Japanese version of the 42-item psychological well-being scale (PWBS-42): a validation study

The aim of this study was to examine the internal consistency, structural validity, and convergent/known-group validity of the Japanese version of the 42-item Psychological Well-Being Scale (PWBS-42). The PWBS-42 includes six 7-item subscales designed to measure the following dimensions of eudaimonic psychological well-being: 1) autonomy, 2) environmental mastery, 3) personal growth, 4) positive relations with others, 5) purpose in life, and 6) self-acceptance. A questionnaire was administered to 2102 community residents in Tokyo aged 30 or over as a part of the Midlife in Japan (MIDJA) survey, in 2008. The internal consistency reliability was tested using Cronbach’s α. Structural validity was examined using exploratory factor analysis (EFA). Convergent validity was evaluated by calculating correlations of the Japanese PWBS-42 subscales with life satisfaction, negative affect, negative adjectives, positive affect, positive adjectives, self-esteem, and perceived stress scales. Data from 1027 respondents (505 males and 522 females) were analyzed (valid response rate = 56.2%). Cronbach’s α values ranged from 0.70 to 0.78 for five of the subscales, while that for purpose in life was lower (0.57). EFA yielded a five-factor structure: The first two factors consisted of negative and positive items mostly from the environmental mastery, purpose in life, and self-acceptance subscales. The third, fourth, and fifth factors consisted mostly of items from the positive relations with others, autonomy, and personal growth subscales, respectively. As hypothesized, the scores for life satisfaction, negative and positive affect/adjectives, self-esteem and perceived stress were significantly correlated with all subscales of the Japanese PWBS-42. The subscales of the Japanese version of the PWBS-42 showed accep. levels of reliability and support for convergent validity in the Japanese population. The factor structure was slightly different from the theoretical 6-factor model: items of three subscales (environmental mastery, purpose in life, and self-acceptance) loaded together on two factors. This finding may be interpreted in light of the interdependent self construal found in Japan in which these three components could be closely linked.


Background
Eudaimonic well-being has been often discussed as purpose in life, meaning of life [1,2], or as a concept relevant to self-determination theory [3,4]. One of the most famous models of eudaimonic well-being is the psychological well-being (PWB) model, developed by Carol D. Ryff [5]. PWB integrated psychological concepts such as self-actualization [6] and being a fully functioning person [7], distilling six key factors of well-being: autonomy, environmental mastery, personal growth, positive relations with others, purpose in life, and self-acceptance. The association of Ryff's PWB with various health outcomes has been well-documented [8]: overall longer survival [9], low risk of physical diseases such as stroke [10], myocardial infarction [11], and metabolic syndrome [12], and low risk of mental illness such as depression [13,14], insomnia [15] and Alzheimer's disease [16]. In terms of biological functioning, higher levels of PWB associated with lower level of plasma cytokines [17,18].
Since promoting PWB can contribute to people's personal growth, career development, and successful aging, measuring and promoting PWB among the Japanese population might be helpful for solving important issues in life at every stage. PWB has been frequently measured with Ryff's Psychological Well-Being Scales (PWBS) [19]. The reliability and validity of the PWBS has been established in more than 30 languages and across various cultures [20,21].
In Japan, three published articles have examined measurement of PWB in Japanese samples [22][23][24]. However, the measures the researchers used made it difficult to compare scores with those in other languages: two of them were not developed based on Ryff's original items of PWBS [22,23], the other one [24] had 84 items which were rated dichotomously, with categories of "Yes" or "No". However, in cross-cultural cohort studies, it is more common for items to be rated on 6-or 7-point Likert scales. In addition, the measures used in these three studies have limitations in terms of balance between the burden on the participants of completing them and adequate depth of measurement, thus raising questions about the credibility of their assessments of the six well-being constructs. Although the length of the PWBS ranges from 18 [25] to 120 items [19], the ultra-short (18 items) version had psychometric problems of low internal consistency (0.33 < Cronbach's alpha < 0.56) [25], while acceptable criteria is regarded Cronbach's alpha > 0.7. Moreover, the 84-and 120-items versions imposed an undue burden on the respondents [8,19,26,27]. The 42-item PWBS has been recently recognized as achieving the needed balance, and has been adopted in several studies including the Midlife in the United States (MIDUS) survey, a large cohort study in the United States [8,28].
Hence, the purpose of this study is to examine the internal consistency, structural validity, and convergent/ known-group validity of the Japanese version of the 42item PWBS among the general population in Japan. In addition, we considered the interpretability of this measurement.

