- Email: [email protected]
Individual differences in the relationship between domain satisfaction and happiness: The moderating role of domain importance
Personality and Individual Differences 86 (2015) 82–87
Contents lists available at ScienceDirect
Personality and Individual Differences journal homepage: www.elsevier.com/locate/paid
Individual differences in the relationship between domain satisfaction and happiness: The moderating role of domain importance Tim Tiefenbach a,⁎, Florian Kohlbacher b a b
German Institute for Japanese Studies (DIJ), Jochi Kioizaka Bldg. 2F, 7-1 Kioicho, Chiyoda-ku, Tokyo 102-0094, Japan International Business School Suzhou (IBSS), Xi'an Jiaotong-Liverpool University (XJTLU), China
a r t i c l e
i n f o
Article history: Received 27 February 2015 Received in revised form 29 May 2015 Accepted 31 May 2015 Available online xxxx Keywords: Happiness Domain satisfaction Domain importance Interaction effects Value-as-a-moderator hypothesis
a b s t r a c t Previous studies have shown that personality traits account for a substantial amount of variance in individual levels of subjective well-being (SWB). However, these studies are limited in their ability to explain the intraand interindividual differences in the processes of SWB. To redress this shortcoming, researchers have focused on moderators of the relationship between domain satisfaction and global life satisfaction. However, those studies assume only one specific type of interaction pattern for all life domains. Based on a national probability sample from Japan this paper analyzes the role of domain importance in the relationship between domain satisfaction and the overall SWB level. Our study is the first to explore different kinds of interaction patterns in the importance satisfaction moderation of life domains. We identify four different types of domains: (i) domains in which satisfaction correlates with happiness only when the domain is considered as important; (ii) domains in which satisfaction correlates with happiness no matter whether it is considered as important or not; (iii) domains in which the slope of the correlation between satisfaction and happiness increases when it is considered as important and (iv) domains which show no correlation with happiness not matter whether it is considered as important or not. © 2015 Elsevier B.V. All rights reserved.
1. Introduction Over the last four decades happiness studies have become an established field and research on subjective well-being (SWB) now features regularly and prominently in psychology and personality journals (see e.g. Bixter, 2015; Diener, Suh, Lucas, & Smith, 1999; Oishi, Graham, Kesebir, & Galinha, 2013; Weiss, Bates, & Luciano, 2008). Previous studies on SWB have revealed that personality traits account for a substantial amount of variance in individual levels of SWB. This is especially the case for extraversion and neuroticism (Lucas, Diener, Grob, Suh, & Shao, 2000; Schimmack, Oishi, Furr, & Funder, 2004; Weber & Huebner, 2015), but also other personality related variables such as self-esteem (Lucas, Diener, & Suh, 1996), optimism (Bailey, Eng, Frisch, & Snyder, 2007), and positive illusions (Erez, Johnson, & Judge, 1995) show a strong correlation with measures of global life satisfaction. While those variables can explain variance in SWB to a good extent, these studies are limited in their ability to explain the underlying processes of SWB, specifically the intra- and interindividual differences in the processes of SWB (Oishi, Diener, Suh, & Lucas, 1999). In an effort to redress this shortcoming, researchers have zoomed in on the relationship between domain satisfaction and global life satisfaction as it ⁎ Corresponding author. E-mail addresses: [email protected] (T. Tiefenbach), fl[email protected] (F. Kohlbacher).
http://dx.doi.org/10.1016/j.paid.2015.05.040 0191-8869/© 2015 Elsevier B.V. All rights reserved.
cannot be completely explained by traditional personality measures (Schimmack, 2008; Schimmack, Diener, & Oishi, 2002). Oishi et al. (1999) formulated the ‘value-as-a-moderator’ hypothesis and showed that values moderate the relationship between domain satisfaction and global life satisfaction. Another stream of research looks at the role of domain importance in the relationship between domain satisfaction and overall life satisfaction (see Hsieh, 2012b for a review of the literature). Hsieh (2012b) offers preliminary evidence to the effect that domain importance can moderate the relationship between domain satisfaction and global life satisfaction. Viewing importance as a value orientation, this finding can be interpreted as a variant of Oishi et al.'s ‘value-as-a-moderator’ hypothesis – which we will call ‘domain-importance-as-a-moderator’ hypothesis in this paper – even though Hsieh does not explicitly refer to Oishi et al. or use such a terminology. However, all of these approaches work with the underlying assumption that there are interaction effects for all life domains analyzed. I.e. technically, they assume both the coefficient of the interaction term as well as the coefficient of the conditional main effect of satisfaction to be significant and positive. However, previous studies in happiness research have shown that people do not always know what makes them happy. The best proof for this is the misprediction of utility (Frey & Stutzer, 2014; Wilson & Gilbert, 2005). Popular examples are the over-consumption of watching TV (Frey, Benesch, & Stutzer, 2007) and the commuting paradox
T. Tiefenbach, F. Kohlbacher / Personality and Individual Differences 86 (2015) 82–87
(Stutzer & Frey, 2008). Applied to the context of domain satisfaction these findings imply that there is not necessarily an interaction effect between value/importance and satisfaction, at least not in all domains. Conversely, it is also possible that domain satisfaction only predicts global happiness if that domain is considered as important, which would lead to a significant interaction effect with the conditional main effect of satisfaction being not-significant. Preliminary evidence for those theoretical considerations is provided by Yoon and Lee (2010). Applying the ‘value-as-a-moderator’ model to the context of social connectedness they find in one model only the interaction effect, and in another model only the conditional main effect to be significant. Unfortunately, they fail to discuss the implications of their findings for the ‘value-as-a-moderator’ model. In a similar vein, Lent et al. (2005) did not find support for the ‘domain-importance-as-a-moderator’ hypothesis. Moreover, all empirical studies mentioned above also suffer from technical and sample-based shortcomings. The ‘value-as-a-moderator’ hypothesis (Oishi et al., 1999) e.g. has only been tested in an underspecified model setting focusing only on a limited number of domains.1 A validation of the hypothesis in a well specified prediction model of global happiness or life satisfaction, encompassing all major life domains, is still missing. Hsieh (2012b) merely says that importancesatisfaction interactions can increase the variances explained (ΔR2) in a global life satisfaction model, but fails to discuss the specific patterns of interaction which lead to this increase. This paper investigates the validity of the ‘domain-importance-as-amoderator’ hypothesis in the context of a fully specified happiness prediction model. As a guiding hypothesis we assume that both the interaction effects of domain importance-satisfaction as well as the conditional main effect of domain satisfaction are significant predictors of happiness (see Fig. 1 for our conceptual model). However, we expect that this hypothesis can only be confirmed in certain domains. Due to a lack of theory and empirical evidence we are not able to formulate specific hypothesis about each domain, making our research exploratory in nature. We contribute to knowledge in two ways: By showing that the ‘domain-importance-as-moderator’ hypothesis does not hold for all life domains we overcome the simplistic assumption and misspecifications of previous studies and reveal different types of relationships depending on the domain in question. Finally, by using a national probability sample we also overcome shortcomings of previous studies related to non-representative samples of small size.
2. Methods, data and strategy The special nature of our research question required us to find reliable survey data that not only contains measures of happiness and domain satisfaction, but also measures of domain importance. While the former can be found in several large scale surveys (e.g. British Household Panel Study, German Socio-Economic Panel), the latter is not included in any major panel study (see for example the Cross National Equivalent File, CBEF). We identified the Japanese National Survey on Lifestyle Preferences (NSLP; in Japanese: kokumin seikatsu senkōdo chōsa) of the year 2010 as – to our knowledge – the only government commissioned, publicly available dataset that contains measures of domain importance and domain satisfaction.2 The population of the survey includes 4000 men and women in Japan between 15 and 80 years of age and the sample is generated via a two-stage randomized stratified procedure. Due to the relatively high response rate (72.5%), there are 2900 completed questionnaires available for analysis. 1 Oishi et al. (1999) only consider three randomly chosen domains (“grades”,“family”, and “social life and friends”), whereas Yoon and Lee (2010) only analyze the domain of social connectedness. 2 For more information on the NSLP see http://www5.cao.go.jp/seikatsu/senkoudo/ senkoudo.html. For a happiness analysis of the NSLP 2011 data see Tiefenbach and Kohlbacher (2015).
83
Fig. 1. Conceptual model.
