In this study, we conducted an initial evaluation of the J-MVS, a new Japanese version of the MVS that we developed with data from 500 people in Japan, while controlling for age and sex. No materialism scale has been examined for use in Japan in any previous research. The following is a discussion of each of the research questions proposed at the beginning of this paper.
The J-MVS was considered to have a dual-factor structure based on the following points: (1) the triple-factor structure, which was the same as the original 18-item version and the subsequently proposed 15-item version, had a low goodness-of-fit; (2) the J-MVS-A6 and J-MVS-P6 were the most dominant among the models compared in this study. The content of those versions was derived through exploratory factor analyses, and they were shown to meet the various goodness-of-fit indices.
Both the J-MVS-A6 and J-MVS-P6 were shown to have sufficient internal consistency. The discriminant validity of the success/centrality domain and the happiness domain was also demonstrated. The convergent validity of the J-MVS-P6 was slightly below the criterion but was considered to be relatively good when referring to previous reports from other countries. Based on the factor loadings reported in the German version [11], the AVE in the success/centrality domain was estimated to be 0.41, and, in Watchravesringkan’s validation in the United States and Thailand, the AVE in the success domain was less than 0.50 [15]. The AVE calculated from the factor loadings based on Richins’ original 18-item [8] exploratory factor analysis was 0.36–0.41 for each domain, whereas the J-MVS-P6 showed an AVE of 0.45.
Regarding CR, the J-MVS-P6 met the criterion, while the J-MVS-A6 had CR close to the criteria (0.698 in the success/centrality domain). In both these scales, the goodness-of-fit indices of the dual-factor structure were high, and although some of the validity and reliability indices did not fully meet the criterion mentioned above, they were considered acceptable at this stage when compared to existing reports. Nevertheless, the fact that there are many items with low factor loadings and insufficient validity indices in translated versions of the MVS is an issue, not only for the developed J-MVS but also for other language versions of the MVS in general.
The reasons for the J-MVS being judged as having a dual-factor structure, not a triple-factor structure as in the original version, require further investigation. Griffin et al. suggested that the triple-factor structure was not replicated in France and Russia for the MVS because the inter-factor correlations were high, and there may have been a lack of discrimination between the factors [10]. They also suggested that, according to previous studies, RWI can cause problems in cross-cultural studies, potentially leading some items to not have sufficient factor loadings [10]. In this study, the above situation may have contributed to the dual-factor structure. Furthermore, the division of materialism into two factors may be consistent with the ideas described in the “Dual Model of Materialism” [56, 57] by Sirgy et al. In this model, the two aspects of materialism, success materialism and happiness materialism, are clearly distinguished. Success materialism is related to economic motives and positively influences life satisfaction, while happiness materialism may increase dissatisfaction with the current standard of living [56, 57].
When scales developed in Western countries are translated, the responses of individuals in Asia differ between PWI and RWI, and one of the proposed solutions for this conundrum was the deletion of RWI [12]. Hence, to construct a more appropriate scale and verify the external criterion-referenced validity, we used both the J-MVS-A6, which was obtained by repeating the exploratory factor analysis for the original 18 items, and the J-MVS-P6, which was obtained by repeating the exploratory factor analysis for the items without the RWI.
When the J-MVS-P6 was compared with the J-MVS-A6, the correlation level with personality (social comparison orientation, neuroticism) and subjective well-being indices were almost the same or slightly higher in the J-MVS-P6. Therefore, the external criterion-referenced validity of the J-MVS-P6 was the same or slightly higher than that of the J-MVS-A6. In addition, Wong et al. discussed that the questionnaire composed only of PWI is less likely to confuse respondents [12]. Furthermore, in the J-MVS-P6, the number of measurement items is well balanced, with three items in each domain. We therefore propose the J-MVS-P6, which consists of six items (three items from the success/centrality domain and three items from the happiness domain) and has no RWI, as a scale to measure materialism in Japan.
The distribution of scores was close to a normal distribution, making the scale easy to use in research. However, it should be noted that there are some disadvantages of using PWI alone, such as the increase in straight-line responses and the possibility of agreement bias [12, 13]. The results of the measurement invariance test supported configural invariance, partial metric invariance (i.e., the equivalence of factor loadings), and partial scalar invariance (i.e., the equivalence of intercepts between sex groups). Regarding age groups, although the measurement invariance test was conducted between relatively small-sized sample groups (100 each), it was supported up to a partial metric invariance, indicating the equivalence of factor loadings. This is considered to confirm a degree of measurement invariance.
Regarding the relationship between the MVS and age, a meta-analysis of 23 existing studies reported a mean correlation coefficient of − 0.16 with a 95% confidence interval of − 0.14 to − 0.18 [19]. The correlation coefficient between the J-MVS-P6 and age was − 0.28, indicating a higher correlation than that in prior research. Regarding the association between the J-MVS-P6 and ability comparison in the INCOM, the association was high with r = 0.51; this number is close to the value reported by Kim et al. (r = 0.60) [24].
