Methods
Sample
Data were collected from a sample of the general population (N = 4,407). Participants were from research panels (i.e., groups of people that agree to regularly participate in social surveys) of Millward Brown, a survey company that operates worldwide. We selected countries from diverse linguistic, religious, and cultural backgrounds. From Europe, we included Spain (Southern Europe), Germany (Central Europe), Sweden (Northern Europe), and Russia (Eastern Europe); from Asia, we chose Turkey (Western Asia), India (Central Asia), and Japan (Eastern Asia); from the Americas, we decided on the USA (predominantly English-speaking) and Mexico (predominantly Spanish-speaking). We also covered major religious traditions: Islam (Turkey), Catholic Christianity (Spain, Mexico, and the USA), Eastern Orthodox Christianity (Russia), Protestant Christianity (Germany and the USA), Hinduism (India), and Buddhism (Japan).
The composition of the samples was heterogeneous in terms of sex, socioeconomic status, and education level. Data from 355 participants (8.05%) were removed due to invalid responses (e.g., missing values or inconsistencies in information on age). The results reported here refer to the rest of the sample (N = 4,052; 2,041 male) with ages ranging from 16 to 60 years (M = 34.30 years, SD = 10.45). Regarding education level, 0.3% reported not having completed elementary school, 1.9% only completed elementary school, 8.9% completed middle school studies, 34.7% completed high school, and 54.2% obtained a college degree. Regarding living arrangements, 66.6% reported to be living with a partner. Table 1 shows the demographic characteristics of each country's sample.
Instruments
A) Integrative well-being: Scale development.
1. Remembered well-being. We generated items that had similar content to those included in well-known validated measures of well-being. After assessment by subject matter experts, an initial pool of 21 items was created to assess four domains of remembered well-being (i.e., general, eudaimonic, hedonic, and social well-being). Each domain or subdomain (eudaimonic well-being has six subdomains and hedonic well-being has two subdomains) consisted of at least two items. Item translation followed the standard guidelines of translation and back-translation procedures.
General well-being. We included two items related to global satisfaction with life and one item of vitality as it is closely associated with eudaimonic functioning.
Eudaimonic well-being. Items covering optimal psychological functioning were derived from Ryff's psychological well-being model. We put together a list of 12 items addressing the following subdomains that are equivalent to Ryff's six areas of psychological well-being: life meaning, self-acceptance, personal growth, relatedness, perceived control, and autonomy.
Hedonic well-being. Affective state was assessed with items reflecting the frequency of positive and negative affect in daily life with two items for each affect type.
Social well-being. Although there are several components of social well-being (see), we selected two items that tap into the global feeling of living in a society that promotes optimal psychological functioning.
Participants were asked to rate each of the 21 statements using a scale from 0 (fully disagree) to 10 (fully agree). (See Appendix for the English version).
2. Experienced well-being. We created a list of 16 items related to specific experiences. To construct this list, we followed an approach similar to the one used in the Gallup-Healthways Well-Being Index which in turn was based on the Day Reconstruction Method. Participants were presented with eight common positive events (e.g., "I hugged someone") and eight negative ones (e.g., "I had an argument with someone") that can be experienced by virtually anyone on a given day in different cultures. Participants were simply asked to state whether these events occurred the day before. The final 10 items and the response format are presented in the Appendix.
B) Validation measures.
In addition to our initial pool of 37 items specifically generated for our scale, the Internet-based survey also included a battery of highly validated well-being measures. These measures were used as criteria to validate the items of our scale which were chosen to produce the best validity results for all languages and countries. We chose convergent validity as the main criterion so that final items (not only total scores, which are typically used in the validation of similar scales) were those that showed the highest mean correlations with their respective validation measures across countries. The following instruments were included to validate the items:
1. Remembered well-being.
General well-being. We used three measures to validate the items examining general well-being: the SWLS, SHS, and Satisfaction With Domains of Life (SWDL). The SWLS includes five items to assess the cognitive component of life satisfaction (e.g., "I am satisfied with my life"); it is the most common scale used to assess global satisfaction with life and has been implemented in several languages and cultures, providing good psychometric indices. Similar to the SWLS, the SHS is a four-item scale that assesses a general appreciation for life and personal feelings of happiness; it has been validated in several countries using different types of samples and results have indicated that the SHS has high internal consistency and sound test-retest reliability. For the SWDL, following published literature on the assessment of life satisfaction, we selected 12 different domains of life (e.g., relationships, family, friends, health, income, city, and country); participants were asked to rate their responses on a scale from 0 (very dissatisfied) to 10 (very satisfied), and a total score of satisfaction with the domains of life was calculated by summing up all the items.