2.1. Measures 2.1.1. Happiness Our explained variable is the current happiness level of the respondent. The corresponding survey item reads: “How happy are you currently?”. Answer options range from 0 to 10 on an 11-point scale. Single item happiness scales are commonly used in happiness research (e.g. Bixter, 2015; Dolan, Peasgood, & White, 2008) and have been shown to be valid and reliable (Abdel-Khalek, 2006). 2.1.2. Domain importance The NSLP 2010 captures domain importance as a dummy variable, asking people “When you evaluated your happiness feeling which of the following items did you consider important? Please check all relevant items”. It then lists nine pre-coded items: ‘financial situation’, ‘employment situation’, ‘health condition’, ‘spending free and satisfying time’, ‘purpose in life’ (regarding work, hobbies and social contribution), ‘family relations’, ‘friendship relations’, ‘relations at the workplace’ and ‘relations to the regional community’. Note that it is necessary to emphasize the precise wording of the question which explicitly asks respondents to evaluate the domains in terms of their importance regarding the overall feeling of happiness. Other surveys, such as the World Values Survey, only ask respondents to “indicate how important it [the respective domain] is in your life”. With this kind of question, the link between importance and overall life satisfaction/happiness is left ambiguous, since some domains might be important in life, but are not related to life satisfaction itself. This is why the Japanese NSLP lends itself so suitable for analyzing the relationship between satisfaction and importance. 2.1.3. Domain satisfaction Finally, respondents were asked to rank their satisfaction with several aspects of their life. The exact wording of the questions reads “How satisfied are you with each of the following items? Please indicate on a scale from ‘satisfied’ to ‘dissatisfied’ the state which comes closest to your personal feelings”. The respondents can then choose on a 1–5 Likert scale their level of satisfaction. The exact scale reads: “satisfied” recoded as “5”, “somewhat satisfied” recoded as “4”, “neither satisfied nor dissatisfied” recoded as “3”, “If anything, dissatisfied” recoded as “2” and “dissatisfied” recoded as “1”. Note that all satisfaction measures have been mean-centered. For this analysis we focus on all the domains listed under the importance ranking question, except for ‘time’, for which no equivalent satisfaction measure was included in the survey. 2.1.4. Control variables Finally, we include a number of control variables that are common in happiness estimations. We control for basic socio-demographic variables (household income, age, age squared, gender) and family relations
84
T. Tiefenbach, F. Kohlbacher / Personality and Individual Differences 86 (2015) 82–87
(cohabitation with spouse, number of children). Descriptive statistics of our main variables are presented in Table A1 in Appendix A. 2.2. Sample and variable selection Studies on domain satisfaction usually rely on a number of major life domains that have been identified in the literature (e.g. Hsieh, 2012a; Loewe, Bagherzadeh, Araya-Castillo, Thieme, & Batista-Foguet, 2014) A well-established index that breaks down global life satisfaction into major life domains is the Personal Well-being Index (PWI) (Cummins, 1996). All available domains in the NSLP 2010 fall into one of the core domains of the PWI so that our data capture the major life domains relevant to human happiness.3 In order to address the problem of domain-absentees (Rojas, 2004) – i.e. respondents for which a domain does not apply – we followed van Praag, Frijters, and Ferrer-i-Carbonell (2003) and chose to analyze working and non-working respondents separately. This allowed us to drop the domains ‘employment situation’ and ‘relations at workplace’ for the non-working sub-sample. After deleting all missing values 1646 observations in the working sample and 864 observations in the non-working sample remained. 2.3. Empirical strategy To explore the moderating role of domain importance we apply a bottom-up approach (Schimmack, 2008) predicting happiness based on domain satisfaction, domain importance and the respective interaction terms. Although there is no final agreement on the direction of causality, many studies assume a bottom-up relationship running from domain satisfaction to global life satisfaction (e.g. Easterlin, 2006; Loewe et al., 2014; van Praag et al., 2003). Reviewing several studies on the structure of well-being Schimmack finds only ‘modest evidence’ for top-down models, while acknowledging that most of the findings “are consistent with bottom-up theories” (Schimmack, 2008, p. 108). Following Oishi et al. (1999) we apply hierarchical regression to our data which allows us to test whether the inclusion of satisfactionimportance interaction terms increases the variance explained of our prediction model. In a first step we regress global happiness on only the control variables (block 1). In the next steps we included measures of domain satisfactions (block 2), and the importance ratings (block 3). In a final step we included the interaction terms of domain satisfaction and domain importance (block 4). Note that we centered domain satisfactions around the respective means first, and then formed the interaction terms by multiplying the centered values with the binary importance variables.
Table 1 Hierarchical regression: Happiness scores predicted by satisfaction, importance and the satisfaction/importance interaction term of all life domains and controls. Variables
Model 1: working
Model 2: non-working
Coef
SE
t
Coef
SE
t
Block 1: controls Household Income Age Age-squared Cohabitation with spouse Female Number of children
0.01 0.02 −0.00 0.25 0.13 0.07
0.00 0.02 0.00 0.10 0.08 0.04
3.16⁎⁎ 1.07 −1.69 2.40⁎
0.00 −0.04 0.00 0.44 0.43 0.11
0.00 0.02 0.00 0.16 0.12 0.06
1.13 −1.90 1.35 2.70⁎⁎ 3.64⁎⁎⁎
Block 2: satisfaction Financial satisfaction Health satisfaction Purpose in life satisfaction Family satisfaction Friends satisfaction Region satisfaction Job satisfaction Workplace satisfaction
0.07 0.13 0.31 0.26 0.04 −0.01 0.07 0.03
0.08 0.07 0.07 0.09 0.07 0.07 0.06 0.06
0.91 1.83 4.39⁎⁎⁎ 3.00⁎⁎ 0.52 −0.21 1.08 0.48
0.10 0.09 0.21 0.43 0.06 0.02
0.08 0.09 0.08 0.14 0.09 0.08
Block 3: importance Financial importance Health importance Purpose in life importance Family importance Friends importance Region importance Job importance Workplace importance
−0.40 0.20 0.07 0.48 0.04 0.01 −0.01 −0.11
0.08 0.09 0.08 0.09 0.09 0.14 0.08 0.09
−4.75⁎⁎⁎ 2.31⁎ 0.95 5.38⁎⁎⁎
−0.16 −0.18 0.15 0.59 0.27 0.23
0.11 0.12 0.11 0.13 0.12 0.16
−1.44 −1.52 1.29 4.42⁎⁎⁎ 2.32⁎
Block 4: interaction Finance interaction Health interaction Purpose in life interaction Family interaction Friends interaction Region interaction Job interaction Workplace interaction Observations
0.42 0.09 −0.00 0.45 0.02 0.17 0.04 0.18 1646
0.09 0.09 0.09 0.10 0.11 0.15 0.08 0.09
0.37 0.21 0.02 0.15 0.15 0.18
0.10 0.11 0.12 0.15 0.13 0.23
3.67⁎⁎⁎ 1.89 0.20 1.00 1.19 0.82
2
Model statistics
R
Block 1: controls Block 2: satisfaction Block 3: importance Block 4: interaction
0.11 0.46 0.49 0.51
1.72 1.97⁎
0.44 0.09 −0.07 −1.28 4.76⁎⁎⁎ 1.02 −0.02 4.56⁎⁎⁎ 0.16 1.14 0.51 2.02⁎
1.84
1.22 0.96 2.54⁎ 3.17⁎⁎ 0.65 0.27
1.44
864 ΔR
2
0.35 0.03 0.03
F
R2
36.13⁎⁎⁎ 115.38⁎⁎⁎ 10.86⁎⁎⁎ 8.31⁎⁎⁎
0.14 0.42 0.45 0.47
ΔR2
F
0.27 0.04 0.02
22.17⁎⁎⁎ 62.81⁎⁎⁎ 9.15⁎⁎⁎ 4.67⁎⁎⁎
All models are calculated with robust standard errors. ⁎⁎⁎ p b 0.001. ⁎⁎ p b 0.01 ⁎ p b 0.05.
3. Results Results of our hierarchical regression models are reported in Table 1. Each stepwise inclusion of the four blocks of variables leads to a statistically significant increase in the variance explained. As predicted by the ‘domain-importance-as-a-moderator’ model, including the interaction terms in the regression significantly explained variance of life satisfaction, above and beyond the main effects of domain satisfactions and domain importance (ΔR2 = 0.03 in the working and ΔR2 = 0.02 nonworking samples). The fully specified models show a high goodness of fit by explaining almost 51% and 47% of the variance in individual happiness levels among workers and non-workers, respectively. Looking at the block of domain satisfaction variables we find that both in the working and non-working sub-samples only the domains of ‘purpose in life’ and ‘family’ have a significant conditional main effect. 3 The core domains of the PWI are as follows: 1. ‘standard of living’; 2. ‘health’; 3. ‘achieving in life’, 4; ‘personal relationships’, 5; ‘community’. Their respective representation in the NSLP 2010 are as follows: 1. ‘financial situation’ and ‘employment situation’; 2. ‘health condition’; 3. ‘purpose in life’; 4. ‘family relations’, ‘friendship relations’ and ‘workplace relations’; and 5. ‘relations to the regional community’.
