, gender, SES, and depressive symptom severity at baseline as the predictors. Prior to fitting the regression model, we checked for multicollinearity among the independent variables: no predictors needed to be removed due to multicollinearity (VIF\u0026gt;10; Menard, 2010). As is standard, we estimated the variability accounted for by the model (via McFadden\u0026rsquo;s \u003cem\u003eR\u003csup\u003e2\u003c/sup\u003e\u003c/em\u003e) and the impact of significant predictor variables on the likelihood that a person belonged to the stable well-being trajectory (i.e. relative risk ratio).\u003c/p\u003e"},{"header":"RESULTS","content":"\u003cp\u003e\u003cstrong\u003eModeling Well-being Over Time\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eBased on results when fitting 1-class baseline growth models (\u003cstrong\u003eTable 2\u003c/strong\u003e),\u0026nbsp;we fit both quadratic and sinusoidal models.\u0026nbsp;Examination of -2LL estimates provided clear support for quadratic over the sinusoidal models but supported either three or six classes depending on the approach used for\u0026nbsp;interpretation (\u003cstrong\u003eFigure 1a\u003c/strong\u003e). Given this, we fit quadratic GMMs\u0026nbsp;with up to seven classes to the full data\u0026nbsp;(\u003cstrong\u003eTable 3\u003c/strong\u003e) and planned to use model fit indices to decide between the\u0026nbsp;3- and 6-class models.\u0026nbsp;Unfortunately, none of the models fit demonstrated adequate entropy (\u0026gt;.80), even when refit with variance constrained across random intercepts.\u003c/p\u003e\n\u003cp\u003eHaving noted large variance in the intercept for each class across models, we plotted the mean trajectories of each latent class grouped by initial well-being level (\u003cstrong\u003eFigure S2\u003c/strong\u003e). These plots revealed heterogeneity within latent classes, such that phenomenologically distinct trajectories were grouped together (e.g., each model identified a latent class that included people experiencing increasing and then sustained well-being and people sustaining relatively high well-being). Subsequent inspection of class-level entropies and individual-level posterior probabilities for class assignment revealed that average class confidence was acceptable for two classes (M\u003csub\u003ePP\u003c/sub\u003e: .80 and .72) but not the other (M\u003csub\u003ePP\u003c/sub\u003e=.61), and class confidence was ambiguous\u003csup\u003e[1]\u0026nbsp;\u003c/sup\u003efor 505 (20.9%) participants.\u003c/p\u003e\n\u003cp\u003eThis indicated a need for alternative modeling approaches. To address model misfit driven by heterogeneity in baseline well-being, we reduced within-class variability by refitting models using well-being change scores. Across models, information criterion values reduced and entropy values increased, suggesting fit improvement. As before, examination of -2LL estimates provided clear support for quadratic over the sinusoidal models; however, examination of -2LL estimates now also provided clear support for a 4-class model (\u003cstrong\u003eFigure 1b\u003c/strong\u003e). Given this, we fit quadratic GMMs with up to five classes to the full data (\u003cstrong\u003eTable 3\u003c/strong\u003e).\u003c/p\u003e\n\u003cp\u003eResults were consistent with the five-fold cross-validation method for class enumeration: all information criterion indicated that a 4-class model was most appropriate and likelihood ratio tests supported a 4-class model. Variance in the intercept and heterogeneity within latent classes were more reasonable (i.e. lower) and plots of mean trajectories for each latent class yielded classes with logical group differences, even when grouped by initial well-being level (\u003cstrong\u003eFigure 2\u003c/strong\u003e). Entropy was also much improved (.636 vs .491) and class-level entropies increased such that all were acceptable (\u0026gt;.70). Further, the number of ambiguously classified participants decreased (484 vs 505), the classification certainty for ambiguously classified participants improved (M\u003csub\u003eMAX(PP)\u003c/sub\u003e of .58, .58, .56, and .57 vs M\u003csub\u003eMAX(PP)\u003c/sub\u003e of .52, .52, and .48), and plots of ambiguously classified participants\u0026rsquo; trajectories generally supported the assumption that they were correctly classified (\u003cstrong\u003eFigure S3\u003c/strong\u003e). Overall, model fit improved across multiple metrics.\u003c/p\u003e\n\u003cp\u003eConsistent with the idea that a persistent well-being group would predominate, class 1 encompassed most participants (N\u003csub\u003e1\u003c/sub\u003e=1,443, 59.63%), and members of this class had good probability of being assigned to it (M=.79, SD=.12). Class members tended to report intermediate levels of initial well-being (M=0.11, SD=.42), and their well-being levels were relatively stable over time. There was significant within-group variation in intercept but not slope, indicating that class members differed in their initial well-being levels but not in how their well-being changed over time. Thus, we interpret Class 1 as encompassing a group of people experiencing persistent well-being.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTo further characterize well-being levels in this group, we used thresholds calculated from baseline data for people who reported minimal or no baseline depressive symptoms: scores within the bottom quartile were considered low, within the interquartile range, average, and within the upper quartile, high. Interestingly, Class 1 was far from monolithic. In fact, it included people at every level of well-being, with 75% maintaining average or relatively high levels of well-being (N\u003csub\u003e1.mid\u003c/sub\u003e=720, 49.90%; N\u003csub\u003e1.high\u003c/sub\u003e=361, 25.02%) but also with a quarter maintaining relatively low levels of well-being (N\u003csub\u003e1.low\u003c/sub\u003e=362, 25.09%). Furthermore, although class members sustained their well-being over time (stable average of well-being), being in class 1 did not preclude intra-individual variability. It was actually the minority of Class 1 members (n=413, 28.62%) who demonstrated perfect intra-individual consistency (i.e. no quartile change in well-being scores across\u0026nbsp;the 5 timepoints).\u003c/p\u003e\n\u003cp\u003eClass 4 was the next largest group, encompassing just under a third of participants (N\u003csub\u003e4\u003c/sub\u003e=736, 30.41%), and class members had similarly good probability of being assigned to it (M=.78, SD=.12). Class members tended to report low initial well-being (M=-0.26, SD=.40), and exhibited gradual increases in well-being and slowing of improvement over time. Significant within-group variation in intercept and little variation in slope indicated that although participants had different baseline well-being levels, their well-being generally followed a similar slow growth trajectory. Non-significant covariance between intercept and slope indicated that initial well-being level was not associated with different rates of improvement. Thus, Class 4 encompasses a group of people experiencing gradually increasing well-being.\u003c/p\u003e\n\u003cp\u003eAlmost all members (94.97%) of Class 4 began with low or average levels of well-being (N\u003csub\u003e4.low\u003c/sub\u003e=440, 59.78%; N\u003csub\u003e4.mid\u003c/sub\u003e=259, 35.19%), but a small minority had relatively high initial well-being (N\u003csub\u003e4.high\u003c/sub\u003e=37, 5.03%). Of those with low initial well-being, most (n=265, 60.23%) achieved average levels of well-being and some (n=112, 25.45%) achieved relatively high levels of well-being. Around half (n=221, 58.62%) sustained the highest level of well-being they achieved, with 52 (12.27%) sustaining relatively high levels of well-being. The remainder experienced regression toward their baseline well-being level, with 100 (37.74%) regressing to average well-being levels and 14 (5.28%) regressing back to relatively low well-being levels. Of those with average initial well-being, 205 (79.15%) achieved relatively high levels of well-being, and around half (n=109, 53.17%) sustained that new, relatively high level of well-being. Others (n=73, 35.61%) regressed toward their baseline well-being level, and some (n=23, 11.22%) experienced significant declines in well-being such that their last documented level of well-being was relatively low.\u003c/p\u003e\n\u003cp\u003eIn comparison, far fewer participants (N\u003csub\u003e2\u003c/sub\u003e=183, 7.56%) belonged to Class 2. Still, class members demonstrated good probability of being assigned to it (M=.79, SD=.16). Class members tended to report relatively high initial well-being (M=0.37, SD=.15), and demonstrated steep declines in well-being over time. Significant within-group variation in intercept, minimal variation in slope, and nonsignificant covariance between intercept and slope indicated that while participants in this class had different baseline well-being levels, they experienced similar rates of decline. Overall, Class 2 encompasses a group of people experiencing declining well-being.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAlmost all people (90.71%) in Class 2 began with high or average levels of well-being (N\u003csub\u003e2.high\u003c/sub\u003e=91, 49.73%; N\u003csub\u003e2.mid\u003c/sub\u003e=75, 40.98%), but some had relatively low baseline well-being (N\u003csub\u003e2.low\u003c/sub\u003e=17, 9.29%). Of those who began with high well-being, 18 (19.78%) went on to experience more typical (or average) levels of well-being and 73 (80.22%) went on to experience relatively low well-being. Six (6.59%) managed to recover their initial, higher level of well-being, 20 (21.98%) experienced improvements in well-being after their initial decline, and 65 (69.75%) maintained their new, lower level of well-being. Of those who began with average well-being, all went on to experience relatively low levels of well-being and 56 (74.67%) maintained that new, relatively low level of well-being. That said, 16 (21.33%) recovered their initial level of well-being and 3 (4%) went on to experience relatively high levels of well-being.