Regarding the correlation between the MVS and neuroticism (one the Big Five personality traits), it was previously reported to be r = 0.22–0.24 for the overall scale [23], r = 0.33 for the overall scale, r = 0.36 for the happiness domain, and r = 0.12 for the centrality domain [58]. In the present study, positive correlations with neuroticism were found in the J-MVS-P6 overall score (r = 0.27), the happiness domain score (r = 0.23), and the success/centrality domain score (r = 0.14). As for the correlations of the MVS with agreeableness and conscientiousness, negative correlations were suggested in previous studies [21, 22]. In this study, the J-MVS-P6 overall scores and the happiness domain scores were also negatively correlated with agreeableness scores; the J-MVS-P6 happiness domain score was also negatively correlated with the conscientiousness score. Many of the previously reported tendencies in the relationship between materialism, age, and personality were replicated. Negative, weak positive, and insignificant correlation coefficients have been reported for the MVS with agreeableness, extraversion, and openness, respectively [22]. In the J-MVS-P6, no significant correlations were found with agreeableness and extraversion, and a weak positive correlation with openness was found in the success and centrality domain. It has been reported that low materialism groups have higher openness [21], and based on regression analysis, it has also been shown that low openness predicts high materialism [22]. Therefore, the positive correlation with openness in the success and centrality domains of the J-MVS-P6 may be a Japanese-specific tendency. Kilbourne et al. hypothesized that openness is positively related to variety and excitement associated with consumption, which leads to higher materialism [59]. The positive association between openness and materialism found in the present study may be consistent with the mechanism hypothesized by Kilbourne et al. [59].
The J-MVS-P6 scores were negatively correlated with various well-being indicator scores (except for self-rated health), and the negative correlation was particularly large in the happiness domain. The mean correlation coefficient between materialism and well-being was reported as − 0.15 in a meta-analysis by Dittmar et al. [18]. The correlation coefficients between the J-MVS-P6 overall score and self-rated happiness, SHS, and SWLS scores ranged from − 0.29 to − 0.34; those with the happiness domain ranged from − 0.39 to − 0.42, which were larger than those reported in the meta-analysis [18]. These results suggest that materialism and well-being may be more closely related in Japan.
The correlation between the J-MVS-P6 overall score and PANAS score was r = 0.33, which was comparable to that in a previous report (r = 0.35) [27]. For the correlation with self-rated stress, the r = 0.23, which was similar to the significant correlation coefficient of 0.20 with the MVS score reported in Burroughs and Rindfleisch’s study [26]. However, and in contrast to the results found in Western countries and Korea [18, 28], the J-MVS-P6 score was shown to be nearly uncorrelated with self-rated general health.
The J-MVS-P6 and the MLQ Search scores were positively correlated, and a higher positive correlation was found with scores for the success/centrality domain. It may be that the tendency to search for meaning in life may be higher among people who consider material wealth to be a success and a central value in life. The higher the tendency to regard material possessions and consumption as life success or as a central value in life, the more likely it is to lead to dissatisfaction with psychological needs [25]; this may increase the tendency to search for meaning in life.
In contrast to the results from a study conducted with American people, the values for the MLQ Search and MLQ Presence were positively correlated (r = 0.24) among Japanese people [41]. In the present study, both were highly and positively correlated (r = 0.44); however, the J-MVS-P6 score showed different associations with MLQ Search and MLQ Presence scores. Specifically, the happiness domain and MLQ Presence scores showed a negative correlation. These results suggest that people with high materialism tend to have a weaker sense of presence of meaning in their lives. Further, the higher the tendency to think of material possession and consumption as happiness, the more likely it is that the presence of meaning in life is weakened due to dissatisfaction with psychological needs. The evidence on the relationship between the MLQ Presence and the MLQ Search differs between the United States, which has a high independent construal of self, and Japan, which has a high interdependent construal of self [41]. The different tendencies between the MLQ Presence and the MLQ Search scores in the success/centrality domain and the happiness domain of materialism may be based on differences in the cultural construal of self.
Limitations
To ensure the equivalence of the translated scale with the original scale, we obtained the cooperation of three people and performed a back-translation process. However, there may have been a tendency to be overly literal in the interpretations. This tendency may be related to the low factor loadings on some of the scale items. For future scale translations, the challenge will be to make the wording more natural and easier to understand while ensuring equivalence, for example, by utilizing more human resources or AI.
Additionally, respondents were limited to those who responded voluntarily to the online agency’s call for responses, use the Internet in their daily lives, and are registered as survey company monitors. At the same time, the weakness of the argument regarding sample size could be pointed out, and it might have been better to determine the sample size using a power analysis approach. Therefore, the results may be limited in their generalizability for the Japanese population. The situation of those who are not proficient or comfortable with computers or other online (internet) interfaces and those who do not intend to register as a survey company’s monitor will be important data and should be investigated in the future. It will be necessary to conduct interviews or paper-based surveys to analyze subjects according to their demographics.
In addition, to verify the external referenced-criterion validity of the scale, many indicators were incorporated into the questionnaire, but the large number of questions may have reduced the quality of the responses. In the future, it would be desirable to conduct a survey with the number of questions below a certain level.
Furthermore, while sufficient reliability was confirmed for the well-being and social comparison orientation measures, the reliability of the Big Five scale with two items for each personality trait was low. This also affects the reliability of the correlation with the J-MVS-P6. Future researchers will have to tackle the challenge of using a more reliable version of the Big Five questionnaire instead of a shortened version, such as the TIPI-J.