Do Romantic Relationships Promote Happiness? Relationships’ Characteristics as Predictors of Subjective Well-Being

This study aimed to investigate the predictive power of aspects of the romantic relationship on subjective well-being, beyond what is explained by sociodemographic and personality variables. Participants were 490 heterosexual adults (68.8% women), all involved in a monogamous romantic relationship. Romantic relationship variables were substantial predictors of the three components of subjective well-being, explaining 21% of the variance in life satisfaction, 19% of the variance in positive affect, and 15% of the variance in negative affect, in addition to sociodemographic variables and personality factors. Still, relationship satisfaction was one of the main predictors of subjective well-being. The results highlight the importance of romantic relationships over subjective well-being, suggesting that cultivating satisfying romantic relationships contributes to a happier life.

(r = -.24) or with negative affect (r = -.23). Extroversion and agreeableness were the factors that showed the highest correlation with positive affect (r = .20;r = .17, respectively;DeNeve & Cooper, 1998). Longitudinal studies revealed that the main predictor of the components of subjective well-being is neuroticism; however, extroversion, agreeableness, and conscientiousness were important predictors of the three components of SWB (Soto, 2015).

Romantic Relationship Variables and SWB
Romantic relationship variables have also been associated with SWB, especially relationship status. For example, studies have found that married people have higher levels of subjective well-being than other marital statuses (e.g., Argyle, 1999;Diener, Gohm, Suh, & Oishi, 2000;Myers, 1999). However, these findings should be carefully analyzed. Because of hedonic adaptation, life satisfaction after marriage tends to return to premarital levels (Stutzer & Frey, 2006; see Diener, Lucas, & Oishi, 2018). Still, the higher SWB levels of married people seem to stem partly from the fact that people with high levels of subjective well-being most often marry and stay in their relationships. People with low levels of subjective well-being remain single or divorce (Lucas, 2007;Stutzer & Frey, 2006) Faced with the frequency of consensual unions (without marriage) in western cultures, researchers have investigated differences in well-being between people married and those in a consensual union. Diener et al. (2000) found that married people have higher levels of subjective well-being than cohabiting people. However, this association between relationship status and well-being appears to be controlled by other variables. For example, after controlling for the effect of participant selection biases (characteristics that are associated with the propensity to marry, such as parental divorce and parental financial income) and relationship satisfaction, Perelli-Harris, Hoherz, Lappegård, and Evans (2019) found no differences between married people and those in consensual union (cohabitants).
The length of the romantic relationship can also be a factor that impacts subjective well-being. However, few studies investigated the impact of relationship length and SWB, and there is no consensus on the relationship between these variables. While Ventegodt, Flensborg-Madsen, Andersen, and Merrick (2008) found a weak but significant association between relationship length and quality of life, Rhoades, Kamp Dush, Atkins, Stanley, and Markman (2011) reported that relationship duration was not a significant predictor of either life satisfaction or psychological distress after the couple's separation. Also, studies that examined the association between relationship length and marital satisfaction have shown a decreasing linear relationship between these variables (for a review, see Karney, 2015).
People who are in happy relationships report high levels of subjective well-being (Kamp Dush & Amato, 2005).
A meta-analysis showed that quality of the romantic relationship is significantly associated with well-being (Proulx, Helms, & Buehler, 2007). Researchers have consistently detected an association between relationship satisfaction and SWB (e.g., Cao, Wang, & Wang, 2020;Love & Holder, 2016;Londero-Santos, Pereira Neto, & Natividade, 2017). Besides that, a longitudinal study suggests that relationship satisfaction has a causal influence on life satisfaction (e.g., Headey, Veenhoven, & Wearing, 1991). Other studies also highlight the increased importance of romantic relationships on people's well-being, as evidenced by the fact that in recent years there has been an increase in the association between relationship satisfaction and well-being indicators (see Braithwaite & Holt-Lunstad, 2017;Finkel, Cheung, Emery, Carswell, & Larson, 2015).
Londero-Santos, Natividade, & Féres-Carneiro A possible explanation for the observed relationships between relationship satisfaction and SWB may be given by theories using bottom-up processing models (see Schimmack, 2008). Bottom-up processing models suggest that subjective well-being derives from a sum of pleasant and unpleasant moments and experiences.
The satisfaction of goals theories, which are based on these processing models, for example, asserts that satisfaction with one's own life stems from the satisfaction of one's own needs, desires, and goals (see Diener et al., 1999;Diener et al., 2018b). Good romantic relationships would therefore satisfy the basic needs, desires, and goals of individuals. Diener, Scollon, and Lucas (2009) assume that if it were possible to access all domains of a person's life, it would be possible to infer that person's assessment of his/her own life (life satisfaction) using bottom-up processing. Thus, the attitude to life itself (i.e., life satisfaction) derives from attitudes towards specific aspects of life (for a review, see Schimmack, 2008). The individual, therefore, would be more satisfied with his/her own life, the more positive attitudes he/she has about specific aspects of his/her life (e.g., satisfaction with his/her romantic relationship).
Thus, the greater the accessibility and intensity of attitudes of specific aspects of life, the more these aspects will be used to assess satisfaction with one's own life (Schwarz & Strack, 1999). Indeed, Diener et al. (2009) state that the assessment of life satisfaction stems from information that is salient in the individual's memory at the time of the assessment. Diener et al. (2009) included in the definition of subjective well-being satisfaction with specific aspects of life. Therefore, satisfaction with a romantic relationship should be one of those specific domains that would contribute to assessing the overall aspect of life satisfaction.
Many variables explain variations in well-being levels, and there is no consensus on all those that best predict SWB. Besides, culture plays a critical role in SWB (see Suh & Choi, 2018). Those open questions emphasize the importance of research in different cultures to provide a complete understanding of the factors that impact well-being. Results from studies in different cultures may allow the identification of specific and universal SWB predictors. Therefore, this study was developed to investigate the predictive power of romantic relationship variables on subjective well-being beyond what is explained by sociodemographic and personality variables.