\u003c/p\u003e\n\u003cp\u003eFinally, the smallest group was Class 3 (N\u003csub\u003e3\u003c/sub\u003e=58, 2.4%). Class members demonstrated high probability of being assigned to it (M=.82, SD=.17) and low probability of being assigned to other classes. Like members of Class 4, members of Class 3 tended to report low initial well-being (M=-0.78, SD=.43). However, their average initial well-being scores were much lower (-0.78 vs -0.26), and they demonstrated more marked increases in well-being than those in Class 4. They were also more likely to regress toward their initial well-being level than participants in Class 4. As in other classes, participants experienced similarly strong upward trajectories in well-being regardless of differences in baseline well-being. Overall, Class 3 encompasses a small group of people experiencing marked increases in well-being. Almost all people in this class (n=57, 98.28%) began with low levels of well-being: of those, a third (n=19, 33.33%) achieved average levels of well-being, and most (n=36, 63.16%) achieved relatively high levels of well-being. Half (n=28, 49.12%) sustained the highest level of well-being they achieved, with 18 sustaining relatively high levels of well-being.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eIdentifying Predictors of Well-being Trajectories\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAs baseline depressive symptoms and well-being were highly correlated (\u003cem\u003er\u003c/em\u003e=-0.76), we included baseline well-being level in analyses (\u003cstrong\u003eAppendix SA\u003c/strong\u003e includes results without this adjustment\u003csup\u003e[2]\u003c/sup\u003e). Additionally, because 20% of participants (n=484) were classified with uncertainty, we ran models with full and reduced datasets: results were consistent across both. Sex and age at baseline were not significant predictors of well-being trajectory class, but baseline SES, depressive symptom severity, and well-being level were (\u003cstrong\u003eTable 4; Figure 3\u003c/strong\u003e). Higher SES at baseline was associated with greater odds of gradual well-being improvement and lower odds of declining well-being.\u003c/p\u003e\n\u003cp\u003eImportantly, baseline depressive symptoms were predictive of well-being class membership. When accounting for baseline well-being level, more severe baseline depressive symptoms were associated with increased odds of being in \u003cem\u003eeither\u003c/em\u003e the declining well-being or the markedly increasing well-being classes: a one-unit increase in baseline depressive symptom severity was associated with a 1.10-fold increase in the odds of declining well-being class membership and a 1.20-fold increase in the odds of markedly increasing well-being class membership. Higher baseline well-being was associated with increased odds of being in the declining well-being class and reduced the odds of being in the increases in well-being class. Each quartile increase in baseline well-being was associated with a 3.47-fold increase in the odds of being in the declining well-being class, and a respective 98% and 69% reduction in the odds of being in the marked or gradually increasing well-being classes. Overall, these findings indicate that baseline well-being is the strongest predictor of well-being change \u003cem\u003edirection\u003c/em\u003e, whereas baseline depressive symptom severity is a strong predictor of well-being change \u003cem\u003elikelihood\u003c/em\u003e. People with lower baseline well-being were more likely to experience improvements in well-being, and people with elevated baseline depressive symptoms were more likely to experience \u003cem\u003echange\u0026nbsp;\u003c/em\u003ein well-being. Thus, people with both low well-being and elevated depressive symptoms at baseline had the highest odds of experiencing marked improvements in well-being. In this sample, 57% of people with both low well-being and elevated depressive symptoms at baseline experienced increases in well-being.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eUnderstanding Transitions from Depression Symptoms to Well-being\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThree hundred and ninety-six people exhibited elevated depressive symptoms at baseline (i.e. CESD-11\u0026ge;8.289). While baseline depressive symptom severity was a robust predictor of less stable well-being over time, many people with elevated baseline depressive symptoms (n=169, 42.67%) exhibited persistent well-being. Of those in the persistent well-being class, most began with low well-being (n=146, 85%) and experienced stable persistent low well-being (n=98, 57%), but others experienced high intra-individual variability despite demonstrating persistent well-being (n=70, 41%), and five reported one or more instances of relatively high well-being. Fifty-six other people with elevated baseline depressive symptoms also experienced at least one instance of relatively high well-being over the five assessment points. In total, 15.4% of people with elevated baseline depressive symptoms went on to report at least one instance of elevated well-being and 2.5% sustained the elevated level of well-being they achieved (\u003cstrong\u003eFigure 4\u003c/strong\u003e).\u003c/p\u003e"},{"header":"DISCUSSION","content":"\u003cp\u003eHow consistent is human well-being over decades? Supportive of perspectives that view consistency as normative, persistent well-being was the most common trajectory in this sample of U.S. adults, with around 60% of the sample demonstrating little change from baseline well-being 25 years later. However, results also lend credence to perspectives that highlight intra-individual change in well-being, as only 23% of people maintained their baseline well-being level at all five timepoints and we identified three other meaningful classes: gradually increasing well-being, declining well-being, and markedly increasing well-being. Thus\u0026mdash;although persistent well-being is common\u0026mdash;it does not necessarily indicate \u003cem\u003estable\u003c/em\u003e well-being. When well-being is assessed at enough time points and with short enough lags, meaningful intra-individual change (i.e. instability) in well-being is common.\u003c/p\u003e\n\u003cp\u003eWhile persistent well-being was generally the healthiest class, membership in this class did not guarantee high levels of well-being. A quarter of people in the persistent well-being class reported consistently low well-being. Although often treated as part of a psychologically \u0026ldquo;healthy\u0026rdquo; reference group, this subgroup\u0026rsquo;s existence highlights the need to distinguish persistent or stable well-being that might reflect resilience and that which might reflect enduring difficulty (i.e. languishing). That over half of those experiencing persistently low well-being did not report elevated baseline depressive symptoms, supports the dual continua model and peril of equating the absence of well-being with the presence of psychopathology (Keyes, 2005; Westerhof \u0026amp; Keyes, 2010).\u003c/p\u003e\n\u003cp\u003eDemographic features had a muted role in predicting well-being trajectories. Despite robust prior association between SES and well-being (Navarro-Carrillo et al., 2020; Tan et al., 2020), higher SES was only modestly associated with reduced odds of declining well-being. Thus, although elevated SES may increase access to helpful resources and decrease exposure to chronic stressors, these changes seem to protect well-being over time rather than enhance it.\u003c/p\u003e\n\u003cp\u003eIn contrast, baseline well-being level and depressive symptom severity were both strong predictors of well-being trajectory. Indeed, both low baseline well-being and elevated baseline depressive symptoms were associated with subsequent improvements in well-being; however, those with elevated baseline depressive symptoms were also less likely to sustain well-being improvements. These differences between the gradually and markedly improving classes further indicate that depressive symptoms may predict less stable well-being trajectories.\u003c/p\u003e\n\u003cp\u003eWhy depressive symptom severity is associated with less stable well-being trajectories is difficult to explain from our research design, but prior evidence on the heterogeneity of depression (Goldberg, 2011; Lynall \u0026amp; McIntosh, 2023) may help. People who experience depression with more cognitive symptoms may have higher odds of experiencing declining well-being (Liao et al., 2022). Likewise, depressed people prone to heightened affective variability may have higher odds of experiencing early well-being gains followed by reversal (Bylsma et al., 2011). In contrast, people whose symptoms began in response to addressable life stressors, may be more likely to experience well-being turning points (Monroe et al., 2019). Other factors\u0026mdash;such as treatment-seeking, effective emotion regulation, and engagement in health-related behaviors (e.g., regular exercise)\u0026mdash;may also drive well-being change direction among people with elevated baseline depressive symptoms. Understanding which of these factors is most decisive may provide leverage for initiating improvements. Regardless, the fact that both depressive symptom severity and well-being were independent significant predictors of well-being trajectory supports recent calls for depression treatment to go beyond symptom reduction by incorporating activities that help clients recognize, cultivate, and sustain well-being (Fava et al., 2017; 1998).