Eudaimonic well-being. We used Ryff's Scales of Psychological Well-Being (SPWB;) to validate the items measuring eudaimonic well-being. Although there are several versions of the scale, we used the 39-item version, a questionnaire that covers the six areas of psychological well-being proposed in Ryff's model with six to eight items per area (environmental control, autonomy, positive relationships, purpose in life, personal growth, and self-acceptance).
Hedonic well-being. We used the PANAS to validate the items associated with hedonic well-being. This 20-item scale, assessing 10 positive and 10 negative emotions, is the most commonly used scale to assess positive and negative affect. It has been adapted for use in several languages and cultures.
Social well-being. We used the SWDL item that assesses satisfaction with one's own country as well as total scores from the SWLS and SHS to validate the items related to global social well-being.
2. Experienced well-being. Here, we sought to determine the participant's satisfaction with the previous day. To validate our 10-item measure of experienced well-being (i.e., experiences that occurred the day before), we included a question aimed at assessing the participant's overall well-being experienced the day before (i.e., "How did you feel yesterday?") rated on a Likert scale from 0 (very badly) to 4 (great).
Aside from this set of questionnaires addressing remembered and experienced well-being, we included two additional questions on health issues in the web-based interview for further validation purposes. As perceived health is a consistent proxy of happiness and well-being, participants were asked to rate their health (i.e., "How, in general, would you rate your health at this moment?") on a Likert scale from 0 (very poor health) to 10 (very good health). In a second question, they were asked to rate their sleep quality (i.e., "How much rest do you get when you sleep?"), which has also been linked to subjective well-being, on a Likert scale from 0 (none at all) to 10 (total rest).
Procedure
Data collection occurred between December 1 and 15, 2009. Using the Computer-Assisted Web Interviewing (CAWI) technique, all questionnaires were programmed into a web-based application with the content translated into seven different languages, and data were collected in an online database. Participants were invited via email and received a small incentive for their participation (i.e., they received points, which each had a value of $5, that could be accumulated and exchanged for an object from a list of goods). The percentage of panelists who initially agreed to participate but later declined to do so was 21.7%. Finally, 26.4% of the panelists initially entered the study were screened out early on because of inability to fulfill the panel quota requirements (sex, age, and location). The average time to answer the complete questionnaire was 21.0 (± 4.09) minutes.
Analytic Strategy
Our aim was to create a scale based on items from the initial pool that maximized overall convergent validity across countries. First, each of the 21 initial remembered well-being items was correlated with its respective validation criterion (e.g., positive affect items were correlated with the positive subscale of the PANAS). Items showing the highest mean correlation across countries were chosen for inclusion in the final scale. A similar procedure was followed to select the experienced well-being items. An initial pool of 16 common experiences with potential to have an emotional impact (8 positive, 8 negative) was created by the authors. The validation criterion for this pool was participants' evaluation of their overall satisfaction with the day before. Given the dichotomous nature of the experienced well-being items (i.e., yes/no responses), final item selection was based on Cramer's V (a commonly-used measure of association for the chi-squared test). The ten experienced well-being items (5 positive, 5 negative) with the highest effect sizes across countries were chosen for inclusion in the final scale. Reliability was examined by internal consistency (Cronbach's alpha).
Structural validity was assessed using principal components factor analysis, and the number of factors was determined through the Velicer's minimum average partial test. Incremental validity was tested with a series of regression analyses examining the predictive value of our scale above and beyond the SHS, SWLS, six SPWB subscales, and two PANAS subscales, using subjective health and sleep quality as the criteria.