Method Participants
Participants were 490 heterosexual adults, with a mean age of 33.9 years (SD = 9.66), ranging from 18 to 69 years. Most participants were women, 68.8% (n = 337). All participants were involved in a monogamous romantic relationship. The mean relationship length was 9.45 years (SD = 8.67). Most participants were married, 45.5%; 27.1% of them lived in consensual union without marriage (with or without legal recognition), and the remaining 27.4% were in a committed relationship or engaged. Most participants, 73.9% (n = 362), reported cohabiting with their spouse or partner. Of the total participants, 33.1% reported having children.
Most participants lived in the southern (41.4%) and southeast (37.1%) regions of Brazil, the others were from the northeast (11.6%), central west (6.1%), northern (1.0%), and the others were outside Brazil when they responded the questionnaire (2.7%). Regarding education, 42% participants reported having completed graduate studies (in particular, 17.3% higher education specialization, 17.6% master degree, and 7.1% doctoral degree). Other 22.8% of participants were attending or attended (without having completed) graduate studies; 20.0% of the participants reported having just completed higher education; 14.1% said they had incomplete Romantic Relationship and Subjective Well-Being higher education, and 1.1% reported having just completed high school. The mean of monthly per capita income of the participants was R$ 4,098.17 (SD = 3,508.23), approximately US$ 1,280.00. Regarding religious practice, 36.5% declared to have no religion, 29.4% declared to be a non-practicing religious, and 34.1% declared to be religious practicing people.

Instruments
Data were collected through an online questionnaire available on the internet containing sociodemographic questions (e.g., gender, sexual orientation, education, relationship length, religiosity, family income, cohabitation of partner with participant). The questionnaire also contained scales to measure relationship satisfaction, personality and subjective well-being (positive and negative affect, and life satisfaction).

