BACKGROUND

The aim of the present study was to explore in more detail the relationship between subjective well-being (SWB) and depression. According to Diener’s well-recognized (1984) definition, SWB involves high positive affect, low negative affect and high satisfaction with life, which implicitly suggests an inverse relationship with depression. Some authors have even conceived depression as the loss of subjective well-being (Gargiulo & Stokes, 2009). Many researchers have found that the lower subjective well-being is, the higher is the severity of depression (Goldberg & Harrow, 2005; Wood & Joseph, 2010). This effect is consistent between age groups and cultures (Carandang et al., 2019; Derdikman-Eiron et al., 2011; Rao et al., 2017; Van Beveren et al., 2018).

This seemingly clear issue becomes more complex when: (a) depression is conceived as a personality dimension (depressiveness) rather than in clinical terms and when (b) subcomponents of this construct are considered. Such conceptualization of depressiveness draws on Cattell and Scheier’s (1960) distinction between state and trait aspects of emotional experience as well as on Spielberger’s (2006, p. 301) theorizing that “individual differences in anxiety, anger and depression as personality traits can be evaluated in terms of frequency that these emotions are experienced”. The present paper is focused on trait depression, i.e. depression conceived as a personality dimension rather than a clinical condition.

Given that “traits are presumed to be relatively enduring psychological characteristics that influence people’s thoughts, feelings and behaviours” (Nezlek, 2007, p. 791), trait depression is defined as a personality disposition to experience depressive emotions across contexts and situations. In terms of the Five Factor Model of personality, trait depression refers to the depression facet of neuroticism, which is defined as “the disposition to experience sadness, hopelessness, and loneliness” (McCrae & Costa, 2003, p. 48). Depressiveness here is conceived as a dispositional form of the emotional state of sorrow, and high levels of this trait are related to feelings of guilt and diminished self-worth.

Most conceptualizations of depression, regardless of whether they refer to a clinical condition or a personality trait, are focused on its dysthymic aspect, which refers to experiencing negative emotional states and/or negative symptoms. In contrast, Spielberger et al. (2003) and Spielberger and Reheiser (2009) argued that trait depression should be described in terms of two interrelated but distinct facets called dysthymia and euthymia. They refer to separate – negative vs positive – aspects of affectivity associated with depression. Dysthymia is a depression-presence factor, while euthymia is a depression-absence factor. Whereas dysthymia is strongly emphasized in the depression literature, the euthymia aspect captures a neglected facet of depressiveness. It refers to depression in an indirect way, by pointing to deficits in positive emotions. This emphasis on positive experiences is clearly reflected in the etymology of the term “euthymia”. It comes from the combination of the Greek words “eu”, which means “well”, and “thymeo”, meaning “soul, emotion”, or from the verb “euthymeo”, which means “I am happy” (Fava & Bech, 2016, p. 1).

The distinction between dysthymia and euthymia is particularly relevant for studying depressiveness in non-clinical samples, because it broadens the spectrum of depression-related experiences. While dysthymia refers to negative emotional experiences typical for higher levels of depression including its clinical symptoms, the reversed euthymia factor is more sensitive to milder levels of depressiveness (Spielberger et al., 2003). This two-facet conceptualisation of depression is thus particularly useful as a theoretical framework for a more detailed analysis of the relationship between depression and subjective well-being in the general population (as contrasted with clinically depressed individuals).

We hypothesised that for the general population the euthymia facet of depression is a more important predictor of SWB than the dysthymia facet. Among the subcomponents of SWB hedonic well-being is usually defined in terms of experiencing positive affective states or subjective happiness (Diener, 1984; Sin & Lyubomirsky, 2009; Spinhoven et al., 2015). Similarly, hedonic balance defined as perceived experience of positive affect relative to negative affect (Allen et al., 2016; Stillmaker & Kasser, 2013) is considered to be a strong indicator of the emotional component of well-being (Caprara & Steca, 2006; Schimmack, 2003). Likewise, the preponderance of positive over negative affect along with the focus on an overall affective appraisal of life is the key element of happiness (Ryan & Deci, 2001; Sin & Lyubomirsky, 2009). Happy individuals with a positive view of life are characterized by a higher level of life satisfaction and psychological well-being including purpose of life and self-acceptance. This suggests that it is the presence of positive psychological resources, rather that the absence of negative affect, which protects against the occurrence of depressive symptoms (Xu et al., 2015). We thus expected that for the non-clinical population euthymia is a more important predictor of SWB that dysthymia.