Study design and participants
This is a validation study for a self-report measure of well-being. The Midlife in Japan (MIDJA) study surveyed 2102 community residents aged 30-79 years old living in the 23 wards of Tokyo city from April to September 2008 [29]. The responses were collected from 1027 adults (505 men and 522 women; age range = 30 to 79 years). The overall response rate was 56.2%. It was found that 275 participants were non-eligible (i.e., undelivered) due to having moved, unknown address, being absent during the time of the survey, illness, injury, being hospitalized or being deceased, which described in detail elsewhere (data available upon request). The primary objective of the MIDJA is to compare the Japanese sample (MIDJA) with the United States sample (MIDUS) to test hypotheses linking constructs of interdependence and independence to well-being and health. All respondents completed a self-administered questionnaire. The Research Ethics Committee of the Graduate School of Medicine in The University of Tokyo approved this study (no. 1691- [5]).
The study was reported according to the Consensusbased Standards for the Selection of Health Measurement Instruments (COSMIN) guideline, which is used to improve the quality of efforts to develop health-related selfreport measurement instruments [30]. Each finding for properties of measurement in this study was assessed by the COSMIN Risk of Bias checklist (Additional file 1).

Psychological well-being scale (PWBS)
With permission from Professor Ryff, the developer of the PWBS, we translated the 42 items of the PWBS into Japanese. The PWBS [19] originally comprised six subscales with seven items each to measure the six factors: 1) autonomy (e.g., I am not afraid to voice my opinions, even when they are in opposition to the opinions of most people.); 2) environmental mastery (e.g., In general, I feel I am in charge of the situation in which I live.); 3) personal growth(e.g., I think it is important to have new experiences that challenge how I think about myself and the world.); 4) positive relations with others(e.g., Most people see me as loving and affectionate.); 5) purpose in life(e.g., I have a sense of direction and purpose in life.); and 6) self-acceptance(e.g., In general, I feel confident and positive about myself.). Response categories for these items are on a seven-point scale: ranging from 1 ("Strongly disagree") to 7 ("Strongly agree"). The scores for six subscales were calculated as averages; higher scores mean greater psychological well-being.
The authors complied with the International Society for Pharmacoeconomics and Outcomes Research (ISPOR) taskforce guidelines [31], which constitute the standard procedure for translation and adaptation of self-reporting scales in other languages. The forward (from English to Japanese) translation of the PWBS 42 items was prepared by Prof. Karasawa, who is a research member of the MIDJA, with considerable knowledge about the theoretical background of PWB. The forward translation was reviewed and revised by colleagues at The University of Tokyo to confirm its relevance and meaningfulness. After these consultations with researchers, the slightly amended version was back-translated into English by an independent translator who was blind to the original English version and reviewed by research members of MIDJA. The original developer, Prof. Ryff, reviewed this back translation and further corrections were made based on her suggestions.

Life satisfaction (6 items)
Life satisfaction was measured by the newly developed six items scale for the MIDUS/MIDJA (e.g., "How would you rate your life overall these days"). Assessing overall, work, health, family (the mean of ratings for relationship with spouse/partner and relationship with children) and financial satisfaction each question asked participants about their satisfaction with their life on current days on a scale ranging from 0 (the worst possible) to 10 (the best possible). The overall score was constructed by calculating the mean of the items. The scale was computed for cases that had valid values for at least one item on the scale. Higher scores reflect higher levels of overall life satisfaction, meaning high hedonic well-being. In this study sample, the internal consistency reliability of the scale (Cronbach's alpha) was 0.75 [32].

Positive and negative affect/adjective
Positive and negative adjectives were measured by four and five items of each (e.g., enthusiastic, attentive, proud, active, afraid, jittery, irritable, ashamed, upset) from The Positive and Negative Affect Schedule (PANAS) [33], which is a widely used mood measurement and regarded as one of the indicators of hedonic well-being. In addition, positive affect was each measured by six items (i.e., cheerful, in good spirits, extremely happy, calm and peaceful, satisfied, full of life), because such low arousal positive feeling has been reported to have stronger link with health among Japanese people, other than high arousal feelings, measured by PANAS [34]. The negative affect was also measured by an original six-item scale consisting one item on "nervousness" from PANAS and items from other sources (i.e., so sad nothing could cheer you up, restless or fidgety, hopeless, that everything was an effort, worthless). Additional items both in positive and negative affect were retrieved from the previous study [35]. This study used the past 30 days as the time frame (i.e.., During the past 30 days, how much of the time did you feel…). All items were rated on a 5-point Likert-type scale, ranging from 1 "None of the time" to 5 "All of the time". Scales were constructed by calculating the mean across each set of items. The scales were computed for cases that had valid values for at least one item on the scale. In this study sample, the internal consistency reliability (Cronbach's alpha) of the scales for 1) positive affect, 2) positive adjective, 3) negative affect, 4) negative adjective was 1) 0.93, 2) 0.80, 3) 0.86 and 4) 0.82 respectively [32]. We used these four sub-scales as a measurement of affect to keep their independent construction. Positive affect and adjective in this questionnaire is well validated elsewhere [34].