\u003c/p\u003e\n\u003cp\u003eOf course, this study\u0026rsquo;s results should be considered in the context of several limitations. \u0026nbsp;This was a secondary analysis of archival data, which constrained the measurement instruments available. Nevertheless, our well-being battery included all facets of Diener\u0026rsquo;s (1984) model of well-being and several key facets of Ryff\u0026rsquo;s (1989) model of well-being and demonstrated acceptable reliability. Second, while the ACL data allowed us to model five timepoints over 25 years, the time between waves increased in a near exponential pattern, with the longest time lag being 10 years. This complicates claims that similar well-being scores across later waves indicate well-being \u003cem\u003estability\u003c/em\u003e, as the increasing interval between waves decreases the certainty that a person\u0026rsquo;s well-being did not change between assessments. Although the intervals between waves in the ACL are not unusual, it would be valuable if future studies examined well-being trajectories over smaller, more uniform intervals. Finally, the ACL used a cutoff from a validated measure of self-reported depressive symptom severity rather than clinical diagnostic interviews to identify those with significant depression. Although self-report measures are more efficient for large-scale data collection, they may be influenced by individual differences in insight and recall bias and may not differentiate transient distress from clinically significant depression.\u003c/p\u003e\n\u003cp\u003eDespite these limitations, our analyses provide novel insights on typical well-being trajectories and how these trajectories may be altered by depression. We were able to interpret all classes identified in terms of well-being thresholds created using a well-accepted population-based approach and provided initial insight into how often people with depression sustain subsequently achieved high well-being. Finally, while generalizability is always uncertain, it should be underscored that the ACL was a nationally representative longitudinal cohort study of noninstitutionalized, English-speaking adults recruited via stratified, multistage area probability sampling with an oversampling African Americans and people 65 or older.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eBylsma, L. M., et al. (2011). \u0026quot;Emotional reactivity to daily events in major and minor depression.\u0026quot; Journal of Abnormal Psychology 120(1): 155.\u003c/li\u003e\n\u003cli\u003eChen, F. F. (2007). Sensitivity of Goodness of Fit Indexes to Lack of Measurement Invariance. \u003cem\u003eStructural Equation Modeling: A Multidisciplinary Journal, 14\u003c/em\u003e(3), 464-504. https://doi.org/10.1080/10705510701301834\u003c/li\u003e\n\u003cli\u003eCheung, G. W., \u0026amp; Rensvold, R. B. (2002). Evaluating Goodness-of-Fit Indexes for Testing Measurement Invariance. \u003cem\u003eStructural Equation Modeling: A Multidisciplinary Journal, 9\u003c/em\u003e(2), 233-255. https://doi.org/10.1207/s15328007sem0902_5\u003c/li\u003e\n\u003cli\u003eChevance, A., Ravaud, P., Tomlinson, A., Le Berre, C., Teufer, B., Touboul, S., Fried, E. I., Gartlehner, G., Cipriani, A., \u0026amp; Tran, V. T. (2020, 2020/08/01/). Identifying outcomes for depression that matter to patients, informal caregivers, and health-care professionals: qualitative content analysis of a large international online survey. \u003cem\u003eThe Lancet Psychiatry, 7\u003c/em\u003e(8), 692-702. https://doi.org/https://doi.org/10.1016/S2215-0366(20)30191-7\u003c/li\u003e\n\u003cli\u003eCintron, D. W., \u0026amp; Ong, A. D. (2024). Trajectories of affective well-being and survival in middle-aged and older adults. \u003cem\u003eEmotion, 24\u003c/em\u003e(5), 1149-1156. https://doi.org/10.1037/emo0001341\u003c/li\u003e\n\u003cli\u003eDemyttenaere, K., Donneau, A.-F., Albert, A., Ansseau, M., Constant, E., \u0026amp; Van Heeringen, K. (2015). What is important in being cured from depression? Discordance between physicians and patients (1). \u003cem\u003eJournal of Affective Disorders, 174\u003c/em\u003e, 390-396. https://doi.org/10.1016/j.jad.2014.12.004\u003c/li\u003e\n\u003cli\u003eDevendorf, A. R., Rum, R., Kashdan, T. B., \u0026amp; Rottenberg, J. (2022). Optimal Well-Being After Psychopathology: Prevalence and Correlates. \u003cem\u003eClinical Psychological Science, 10\u003c/em\u003e(5), 981-996. https://doi.org/10.1177/21677026221078872\u003c/li\u003e\n\u003cli\u003eDiener, E. (1984). Subjective Well-being. \u003cem\u003ePsychological Bulletin, 95\u003c/em\u003e(3), 542-575.\u003c/li\u003e\n\u003cli\u003eDisabato, D., Goodman, F., \u0026amp; Kashdan, T. B. (2019). A hierarchical framework for the measurement of well-being.\u003c/li\u003e\n\u003cli\u003eEid, M., \u0026amp; Diener, E. (2004). Global Judgments of Subjective Well-Being: Situational Variability and Long-Term Stability. \u003cem\u003eSocial Indicators Research, 65\u003c/em\u003e(3), 245-277. https://doi.org/10.1023/b:soci.0000003801.89195.bc\u003c/li\u003e\n\u003cli\u003eFava, G. A., Cosci, F., Guidi, J., \u0026amp; Tomba, E. (2017). Well-being therapy in depression: New insights into the role of psychological well-being in the clinical process. \u003cem\u003eDepression and Anxiety, 34\u003c/em\u003e(9), 801-808. https://doi.org/10.1002/da.22629\u003c/li\u003e\n\u003cli\u003eFava, G. A., Rafanelli, C., Cazzaro, M., Conti, S., \u0026amp; Grandi, S. (1998). Well-being therapy. A novel psychotherapeutic approach for residual symptoms of affective disorders. \u003cem\u003ePsychological Medicine, 28\u003c/em\u003e(2), 475-480. https://doi.org/10.1017/s0033291797006363\u003c/li\u003e\n\u003cli\u003eFrijda, N. H. (1988). The laws of emotion. \u003cem\u003eAmerican Psychologist, 43\u003c/em\u003e, 349-358. https://doi.org/10.1037/0003-066X.43.5.349\u003c/li\u003e\n\u003cli\u003eFujita, F., \u0026amp; Diener, E. (2005). Life Satisfaction Set Point: Stability and Change. \u003cem\u003eJournal of Personality and Social Psychology, 88\u003c/em\u003e, 158-164. https://doi.org/10.1037/0022-3514.88.1.158\u003c/li\u003e\n\u003cli\u003eGallagher, M. W., Lopez, S. J., \u0026amp; Preacher, K. J. (2009, Aug). The hierarchical structure of well-being. \u003cem\u003eJournal of Personality, 77\u003c/em\u003e(4), 1025-1050. https://doi.org/10.1111/j.1467-6494.2009.00573.x\u003c/li\u003e\n\u003cli\u003eGoldberg, D. (2011). The heterogeneity of \u0026ldquo;major depression\u0026rdquo;. \u003cem\u003eWorld Psychiatry, 10\u003c/em\u003e(3), 226-228. https://doi.org/10.1002/j.2051-5545.2011.tb00061.x\u003c/li\u003e\n\u003cli\u003eGrant, F., Guille, C., \u0026amp; Sen, S. (2013). Well-Being and the Risk of Depression under Stress. \u003cem\u003ePloS One, 8\u003c/em\u003e(7), e67395. https://doi.org/10.1371/journal.pone.0067395\u003c/li\u003e\n\u003cli\u003eHeadey, B., \u0026amp; Muffels, R. (2016, 2016/11/01). Towards a Theory of Medium Term Life Satisfaction: Two-Way Causation Partly Explains Persistent Satisfaction or Dissatisfaction. \u003cem\u003eSocial Indicators Research, 129\u003c/em\u003e(2), 937-960. https://doi.org/10.1007/s11205-015-1146-8\u003c/li\u003e\n\u003cli\u003eHooper, D., Coughlan, J., \u0026amp; Mullen, M. (2008, 11/30). Structural Equation Modeling: Guidelines for Determining Model Fit. \u003cem\u003eThe Electronic Journal of Business Research Methods, 6\u003c/em\u003e.\u003c/li\u003e\n\u003cli\u003eHuppert, F. A. (2009). Psychological Well-being: Evidence Regarding its Causes and Consequences. \u003cem\u003eApplied Psychology: Health and Well-Being, 1\u003c/em\u003e(2), 137-164. https://doi.org/10.1111/j.1758-0854.2009.01008.x\u003c/li\u003e\n\u003cli\u003eJoshanloo, M. (2022). Longitudinal Relations Between Depressive Symptoms and Life Satisfaction Over 15 Years. \u003cem\u003eApplied Research in Quality of Life, 17\u003c/em\u003e(5), 3115-3130. https://doi.org/10.1007/s11482-022-10055-x\u003c/li\u003e\n\u003cli\u003eJoshanloo, M., \u0026amp; Blasco-Belled, A. (2023, Jan 29). Reciprocal Associations between Depressive Symptoms, Life Satisfaction, and Eudaimonic Well-Being in Older Adults over a 16-Year Period. \u003cem\u003eInternational Journal of Environmental Research and Public Health, 20\u003c/em\u003e(3). https://doi.org/10.3390/ijerph20032374\u003c/li\u003e\n\u003cli\u003eKeyes, C., \u0026amp; Annas, J. (2009). Feeling good and functioning well: Distinctive concepts in ancient philosophy and contemporary science. \u003cem\u003eThe Journal of Positive Psychology, 4\u003c/em\u003e(3), 197-201.\u003c/li\u003e\n\u003cli\u003eKeyes, C. L. M. (2005). \u003cem\u003eMental Illness and/or Mental Health? Investigating Axioms of the Complete State Model of Health\u003c/em\u003e [doi:10.1037/0022-006X.73.3.539]. American Psychological Association.\u003c/li\u003e\n\u003cli\u003eKeyes, C. L. M., Dhingra, S. S., \u0026amp; Simoes, E. J. (2010). Change in Level of Positive Mental Health as a Predictor of Future Risk of Mental Illness. \u003cem\u003eAmerican Journal of Public Health, 100\u003c/em\u003e(12), 2366-2371. https://doi.org/10.2105/ajph.2010.192245\u003c/li\u003e\n\u003cli\u003eKeyes, C. L. M., Shmotkin, D., \u0026amp; Ryff, C. D. (2002). Optimizing well-being: The empirical encounter of two traditions. \u003cem\u003eJournal of Personality and Social Psychology, 82\u003c/em\u003e(6), 1007-1022. https://doi.org/10.1037/0022-3514.82.6.1007\u003c/li\u003e\n\u003cli\u003eKeyes, C. L. M., \u0026amp; Simoes, E. J. (2012). To Flourish or Not: Positive Mental Health and All-Cause Mortality. \u003cem\u003eAmerican Journal of Public Health, 102\u003c/em\u003e(11), 2164-2172. https://doi.org/10.2105/ajph.2012.300918\u003c/li\u003e\n\u003cli\u003eKim, H.-Y. (2013). Statistical notes for clinical researchers: assessing normal distribution (2) using skewness and kurtosis. \u003cem\u003eRestorative Dentistry \u0026amp; Endodontics, 38\u003c/em\u003e(1), 52. https://doi.org/10.5395/rde.2013.38.1.52\u003c/li\u003e\n\u003cli\u003eKohout, F. J., Berkman, L. F., Evans, D. A., \u0026amp; Cornoni-Huntley, J. (1993, May). Two shorter forms of the CES-D (Center for Epidemiological Studies Depression) depression symptoms index. \u003cem\u003eJournal of Aging and Health, 5\u003c/em\u003e(2), 179-193. https://doi.org/10.1177/089826439300500202\u003c/li\u003e\n\u003cli\u003eKoivumaa-Honkanen, H., Kaprio, J., Honkanen, R., Viinam⍰Ki, H., \u0026amp; Koskenvuo, M. (2004). Life satisfaction and depression in a 15-year follow-up of healthy adults. \u003cem\u003eSocial Psychiatry and Psychiatric Epidemiology, 39\u003c/em\u003e(12), 994-999. https://doi.org/10.1007/s00127-004-0833-6\u003c/li\u003e\n\u003cli\u003eKorkmaz S, G. D., Zararsiz G. (2014). MVN: An R Package for Assessing Multivariate Normality.