One may wonder whether this relatively less-known dysthymia–euthymia distinction is not redundant with the well-known distinction between positive and negative affect (PA, NA; Watson & Tellegen, 1985). Thus, we controlled for the basic levels of PA and NA. We expected that the predictive role of euthymia over dysthymia for SWB would be retained when PA and NA were taken into consideration.

PARTICIPANTS AND PROCEDURE

PARTICIPANTS

A total of 213 community dwelling adults (females n = 119, 56%), aged 19-61 (Mage = 35.85, SD = 11.30) participated in the study. All participants were informed about the aim and procedure of the study. Those who gave their agreement to participate completed self-report measures (as listed below) in the paper-pencil format. They did not receive compensation for their participation.

MEASURES

Trait depression. In line with the focus of this study on trait depression, the two components of depressiveness, i.e. trait dysthymia and trait euthymia, were measured by the trait scales from the State-Trait Depression Scales (STDS; Krohne et al., 2002; Spielberger et al., 2003) in the Polish adaptation by Bąk and Oleś (2005) (see also Bąk & Łysiak, 2018). The development of the STDS was guided by the need for more efficient measures of different degrees of state and trait depression in non-clinical samples. The existing measures of depression, widely used in psychological research (e.g. Beck Depression Inventory), are effective in screening the general population for people with clinical depression. However, they are unable to differentiate between milder levels of depressiveness within non-clinical samples (Krohne et al., 2002). The STDS was designed to do so by including both depression-present (dysthymia) and depression-absent (euthymia) items. This drew on the concept of item-intensity specificity, stating that individual items are more effective in measuring a given construct at some levels of underlying theoretical dimension than at other levels (Spielberger et al., 2003). With reference to trait depression “the concept of item-intensity specificity calls attention to the fact that depression-absent items (…) when reverse scored, are more sensitive for assessing lower levels of depression, whereas depression-present items (…) are more effective measures of higher levels of depression” (Spielberger et al., 2003, p. 230).

The STDS trait depression scale is composed of 10 dysthymia items (e.g. “I feel miserable”, “I feel low”) and 10 euthymia items (e.g. “I feel strong”, “I feel happy”). Participants responded to each item using a 4-point frequency scale ranging from 1 (almost never) to 4 (almost always) to indicate how they generally feel. The reliability of dysthymia and euthymia subscales for the current sample were α = .91, 95% CI [.89, .93] and α = .89, 95% CI [.87, .91], respectively.

Subjective well-being. Subjective well-being (SWB) was operationalized in terms of satisfaction with life measured by the Satisfaction with Life Scale (SWLS; Diener et al., 1985) in the Polish adaptation by Juczyński (2012). The SWLS is a 5-item scale designed to measure global cognitive judgments of one’s satisfaction with life. Participants indicate how much they agree with each item (e.g. “In most ways my life is close to my ideal”) using a 7-point scale ranging from 1 (strongly disagree) to 7 (strongly agree). The reliability of the SWLS score for the current sample was α = .80, 95% CI [.75, .84].

Positive and negative affect. The basic affect was measured with the Positive and Negative Affect Schedule (PANAS; Watson et al., 1988), which in the Polish adaptation consists of 20 or 30 items, depending of the version of PANAS (Brzozowski, 2010). We used a longer version composed of 15 positive affect items (e.g. strong, active) and 15 negative affect items (e.g. distressed, upset), to which participants respond on a 5-point scale ranging from 1 (very slightly or not at all) to 5 (extremely). The reliability of PA and NA subscales was α = .86, 95% CI [.84, .89] and α = .78, 95% CI [.74, .82] respectively.

RESULTS

Table 1 presents means, standard deviations, and reliability coefficients for all scales included in the study as well as intercorrelations between the variables. To test the hypothesis that euthymia predominates over dysthymia in predicting satisfaction with life and that this effect is retained when positive and negative affect are controlled, we conducted a hierarchical regression analysis with SWLS as the dependent variable. Checking for collinearity of predictors the appropriate diagnostics showed that the assumptions were met (VIF ranged between 1.07 and 2.03; tolerance ranged between .49 and .94). We entered the predictor variables in three blocks (steps) to control for their independent role in predicting the outcome variable. NA and PA were entered as the first block, followed by dysthymia in the second step and finally euthymia in the third step (see Table 2).