Self-esteem (7 items)
Self-esteem was measured with a scale consisting of seven items taken from Rosenberg's self-esteem scales [36] (e.g., "I am able to do things as well as most people"). A response to each item was scored on a seven-point scale ranging from 1 (agree strongly) to 7 (disagree strongly). In this study sample, Cronbach's alpha for men and women was 0.63 and 0.69 [37], respectively.

Perceived stress (10 items)
Perceived stress was measured with the Perceived Stress Scale (PSS) [38,39], which asked 10 items about their recognized stressors on these 30 days (e.g., "felt that you were unable to control the important things in your life?"). Each item was rated on five-point scale ranging from 1 (Never) to 5 (Very often). In this study sample, Cronbach's alpha was 0.76 [37].

Demographic variables
A questionnaire was administered to assess demographic variables, including gender (male or female), age, education status (Junior high school, some high school, high school, vocational school, 2 year college, some college, 4 or 6 year college or over graduate school), marital status (married, divorced/widowed or single) and work status (Have a paid job or Do not have a paid job).

Statistical analysis
Statistical significance was defined as p < 0.05. All the statistical analyses were performed using SPSS 26.0, Japanese version (SPSS Inc., Chicago, IL).
Before conducting the analyses, the scores of some items were reversed as recommended in Ryff's original PWBS [19]. For an item with a missing value, the mean value of the completed items was imputed. The scales were computed for cases that had valid values for at least one item in each domain following the documentation of "Scales and Constructed Variables in MIDJA" guide [32].
To assess the internal consistency of the Japanese PWBS, Cronbach's alpha (α) coefficients were calculated for each of the six subscales. To assess structural validity, confirmatory factor analysis (CFA) was conducted to test the goodness of fit for the existing structure of psychological well-being, and an exploratory factor analysis (EFA) was conducted in case the CFA showed a poor fit. Based on previous research, we hypothesized a six-factor structure. In addition, we extracted factors with eigenvalues of more than 1.0, following the Kaiser-Guttman "Eigenvalues greater than one" criterion [40][41][42], using robust maximum likelihood estimation. In the two EFAs, the promax rotation method was adopted to calculate factor loadings for the items. As a hypothesis test for convergent validity, Pearson's correlation coefficients (rs) were calculated between each score of the PWBS and life satisfaction, positive and negative affect/adjectives, and perceived stress, which was considered to have moderate-to-strong associations with psychological wellbeing. Positive and moderate correlations would be expected with life satisfaction, positive affect/adjectives, and self-esteem. Negative and moderate correlations would be expected with negative affect/adjectives and perceived stress. Although these measures consisted of distinct components of well-being, prior research revealed relations among evaluative, hedonic and eudaimonic well-being [43]. In addition, a one-way factorial ANOVA was performed as another hypothesis test for known-group validity to compare the scores of the six subscales of the Japanese PWBS when stratified based on demographic and occupational variables (sex, age, marital status and working status). For these demographic and occupational variables, we observed significant differences from the PWB scores in previous studies (e.g., high personal growth and purpose in life in the elderly, high positive relationship with others in women) [25].

Characteristics of participants
We received responses from 1027 members (56.2%) of the target sample. The proportion of males was 49.2%, and average age was 54.4 years old. The demographic characteristics of the participants are shown in Table 1. Many of the participants were married (69.1%) and had a paid job (71.6%). Table 2 shows mean scores and Cronbach's alphas (α) for the PWBS-42 subscales. The Cronbach's α coefficients ranged from 0.70 to 0.78, except for that for Purpose in life (α = 0.57).

