People adapt to temporal changes in socio-cultural environments. Thus, by examining how psychological tendencies and socio-cultural environments have changed over time, we can better understand how people adapt to socio-cultural environments (e.g., Greenfield, 2016 ; Twenge, 2015 ).
In the current research, we focus on self-esteem, which is an important psychological concept (e.g., Ogihara, 2016a ). Self-esteem is one of the most frequently researched topics in psychology presumably because it is one of the most basic and important psychological concepts. Definitions of self-esteem vary across scientists, but a common aspect is that it refers to the positivity of a person’s global evaluations of the self (e.g., Baumeister, Campbell, Krueger, & Vohs, 2003 ). Self-esteem is closely related to psychological/physical health (e.g., Baumeister et al., 2003 ) and socio-economic status (e.g., Twenge & Campbell, 2002 ). It also works as a signal of social standing, which encourages adaptive behaviors (e.g., Leary, Tambor, Terdal, & Downs, 1995 ).
1.1. Temporal Changes in Self-Esteem in the U.S.
Twenge & Campbell (2001) conducted a cross-temporal meta-analysis of scores from the Rosenberg Self-Esteem Scale (RSES; 1965) and Coopersmith Self-Esteem Inventory (Coopersmith, 1967) . They showed that self-esteem increased among college students between 1968 and 1994, and among children and early adolescents between 1980 and 1993. In addition, they indicated that for children, these changes were positively associated with having comfortable socio-economic situation (e.g., negatively correlated with unemployment and divorce rates).
Gentile, Twenge, & Campbell (2010) also conducted a cross-temporal meta-analysis on RSES scores (Rosenberg, 1965) between 1988 and 2008. They indicated that self-esteem increased among middle school, high school and college students over this time.
However, little research has investigated temporal changes in self-esteem in cultures outside the European American cultural context. To present a more general theory, it is necessary to examine other cultures as well. Therefore, we examined temporal changes in self-esteem in Japan.
1.2. Temporal Changes in Self-Esteem in Japan
In Japan, Oshio, Okada, Mogaki, Namikawa, & Wakita (2014) conducted a cross- temporal meta-analysis on the scores of the RSES (Rosenberg, 1965) between 1984 and 2010. They found that self-esteem decreased among middle and high school students (analyzed together), college students and adults (aged 18 to 60, excluding college students).
Yet, their research 1) did not examine self-esteem at developmental stages prior to middle school, 2) did not separate middle schoolers from high schoolers in their analysis, and 3) analyzed data from samples that might not be representative of Japan as a whole. Therefore, Ogihara, Uchida, & Kusumi (2016a) examined temporal changes in the self-esteem of elementary and middle school students separately between 1999 and 2006 by analyzing highly representative large-sample time-series data. They found that self-esteem decreased among both groups from 1999 to 2006, regardless of gender.
These findings are thought-provoking because at first glance, they appear to contradict previous literature. It has been suggested that Japanese culture has become more individualistic at least in some respects (Hamamura, 2012; Ogihara, 2016b; Ogihara, Fujita, Tominaga, Ishigaki, Kashimoto, Takahashi, Toyohara, & Uchida, 2015; Ogihara, Uchida, Kusumi, 2015) . Given this, it could be expected that self-esteem would have increased because self-esteem and individualistic tendencies are positively correlated (e.g., Heine, Lehman, Markus, & Kitayama, 1999; Singelis, Bond, Sharkey, & Lai, 1999 ). Indeed, that is the case in the U.S. where self-esteem has risen (e.g., Gentile et al., 2010; Twenge & Campbell, 2001 ) as the culture has become more individualistic (e.g., DeWall, Pond Jr., Campbell, & Twenge, 2011; Greenfield, 2013; Twenge, Abebe, & Campbell, 2010; Twenge, Campbell, & Gentile, 2012, 2013 ). Thus, while Japanese people should show increased self-esteem, it has actually decreased. This may be because Japanese culture has not been individualistic either historically or traditionally. Rather, it has undergone a rapid change towards greater individualism, which might have caused conflicts and difficulties in adapting to it (Ogihara & Uchida, 2014; Ogihara, Uchida & Kusumi, 2014) .