\u003cem\u003e, 6\u003c/em\u003e(2), 151\u0026ndash;162. https://journal.r-project.org/archive/2014-2/korkmaz-goksuluk-zararsiz.pdf\u003c/li\u003e\n\u003cli\u003eKyriazos, T. A. (2018). Applied Psychometrics: The 3-Faced Construct Validation Method, a Routine for Evaluating a Factor Structure. \u003cem\u003ePsychology (Savannah, Ga.), 09\u003c/em\u003e(08), 2044-2072. https://doi.org/10.4236/psych.2018.98117\u003c/li\u003e\n\u003cli\u003eLamers, S. M. A., Westerhof, G. J., Glas, C. A. W., \u0026amp; Bohlmeijer, E. T. (2015). The bidirectional relation between positive mental health and psychopathology in a longitudinal representative panel study. \u003cem\u003eThe Journal of Positive Psychology, 10\u003c/em\u003e(6), 553-560. https://doi.org/10.1080/17439760.2015.1015156\u003c/li\u003e\n\u003cli\u003eLee, D. Y., \u0026amp; Harring, J. R. (2023). Handling Missing Data in Growth Mixture Models. \u003cem\u003eJournal of Educational and Behavioral Statistics\u003c/em\u003e, 107699862211491. https://doi.org/10.3102/10769986221149140\u003c/li\u003e\n\u003cli\u003eLiao, Y., Zhang, H., Guo, L., Fan, B., Wang, W., Teopiz, K. M., Lui, L. M. W., Lee, Y., Li, L., Han, X., Lu, C., \u0026amp; McIntyre, R. S. (2022, 2022/10/15/). Impact of cognitive-affective and somatic symptoms in subthreshold depression transition in adults: Evidence from Depression Cohort in China (DCC). \u003cem\u003eJournal of Affective Disorders, 315\u003c/em\u003e, 274-281. https://doi.org/https://doi.org/10.1016/j.jad.2022.08.009\u003c/li\u003e\n\u003cli\u003eLischetzke, T. and M. Eid (2003). \u0026quot;Is Attention to Feelings Beneficial or Detrimental to Affective Well-Being? Mood Regulation as a Moderator Variable.\u0026quot; Emotion 3(4): 361-377.\u003c/li\u003e\n\u003cli\u003eLiu, Y., Millsap, R. E., West, S. G., Tein, J. Y., Tanaka, R., \u0026amp; Grimm, K. J. (2017, Sep). Testing measurement invariance in longitudinal data with ordered-categorical measures. \u003cem\u003ePsychological Methods, 22\u003c/em\u003e(3), 486-506. https://doi.org/10.1037/met0000075\u003c/li\u003e\n\u003cli\u003eLucas, R. E. (2007). Adaptation and the Set-Point Model of Subjective Well-Being:Does Happiness Change After Major Life Events? \u003cem\u003eCurrent Directions in Psychological Science, 16\u003c/em\u003e(2), 75-79. https://doi.org/10.1111/j.1467-8721.2007.00479.x\u003c/li\u003e\n\u003cli\u003eLykken, D., \u0026amp; Tellegen, A. (1996). Happiness Is a Stochastic Phenomenon. \u003cem\u003ePsychological Science, 7\u003c/em\u003e(3), 186-189. https://doi.org/10.1111/j.1467-9280.1996.tb00355.x\u003c/li\u003e\n\u003cli\u003eLynall, M.-E., \u0026amp; McIntosh, A. M. (2023). The Heterogeneity of Depression. \u003cem\u003eAmerican Journal of Psychiatry, 180\u003c/em\u003e(10), 703-704. https://doi.org/10.1176/appi.ajp.20230574\u003c/li\u003e\n\u003cli\u003eMaier, M. (2022). \u003cem\u003eCompanion Package to the Book \u0026ldquo;R: Einf\u0026uuml;hrung durch angewandte Statistik\u0026rdquo;.\u003c/em\u003e\u003c/li\u003e\n\u003cli\u003e\u003cem\u003e. \u003c/em\u003eIn (Version 0.9.4) https://CRAN.R-project.org/package=REdaS\u003c/li\u003e\n\u003cli\u003eMenard, S. (2010). \u003cem\u003eLogistic Regression: From Introductory to Advanced Concepts and Applications\u003c/em\u003e https://doi.org/10.4135/9781483348964\u003c/li\u003e\n\u003cli\u003eMonroe, S. M., Anderson, S. F., \u0026amp; Harkness, K. L. (2019). Life stress and major depression: The mysteries of recurrences. \u003cem\u003ePsychological Review, 126\u003c/em\u003e(6), 791-816. https://doi.org/10.1037/rev0000157\u003c/li\u003e\n\u003cli\u003eNational Academies of Sciences, E., \u0026amp; Medicine. (2022). \u003cem\u003eOntologies in the Behavioral Sciences: Accelerating Research and the Spread of Knowledge\u003c/em\u003e. The National Academies Press. https://doi.org/doi:10.17226/26464\u003c/li\u003e\n\u003cli\u003eNavarro-Carrillo, G., Alonso-Ferres, M., Moya, M., \u0026amp; Valor-Segura, I. (2020, 2020-June-10). Socioeconomic Status and Psychological Well-Being: Revisiting the Role of Subjective Socioeconomic Status [Original Research]. \u003cem\u003eFrontiers in Psychology, 11\u003c/em\u003e. https://doi.org/10.3389/fpsyg.2020.01303\u003c/li\u003e\n\u003cli\u003eNesse, R. M. (2000). Is Depression an Adaptation? \u003cem\u003eArchives of General Psychiatry, 57\u003c/em\u003e(1), 14. https://doi.org/10.1001/archpsyc.57.1.14\u003c/li\u003e\n\u003cli\u003eO\u0026apos;Hara, M. W., Kohout, F. J., \u0026amp; Wallace, R. B. (1985). Depression among the Rural Elderly A Study of Prevalence and Correlates. \u003cem\u003eThe Journal of nervous and mental disease, 173\u003c/em\u003e(10), 582-589. https://journals.lww.com/jonmd/fulltext/1985/10000/depression_among_the_rural_elderly_a_study_of.2.aspx\u003c/li\u003e\n\u003cli\u003ePark, C. L., Kubzansky, L. D., Chafouleas, S. M., Davidson, R. J., Keltner, D., Parsafar, P., Conwell, Y., Martin, M. Y., Hanmer, J., \u0026amp; Wang, K. H. (2023). Emotional Well-Being: What It Is and Why It Matters. Affective Science, 4(1), 10-20. https://doi.org/10.1007/s42761-022-00163-0\u003c/li\u003e\n\u003cli\u003ePanaite, V., Rottenberg, J., \u0026amp; Bylsma, L. M. (2020). Daily Affective Dynamics Predict Depression Symptom Trajectories Among Adults with Major and Minor Depression. \u003cem\u003eAffective Science, 1\u003c/em\u003e(3), 186-198. https://doi.org/10.1007/s42761-020-00014-w\u003c/li\u003e\n\u003cli\u003ePearlin, L. I., \u0026amp; Schooler, C. (1978). The Structure of Coping. \u003cem\u003eJournal of Health and Social Behavior, 19\u003c/em\u003e(1), 2-21. https://doi.org/10.2307/2136319\u003c/li\u003e\n\u003cli\u003ePinquart, M., \u0026amp; S\u0026ouml;rensen, S. (2000). Influences of socioeconomic status, social network, and competence on subjective well-being in later life: a meta-analysis. \u003cem\u003ePsychology and Aging, 15\u003c/em\u003e(2), 187.\u003c/li\u003e\n\u003cli\u003eProust-Lima, C., Philipps, V., \u0026amp; Liquet, B. (2015). Estimation of extended mixed models using latent classes and latent processes: the R package lcmm. \u003cem\u003earXiv preprint arXiv:1503.00890\u003c/em\u003e.\u003c/li\u003e\n\u003cli\u003eRadloff, L. S. (1977). The CES-D Scale. \u003cem\u003eApplied Psychological Measurement, 1\u003c/em\u003e(3), 385-401. https://doi.org/10.1177/014662167700100306\u003c/li\u003e\n\u003cli\u003eRam, N., \u0026amp; Grimm, K. J. (2009). Methods and Measures: Growth mixture modeling: A method for identifying differences in longitudinal change among unobserved groups. \u003cem\u003eInternational Journal of Behavioral Development, 33\u003c/em\u003e(6), 565-576. https://doi.org/10.1177/0165025409343765\u003c/li\u003e\n\u003cli\u003eRevelle, W. (2024). \u003cem\u003epsych: Procedures for Psychological, Psychometric, and Personality Research. \u003c/em\u003eIn \u003cem\u003eR package\u003c/em\u003e (Version 2.4.1) https://cran.r-project.org/package=psych\u003c/li\u003e\n\u003cli\u003eRidge, D., \u0026amp; Ziebland, S. (2006). \u0026ldquo;The Old Me Could Never Have Done That\u0026rdquo;: How People Give Meaning to Recovery Following Depression. \u003cem\u003eQualitative Health Research, 16\u003c/em\u003e(8), 1038-1053. https://doi.org/10.1177/1049732306292132\u003c/li\u003e\n\u003cli\u003eRosenberg, M. (1965). \u003cem\u003eSociety and the Adolescent Self-image\u003c/em\u003e. Princeton University Press.\u003c/li\u003e\n\u003cli\u003eRosseel, Y. (2012, 05/24). lavaan: An R Package for Structural Equation Modeling. \u003cem\u003eJournal of Statistical Software, 48\u003c/em\u003e(2), 1 - 36. https://doi.org/10.18637/jss.v048.i02\u003c/li\u003e\n\u003cli\u003eRottenberg, J., Devendorf, A. R., Panaite, V., Disabato, D. J., \u0026amp; Kashdan, T. B. (2019). Optimal Well-Being After Major Depression. \u003cem\u003eClinical Psychological Science, 7\u003c/em\u003e(3), 621-627. https://doi.org/10.1177/2167702618812708\u003c/li\u003e\n\u003cli\u003eRyan, R. M., \u0026amp; Deci, E. L. (2001). On happiness and human potentials: A review of research on hedonic and eudaimonic well-being. \u003cem\u003eAnnual Review of Psychology, 52\u003c/em\u003e, 141.\u003c/li\u003e\n\u003cli\u003eRyff, C. D. (1989). Happiness is everything, or is it? Explorations on the meaning of psychological well-being. \u003cem\u003eJournal of Personality and Social Psychology, 57\u003c/em\u003e(6), 1069.\u003c/li\u003e\n\u003cli\u003eRyff, C. D. (1995). Psychological Well-Being in Adult Life. \u003cem\u003eCurrent Directions in Psychological Science, 4\u003c/em\u003e(4), 99-104. http://www.jstor.org/stable/20182342\u003c/li\u003e\n\u003cli\u003eRyff, C. D. (2023). Flotsam, Jetsam, and Forward-Moving Vessels on the Sea of Well-Being. Affective Science, 4(1), 49-51. https://doi.org/10.1007/s42761-022-00162-1\u003c/li\u003e\n\u003cli\u003eRyff, C. D., \u0026amp; Keyes, C. L. M. (1995). The structure of psychological well-being revisited. \u003cem\u003eJournal of Personality and Social Psychology, 69\u003c/em\u003e(4), 719.\u003c/li\u003e\n\u003cli\u003eRyff, C. D., Radler, B. T., \u0026amp; Friedman, E. M. (2015). Persistent psychological well-being predicts improved self-rated health over 9\u0026ndash;10\u0026thinsp;years: Longitudinal evidence from MIDUS. \u003cem\u003eHealth Psychology Open, 2\u003c/em\u003e(2), 205510291560158. https://doi.org/10.1177/2055102915601582\u003c/li\u003e\n\u003cli\u003eSamuelson, W., \u0026amp; Zeckhauser, R. (1988, 1988/03/01). Status quo bias in decision making. \u003cem\u003eJournal of Risk and Uncertainty, 1\u003c/em\u003e(1), 7-59. https://doi.org/10.1007/BF00055564\u003c/li\u003e\n\u003cli\u003eSwann, W. B., \u0026amp; Read, S. J. (1981, 1981/07/01/). Self-verification processes: How we sustain our self-conceptions. \u003cem\u003eJournal of Experimental Social Psychology, 17\u003c/em\u003e(4), 351-372. https://doi.org/https://doi.org/10.1016/0022-1031(81)90043-3\u003c/li\u003e\n\u003cli\u003eTan, J. J. X., Kraus, M. W., Carpenter, N. C., \u0026amp; Adler, N. E. (2020). The association between objective and subjective socioeconomic status and subjective well-being: A meta-analytic review. \u003cem\u003ePsychological Bulletin, 146\u003c/em\u003e(11), 970-1020. https://doi.org/10.1037/bul0000258\u003c/li\u003e\n\u003cli\u003eThompson, R. J., et al. (2013). \u0026quot;The Role of Attention to Emotion in Recovery from Major Depressive Disorder.