Romantic Relationship Satisfaction Level Scale -Revised (RRLS-R)
Relationship satisfaction was measured using the RRLS-R (Londero-Santos, Natividade, & Féres-Carneiro, 2021). RRLS-R is a revised version of the Rusbult, Martz, and Agnew (1998) scale. The instrument contains five affirmative items to be answered on a nine-point agreement scale, such that 0 = Strongly Disagree and 8 = Strongly Agree, the higher the scores, the greater the satisfaction with the romantic relationship. Londero-Santos et al. (2021) found evidence of validity based on internal structure, on relations to other variables, and convergent validity. Also, an item response theory analysis shown adequate parameters for the items, the items were highly discriminatory and covering a large extent of the latent construct. The scale alpha coefficient was .91 (Londero-Santos et al., 2021). In this study, the total omega reliability coefficient was .91.

Personality Descriptors Scale -Reduced (PDS-R)
PDS-R (Natividade & Hutz, 2015) assesses the big five personality factors: extroversion, agreeableness, conscientiousness, neuroticism, and openness. The scale is made up of 20 adjectives or small expressions, four for each factor; so that participants answer how much they agree that each one describes them properly, on a seven-point scale. The higher the participants' scores on each factor, the higher the intensity of the factors. In the study of Natividade and Hutz (2015), the scale presented adequate evidence of validity based on internal structure, relations to other variables, and convergent validity. Also, alpha coefficients ranged from .59 to .84 among the factors, and test-retest correlations ranged from .69 to .81 (Natividade & Hutz, 2015). In this study, the total omega reliability coefficient ranged from .65 to .87 among the factors.

Positive and Negative Affect Scale (PANAS)
PANAS  is formed by two orthogonal factors, positive affect and negative affect, and it is an adaptation of the scale originally proposed by Watson and Clark (1994). Each factor consists of 10 adjectives that represent the subjects' emotions and moods. The intensity of affects is judged by the participants on a five-point scale. In this study, the total omega reliability coefficient was .90 for Negative Affect and .88 for Positive Affect.

Life Satisfaction Scale
Life Satisfaction Scale ) -an adaptation of the satisfaction with life scale, originally proposed by Diener, Emmons, Larsen, and Griffin (1985). This scale consists of five items that globally assess the cognitive aspects of subjective well-being. Items are answered on a seven-point Likert scale. The scale presented adequate internal consistency, with an alpha coefficient of .87, in addition to presenting scalar Londero-Santos, Natividade, & Féres-Carneiro invariance across gender (Zanon, Bardagi, Layous, & Hutz, 2014). In this study, the total omega reliability coefficient was .88.
The questionnaire had different versions so that words, when necessary, were flexed according to the gender of the participant and his/her respective partner, in order to avoid bias in the answer (Natividade, Barros, & Hutz, 2012). That is important because in the Portuguese language some adjectives are flexed according to gender, and that could influence responses from men and women. Also, along with the scale items, there were control questions to check the fidelity of the items' answers.

Procedures Data Collection
Participants were recruited through email invitations and social networking sites. The invitation emails were sent to a list of individuals who had previously agreed to participate in our laboratory research. Concerning social networks, invitations were posted on the researchers' websites, research authors' laboratory websites, and social scientific research-related-websites. Those who accepted to participate had to click on the address link provided in the invitation and were directed to the questionnaire. This research was sent to the Ethics Committee on Human Research and obtained a favorable agreement.

Data Analysis
Data cleaning was performed, excluding incorrect answers to control questions. Then, data scatter plots were constructed in order to check the distribution of variables visually. As they had substantial negative asymmetry, the variables per capita income and relationship length were transformed into logarithms. The dichotomous variables gender, having children, relationship status (consensual union, married, cohabitation) were transformed into dummy variables of 0 (being male; having no children; no living in a consensual union, no married, no cohabitation with a romantic partner) and 1 (being female; having children; living in a consensual union, married, cohabitation with a romantic partner). The religious practice variable was coded as 0 = no religion, 1 = religious non-practicing, and 2 = religious practicing. Subsequently, the correlation coefficients between the variables were calculated. Finally, we tested the predictive power of romantic relationship-related variables (relationship length, relationship status, and relationship satisfaction) on subjective well-being by controlling the effects of sociodemographic variables (gender, age, per capita income, and having children) and the big five factors through multiple hierarchical regressions (Enter method). Only the variables that showed a statistically significant correlation with the predicted variable were included in the regression models. Preliminary analyses were conducted to ensure no violation of the assumptions of normality, linearity, multicollinearity, and homoscedasticity.