In the block of interaction effects only the three domains ‘finance’, ‘family’ and ‘workplace’ show a significant coefficient in the working sample, whereas in the non-working sample only ‘finance’ has a significant importance-satisfaction interaction term. Taken together we find very different patterns of interactions. The coefficient of the conditional main effect of financial satisfaction is not significant, while its interaction term is significant. This means that finance shows a positive correlation with happiness only for respondents who consider it as important. In the case of family relations both the interaction term and the conditional main effect of family satisfaction have positive coefficients — at least in the working sample. This indicates that respondents who do not consider family as important still show a positive correlation between family satisfaction and happiness. However, respondents who think family relations as important show a significantly higher correlation between family satisfaction and happiness. Finally, purpose in life shows no significant interaction effect between satisfaction and importance. The conditional main effect, however, is significant, indicating that satisfaction with ‘purpose in life’ is correlated with happiness no matter whether it is considered as
T. Tiefenbach, F. Kohlbacher / Personality and Individual Differences 86 (2015) 82–87 Table 2 A systematic classification of patterns of interactions in different life domains. Conditional main effect of domain satisfaction Satisfaction–importance interaction effect
Significant
Not significant
Significant Not significant
Familya Purpose in life, familya
Finance, workplace Friends, region, job, health
a
NB.: Family shows different patterns in the working and non-working sample.
important or not. Finally, the domains ‘friends’, ‘region’, ‘job’ and ‘health’ do not have a significant interaction term and also their conditional main effect of domain satisfaction is not significant. Table 2 summarizes our findings. Note that the conditional main effects of importance ratings also correlate with happiness. As Block 3 in Table 1 shows, respondents in the working sample who consider family and health as important are happier, whereas respondents who consider finance as important are significantly less happy. In the non-working sample we find positive correlations for respondents who consider family and friends as important to their personal feelings of happiness. As domain satisfaction values are mean-centered the effects of domain importance are represented at the mean for each respective domain. 4. Discussion This paper examined the validity of the ‘domain-importance-as-amoderator’ hypothesis in the context of a fully specified happiness prediction model. As expected – but contrary to Oishi et al. and Hsieh – we find that moderation does not occur in all domains. Indeed, the original interaction pattern assumed by Oishi et al. (1999) holds only true for the domain of family relations in the working sample. Here both the conditional main effect of satisfaction with family as well as the interaction effect of importance of family relations and family satisfaction are significant. All other domains analyzed showed different patterns of interaction which have not yet been explained in the literature. Our results regarding the ‘purpose or meaning in life’ domain stand in stark contrast to findings by Steger, Oishi, and Kesebir (2011). Steger et al. show that the relationship between ‘presence of meaning in life’ and SWB is moderated by the ‘search for meaning in life’. Considering the ‘presence of meaning in life’ as a proxy for satisfaction and the ‘search for meaning in life’ as a proxy for importance, their findings can be interpreted in favor of the ‘domain-importance-as-a moderator’ hypothesis, since both the conditional main effect of satisfaction as well as its interaction term with importance are positively correlated with SWB. Our findings, however, suggest that in the case of purpose in life, only the conditional main effect, but not the interaction effect is a unique predictor of happiness. Yoon and Lee (2010) report slightly different results regarding the domain of social connectedness. In their life satisfaction model they find only the conditional main effect of social connectedness, but not its interaction effect with importance to be significant. In our model social connectedness is represented by family and friends relationships. Whereas the latter does not seem to be a unique predictor of happiness, we find that the interaction patterns for family relations differ depending on whether the respondents are working or non-working, a variable that Yoon and Lee fail to control for. The differing results compared to Oishi et al., Steger et al. and Yoon and Lee are likely due to differences in the methodological approaches. Unlike previous studies we do not analyze single domains, but rather rely on a fully specified happiness prediction model, encompassing all major life domains. Further, whereas previous studies used nonrepresentative samples (mostly students) of small size (n b 150) our data is based on a large-scale national probability sample. Finally, including variables for household income and marital status, our study uses more control variables than previous studies, lending additional
85
confidence to the findings. Thus, our results provide more general evidence that domain importance can moderate the relationship between domain satisfaction and global happiness, but that there are different patterns of interaction. Further, we show that the rather simplistic assumption of the ‘value-as-a-moderator hypothesis’, that the conditional main effect and the interaction with importance are both significant predictors of global life satisfaction, does not hold in most cases. Up to now, the interaction between domain importance and satisfaction has only been studied in the context of increasing the model fit of happiness prediction models (Hsieh, 2012b). Our study confirms the preliminary findings of Hsieh by showing that adding interaction effects increases the explained variance in happiness, over and above the main effects of domain satisfactions and domain importance. However, our study extends well beyond the findings of Hsieh (2012b) by revealing the different patterns of interaction which underlie this increase in explained variance. Due to its unique pattern of interaction the domain of purpose in life is of special interest for further research. Given that many individuals do not consider purpose in life as important, although it is significantly correlated with individual happiness, we expect to see people mispredicting their future utility in various decisions that affect their purpose in life. Overall, our study shows that misprediction of utility is not only a phenomenon of individual actions, but also occurs on the meta-level of life domains. In a similar vein, our finding that individuals who consider finance as important are systematically less happy, is in line with studies reporting that people tend to overvalue extrinsic desires such as income and status (Frey & Stutzer, 2014). Finally, misprediction of utility can also be observed in the domain of health. Although importance of health is in both samples considered as the most important domain (see Table A1), neither satisfaction with health nor its interaction effect with importance of health show a significant correlation with happiness. This result is in line with studies which find that individuals perceive health more important than it actually is Hsieh (2008). 5. Limitations and outlook There are a number of limitations to our study which should be mentioned. Most limitations mainly revolve around the measurement instruments used in the NSLP survey. As described above, not only our outcome variable, but also all of our main variables are single item measures. Future studies should re-examine the importance satisfaction interaction by using more sophisticated psycho-metric approaches that can take measurement error into account and ideally also account for response biases. Finally, our happiness and satisfaction variables only tap state affects, and therefore do not pick-up trait affects. Although better measurement instruments and methods are in any case desirable, the findings of our study have to be evaluated in light of existing studies of which many focus on state effects measured by similar instruments (e.g. Easterlin, 2006; Hsieh, 2012b; van Praag et al., 2003). Further, single item measures of domain importance have been validated in widely used large sample surveys, such as the World Value Survey. Our study shows that the interindividual differences in the processes of SWB are more complex than previous studies assumed. Building on the different patterns of interaction between domain satisfaction and domain importance and their relation to global happiness, it is up to future research to link those results to traditional measures of personality traits. Acknowledgment We thank Kristopher Preacher, Andrew Hayes and David Kenny for their insightful comments on probing multiple unrelated interaction effects in different model settings. Of course, all errors and omissions are our own. An earlier version of this paper was presented at the third World Congress on Positive Psychology 2013, Los Angeles and at the International Conference on Happiness in October 2014 at EHESS, Paris.
86
T. Tiefenbach, F. Kohlbacher / Personality and Individual Differences 86 (2015) 82–87
Appendix A
Table A1 Descriptive statistics of main variables. Variable
Working Obs.
Non-Working Mean
Std. Dev.
Obs.