\u0026quot; Depression Research and Treatment 2013: 1-6. https://dx.doi.org/10.1155/2013/540726\u003c/li\u003e\n\u003cli\u003evan Buuren, S., \u0026amp; Groothuis-Oudshoorn, K. (2011). mice: Multivariate Imputation by Chained Equations in R. \u003cem\u003eJournal of Statistical Software, 45\u003c/em\u003e(3). https://www.jstatsoft.org/v45/i03/ \u003c/li\u003e\n\u003cli\u003eVilagut, G., Forero, C. G., Barbaglia, G., \u0026amp; Alonso, J. (2016). Screening for Depression in the General Population with the Center for Epidemiologic Studies Depression (CES-D): A Systematic Review with Meta-Analysis. \u003cem\u003ePloS One, 11\u003c/em\u003e(5), e0155431. https://doi.org/10.1371/journal.pone.0155431\u003c/li\u003e\n\u003cli\u003eWest, S. G., Finch, J. F., \u0026amp; Curran, P. J. (1995). Structural equation models with nonnormal variables: Problems and remedies.\u003c/li\u003e\n\u003cli\u003eWesterhof, G. J., \u0026amp; Keyes, C. L. M. (2010). Mental Illness and Mental Health: The Two Continua Model Across the Lifespan. \u003cem\u003eJournal of Adult Development, 17\u003c/em\u003e(2), 110-119. https://doi.org/10.1007/s10804-009-9082-y\u003c/li\u003e\n\u003cli\u003eWood, A. M., \u0026amp; Joseph, S. (2010, 2010/05/01/). The absence of positive psychological (eudemonic) well-being as a risk factor for depression: A ten year cohort study. \u003cem\u003eJournal of Affective Disorders, 122\u003c/em\u003e(3), 213-217. https://doi.org/https://doi.org/10.1016/j.jad.2009.06.032\u003c/li\u003e\n\u003cli\u003eWood, A. M., \u0026amp; Tarrier, N. (2010, 2010/11/01/). Positive Clinical Psychology: A new vision and strategy for integrated research and practice. \u003cem\u003eClinical Psychology Review, 30\u003c/em\u003e(7), 819-829. https://doi.org/https://doi.org/10.1016/j.cpr.2010.06.003\u003c/li\u003e\n\u003cli\u003eZimmerman, M., McGlinchey, J. B., Posternak, M. A., Friedman, M., Attiullah, N., \u0026amp; Boerescu, D. (2006). How Should Remission From Depression Be Defined? The Depressed Patient\u0026rsquo;s Perspective. \u003cem\u003eAmerican Journal of Psychiatry, 163\u003c/em\u003e(1), 148-150. https://doi.org/10.1176/appi.ajp.163.1.148\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Declarations","content":"\u003cp\u003eAs defined at 45 CFR 46.102, the analysis of de-identified, publicly available data does not constitute human subjects research and thus does not require IRB review.\u003c/p\u003e"},{"header":"Footnotes","content":"\u003cp\u003e\u003csup\u003e1\u003c/sup\u003e\u003cimg width=\"207\" height=\"17\" src=\"data:image/png;base64,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\" alt=\"image\"\u003e\u003c/p\u003e\n\u003cp\u003e\u003csup\u003e2\u0026nbsp;\u003c/sup\u003eThe predictive effect of baseline depressive symptoms differed across models. Without baseline well-being level included, more severe baseline depressive symptoms were associated with \u003cem\u003edecreased\u003c/em\u003e odds of experiencing declining well-being and \u003cem\u003eincreased\u003c/em\u003e odds of experiencing either gradually or markedly increasing well-being.\u003c/p\u003e"},{"header":"Tables","content":"\u003cp\u003e\u003cstrong\u003eTable 1.\u003c/strong\u003e\u003cem\u003e\u0026nbsp;Final Sample Characteristics at Baseline (\u003c/em\u003eN=2,420\u003cem\u003e)\u003c/em\u003e\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"621\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 411px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eCharacteristic\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 210px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eN (%)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 411px;\"\u003e\n \u003cp\u003eSex (male)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 210px;\"\u003e\n \u003cp\u003e879 (36.32%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 411px;\"\u003e\n \u003cp\u003eRace\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 210px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 411px;\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp;Caucasian\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 210px;\"\u003e\n \u003cp\u003e1,658 (68.51%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 411px;\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp;African American\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 210px;\"\u003e\n \u003cp\u003e693 (28.64%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 411px;\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp;Other\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 210px;\"\u003e\n \u003cp\u003e69 (2.85%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 411px;\"\u003e\n \u003cp\u003eSES\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 210px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 411px;\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp;Low SES\u003c/p\u003e\n \u003cul\u003e\n \u003cli\u003e\u0026ge;11 years education and family income \u0026lt;$20,000\u003c/li\u003e\n \u003c/ul\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 210px;\"\u003e\n \u003cp\u003e536 (22.15%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 411px;\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp;Lower-Middle SES\u003c/p\u003e\n \u003cul\u003e\n \u003cli\u003e\u0026ge;11 years education and family income \u0026gt;$20,000\u003c/li\u003e\n \u003cli\u003e\u0026le;12 years of education and family income \u0026lt;$20,000)\u003c/li\u003e\n \u003c/ul\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 210px;\"\u003e\n \u003cp\u003e749 (30.95%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 411px;\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp;Upper-Middle SES\u003c/p\u003e\n \u003cul\u003e\n \u003cli\u003e12-15 years education and family income \u0026gt;$20,000\u003c/li\u003e\n \u003c/ul\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 210px;\"\u003e\n \u003cp\u003e795 (32.85%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 411px;\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp;High SES\u003c/p\u003e\n \u003cul\u003e\n \u003cli\u003e\u0026ge;16 years education and family income \u0026gt;$20,000\u003c/li\u003e\n \u003c/ul\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 210px;\"\u003e\n \u003cp\u003e340 (14.05%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 411px;\"\u003e\n \u003cp\u003eAge (\u003cem\u003eM, SD\u003c/em\u003e)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 210px;\"\u003e\n \u003cp\u003e49.15 (15.88)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 411px;\"\u003e\n \u003cp\u003eBaseline Depressive Symptom Severity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 210px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 411px;\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp;Likely Not Depressed\u003c/p\u003e\n \u003cul\u003e\n \u003cli\u003eCES-D-11\u0026le;8.289 (n, \u003cem\u003e%\u003c/em\u003e;\u003cem\u003e\u0026nbsp;M, SD)\u003c/em\u003e\u003c/li\u003e\n \u003c/ul\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 210px;\"\u003e\n \u003cp\u003e2020 (83.47%); 3.16 (2.47)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 411px;\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp;Likely Depressed\u003c/p\u003e\n \u003cul\u003e\n \u003cli\u003eCES-D-11\u0026gt;8.289 (n, \u003cem\u003e%\u003c/em\u003e;\u003cem\u003e\u0026nbsp;M, SD)\u003c/em\u003e\u003c/li\u003e\n \u003c/ul\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 210px;\"\u003e\n \u003cp\u003e396 (16.36%); 11.37 (2.51)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 411px;\"\u003e\n \u003cp\u003eWell-being\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 210px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 411px;\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp;Low Well-being\u003c/p\u003e\n \u003cul\u003e\n \u003cli\u003eWBLV \u0026lt; -.16\u003c/li\u003e\n \u003c/ul\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 210px;\"\u003e\n \u003cp\u003e876 (36.20%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 411px;\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp;Lower-Typical Well-being\u003c/p\u003e\n \u003cul\u003e\n \u003cli\u003e-.16\u0026le;WBLV\u0026lt;.13\u003c/li\u003e\n \u003c/ul\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 210px;\"\u003e\n \u003cp\u003e517 (21.36%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 411px;\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp;Upper-Typical Well-being\u003c/p\u003e\n \u003cul\u003e\n \u003cli\u003e.13\u0026le;WBLV\u0026lt;.39\u003c/li\u003e\n \u003c/ul\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 210px;\"\u003e\n \u003cp\u003e538 (22.23%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 411px;\"\u003e\n \u003cp\u003e\u0026nbsp; \u0026nbsp;Optimal Well-being\u003c/p\u003e\n \u003cul\u003e\n \u003cli\u003eWBLV\u0026ge;.39\u003c/li\u003e\n \u003c/ul\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 210px;\"\u003e\n \u003cp\u003e489 (20.21%)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eNote.\u003c/em\u003e\u003c/strong\u003e CES-D=depressive symptom severity score. WBLV=well-being latent variable score. Thresholds for well-being classes were determined by identifying the 25\u003csup\u003eth\u003c/sup\u003e, 50\u003csup\u003eth\u003c/sup\u003e, and 75\u003csup\u003eth\u003c/sup\u003e quartiles of well-being among people who were likely not depressed at baseline.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 2.