Results
Initially, we calculated Pearson's correlation coefficients, as can be seen in Table 1. In the regression analyses, we included those variables that significantly correlated with the factors of subjective well-being. Thereby, we investigated the predictive power of variables concerning the romantic relationship on subjective well-being, after controlling for the influence of sociodemographic and personality variables. To this end, three hierarchical regression analyses were run, one for each component of subjective well-being, which included sociodemo-Romantic Relationship and Subjective Well-Being graphic variables in the first block, personality factors in the second block, and romantic relationship variables in the third block. Results can be seen in Table 2, Table 3 and Table 4.     conscientiousness, openness, and relationship satisfaction were statistically significant (see Table 2). Relationship satisfaction (β = .46) explained the most unique variance in life satisfaction, followed by neuroticism (β = -.21).

Negative Affect
Religious practice, at Step 1, significantly accounted for 10.1% of the variance in positive affect, F(1, 488) = 5.02, p = .026. When entered into Step 2, agreeableness, neuroticism, and conscientiousness resulted in a significant increase in the variance of positive affect, R 2 change = .22, F change = (3, 485) = 45.97, p < .001, and the entire model remained significant, F(4, 485) = 36.08, p < .001. At Step 2, the religious practice remained a significant predictor of negative affect. For the personality, only neuroticism was significant predictors of negative affect (see Table 4). After entry of consensual union and relationship satisfaction at Step 3 the total variance explained by the model as a whole was 38%, F(6, 483) = 49.25, p < .001. The relationship variables explained an additional 15% of the variance in positive affect, R 2 change = .15, F change (2, 483) = 58.48, p < .001.
In the final model, religious practice, neuroticism, and relationship satisfaction were statistically significant.