Mean
Min
Max
1.10 1.10 0.87 0.87 0.82 0.68
1 1 1 1 1 1
5 5 5 5 5 5
0.62 0.71 0.31 0.69 0.45 0.12
0.49 0.46 0.46 0.46 0.50 0.33
0 0 0 0 0 0
1 1 1 1 1 1
864 864 864 864 864 864
1.91 2.45 1.09 2.82 1.88 0.44
1.74 1.84 1.71 2.02 2.14 1.19
0 0 0 0 0 0
5 5 5 5 5 5
864 864 864 864 864 864
36.27 50.08 2958.93 0.66 0.67 1.27
24.27 21.25 1988.25 0.48 0.47 1.07
Satisfaction Financial satisfaction Health satisfaction Purpose in life satisfaction Family satisfaction Friends satisfaction Region satisfaction Job satisfaction Workplace satisfaction
1646 1646 1646 1646 1646 1646 1646 1646
2.90 3.61 3.41 4.05 3.99 3.23 3.40 3.60
1.12 0.96 0.86 0.90 0.79 0.70 1.07 0.93
864 864 864 864 864 864
3.08 3.45 3.29 4.00 3.92 3.20
Importance Financial importance Health importance Purpose in life importance Family importance Friends importance Region importance Job importance Workplace importance
1646 1646 1646 1646 1646 1646 1646 1646
0.69 0.70 0.37 0.69 0.34 0.09 0.55 0.24
0.46 0.46 0.48 0.46 0.47 0.29 0.50 0.43
864 864 864 864 864 864
Interaction Finance interaction Health interaction Purpose in life interaction Family interaction Friends interaction Region interaction Job interaction Workplace interaction
1646 1646 1646 1646 1646 1646 1646 1646
2.00 2.57 1.32 2.85 1.43 0.34 1.59 0.87
1.65 1.87 1.80 2.06 2.05 1.08 1.69 1.69
Controls Household income Age Age-squared Married Female Number of children
1646 1646 1646 1646 1646 1646
44.11 46.83 2385.21 0.74 0.44 1.47
25.15 13.88 1330.78 0.44 0.50 1.17
References Abdel-Khalek, A. M. (2006). Measuring happiness with a single-item scale. Social Behavior and Personality, 34(2), 139–150. Bailey, T. C., Eng, W., Frisch, M. B., & Snyder, C. R. (2007). Hope and optimism as related to life satisfaction. The Journal of Positive Psychology, 2(3), 168–175. http://dx.doi.org/10. 1080/17439760701409546. Bixter, M. T. (2015). Happiness, political orientation, and religiosity. Personality and Individual Differences, 72, 7–11. http://dx.doi.org/10.1016/j.paid.2014.08.010. Cummins, R. A. (1996). The domains of life satisfaction: An attempt to order chaos. Social Indicators Research, 38(3), 303–328. http://dx.doi.org/10.1007/BF00292050. Diener, E., Suh, E. M., Lucas, R. E., & Smith, H. L. (1999). Subjective well-being: Three decades of progress. Psychological Bulletin, 125(2), 276–302. Dolan, P., Peasgood, T., & White, M. (2008). Do we really know what makes us happy? A review of the economic literature on the factors associated with subjective wellbeing. Journal of Economic Psychology, 29(1), 94–122. http://dx.doi.org/10.1016/j. joep.2007.09.001. Easterlin, R. A. (2006). Life cycle happiness and its sources. Journal of Economic Psychology, 27(4), 463–482. http://dx.doi.org/10.1016/j.joep.2006.05.002. Erez, A., Johnson, D. E., & Judge, T. A. (1995). Self-deception as a mediator of the relationship between dispositions and subjective well-being. Personality and Individual Differences, 19(5), 597–612. http://dx.doi.org/10.1016/0191-8869(95)00088-N. Frey, B. S., Benesch, C., & Stutzer, A. (2007). Does watching TV make us happy? Journal of Economic Psychology, 28(3), 283–313. http://dx.doi.org/10.1016/j.joep.2007.02.001. Frey, B. S., & Stutzer, A. (2014). Economic consequences of mispredicting utility. Journal of Happiness Studies, 15(4), 937–956. http://dx.doi.org/10.1007/s10902-013-9457-4. Hsieh, C. -M. (2008). The relative importance of health. Social Indicators Research, 87(1), 127–137. http://dx.doi.org/10.1007/s11205-007-9162-y. Hsieh, C. -M. (2012a). Importance is not unimportant: The role of importance weighting in QOL measures. Social Indicators Research, 109(2), 267–278. http://dx.doi.org/10. 1007/s11205-011-9900-z. Hsieh, C. -M. (2012b). Should we give up domain importance weighting in QoL measures? Social Indicators Research, 108(1), 99–109. http://dx.doi.org/10.1007/ s11205-011-9868-8.
Std. Dev.
4.167 15 225 0 0 0
100 80 6400 1 1 9
Lent, R. W., Singley, D., Sheu, H. -B., Gainor, K. A., Brenner, B. R., Treistman, D., et al. (2005). Social cognitive predictors of domain and life satisfaction: Exploring the theoretical precursors of subjective well-being. Journal of Counseling Psychology, 52(3), 429–442. http://dx.doi.org/10.1037/0022-0167.52.3.429. Loewe, N., Bagherzadeh, M., Araya-Castillo, L., Thieme, C., & Batista-Foguet, J. M. (2014). Life domain satisfactions as predictors of overall life satisfaction among workers: evidence from Chile. Social Indicators Research, 118(1), 71–86. http://dx.doi.org/10. 1007/s11205-013-0408-6. Lucas, R. E., Diener, E., Grob, A., Suh, E. M., & Shao, L. (2000). Cross-cultural evidence for the fundamental features of extraversion. Journal of Personality and Social Psychology, 79(3), 452–468. http://dx.doi.org/10.1037//0022-3514.79.3.452. Lucas, R. E., Diener, E., & Suh, E. (1996). Discriminant validity of well-being measures. Journal of Personality and Social Psychology, 71(3), 616–628. http://dx.doi.org/10. 1037/0022-3514.71.3.616. Oishi, S., Diener, E., Suh, E., & Lucas, R. E. (1999). Value as a moderator in subjective wellbeing. Journal of Personality, 67(1), 157–184. http://dx.doi.org/10.1111/1467-6494. 00051. Oishi, S., Graham, J., Kesebir, S., & Galinha, I. C. (2013). Concepts of happiness across time and cultures. Personality and Social Psychology Bulletin, 39(5), 559–577. http://dx.doi. org/10.1177/0146167213480042. Rojas, M. (2004). The complexity of well-being: A life satisfaction conception and a domainof-life approach. Paper for the International Workshop on Researching Well-being in Developing Countries (Retrieved from http://unpan1.un.org/intradoc/groups/public/ documents/ICAP/UNPAN029029.pdf). Schimmack, U. (2008). The structure of well-being. In M. Eid, & R. J. Larsen (Eds.), The science of subjective well-being (pp. 97–123). New York: Guilford Press. Schimmack, U., Diener, E., & Oishi, S. (2002). Life-satisfaction is a momentary judgment and a stable personality characteristic: The use of chronically accessible and stable sources. Journal of Personality, 70(3), 345–384. http://dx.doi.org/10.1111/1467-6494.05008. Schimmack, U., Oishi, S., Furr, R. M., & Funder, D. C. (2004). Personality and life satisfaction: A facet-level analysis. Personality and Social Psychology Bulletin, 30(8), 1062–1075. http://dx.doi.org/10.1177/0146167204264292. Steger, M. F., Oishi, S., & Kesebir, S. (2011). Is a life without meaning satisfying? The moderating role of the search for meaning in satisfaction with life judgments. The Journal
T. Tiefenbach, F. Kohlbacher / Personality and Individual Differences 86 (2015) 82–87 of Positive Psychology, 6(3), 173–180. http://dx.doi.org/10.1080/17439760.2011. 569171. Tiefenbach, T., & Kohlbacher, F. (2015). Happiness in Japan in Times of Upheaval: Empirical Evidence from the National Survey on Lifestyle Preferences. Journal of Happiness Studies, 16(2), 333–366. http://dx.doi.org/10.1007/s10902-014-9512-9. Stutzer, A., & Frey, B. S. (2008). Stress that doesn't pay: The commuting paradox. Scandinavian Journal of Economics, 110(2), 339–366. http://dx.doi.org/10.1111/j. 1467-9442.2008.00542.x. van Praag, B., Frijters, P., & Ferrer-i-Carbonell, A. (2003). The anatomy of subjective wellbeing. Journal of Economic Behavior & Organization, 51(1), 29–49. http://dx.doi.org/10. 1016/S0167-2681(02)00140-3.
87
Weber, M., & Huebner, E. S. (2015). Early adolescents' personality and life satisfaction: A closer look at global vs. domain-specific satisfaction. Personality and Individual Differences, 83, 31–36. http://dx.doi.org/10.1016/j.paid.2015.03.042. Weiss, A., Bates, T. C., & Luciano, M. (2008). Happiness is a personal(ity) thing: The genetics of personality and well-being in a representative sample. Psychological Science, 19(3), 205–210. http://dx.doi.org/10.1111/j.1467-9280.2008.02068.x. Wilson, T. D., & Gilbert, D. T. (2005). Affective forecasting knowing what to want. Current Directions in Psychological Science, 14, 131–134. Yoon, E., & Lee, R. M. (2010). Importance of social connectedness as a moderator in Korean immigrants' subjective well-being. Asian American Journal of Psychology, 1(2), 93–105. http://dx.doi.org/10.1037/a0019964.