\u003c/strong\u003e \u003cem\u003eFit Statistics for Baseline Growth Model and Growth Mixture Models\u0026nbsp;\u003c/em\u003e(N=2420)\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"100%\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 98px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 78px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e1-Class Linear\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 111px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e1-Class Exponential\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e1-Class Logarithmic\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e1-Class Sinusoidal\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e1-Class Quadratic\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 98px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eNumber of Parameters\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 78px;\"\u003e\n \u003cp\u003e6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 111px;\"\u003e\n \u003cp\u003e7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e10\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e10\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e7\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 98px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eLL\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 78px;\"\u003e\n \u003cp\u003e-12,666\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 111px;\"\u003e\n \u003cp\u003e-19,307\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e-12,658\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e-12,654\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e-12,661\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 98px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eAIC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 78px;\"\u003e\n \u003cp\u003e25,344\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 111px;\"\u003e\n \u003cp\u003e38,627\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e25,335\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e25,327*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e25,336\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 98px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eBIC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 78px;\"\u003e\n \u003cp\u003e25,379\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 111px;\"\u003e\n \u003cp\u003e38,668\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e25,393\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e25,386\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e25,376*\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 98px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eCAIC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 78px;\"\u003e\n \u003cp\u003e25,344\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 111px;\"\u003e\n \u003cp\u003e38,628\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e25,336\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e25,328*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e25,336\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 98px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSABIC\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 78px;\"\u003e\n \u003cp\u003e25,360\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 111px;\"\u003e\n \u003cp\u003e38,646\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e25,362\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e25,354*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e25,354*\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 98px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eEntropy\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 78px;\"\u003e\n \u003cp\u003e1.00\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 111px;\"\u003e\n \u003cp\u003e1.00\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 127px;\"\u003e\n \u003cp\u003e1.00\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 94px;\"\u003e\n \u003cp\u003e1.00\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 117px;\"\u003e\n \u003cp\u003e1.00\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eNote.\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e(C)AIC = (Corrected) Akaike Information Criterion. (SA)BIC = (Sample-Adjusted) Bayes Information Criterion. VLMR LRT = Voung-Lo-Mendell-Rubin Likelihood Ratio Test. *model of best fit for each information criterion\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 3.\u0026nbsp;\u003c/strong\u003e\u003cem\u003eFit Statistics for Growth Mixture Models using Well-being Change Scores\u0026nbsp;\u003c/em\u003e(N=2420)\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"100%\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 145px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 92px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e1-Class\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e2-Class\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e3-Class\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e4-Class\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 102px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e5-Class\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"2\" valign=\"top\" style=\"width: 229px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSample Size\u003csup\u003eA\u003c/sup\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 8px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 93px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 93px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 145px;\"\u003e\n \u003cp\u003eN\u003csub\u003ec=1\u003c/sub\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 92px;\"\u003e\n \u003cp\u003e2,420\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e1,499\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e1,751\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e1,443\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 102px;\"\u003e\n \u003cp\u003e1,356\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 145px;\"\u003e\n \u003cp\u003eN\u003csub\u003ec=2\u003c/sub\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 92px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e921\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e336\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e736\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 102px;\"\u003e\n \u003cp\u003e515\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 145px;\"\u003e\n \u003cp\u003eN\u003csub\u003ec=3\u003c/sub\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 92px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e333\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e183\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 102px;\"\u003e\n \u003cp\u003e318\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 145px;\"\u003e\n \u003cp\u003eN\u003csub\u003ec=4\u003c/sub\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 92px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e58\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 102px;\"\u003e\n \u003cp\u003e188\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 145px;\"\u003e\n \u003cp\u003eN\u003csub\u003ec=5\u003c/sub\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 92px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 102px;\"\u003e\n \u003cp\u003e43\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"2\" style=\"width: 229px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eFit Statistics\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 8px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 99px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 93px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 93px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 102px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 145px;\"\u003e\n \u003cp\u003eNumber of Parameters\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 92px;\"\u003e\n \u003cp\u003e7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e12\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e17\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e22\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 102px;\"\u003e\n \u003cp\u003e27\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 145px;\"\u003e\n \u003cp\u003eLL\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 92px;\"\u003e\n \u003cp\u003e-4,332\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e-4,063\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e-3,948\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e-3,906\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 102px;\"\u003e\n \u003cp\u003e-3,901\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 145px;\"\u003e\n \u003cp\u003eAIC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 92px;\"\u003e\n \u003cp\u003e8,678\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e8,149\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e7,929\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e7,856\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 102px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e7,855\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 145px;\"\u003e\n \u003cp\u003eBIC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 92px;\"\u003e\n \u003cp\u003e8,716\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e8,219\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e8,028\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e7,983\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 102px;\"\u003e\n \u003cp\u003e8,011\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 145px;\"\u003e\n \u003cp\u003eCAIC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 92px;\"\u003e\n \u003cp\u003e8,718\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e8,219\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e8,028\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e7,983\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 102px;\"\u003e\n \u003cp\u003e8,011\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 145px;\"\u003e\n \u003cp\u003eSABIC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 92px;\"\u003e\n \u003cp\u003e8,696\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e8,181\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e7,934\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e7,913\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 102px;\"\u003e\n \u003cp\u003e7,925\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 145px;\"\u003e\n \u003cp\u003eEntropy\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 92px;\"\u003e\n \u003cp\u003e1.00\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e0.480\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e0.612\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.636\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 102px;\"\u003e\n \u003cp\u003e0.551\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 145px;\"\u003e\n \u003cp\u003eAdjusted\u003c/p\u003e\n \u003cp\u003eVLMR LRT\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"2\" style=\"width: 92px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 99px;\"\u003e\n \u003cp\u003e516.28**\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e220.51**\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 93px;\"\u003e\n \u003cp\u003e79.86**\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 102px;\"\u003e\n \u003cp\u003e10.49\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eNote.