Table 1

Descriptive statistics, bivariate correlations, and reliability coefficients

VariablesM95% CISD12345
1. Dysthymia1.75[1.67, 1.82]0.58(.91)
2. Euthymia2.90[2.82, 2.98]0.61−.65(.89)
3. Positive affect3.18[3.09, 3.27]0.66−.55.54(.86)
4. Negative affect2.04[1.92, 2.15]0.83.42−.45−.25(.78)
5. Satisfaction with life4.21[4.06, 4.37]1.13−.39.47.42−.28(.80)

[i] Note. All correlations are significant at p < .001. Cronbach’s alphas are presented in parentheses along the diagonal.

Table 2

Hierarchical multiple regression analysis predicting satisfaction with life (SWLS) – model 1: four predictors

Predictor variableΔR2AICBICB [95% CI]βtp
Step 1.21***8.1218.21
Constant2.68 [1.83, 3.53]
Positive affect0.64 [0.43, 0.85].385.93< .001
Negative affect−0.25 [−0.42, −0.08]−.18−2.86.005
Step 2.02*5.5018.95
Constant3.54 [2.38, 4.69]
Positive affect0.51 [0.26, 0.75].304.07< .001
Negative affect−0.18 [−0.36, 0.01]−.13−1.95.053
Dysthymia−0.33 [−0.63, −0.03]−.17−2.14.033
Step 3.04***−4.0812.73
Constant1.82 [0.31, 3.32]
Positive affect0.38 [0.13, 0.63].222.97.003
Negative affect−0.10 [−0.28, 0.09]−.07−1.02.309
Dysthymia−0.09 [−0.42, 0.23]−.05−0.56.573
Euthymia0.54 [0.23, 0.85].293.41.001
Total R2.25***

[i] Note. AIC – Akaike information criterion, BIC – Bayesian information criterion; *p < .05; ***p < .001.

First, the results showed that both PA and NA are significant predictors of SWLS, though the predictive role of positive affect was twice as strong as the role of negative affect. Those basic temperamental aspects of affect accounted jointly for 21% of variance of satisfaction with life. Entering the dysthymia aspect of trait depression at the second step added 2% of explained variance of SWLS. Dysthymia turned out to be a significant predictor of SWLS independently of PA. Moreover, when dysthymia was entered in the regression model NA was no longer a significant predictor of SWLS.

Entering the euthymia aspect of trait depression in the third step of the regression model added another 4% of variance explained by the predictors and showed that euthymia is the most important predictor of SWLS. The predictive role of NA remained statistically non-significant and additionally dysthymia lost its predictive power for euthymia. PA remained a significant predictor, though its role was weaker compared to both the first and the second steps. It was also noticeably weaker than the predictive role of euthymia.

For the sake of parsimony of the model we finally ran the second regression from which NA and dysthymia were excluded, since they turned out to be statistically non-significant predictors. The results reported in Table 3 showed that euthymia and positive affect contribute independently to predicting SWLS. The higher the levels of euthymia and PA are, the higher is satisfaction with life. However, the predictive role of euthymia is stronger compared to PA. The second model, which includes only two predictors (euthymia aspect of depressiveness and positive affect), can be regarded as a more parsimonious model compared to the original four-predictor model. The simpler model was similarly effective in predicting SWLS (R2 values for both models were equal), and it turned out to be a better model in terms of both the Akaike information criterion and the Bayesian information criterion (see AICs and BICs in Tables 2 and 3).

Table 3

Multiple regression analysis predicting satisfaction with life (SWLS) – model 2: two predictors

Predictor variableB [95% CI]βtp
Constant1.09 [0.36, 1.82]
Positive affect0.40 [0.16, 0.64].233.32.001
Euthymia0.64 [0.38, 0.90].344.85< .001

[i] Note. R2 = .25, p < .001; Akaike information criterion AIC = −6.40, Bayesian information criterion BIC = 3.69.

DISCUSSION

The present study was focused on trait depression as a predictor of subjective well-being. Most conceptualizations of depressiveness, including, in particular, those referring to depression as a clinical condition, are focused on the presence of negative experiences. We, however, adopted Spielberger’s (e.g., Spielberger et al., 2003) proposition to take into account not only the dysthymic aspect of depression but also its euthymic facet, which refers to the deficits of positive experiences. Such conceptualization of trait depression, as the combination of interrelated dysthymia and euthymia facets, is claimed to be particularly relevant for the non-clinical population since it helps in distinguishing between milder levels of depressiveness (Krohne et al., 2002).