Factor structure of PWBS-42 in Japanese
The results of CFA are shown in Table 3   and no item loaded on the sixth factor (data available upon request). Then, we tried conducting another EFA that hypothesized a 5-factor structure with the promax rotation method, which allows factors to correlate with each other, using a robust maximum likelihood estimation. Table 4 shows that the EFA then successfully yielded five factors. The first two factors consisted of negative and positive items mostly from environmental mastery, purpose in life, and self-acceptance. The third, fourth, and fifth factors consisted mostly of items for positive relations with others, autonomy, and personal growth, respectively. EFA according to the Kaiser-Guttman's criterion indicated an eight-factor structure (data available upon request). Authors did not adopt the model because it was difficult to interpret the meaningfulness of the eight factors and there were few items loaded on the subordinate factors. Table 5 shows correlations between the scores of the theoretical subscales that originally assumed on the PWBS-42 and scores of the five factors extracted from the previous EFA. Each factor score was sum of items. There were significant and exclusive correlation between the five factors and the six subscales of the PWBS-42; Factor 2 and Self-acceptance, Factor 3 and Positive relationship with others, Factor 4 and Autonomy, and Factor 5 and Personal growth. Factor 1 showed moderate association with all theoretical six subscales. Table 6 shows correlations between the scores of the six subscales of the PWBS-42 and life satisfaction, positive and negative affect/adjectives, self-esteem, and perceived stress. Although the EFA showed a five-factor structure, we presented average scores and correlations of the original six subscales because it was planned as part of validation and for keeping the international comparability. In addition, we cannot conclude that the original six-factor structure and the observed five-factor structure are definitely different as shown in Table 5. All scales were significantly correlated with all domains of PWB. Scales related to negative feelings (i.e., negative affect, adjective, PSS) were correlated negatively with all domains (− 0.55 < r < − 0.13). Positive scales (i.e., life satisfaction, positive affect, adjectives, self-esteem) were correlated positively with all domains (0.24 < r < 0.71). Table 7 shows the results of ANOVAs to examine the differences between each category for the demographic characteristics (sex, age, marital and working status). One-way ANOVA revealed some significant differences among variables (e.g., high autonomy in males, high personal growth in workers).