\u003c/em\u003e\u003c/strong\u003e A) Estimated counts for the latent classes based on the posterior probabilities. LL = Log-likelihood. (C)AIC = (Corrected) Akaike Information Criterion. (SA)BIC = (Sample-Adjusted) Bayes Information Criterion. VLMR LRT = Voung-Lo-Mendell-Rubin Likelihood Ratio Test; values are for tests of n vs n-1 classes. All information criteria are rounded to the nearest whole number. Bold font indicates the model of best fit for each information criterion. **\u003cem\u003ep\u003c/em\u003e\u0026lt;.001, *\u003cem\u003ep\u003c/em\u003e\u0026lt;.05\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 4.\u003c/strong\u003e\u003cem\u003e\u0026nbsp;Multinomial Logistic Regression Model\u003c/em\u003e\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"101%\" class=\"fr-table-selection-hover\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"4\" style=\"width: 377px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eFull Data (N=2420)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"4\" style=\"width: 395px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eReduced Data (N=1936)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eTerm\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026beta;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSt.\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eErr\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eExp(\u0026beta;)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 129px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e95% CI\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003efor Exp(\u0026beta;)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026beta;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eSt.\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eErr\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eExp(\u0026beta;)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e95% CI\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003efor Exp(\u0026beta;)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003eClass 2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"4\" style=\"width: 377px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003eInt.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e-4.005**\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e0.543\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 129px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003e-4.776**\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0.663\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003eAge\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e-0.017*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e0.005\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e0.98\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 129px;\"\u003e\n \u003cp\u003e0.97, 0.99\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003e-0.016*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0.006\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e0.98\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e0.97, 1.00\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003eSex\u003csub\u003eFemale\u003c/sub\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e0.189\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e0.174\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e1.21\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 129px;\"\u003e\n \u003cp\u003e0.86, 1.70\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003e0.272\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0.212\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e1.31\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e0.87, 1.99\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003eSES\u003csub\u003eLowerMC\u003c/sub\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e-0.432*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e0.215\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e0.65\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 129px;\"\u003e\n \u003cp\u003e0.43, 0.99\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003e-0.249\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0.262\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e0.78\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e0.47, 1.30\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003eSES\u003csub\u003eUpperMC\u003c/sub\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e-0.684*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e0.229\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e0.50\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 129px;\"\u003e\n \u003cp\u003e0.32, 0.79\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003e-0.517\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0.276\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e0.60\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e0.35, 1.03\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003eSES\u003csub\u003eUpper\u003c/sub\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e-0.784*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e0.301\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e0.46\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 129px;\"\u003e\n \u003cp\u003e0.25, 0.82\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003e-0.753*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0.379\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e0.47\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e0.22, 0.99\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003eCES-D\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e-0.095*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e0.031\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e1.10\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 129px;\"\u003e\n \u003cp\u003e1.03, 1.17\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003e-0.105*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0.037\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e1.11\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e1.03, 1.20\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003eWBLV\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003e1.244*\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e0.155\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e3.47\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 129px;\"\u003e\n \u003cp\u003e2.56, 4.71\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003e1.257**\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0.187\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e3.52\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e2.44, 5.07\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003eClass 3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"4\" style=\"width: 377px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003eInt.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e-0.521**\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e1.334\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 129px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003e1.134**\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0.420\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003eAge\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e0.015\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e0.010\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e1.02\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 129px;\"\u003e\n \u003cp\u003e1.00, 1.03\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003e0.008\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0.011\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e1.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e0.99, 1.03\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003eSex\u003csub\u003eFemale\u003c/sub\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e0.007\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e0.321\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e1.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 129px;\"\u003e\n \u003cp\u003e0.54, 1.89\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003e0.126\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0.366\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e1.13\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e0.55, 2.32\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003eSES\u003csub\u003eLowerMC\u003c/sub\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e0.282\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e0.365\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e1.33\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 129px;\"\u003e\n \u003cp\u003e0.65, 2.71\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003e0.059\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0.405\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e1.06\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e0.48, 2.35\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003eSES\u003csub\u003eUpperMC\u003c/sub\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e0.258\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e0.389\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e1.29\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 129px;\"\u003e\n \u003cp\u003e0.60, 2.78\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003e-0.044\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0.425\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e0.96\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e0.42, 2.20\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003eSES\u003csub\u003eUpper\u003c/sub\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e-0.369\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e0.671\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e0.69\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 129px;\"\u003e\n \u003cp\u003e0.19, 2.58\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003e-0.718\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0.800\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e0.49\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e0.10. 