Contents lists available at ScienceDirect
Personality and Individual Differences journal homepage: www.elsevier.com/locate/paid
Individual differences in the relationship between domain satisfaction and happiness: The moderating role of domain importance Tim Tiefenbach a,⁎, Florian Kohlbacher b a b
German Institute for Japanese Studies (DIJ), Jochi Kioizaka Bldg. 2F, 7-1 Kioicho, Chiyoda-ku, Tokyo 102-0094, Japan International Business School Suzhou (IBSS), Xi'an Jiaotong-Liverpool University (XJTLU), China
a r t i c l e
i n f o
Article history: Received 27 February 2015 Received in revised form 29 May 2015 Accepted 31 May 2015 Available online xxxx Keywords: Happiness Domain satisfaction Domain importance Interaction effects Value-as-a-moderator hypothesis
a b s t r a c t Previous studies have shown that personality traits account for a substantial amount of variance in individual levels of subjective well-being (SWB). However, these studies are limited in their ability to explain the intraand interindividual differences in the processes of SWB. To redress this shortcoming, researchers have focused on moderators of the relationship between domain satisfaction and global life satisfaction. However, those studies assume only one specific type of interaction pattern for all life domains. Based on a national probability sample from Japan this paper analyzes the role of domain importance in the relationship between domain satisfaction and the overall SWB level. Our study is the first to explore different kinds of interaction patterns in the importance satisfaction moderation of life domains. We identify four different types of domains: (i) domains in which satisfaction correlates with happiness only when the domain is considered as important; (ii) domains in which satisfaction correlates with happiness no matter whether it is considered as important or not; (iii) domains in which the slope of the correlation between satisfaction and happiness increases when it is considered as important and (iv) domains which show no correlation with happiness not matter whether it is considered as important or not. © 2015 Elsevier B.V. All rights reserved.
1. Introduction Over the last four decades happiness studies have become an established field and research on subjective well-being (SWB) now features regularly and prominently in psychology and personality journals (see e.g. Bixter, 2015; Diener, Suh, Lucas, & Smith, 1999; Oishi, Graham, Kesebir, & Galinha, 2013; Weiss, Bates, & Luciano, 2008). Previous studies on SWB have revealed that personality traits account for a substantial amount of variance in individual levels of SWB. This is especially the case for extraversion and neuroticism (Lucas, Diener, Grob, Suh, & Shao, 2000; Schimmack, Oishi, Furr, & Funder, 2004; Weber & Huebner, 2015), but also other personality related variables such as self-esteem (Lucas, Diener, & Suh, 1996), optimism (Bailey, Eng, Frisch, & Snyder, 2007), and positive illusions (Erez, Johnson, & Judge, 1995) show a strong correlation with measures of global life satisfaction. While those variables can explain variance in SWB to a good extent, these studies are limited in their ability to explain the underlying processes of SWB, specifically the intra- and interindividual differences in the processes of SWB (Oishi, Diener, Suh, & Lucas, 1999). In an effort to redress this shortcoming, researchers have zoomed in on the relationship between domain satisfaction and global life satisfaction as it ⁎ Corresponding author. E-mail addresses: [email protected] (T. Tiefenbach), fl[email protected] (F. Kohlbacher).
http://dx.doi.org/10.1016/j.paid.2015.05.040 0191-8869/© 2015 Elsevier B.V. All rights reserved.
cannot be completely explained by traditional personality measures (Schimmack, 2008; Schimmack, Diener, & Oishi, 2002). Oishi et al. (1999) formulated the ‘value-as-a-moderator’ hypothesis and showed that values moderate the relationship between domain satisfaction and global life satisfaction. Another stream of research looks at the role of domain importance in the relationship between domain satisfaction and overall life satisfaction (see Hsieh, 2012b for a review of the literature). Hsieh (2012b) offers preliminary evidence to the effect that domain importance can moderate the relationship between domain satisfaction and global life satisfaction. Viewing importance as a value orientation, this finding can be interpreted as a variant of Oishi et al.'s ‘value-as-a-moderator’ hypothesis – which we will call ‘domain-importance-as-a-moderator’ hypothesis in this paper – even though Hsieh does not explicitly refer to Oishi et al. or use such a terminology. However, all of these approaches work with the underlying assumption that there are interaction effects for all life domains analyzed. I.e. technically, they assume both the coefficient of the interaction term as well as the coefficient of the conditional main effect of satisfaction to be significant and positive. However, previous studies in happiness research have shown that people do not always know what makes them happy. The best proof for this is the misprediction of utility (Frey & Stutzer, 2014; Wilson & Gilbert, 2005). Popular examples are the over-consumption of watching TV (Frey, Benesch, & Stutzer, 2007) and the commuting paradox
T. Tiefenbach, F. Kohlbacher / Personality and Individual Differences 86 (2015) 82–87
(Stutzer & Frey, 2008). Applied to the context of domain satisfaction these findings imply that there is not necessarily an interaction effect between value/importance and satisfaction, at least not in all domains. Conversely, it is also possible that domain satisfaction only predicts global happiness if that domain is considered as important, which would lead to a significant interaction effect with the conditional main effect of satisfaction being not-significant. Preliminary evidence for those theoretical considerations is provided by Yoon and Lee (2010). Applying the ‘value-as-a-moderator’ model to the context of social connectedness they find in one model only the interaction effect, and in another model only the conditional main effect to be significant. Unfortunately, they fail to discuss the implications of their findings for the ‘value-as-a-moderator’ model. In a similar vein, Lent et al. (2005) did not find support for the ‘domain-importance-as-a-moderator’ hypothesis. Moreover, all empirical studies mentioned above also suffer from technical and sample-based shortcomings. The ‘value-as-a-moderator’ hypothesis (Oishi et al., 1999) e.g. has only been tested in an underspecified model setting focusing only on a limited number of domains.1 A validation of the hypothesis in a well specified prediction model of global happiness or life satisfaction, encompassing all major life domains, is still missing. Hsieh (2012b) merely says that importancesatisfaction interactions can increase the variances explained (ΔR2) in a global life satisfaction model, but fails to discuss the specific patterns of interaction which lead to this increase. This paper investigates the validity of the ‘domain-importance-as-amoderator’ hypothesis in the context of a fully specified happiness prediction model. As a guiding hypothesis we assume that both the interaction effects of domain importance-satisfaction as well as the conditional main effect of domain satisfaction are significant predictors of happiness (see Fig. 1 for our conceptual model). However, we expect that this hypothesis can only be confirmed in certain domains. Due to a lack of theory and empirical evidence we are not able to formulate specific hypothesis about each domain, making our research exploratory in nature. We contribute to knowledge in two ways: By showing that the ‘domain-importance-as-moderator’ hypothesis does not hold for all life domains we overcome the simplistic assumption and misspecifications of previous studies and reveal different types of relationships depending on the domain in question. Finally, by using a national probability sample we also overcome shortcomings of previous studies related to non-representative samples of small size.
2. Methods, data and strategy The special nature of our research question required us to find reliable survey data that not only contains measures of happiness and domain satisfaction, but also measures of domain importance. While the former can be found in several large scale surveys (e.g. British Household Panel Study, German Socio-Economic Panel), the latter is not included in any major panel study (see for example the Cross National Equivalent File, CBEF). We identified the Japanese National Survey on Lifestyle Preferences (NSLP; in Japanese: kokumin seikatsu senkōdo chōsa) of the year 2010 as – to our knowledge – the only government commissioned, publicly available dataset that contains measures of domain importance and domain satisfaction.2 The population of the survey includes 4000 men and women in Japan between 15 and 80 years of age and the sample is generated via a two-stage randomized stratified procedure. Due to the relatively high response rate (72.5%), there are 2900 completed questionnaires available for analysis. 1 Oishi et al. (1999) only consider three randomly chosen domains (“grades”,“family”, and “social life and friends”), whereas Yoon and Lee (2010) only analyze the domain of social connectedness. 2 For more information on the NSLP see http://www5.cao.go.jp/seikatsu/senkoudo/ senkoudo.html. For a happiness analysis of the NSLP 2011 data see Tiefenbach and Kohlbacher (2015).
83
Fig. 1. Conceptual model.