Discussion
The results showed an acceptable level of internal consistency reliability of the newly translated Japanese version of the PWBS-42, although careful consideration of the results was need given that factor analysis showed a five-factor structure. Cronbach's alpha coefficient was lower for the scales for purpose in life, possibly because one item of "Done all there is to do in life" (the item P7) behaved differently than the other items. The results of CFA did not show a good model fit for the previously proposed six-factor structure model. The correlation analysis showed theoretically high correlation with other scales, supporting convergent validity. The ANOVA confirmed the hypothesized differences according to demographic characteristics between known independent groups (population in U.S. [25] vs. Japan), suggesting good known-group validity, as described in detail below.
We found that the original six-factor model [5] did not fit well with the present data in the CFA. Rather, the EFA suggested a five-factor structure. The first two factors  consisted of negatively and positively worded items mostly from environmental mastery, purpose in life, and selfacceptance, as well as some items from the other subscales. The result for the correlation analysis also showed that Factor 1 and 2 was highly related (r > 0.8) with environmental mastery and self-acceptance. Factor 1 had high correlation (r > 0.7) with most of PWB subscales instead of autonomy and purpose in life. The highest correlation among them was with environmental mastery, identifying the EFA result that the top three items of high loading score in Factor 1 were all about environmental mastery. The correlation of Factor 2 and self-acceptance may possibly be due to relatively high loading scores of S7 and S2.
Although purpose in life did not show high correlation with any factors, it might be due to its low internal consistency. We discuss two aspects of the possible reasons the factor analysis yielded these results.
First, it seems that respondents in this sample of Japanese community residents perceived three constructs measured by these subscales (i.e., environmental mastery, purpose in life, and self-acceptance) quite similar to each other. The findings contract to previous observations in Western countries that these constructs are separate in factor analyses [25,44]. Among Eastern cultures, the self is viewed as interdependent, rather than independent as it is in Western culture. Experiencing interdependence entails seeing oneself as part of a network of social relationships and recognizing that one's behavior is determined by the others' perspectives or expectations. The fundamental connectedness of the human community with the surrounding context "self-inrelation-to-other" is focal in individual experiences in Eastern cultures [45]. With the interdependent self, wellbeing is realized in one's relationship with surrounding SD standard deviation a Participants who had missing data were excluded from each analysis ** p < 0.01, * < 0.05 Table 6 Pearson's correlation coefficients a between each subscale on the PWB and Life Satisfaction, Negative Affect, Negative Adjectives, Positive Affect, Positive Adjectives, Self-esteem, Percieved Stress Scale (PSS) among Japanese population realities, both social and non-social [46]. Thus, environmental mastery, being able to take adaptive and situationally bound actions in the social context, might be closely linked with self-acceptance in a country with an Eastern culture, such as Japan. In addition, also in line with the construal of interdependent cultures, the self becomes most meaningful (i.e., high purpose in life) when it is cast in the appropriate social settings [45]. Japanese people are motivated to find a way to fit in with relevant surroundings by fulfilling obligations to become part of various interpersonal relationships [45]. Purpose in life may be also closely related to environmental mastery and self-acceptance in Japanese culture. In Japanese culture, these three components of PWB may come together and be hard to distinguish. Second, negative and positive items were divided into two different factors in this sample. The same pattern was shown in a previous study with a UK cohort [47]. There are two reasons that the PWB scale includes negative/positive distinction of wordings; (1) assessing constructs with items framed positively and negatively is needed as a check on acquiescence response bias (i.e., the tendency to agree with everything, perhaps due to not even reading the items); (2) items for the PWB scales were generated based on the theoretical definitions of each construct which included a definition of both high and low scorers for each dimension (i.e., items intended to capture both having and not having the various aspects of well-being) [48]. The use of both negatively and positively worded items is expected to increase the validity of what is meant by having a high score on any dimension of well-being because to get a high score, a person has to both agree with positively worded items and disagree with negatively worded items. The pattern that negatively and positively worded items load on different factors in a factor analysis has been observed for several mental health and psychological well-being scales in Japan [49][50][51][52], which is interpreted as differential response styles to negatively and positively worded items in the Japanese population. Thus, the finding for the two factors may not necessarily imply that there are two distinct unique constructs of psychological well-being in the Japanese population. The two factors observed in the exploratory factor analysis may be just due to a reporting style, but still reflect the same construct.
Most items for the remaining three subscales (i.e., positive relations with others, autonomy, and personal growth) loaded on separate factors that seem to reflect the corresponding dimensions of the PWB. The results for the correlation analysis also suggested that Factor 3, 4, and 5 had high to moderate correlation with positive relationship with others, autonomy, and personal growth respectively. The observed five factor-structure in this study seem compatible with the original six-factor model [19] to some extent, while the three subscales (i.e, environmental mastery, purpose in life, and self-acceptance) were collapsed to represent a single combined dimension in this sample. It is possible that the factor structure is affected by the characteristics (e.g., gender, age groups, and urban/rural difference) of the population and the cultural context. The authors could not conclude from this study that the fivefactor structure fit to the Japanese PWBS-42 rather than the original six-factor model [19]. Further research is needed to re-examine the factor structure of the Japanese PWBS-42 in a sample with a broader age range and from non-urban settings to know if the pattern could be generalized to the whole Japanese population.
The results for the correlation analysis indicated that all the PWBS-42 correlated highly with other psychometrics variables measured by other scales in the theoretically expected direction, supporting its convergent validity. As shown by ANOVA, there were some significant demographic differences in scores on the Japanese version of the PWBS-42. Specifically, some differences, such as decremental age profiles (with scores for the older people significantly lower than those for younger people) for personal growth and purpose in life, sex profile (with women scoring higher than men) of positive relationship with others, were coincident with previous research in the U.S. [25]. Otherwise, high autonomy in men seems peculiar to Japanese men. Marriage had an advantage over being divorced or single in many subscales, except for autonomy. Previous studies suggested that marriage is a powerful predictor of high well-being [53,54]. Having paid work was related to high scores for personal growth and purpose in life. These two subscales were found to contribute to work and career related outcomes [55]. The findings supported the view that using six subscales for the Japanese population was appropriate, suggesting known-group validity compared with other population. Regarding interpretability of this scale, the result showed relatively low PWB scores among Japanese sample (30.6 < mean < 33.8) compared with the U.S. sample (37.1 < mean < 40.6). If we need to categorize individuals into low, optimal, and excessively high scores for each dimension of the PWB, such as in assessment for well-being therapy [56], this cultural difference must be considered.

Limitations
This study has some limitations. First, the MIDJA population consisted of people over 30, so young population under 29 cannot be considered. Second, the translation process is not partially appropriate. Third is that our Japanese participants treated item P7 (see Table 4) as positive even though it is negative phrased (and reversed-scored). We kept this item and its original scoring in the Japanese version of the PWBS-42 in this study to maintain consistency with the original PWBS; however, that practice should be reconsidered in future research. Fourth, some scales which were used in convergent/known-class validity analysis are biased due to insufficient consideration of reliability and validity. Finally, some important properties of the scale were not investigated in the study due to the lack of repeated measurement: test-retest reliability, measurement error, predictive validity and responsiveness.

Conclusion
The Japanese version of the PWBS-42 showed acceptable reliability and convergent validity when applying the original scoring method for the six subscales. However, the factorial validity based on the original six-factor model was not well supported in this study, while most factors observed in this study seem compatible with the original factors. A further examination is needed to know if the factor structure of the PWBS-42 is different for the Japanese population.