Discussion
This study aimed to investigate the predictive power of variables concerning romantic relationship, such as length of the relationship, relationship status (consensual union, married, cohabitation with a partner), and satisfaction with the romantic relationship, on subjective well-being, beyond what is explained by sociodemographic (gender, age, per capita income, religious practice and having children) and personality (extroversion, agreeableness, neuroticism, conscientiousness, openness to experience) variables. The results revealed that the romantic relationship variables are important predictors of SWB, adding explanation beyond the sociodemographic and personality variables.
However, neither relationship length nor relationship status (consensual union, marriage, cohabitation) impacted SWB, although previous studies have shown an association between relationship status (e.g., married) and life satisfaction (e.g., Lucas, 2007;Luhmann, Hofmann, Eid, & Lucas, 2012). Among the variables of the romantic relationship, the only significant predictor of subjective well-being was relationship satisfaction.
Comparing the standardized weights of all predictors SWB included in this study, relationship satisfaction had This study's findings of the impact of relationship satisfaction on SWB are consistent with other researches (e.g., Hudson, Lucas, & Donnellan, 2020;Londero-Santos et al., 2017;Margelisch, Schneewind, Violette, & Romantic Relationship and Subjective Well-Being Perrig-Chiello, 2017). Although we have not found other studies that have investigated the predictive power of relationship satisfaction on SWB, after controlling for the effect of sociodemographic and personality variables, the present study revealed that relationship satisfaction is a crucial variable to predict subjective well-being.
These results suggest that to evaluate their own life, individuals rely on assessments of specific aspects of life, in this case, the romantic relationship. This provides additional evidence for theories based on bottom-up processing models, which assume that life situations and events influence subjective well-being (see Diener et al., 1999;Diener et al., 2009;Diener et al., 2018b;Schimmack, 2008). Maintaining a positive attitude towards the romantic relationship (satisfaction with the romantic relationship) therefore seems to positively impact the attitude towards one's own life (life satisfaction). The results suggest that attitudes towards romantic relationships are taken into account when forming attitudes toward life itself. Nevertheless, it should be emphasized that this is a cross-sectional study, making it impossible to identify the causal relationship of these variables.
Still, it is possible that the relationship between SWB and relationship satisfaction may have occurred due to the moderation/mediation of other variables that were not considered in this study or that were both determined by other variables. Longitudinal studies may be conducted to investigate the variation in SWB and relationship satisfaction over time.
In addition to relationship satisfaction, sociodemographic and personality variables also predict subjective well-being. The results suggest that sociodemographic variables predict life satisfaction, positive affect, and negative affect from the tested models. Per capita income was a positive predictor of life satisfaction. This result is in line with previous studies (e.g., Diener et al., 2013;Stevenson & Wolfers, 2013). One possible explanation for the positive effects of per capita income on life satisfaction is that economic resources make it possible for people to improve their lives (see Diener et al., 2018b).
Religious practice was also a positive predictor of life satisfaction, positive affect, and negative affect. The more involved with religion, the higher the subjective well-being levels of the participants. This result supports other studies that found a positive relationship between religiosity and subjective well-being (e.g., Fleury et al., 2018; for a review, see Diener et al., 1999;Kim-Prieto & Miller, 2018). Religious people tend to have a greater sense of life and better-coping strategies, and religious practice can be an important source of social support, i.e., being involved in religious practice, especially if done in a group context, would increase social contact, and the availability of social support received (for a review, see Kim-Prieto & Miller, 2018).
Regarding personality traits, life satisfaction was predicted negatively by neuroticism and positively by conscientiousness and openness. Positive affect was predicted positively by extroversion, conscientiousness, and openness, and negatively by neuroticism. Finally, negative affect was predicted positively by neuroticism. Among personality factors, neuroticism was the main explanatory variable for life satisfaction and negative affect, consistent with that found in the meta-analysis of DeNeve and Cooper (1998). Thus, having high levels of neuroticism seems to predispose the individual to poorly evaluate his own life and experience more negative affections (the negative correlations between neuroticism and relationship satisfaction also exemplify this).
The results of this study also showed the importance of the factors conscientiousness and openness in the prediction of life satisfaction and positive affect. These two factors are rarely studied as predictors of SWB (Lucas, 2018), and the results are not consistent (e.g., Anglim & Grant, 2016;González Gutiérrez et al., 2005;Malkoç, 2011;Noronha et al., 2016;Natividade et al., 2019). The divergent results can be explained by the different set of variables used in the regression analyzes and the different instruments used to assess the five personality factors (each instrument may evaluate different aspects of personality factors). Therefore, more studies should be carried out to investigate the impact of personality on the SWB in Brazilian samples.
In addition, in this research, only linear relationships between the study variables were investigated. Variables that did not show a statistically significant correlation with the predicted variable were not included in the regression analysis. These variables may have a non-linear relationship with the variables of SWB. Studies have found a positive quadratic relationship between age and well-being (for a review, see Lansford, 2018).
Therefore, studies should be conducted to investigate the non-linear relationship of variables with subjective well-being.
Given the above, our results showed important predictors of SWB, especially personality factors and relationship satisfaction. It can be concluded that, from the results of this study, being satisfied with one's own romantic relationship proved beneficial for the subjective well-being of the individual. Still, SWB seems to reflect the assessment of important aspects of life, such as romantic relationships. Therefore, it is important to cultivate healthy and happy romantic relationships to live a happier and more fulfilling life. These findings may support public policy and clinical practice, as they suggest that helping people improve their romantic relationships can positively impact subjective well-being.

Funding
The authors have no funding to report.