2.1. Measures 2.1.1. Happiness Our explained variable is the current happiness level of the respondent. The corresponding survey item reads: “How happy are you currently?”. Answer options range from 0 to 10 on an 11-point scale. Single item happiness scales are commonly used in happiness research (e.g. Bixter, 2015; Dolan, Peasgood, & White, 2008) and have been shown to be valid and reliable (Abdel-Khalek, 2006). 2.1.2. Domain importance The NSLP 2010 captures domain importance as a dummy variable, asking people “When you evaluated your happiness feeling which of the following items did you consider important? Please check all relevant items”. It then lists nine pre-coded items: ‘financial situation’, ‘employment situation’, ‘health condition’, ‘spending free and satisfying time’, ‘purpose in life’ (regarding work, hobbies and social contribution), ‘family relations’, ‘friendship relations’, ‘relations at the workplace’ and ‘relations to the regional community’. Note that it is necessary to emphasize the precise wording of the question which explicitly asks respondents to evaluate the domains in terms of their importance regarding the overall feeling of happiness. Other surveys, such as the World Values Survey, only ask respondents to “indicate how important it [the respective domain] is in your life”. With this kind of question, the link between importance and overall life satisfaction/happiness is left ambiguous, since some domains might be important in life, but are not related to life satisfaction itself. This is why the Japanese NSLP lends itself so suitable for analyzing the relationship between satisfaction and importance. 2.1.3. Domain satisfaction Finally, respondents were asked to rank their satisfaction with several aspects of their life. The exact wording of the questions reads “How satisfied are you with each of the following items? Please indicate on a scale from ‘satisfied’ to ‘dissatisfied’ the state which comes closest to your personal feelings”. The respondents can then choose on a 1–5 Likert scale their level of satisfaction. The exact scale reads: “satisfied” recoded as “5”, “somewhat satisfied” recoded as “4”, “neither satisfied nor dissatisfied” recoded as “3”, “If anything, dissatisfied” recoded as “2” and “dissatisfied” recoded as “1”. Note that all satisfaction measures have been mean-centered. For this analysis we focus on all the domains listed under the importance ranking question, except for ‘time’, for which no equivalent satisfaction measure was included in the survey. 2.1.4. Control variables Finally, we include a number of control variables that are common in happiness estimations. We control for basic socio-demographic variables (household income, age, age squared, gender) and family relations
84
T. Tiefenbach, F. Kohlbacher / Personality and Individual Differences 86 (2015) 82–87
(cohabitation with spouse, number of children). Descriptive statistics of our main variables are presented in Table A1 in Appendix A. 2.2. Sample and variable selection Studies on domain satisfaction usually rely on a number of major life domains that have been identified in the literature (e.g. Hsieh, 2012a; Loewe, Bagherzadeh, Araya-Castillo, Thieme, & Batista-Foguet, 2014) A well-established index that breaks down global life satisfaction into major life domains is the Personal Well-being Index (PWI) (Cummins, 1996). All available domains in the NSLP 2010 fall into one of the core domains of the PWI so that our data capture the major life domains relevant to human happiness.3 In order to address the problem of domain-absentees (Rojas, 2004) – i.e. respondents for which a domain does not apply – we followed van Praag, Frijters, and Ferrer-i-Carbonell (2003) and chose to analyze working and non-working respondents separately. This allowed us to drop the domains ‘employment situation’ and ‘relations at workplace’ for the non-working sub-sample. After deleting all missing values 1646 observations in the working sample and 864 observations in the non-working sample remained. 2.3. Empirical strategy To explore the moderating role of domain importance we apply a bottom-up approach (Schimmack, 2008) predicting happiness based on domain satisfaction, domain importance and the respective interaction terms. Although there is no final agreement on the direction of causality, many studies assume a bottom-up relationship running from domain satisfaction to global life satisfaction (e.g. Easterlin, 2006; Loewe et al., 2014; van Praag et al., 2003). Reviewing several studies on the structure of well-being Schimmack finds only ‘modest evidence’ for top-down models, while acknowledging that most of the findings “are consistent with bottom-up theories” (Schimmack, 2008, p. 108). Following Oishi et al. (1999) we apply hierarchical regression to our data which allows us to test whether the inclusion of satisfactionimportance interaction terms increases the variance explained of our prediction model. In a first step we regress global happiness on only the control variables (block 1). In the next steps we included measures of domain satisfactions (block 2), and the importance ratings (block 3). In a final step we included the interaction terms of domain satisfaction and domain importance (block 4). Note that we centered domain satisfactions around the respective means first, and then formed the interaction terms by multiplying the centered values with the binary importance variables.
Table 1 Hierarchical regression: Happiness scores predicted by satisfaction, importance and the satisfaction/importance interaction term of all life domains and controls. Variables
Model 1: working
Model 2: non-working
Coef
SE
t
Coef
SE
t
Block 1: controls Household Income Age Age-squared Cohabitation with spouse Female Number of children
0.01 0.02 −0.00 0.25 0.13 0.07
0.00 0.02 0.00 0.10 0.08 0.04
3.16⁎⁎ 1.07 −1.69 2.40⁎
0.00 −0.04 0.00 0.44 0.43 0.11
0.00 0.02 0.00 0.16 0.12 0.06
1.13 −1.90 1.35 2.70⁎⁎ 3.64⁎⁎⁎
Block 2: satisfaction Financial satisfaction Health satisfaction Purpose in life satisfaction Family satisfaction Friends satisfaction Region satisfaction Job satisfaction Workplace satisfaction
0.07 0.13 0.31 0.26 0.04 −0.01 0.07 0.03
0.08 0.07 0.07 0.09 0.07 0.07 0.06 0.06
0.91 1.83 4.39⁎⁎⁎ 3.00⁎⁎ 0.52 −0.21 1.08 0.48
0.10 0.09 0.21 0.43 0.06 0.02
0.08 0.09 0.08 0.14 0.09 0.08
Block 3: importance Financial importance Health importance Purpose in life importance Family importance Friends importance Region importance Job importance Workplace importance
−0.40 0.20 0.07 0.48 0.04 0.01 −0.01 −0.11
0.08 0.09 0.08 0.09 0.09 0.14 0.08 0.09
−4.75⁎⁎⁎ 2.31⁎ 0.95 5.38⁎⁎⁎
−0.16 −0.18 0.15 0.59 0.27 0.23
0.11 0.12 0.11 0.13 0.12 0.16
−1.44 −1.52 1.29 4.42⁎⁎⁎ 2.32⁎
Block 4: interaction Finance interaction Health interaction Purpose in life interaction Family interaction Friends interaction Region interaction Job interaction Workplace interaction Observations
0.42 0.09 −0.00 0.45 0.02 0.17 0.04 0.18 1646
0.09 0.09 0.09 0.10 0.11 0.15 0.08 0.09
0.37 0.21 0.02 0.15 0.15 0.18
0.10 0.11 0.12 0.15 0.13 0.23
3.67⁎⁎⁎ 1.89 0.20 1.00 1.19 0.82
2
Model statistics
R
Block 1: controls Block 2: satisfaction Block 3: importance Block 4: interaction
0.11 0.46 0.49 0.51
1.72 1.97⁎
0.44 0.09 −0.07 −1.28 4.76⁎⁎⁎ 1.02 −0.02 4.56⁎⁎⁎ 0.16 1.14 0.51 2.02⁎
1.84
1.22 0.96 2.54⁎ 3.17⁎⁎ 0.65 0.27
1.44
864 ΔR
2
0.35 0.03 0.03
F
R2
36.13⁎⁎⁎ 115.38⁎⁎⁎ 10.86⁎⁎⁎ 8.31⁎⁎⁎
0.14 0.42 0.45 0.47
ΔR2
F
0.27 0.04 0.02
22.17⁎⁎⁎ 62.81⁎⁎⁎ 9.15⁎⁎⁎ 4.67⁎⁎⁎
All models are calculated with robust standard errors. ⁎⁎⁎ p b 0.001. ⁎⁎ p b 0.01 ⁎ p b 0.05.
3. Results Results of our hierarchical regression models are reported in Table 1. Each stepwise inclusion of the four blocks of variables leads to a statistically significant increase in the variance explained. As predicted by the ‘domain-importance-as-a-moderator’ model, including the interaction terms in the regression significantly explained variance of life satisfaction, above and beyond the main effects of domain satisfactions and domain importance (ΔR2 = 0.03 in the working and ΔR2 = 0.02 nonworking samples). The fully specified models show a high goodness of fit by explaining almost 51% and 47% of the variance in individual happiness levels among workers and non-workers, respectively. Looking at the block of domain satisfaction variables we find that both in the working and non-working sub-samples only the domains of ‘purpose in life’ and ‘family’ have a significant conditional main effect. 3 The core domains of the PWI are as follows: 1. ‘standard of living’; 2. ‘health’; 3. ‘achieving in life’, 4; ‘personal relationships’, 5; ‘community’. Their respective representation in the NSLP 2010 are as follows: 1. ‘financial situation’ and ‘employment situation’; 2. ‘health condition’; 3. ‘purpose in life’; 4. ‘family relations’, ‘friendship relations’ and ‘workplace relations’; and 5. ‘relations to the regional community’.