2.34\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003eCES-D\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e0.183**\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e0.037\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e1.20\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 129px;\"\u003e\n \u003cp\u003e1.12, 1.29\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003e0.192**\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0.040\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e1.21\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e1.12, 1.31\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003eWBLV\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e-3.838**\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e1.023\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e0.02\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 129px;\"\u003e\n \u003cp\u003e0.00, 0.16\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003e-1.527**\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0.420\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e0.00\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e0.00, 0.00\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003eClass 4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd colspan=\"4\" style=\"width: 377px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003eInt.\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e0.933**\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e0.311\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 129px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003e-1.433**\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0.356\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003eAge\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e0.003\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e0.003\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e1.00\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 129px;\"\u003e\n \u003cp\u003e1.00, 1.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003e-0.0009\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0.004\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e1.00\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e0.99, 1.01\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003eSex\u003csub\u003eFemale\u003c/sub\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e0.064\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e0.104\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e1.07\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 129px;\"\u003e\n \u003cp\u003e0.87, 1.31\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003e0.131\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0.120\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e1.14\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e0.90, 1.44\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003eSES\u003csub\u003eLowerMC\u003c/sub\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e0.129\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e0.139\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e1.14\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 129px;\"\u003e\n \u003cp\u003e0.87, 1.49\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003e0.418\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0.161\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e1.15\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e.084, 1.58\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003eSES\u003csub\u003eUpperMC\u003c/sub\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e0.181\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e0.142\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e1.20\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 129px;\"\u003e\n \u003cp\u003e0.91, 1.58\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003e0.052\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0.164\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e1.05\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e0.76, 1.45\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003eSES\u003csub\u003eUpper\u003c/sub\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e0.308\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e0.174\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e1.36\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 129px;\"\u003e\n \u003cp\u003e0.97, 1.91\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003e0.263\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0.201\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e1.30\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e0.88, 1.93\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003eCES-D\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e0.018\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e0.016\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e1.02\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 129px;\"\u003e\n \u003cp\u003e0.99, 1.05\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003e0.010\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0.018\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e1.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e0.97, 1.05\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003eWBLV\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 86px;\"\u003e\n \u003cp\u003e-1.187**\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 73px;\"\u003e\n \u003cp\u003e0.099\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e0.31\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 129px;\"\u003e\n \u003cp\u003e0.25, 0.37\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 105px;\"\u003e\n \u003cp\u003e-1.438**\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 80px;\"\u003e\n \u003cp\u003e0.115\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 89px;\"\u003e\n \u003cp\u003e0.24\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 122px;\"\u003e\n \u003cp\u003e0.19, 0.30\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eNote\u003c/em\u003e\u003c/strong\u003e. MC=Middle Class. CES-D=depressive symptom severity score. WBLV=Well-being latent variable score. * \u003cem\u003ep\u003c/em\u003e\u0026lt;.05, ** \u003cem\u003ep\u003c/em\u003e\u0026lt;.001\u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"nature-portfolio","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"","title":"Nature Portfolio","twitterHandle":"","acdcEnabled":false,"dfaEnabled":false,"editorialSystem":"ejp","reportingPortfolio":"","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"well-being, depressive symptoms, longitudinal data analysis, growth mixture modeling, American’s Changing Lives (ACL)","lastPublishedDoi":"10.21203/rs.3.rs-7123882/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7123882/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"The extent to which adults sustain psychological well-being across decades—and the role of depression in shaping these trajectories—remains poorly understood. Using data from the 25-year Americans' Changing Lives cohort (N=3,617), we applied growth mixture modeling to identify longitudinal well-being trajectories and assess whether baseline depressive symptoms predict stability versus change. Four distinct patterns emerged: persistent well-being (60% of the sample), declining well-being, gradually increasing well-being, and sharply increasing well-being. While persistent well-being was the mentally healthiest trajectory, many of this group's members still exhibited substantial intra-individual variability in well-being, and some maintained persistently low well-being. Elevated baseline depressive symptoms predicted greater odds of non-stable trajectories, including both decline and marked improvement, independent of baseline well-being level. These findings underscore the dynamic interplay between depressive symptoms and long-term mental health. Results support interventions that go beyond symptom reduction to cultivate the maintenance of psychological well-being across time.","manuscriptTitle":"Depressive Symptoms Predict Divergent Trajectories of Well-being in U. S. Adults","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-07-31 11:36:39","doi":"10.21203/rs.3.rs-7123882/v1","editorialEvents":[],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"nature-mental-health","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"natmentalhealth","sideBox":"Learn more about [Nature Mental Health](https://www.nature.com/natmentalhealth/)","snPcode":"44220","submissionUrl":"https://mts-natmentalhealth.nature.com/cgi-bin/main.plex","title":"Nature Mental Health","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"ejp","reportingPortfolio":"Nature Research","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"4de30a52-87e3-47f0-b342-bdd2ec308ea9","owner":[],"postedDate":"July 31st, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[{"id":51955803,"name":"Scientific community and society/Social sciences/Psychology"},{"id":51955804,"name":"Biological sciences/Psychology"}],"tags":[],"updatedAt":"2026-02-27T17:42:15+00:00","versionOfRecord":[],"versionCreatedAt":"2025-07-31 11:36:39","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7123882","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7123882","identity":"rs-7123882","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}