We postulated that for adults from the general population, euthymia is a more important (compared to dysthymia) predictor of subjective well-being as conceptualized in terms of satisfaction with life. Additionally, we expected that this effect would be relatively independent of the basic level of positive and negative affect. The results of our study fully confirmed our expectations. NA served as a statistically significant predictor of life satisfaction only until dysthymia was taken into account. Still, the predictive role of dysthymia remained only until euthymia was entered as a predictor. When all four predictors (PA, NA, dysthymia, and euthymia) were entered in the regression model only those related to positive affectivity (PA and euthymia) served as statistically significant predictors. Though they both contributed independently to the variance of satisfaction with life explained by the model, the predictive role of euthymia was noticeably stronger.

This is in line with the results of several studies which showed associations between diminished positive emotions and elevated depressive symptoms as independent of negative emotions (Feldman et al., 2008; Nelis et al., 2015; Raes et al., 2012; Werner-Seidler et al., 2013). Still, our results suggest that in addition to temperamental level of PA, more specific positive experiences should be considered to adequately address depression-related issues in the non-clinical population. When one adopts the traditional approach to depressiveness focused on the dysthymic aspect only, one may not be able to differentiate between milder levels of this personality disposition, which will result in limited ability of trait depression to predict important outcomes, such as subjective well-being.

Our findings point to a likely attentional mechanism by which emotions promote well-being and alleviate depression, with a stronger effect of euthymia and PA than dysthymia and NA. The experience of positive affect may increase individuals’ selective attention preferences for positive information, and this broadened attention towards positive information may facilitate emotion regulation by helping individuals to attend to opportunities for reward in their environment. This is in line with the broaden-and-build theory by Fredrickson (e.g., Cohn et al., 2009; Fredrickson, 2001) claiming that positive emotions broaden thought and action repertoires, increase mental flexibility, augment meaning-based coping, and motivate engagement in novel activities and social relationships. This leads to trying new activities, developing new skills, and engaging in more positive social interactions (Garland et al., 2010; Smith & Hollinger-Smith, 2015). Presence of positive emotions extends coping resources, leads to better stress management and restores resilience (Dolphin et al., 2015; Folkman, 1997).

The results of the present study have also applicative potential. Many arguments have been made to suggest that mental illness can or even should be treated by promoting well-being strategies, and that positive psychology approaches may be instrumental in order to delay adverse health outcomes (Lamers et al., 2012; Rao et al., 2017). Enhancing positive emotionality is an important treatment goal for persons with lower levels of happiness (Sin & Lyubomirsky, 2009; Spinhoven et al., 2015). Our findings add to this by pointing out not only the importance of positive interventions but also the need for monitoring the level of state euthymia in the prevention programmes against depression. Decreased state euthymia may serve as a warning signal of an incoming depression episode, which may be detectable before more severe symptoms appear. Preparing for the presence of euthymia, on the other hand, may enhance subjective well-being, which, in light of our study, seems to be more important than dealing with negative emotions. Still, to verify the effectiveness of such preventive programmes future applied studies are needed, assessing and controlling not only the trait but also the state aspect of euthymia.

Although recruiting the sample from community-dwelling adults was fully consistent with the purpose of this particular study, one may wonder if the pattern of results would be different for those who are clinically depressed. Indeed, the results of our other study suggest that this may be the case. For clinically depressed participants dysthymia seemed to be at least as important as euthymia in predicting SWB (Łysiak & Bąk, 2018). Unfortunately, since it was not the main objective of that study, we used very simplified, one-item measures of SWB, which made it difficult to adequately test the above hypothesis. Thus, further studies are needed to directly test the independent role of dysthymia and euthymia in predicting SWB in clinical as compared to non-clinical samples. Moreover, given that all the measures used in the present study were self-report inventories, it would be beneficial in future studies to go beyond self-report and include e.g. physicians’ ratings of symptoms of depression.

Another limitation is the cross-sectional design of our study, which does not allow any cause and effect inferences to be made. We tested the regression model with trait depression as a predictor of SWB, but we are aware that this in fact may be a bidirectional relationship. To explicitly test the hypothesis that euthymia is one of the important causes or mechanisms of satisfaction with life, a longitudinal design would be optimal.