In the block of interaction effects only the three domains ‘finance’, ‘family’ and ‘workplace’ show a significant coefficient in the working sample, whereas in the non-working sample only ‘finance’ has a significant importance-satisfaction interaction term. Taken together we find very different patterns of interactions. The coefficient of the conditional main effect of financial satisfaction is not significant, while its interaction term is significant. This means that finance shows a positive correlation with happiness only for respondents who consider it as important. In the case of family relations both the interaction term and the conditional main effect of family satisfaction have positive coefficients — at least in the working sample. This indicates that respondents who do not consider family as important still show a positive correlation between family satisfaction and happiness. However, respondents who think family relations as important show a significantly higher correlation between family satisfaction and happiness. Finally, purpose in life shows no significant interaction effect between satisfaction and importance. The conditional main effect, however, is significant, indicating that satisfaction with ‘purpose in life’ is correlated with happiness no matter whether it is considered as
T. Tiefenbach, F. Kohlbacher / Personality and Individual Differences 86 (2015) 82–87 Table 2 A systematic classification of patterns of interactions in different life domains. Conditional main effect of domain satisfaction Satisfaction–importance interaction effect
Significant
Not significant
Significant Not significant
Familya Purpose in life, familya
Finance, workplace Friends, region, job, health
a
NB.: Family shows different patterns in the working and non-working sample.
important or not. Finally, the domains ‘friends’, ‘region’, ‘job’ and ‘health’ do not have a significant interaction term and also their conditional main effect of domain satisfaction is not significant. Table 2 summarizes our findings. Note that the conditional main effects of importance ratings also correlate with happiness. As Block 3 in Table 1 shows, respondents in the working sample who consider family and health as important are happier, whereas respondents who consider finance as important are significantly less happy. In the non-working sample we find positive correlations for respondents who consider family and friends as important to their personal feelings of happiness. As domain satisfaction values are mean-centered the effects of domain importance are represented at the mean for each respective domain. 4. Discussion This paper examined the validity of the ‘domain-importance-as-amoderator’ hypothesis in the context of a fully specified happiness prediction model. As expected – but contrary to Oishi et al. and Hsieh – we find that moderation does not occur in all domains. Indeed, the original interaction pattern assumed by Oishi et al. (1999) holds only true for the domain of family relations in the working sample. Here both the conditional main effect of satisfaction with family as well as the interaction effect of importance of family relations and family satisfaction are significant. All other domains analyzed showed different patterns of interaction which have not yet been explained in the literature. Our results regarding the ‘purpose or meaning in life’ domain stand in stark contrast to findings by Steger, Oishi, and Kesebir (2011). Steger et al. show that the relationship between ‘presence of meaning in life’ and SWB is moderated by the ‘search for meaning in life’. Considering the ‘presence of meaning in life’ as a proxy for satisfaction and the ‘search for meaning in life’ as a proxy for importance, their findings can be interpreted in favor of the ‘domain-importance-as-a moderator’ hypothesis, since both the conditional main effect of satisfaction as well as its interaction term with importance are positively correlated with SWB. Our findings, however, suggest that in the case of purpose in life, only the conditional main effect, but not the interaction effect is a unique predictor of happiness. Yoon and Lee (2010) report slightly different results regarding the domain of social connectedness. In their life satisfaction model they find only the conditional main effect of social connectedness, but not its interaction effect with importance to be significant. In our model social connectedness is represented by family and friends relationships. Whereas the latter does not seem to be a unique predictor of happiness, we find that the interaction patterns for family relations differ depending on whether the respondents are working or non-working, a variable that Yoon and Lee fail to control for. The differing results compared to Oishi et al., Steger et al. and Yoon and Lee are likely due to differences in the methodological approaches. Unlike previous studies we do not analyze single domains, but rather rely on a fully specified happiness prediction model, encompassing all major life domains. Further, whereas previous studies used nonrepresentative samples (mostly students) of small size (n b 150) our data is based on a large-scale national probability sample. Finally, including variables for household income and marital status, our study uses more control variables than previous studies, lending additional
85
confidence to the findings. Thus, our results provide more general evidence that domain importance can moderate the relationship between domain satisfaction and global happiness, but that there are different patterns of interaction. Further, we show that the rather simplistic assumption of the ‘value-as-a-moderator hypothesis’, that the conditional main effect and the interaction with importance are both significant predictors of global life satisfaction, does not hold in most cases. Up to now, the interaction between domain importance and satisfaction has only been studied in the context of increasing the model fit of happiness prediction models (Hsieh, 2012b). Our study confirms the preliminary findings of Hsieh by showing that adding interaction effects increases the explained variance in happiness, over and above the main effects of domain satisfactions and domain importance. However, our study extends well beyond the findings of Hsieh (2012b) by revealing the different patterns of interaction which underlie this increase in explained variance. Due to its unique pattern of interaction the domain of purpose in life is of special interest for further research. Given that many individuals do not consider purpose in life as important, although it is significantly correlated with individual happiness, we expect to see people mispredicting their future utility in various decisions that affect their purpose in life. Overall, our study shows that misprediction of utility is not only a phenomenon of individual actions, but also occurs on the meta-level of life domains. In a similar vein, our finding that individuals who consider finance as important are systematically less happy, is in line with studies reporting that people tend to overvalue extrinsic desires such as income and status (Frey & Stutzer, 2014). Finally, misprediction of utility can also be observed in the domain of health. Although importance of health is in both samples considered as the most important domain (see Table A1), neither satisfaction with health nor its interaction effect with importance of health show a significant correlation with happiness. This result is in line with studies which find that individuals perceive health more important than it actually is Hsieh (2008). 5. Limitations and outlook There are a number of limitations to our study which should be mentioned. Most limitations mainly revolve around the measurement instruments used in the NSLP survey. As described above, not only our outcome variable, but also all of our main variables are single item measures. Future studies should re-examine the importance satisfaction interaction by using more sophisticated psycho-metric approaches that can take measurement error into account and ideally also account for response biases. Finally, our happiness and satisfaction variables only tap state affects, and therefore do not pick-up trait affects. Although better measurement instruments and methods are in any case desirable, the findings of our study have to be evaluated in light of existing studies of which many focus on state effects measured by similar instruments (e.g. Easterlin, 2006; Hsieh, 2012b; van Praag et al., 2003). Further, single item measures of domain importance have been validated in widely used large sample surveys, such as the World Value Survey. Our study shows that the interindividual differences in the processes of SWB are more complex than previous studies assumed. Building on the different patterns of interaction between domain satisfaction and domain importance and their relation to global happiness, it is up to future research to link those results to traditional measures of personality traits. Acknowledgment We thank Kristopher Preacher, Andrew Hayes and David Kenny for their insightful comments on probing multiple unrelated interaction effects in different model settings. Of course, all errors and omissions are our own. An earlier version of this paper was presented at the third World Congress on Positive Psychology 2013, Los Angeles and at the International Conference on Happiness in October 2014 at EHESS, Paris.
86
T. Tiefenbach, F. Kohlbacher / Personality and Individual Differences 86 (2015) 82–87
Appendix A
Table A1 Descriptive statistics of main variables. Variable
Working Obs.
Non-Working Mean
Std. Dev.
Obs.
