The current study, was part of a larger cross-cultural investigation of the cultural factors related to happinessFootnote 1 (e.g., societal emotional environment, family happiness, and the valuation of different types of happiness). In the current paper, we focus on levels of four types of happiness and on eight dimensions of self-construal.
Participants and Countries
At the time of writing, our data set contained 13,009 participants from 50 countries: Argentina, Austria, Australia, Brazil, Bhutan, Bulgaria, Canada, Chile, China, Colombia, Croatia, Czech Republic, El Salvador, Estonia, France, Georgia, Germany, Ghana, Greece, Guatemala, Hong Kong, Hungary, Iceland, Indonesia, Iran, Ireland, Italy, Japan, Korea, Lithuania, Luxembourg, Malaysia, Mexico, the Netherlands, Nigeria, Norway, Pakistan, Poland, Portugal, Romania, Russia, Saudi Arabia, Serbia, Slovakia, Switzerland, Taiwan, Turkey, UK, Ukraine, and USA.
We aimed to recruit 200 individuals in each country. Some authors, however, collected more and others collected fewer. Overall, 59.6% of participants identified as female, 39.3% as male, 0.4% as other, and 0.7% left the question about gender blank; the mean age of participants was 25.18 years (SD = 9.51). We mainly collected samples of post-secondary students, but some authors managed to complement their student sample with a general population sample. Supplementary online material contains demographic characteristics by country, means and standard deviations of analyzed variables, as well as items and reliabilities of the scales used in this study.
Materials and Procedure
Levels of Four Types of Happiness: Own Data
We used the Satisfaction with Life Scale (personal SWLS; 5 items; Diener et al. 1985; e.g., You are satisfied with your life; α = .85 for the whole sample, in every country α > .70) to measure actual personal life satisfaction. The Interdependent Happiness Scale (personal IHS; 9 items; Hitokoto and Uchida 2015; e.g., You can do what you want without causing problems for other people; α = .87 for the whole sample, in every country α > .74) was used to measure the actual personal interdependent happiness. As in Krys and collaborators (Krys et al. 2019a, c), we also adapted both measures to assess participants’ views of their family’s happiness by changing the subject of the personal SWLS and personal IHS measures from the individual to their family (e.g., Your family is satisfied with its life for family SWLS; α = .90 for the whole sample, in every country α > .79, and As a family you can do what you want without causing problems for other people for family IHS; α = .91 for the whole sample, in every country α > .87). See supplementary online material for a full list of the original and modified SWLS and IHS items. Following Vignoles et al.’s (2016) approach, participants rated items of happiness and self-construal scales (see below) on a nine-point Likert-type scale with five labelled points: 1 (doesn’t describe me at all), 3 (describes me a little), 5 (describes me moderately), 7 (describes me very well), 9 (describes me exactly).
Self-construals as Markers of Individualistic-Collectivistic Context: Own Data
As a proxy of individualism-collectivism, we used the latest upgraded version of the Vignoles et al. (2016) self-construal scales. Originally, this scale contained seven dimensions: difference vs. similarity (6 items; e.g., You like being different from other people vs. You like being similar to other people; α = .76 for the whole sample, in each country apart from IndonesiaFootnote 2α > .58), self-containment vs. connectedness to others (6 items; e.g., You would not feel personally insulted if someone insulted a member of your family vs. If someone insults a member of your family, you feel as if you have been insulted personally; α = .72 for the whole sample, in forty-five countries2 α > .60), self-direction vs. receptiveness to influence (6 items; e.g., You usually decide on your own actions, rather than follow others’ expectations vs. You usually do what people expect of you, rather than decide for yourself what to do; α = .76 for the whole sample, in every country2 α > .55), self-reliance vs. dependence on others (6 items; e.g., You try to avoid being reliant on others vs. Being able to depend on others is very important to you; α = .78 for the whole sample, in every country2 α > .53), consistency vs. variability (6 items; e.g., You behave in a similar way at home and in public vs. You behave differently when you are with different people; α = .84 for the whole sample, in every country2 α > .66), self-expression vs. harmony (6 items; e.g., You like to discuss your own ideas, even if it might sometimes upset the people around you vs. You try to adapt to people around you, even if it means hiding your feelings; α = .76 for the whole sample, in every country2 α > .55), and self-interest vs. commitment to others (6 items; e.g., You protect your own interests, even if it might sometimes disrupt your family relationships vs. You usually give priority to others, before yourself; α = .69 for the whole sample, in every country2 α > .49). The upgraded version of the Vignoles scale included one additional dimension obtained from the first author of the scale (Vignoles, personal communication): de-contextualized vs. contextualized self (6 items; e.g., Someone could understand who you are without needing to know which social groups you belong to vs. If someone wants to understand who you are, they would need to know which social groups you belong to; α = .74 for the whole sample, in every country2 apart from Saudi Arabia α > .54). For full tables with reliabilities, please see supplementary material.
Individualism-Collectivism Meta-Factor: External Data
During data collection we learned that self-construals are not an ideal proxy of individualism-collectivism. Therefore in the analysis we also included an individualism-collectivism meta-factor extracted from individualism-collectivism measures offered by Hofstede (2001) and Minkov et al. (2017), and from autonomy-embeddedness measures that are reflecting individualism-collectivism in the value taxonomy of cultures offered by Schwartz (2008). We calculated an individualism-collectivism meta-factor by averaging standardized scores for each of these three datasets (α > .86).
Analytic Approach
We used two complementary analytic approaches: country-level correlational analysis and multilevel modelling (MLM). These two approaches employ different dependent variables; country-level happiness for correlational analysis and individual-level happiness for MLM analysis. The purpose of this study is to study country-level associations between happiness and individualism; thus, we describe correlational analysis as the main analytic tool, and MLM as an additional analytic tool.
For the predicted variables, we analyzed the four types of happiness (i.e., personal SWLS, personal IHS, family SWLS, and family IHS). For predicting variables, we used eight dimensions of self-construals, and additionally, the individualism-collectivism meta-factor. When comparing country-level correlation coefficients, we employed the test of the difference between two dependent correlations with one variable in common (Steiger 1980).
In two-level analyses, self-construals served as individual-level predictors (this way we studied the influence of individual mindset on happiness), and country-level aggregates of self-construals served as country-level predictors (this way we studied the influence of cultural context on happiness). Furthermore, we controlled for cross-level interactions between individual mindset and cultural context, and for gender, age and social capital of participants (i.e., education of their parents) at the individual-level of analyses, and for GDP per capita at the country-level of analyses.
We employed in our analyses eight different dimensions of self-construal as predicting variables, and four different types of happiness as predicted variables. Thus, in total, we carried out thirty-two, different two-level analyses; each model with cross-level interactions and with random intercepts and slopes. For the more detailed description of our approach to two-level analyses (e.g., regarding centering decisions), please see the supplementary material.