Mean
Min
Max
1.10 1.10 0.87 0.87 0.82 0.68
1 1 1 1 1 1
5 5 5 5 5 5
0.62 0.71 0.31 0.69 0.45 0.12
0.49 0.46 0.46 0.46 0.50 0.33
0 0 0 0 0 0
1 1 1 1 1 1
864 864 864 864 864 864
1.91 2.45 1.09 2.82 1.88 0.44
1.74 1.84 1.71 2.02 2.14 1.19
0 0 0 0 0 0
5 5 5 5 5 5
864 864 864 864 864 864
36.27 50.08 2958.93 0.66 0.67 1.27
24.27 21.25 1988.25 0.48 0.47 1.07
Satisfaction Financial satisfaction Health satisfaction Purpose in life satisfaction Family satisfaction Friends satisfaction Region satisfaction Job satisfaction Workplace satisfaction
1646 1646 1646 1646 1646 1646 1646 1646
2.90 3.61 3.41 4.05 3.99 3.23 3.40 3.60
1.12 0.96 0.86 0.90 0.79 0.70 1.07 0.93
864 864 864 864 864 864
3.08 3.45 3.29 4.00 3.92 3.20
Importance Financial importance Health importance Purpose in life importance Family importance Friends importance Region importance Job importance Workplace importance
1646 1646 1646 1646 1646 1646 1646 1646
0.69 0.70 0.37 0.69 0.34 0.09 0.55 0.24
0.46 0.46 0.48 0.46 0.47 0.29 0.50 0.43
864 864 864 864 864 864
Interaction Finance interaction Health interaction Purpose in life interaction Family interaction Friends interaction Region interaction Job interaction Workplace interaction
1646 1646 1646 1646 1646 1646 1646 1646
2.00 2.57 1.32 2.85 1.43 0.34 1.59 0.87
1.65 1.87 1.80 2.06 2.05 1.08 1.69 1.69
Controls Household income Age Age-squared Married Female Number of children
1646 1646 1646 1646 1646 1646
44.11 46.83 2385.21 0.74 0.44 1.47
25.15 13.88 1330.78 0.44 0.50 1.17
References Abdel-Khalek, A. M. (2006). Measuring happiness with a single-item scale. Social Behavior and Personality, 34(2), 139–150. Bailey, T. C., Eng, W., Frisch, M. B., & Snyder, C. R. (2007). Hope and optimism as related to life satisfaction. The Journal of Positive Psychology, 2(3), 168–175. http://dx.doi.org/10. 1080/17439760701409546. Bixter, M. T. (2015). Happiness, political orientation, and religiosity. Personality and Individual Differences, 72, 7–11. http://dx.doi.org/10.1016/j.paid.2014.08.010. Cummins, R. A. (1996). The domains of life satisfaction: An attempt to order chaos. Social Indicators Research, 38(3), 303–328. http://dx.doi.org/10.1007/BF00292050. Diener, E., Suh, E. M., Lucas, R. E., & Smith, H. L. (1999). Subjective well-being: Three decades of progress. Psychological Bulletin, 125(2), 276–302. Dolan, P., Peasgood, T., & White, M. (2008). Do we really know what makes us happy? A review of the economic literature on the factors associated with subjective wellbeing. Journal of Economic Psychology, 29(1), 94–122. http://dx.doi.org/10.1016/j. joep.2007.09.001. Easterlin, R. A. (2006). Life cycle happiness and its sources. Journal of Economic Psychology, 27(4), 463–482. http://dx.doi.org/10.1016/j.joep.2006.05.002. Erez, A., Johnson, D. E., & Judge, T. A. (1995). Self-deception as a mediator of the relationship between dispositions and subjective well-being. Personality and Individual Differences, 19(5), 597–612. http://dx.doi.org/10.1016/0191-8869(95)00088-N. Frey, B. S., Benesch, C., & Stutzer, A. (2007). Does watching TV make us happy? Journal of Economic Psychology, 28(3), 283–313. http://dx.doi.org/10.1016/j.joep.2007.02.001. Frey, B. S., & Stutzer, A. (2014). Economic consequences of mispredicting utility. Journal of Happiness Studies, 15(4), 937–956. http://dx.doi.org/10.1007/s10902-013-9457-4. Hsieh, C. -M. (2008). The relative importance of health. Social Indicators Research, 87(1), 127–137. http://dx.doi.org/10.1007/s11205-007-9162-y. Hsieh, C. -M. (2012a). Importance is not unimportant: The role of importance weighting in QOL measures. Social Indicators Research, 109(2), 267–278. http://dx.doi.org/10. 1007/s11205-011-9900-z. Hsieh, C. -M. (2012b). Should we give up domain importance weighting in QoL measures? Social Indicators Research, 108(1), 99–109. http://dx.doi.org/10.1007/ s11205-011-9868-8.
Std. Dev.
4.167 15 225 0 0 0
100 80 6400 1 1 9
Lent, R. W., Singley, D., Sheu, H. -B., Gainor, K. A., Brenner, B. R., Treistman, D., et al. (2005). Social cognitive predictors of domain and life satisfaction: Exploring the theoretical precursors of subjective well-being. Journal of Counseling Psychology, 52(3), 429–442. http://dx.doi.org/10.1037/0022-0167.52.3.429. Loewe, N., Bagherzadeh, M., Araya-Castillo, L., Thieme, C., & Batista-Foguet, J. M. (2014). Life domain satisfactions as predictors of overall life satisfaction among workers: evidence from Chile. Social Indicators Research, 118(1), 71–86. http://dx.doi.org/10. 1007/s11205-013-0408-6. Lucas, R. E., Diener, E., Grob, A., Suh, E. M., & Shao, L. (2000). Cross-cultural evidence for the fundamental features of extraversion. Journal of Personality and Social Psychology, 79(3), 452–468. http://dx.doi.org/10.1037//0022-3514.79.3.452. Lucas, R. E., Diener, E., & Suh, E. (1996). Discriminant validity of well-being measures. Journal of Personality and Social Psychology, 71(3), 616–628. http://dx.doi.org/10. 1037/0022-3514.71.3.616. Oishi, S., Diener, E., Suh, E., & Lucas, R. E. (1999). Value as a moderator in subjective wellbeing. Journal of Personality, 67(1), 157–184. http://dx.doi.org/10.1111/1467-6494. 00051. Oishi, S., Graham, J., Kesebir, S., & Galinha, I. C. (2013). Concepts of happiness across time and cultures. Personality and Social Psychology Bulletin, 39(5), 559–577. http://dx.doi. org/10.1177/0146167213480042. Rojas, M. (2004). The complexity of well-being: A life satisfaction conception and a domainof-life approach. Paper for the International Workshop on Researching Well-being in Developing Countries (Retrieved from http://unpan1.un.org/intradoc/groups/public/ documents/ICAP/UNPAN029029.pdf). Schimmack, U. (2008). The structure of well-being. In M. Eid, & R. J. Larsen (Eds.), The science of subjective well-being (pp. 97–123). New York: Guilford Press. Schimmack, U., Diener, E., & Oishi, S. (2002). Life-satisfaction is a momentary judgment and a stable personality characteristic: The use of chronically accessible and stable sources. Journal of Personality, 70(3), 345–384. http://dx.doi.org/10.1111/1467-6494.05008. Schimmack, U., Oishi, S., Furr, R. M., & Funder, D. C. (2004). Personality and life satisfaction: A facet-level analysis. Personality and Social Psychology Bulletin, 30(8), 1062–1075. http://dx.doi.org/10.1177/0146167204264292. Steger, M. F., Oishi, S., & Kesebir, S. (2011). Is a life without meaning satisfying? The moderating role of the search for meaning in satisfaction with life judgments. The Journal
T. Tiefenbach, F. Kohlbacher / Personality and Individual Differences 86 (2015) 82–87 of Positive Psychology, 6(3), 173–180. http://dx.doi.org/10.1080/17439760.2011. 569171. Tiefenbach, T., & Kohlbacher, F. (2015). Happiness in Japan in Times of Upheaval: Empirical Evidence from the National Survey on Lifestyle Preferences. Journal of Happiness Studies, 16(2), 333–366. http://dx.doi.org/10.1007/s10902-014-9512-9. Stutzer, A., & Frey, B. S. (2008). Stress that doesn't pay: The commuting paradox. Scandinavian Journal of Economics, 110(2), 339–366. http://dx.doi.org/10.1111/j. 1467-9442.2008.00542.x. van Praag, B., Frijters, P., & Ferrer-i-Carbonell, A. (2003). The anatomy of subjective wellbeing. Journal of Economic Behavior & Organization, 51(1), 29–49. http://dx.doi.org/10. 1016/S0167-2681(02)00140-3.
87
Weber, M., & Huebner, E. S. (2015). Early adolescents' personality and life satisfaction: A closer look at global vs. domain-specific satisfaction. Personality and Individual Differences, 83, 31–36. http://dx.doi.org/10.1016/j.paid.2015.03.042. Weiss, A., Bates, T. C., & Luciano, M. (2008). Happiness is a personal(ity) thing: The genetics of personality and well-being in a representative sample. Psychological Science, 19(3), 205–210. http://dx.doi.org/10.1111/j.1467-9280.2008.02068.x. Wilson, T. D., & Gilbert, D. T. (2005). Affective forecasting knowing what to want. Current Directions in Psychological Science, 14, 131–134. Yoon, E., & Lee, R. M. (2010). Importance of social connectedness as a moderator in Korean immigrants' subjective well-being. Asian American Journal of Psychology, 1(2), 93–105. http://dx.doi.org/10.1037/a0019964.