However, it is unclear whether similar decreases in self-esteem occurred during other periods of time. To comprehensively understand cultural changes, it is desirable to examine long-term cultural changes (e.g., Grossmann & Varnum, 2015; Mesoudi, 2011 ). Thus, it is necessary to examine cultural changes in other periods of time and integrate them into the accumulated findings.
1.3. Present Study
The current research examined temporal changes in self-esteem among middle school students in Japan from 1989 to 2002 by analyzing large-sample time-series data. We predicted that self-esteem would decrease from 1989 to 2002, consistent with previous research (Ogihara et al., 2016a, 2016b; Oshio et al., 2014) .
We analyzed data that was collected by the Japan Youth Research Institute in 1989 and 2002 (JYRI; Japan Youth Research Institute, 1990, 2002 ). Data was collected to be proportionate with the types of schools (public:private = 5:1) and types of areas (urban:rural = 6:5) in middle schools in Japan for both 1989 and 2002. Sample sites were 11 (1989 sample) and 12 (2002 sample) schools in 10 prefectures (similar to states in the U.S.). The distributions for all answers for each item were published by the JYRI (raw data at the individual level was not available, nor were results by gender or school grade).
The sample sizes were 1288 (male: 50.2%, female: 48.9%, unknown 0.9%) in 1989 and 1071 (male: 46.8%, female: 52.6%, unknown 0.7%) in 2002. There was no significant difference in the gender ratio between two surveys, χ2(2) = 3.54, p = 0.17. In the 2002 survey, a greater proportion of participants were 2nd-year students and a lower proportion were 3rd-year students compared to the 1989 survey (1989, 1st-year: 31.9%, 2nd- year: 33.8%, 3rd-year: 34.3%; 2002, 1st-year: 30.1%, 2nd-year: 44.2%, 3rd-year: 25.8%; χ2(2) = 31.08, p < 0.001). The students answered paper questionnaires in the classes.
2.3. Question Items
Eight items from the RSES (Rosenberg, 1965) were used to measure self-esteem (see Table 1). Four items were affirmative measures and four items were negative (i.e., reverse-coded). Respondents answered to what extent each sentence applied to them on a 4-point scale (1: applies very much, 2: applies somewhat, 3: does not apply much, 4: does not apply at all). For ease of interpretation, we reversed the scores and subtracted 1 (i.e., 3: applies very much, 2: applies somewhat, 1: does not apply much, 0: does not apply at all).
A summary of the results is shown in Table 1.
3.1. Positive Items
Consistent with our hypothesis, the average scores for three items were lower in 2002 than in 1989. Yet, one item showed no change.
3.2. Negative Items
Consistent with our prediction, self-evaluations on two items were more negative in 2002. But, self-evaluations for the other two items grew less negative.
4.1. Summary of the Results
This study examined how self-esteem changed for Japanese middle schoolers between 1989 and 2002. Previous research showed that self-esteem decreased between 1999 and 2006 (Ogihara et al., 2016a) . However, it was unclear whether such decreases occurred during other periods of time. It is important to examine long-term cultural/social changes (e.g., Grossmann & Varnum, 2015; Mesoudi, 2011 ). Therefore, we analyzed large-sample time-series data collected in 1989 and 2002. Results showed that self- esteem generally decreased among middle school students. But, some items initially
Table 1. Temporal changes in self-esteem among middle school students between 1989 and 2002.
Note: **p < 0.001, *p < 0.05, +p < 0.10.
Two negative items showed a pattern that was opposite to the other items. Previous research has showed that the item (“I wish I could have more respect for myself”) is often dropped from analyses due to weak correlations with other items, implying low validity and reliability (e.g., Abe & Konno, 2007; Ito & Kodama, 2005; Moroi, 1985 ). Indeed, among studies in Oshio et al.’s (2014) meta-analysis, this item was most frequently removed from analysis (in 18 of the 19 studies reporting dropped items1). Further, the JYRI used the Japanese word “jisonshin (自尊心; self-esteem)” in this item, but it is probably not so common for Japanese, especially for middle schoolers. JYRI also used the word “rakugosya (落伍者; one who is a failure)” in the item (“At times I think I am no good at all”). These words are not commonly used by middle schoolers, which might decrease the validity and reliability of these items. Indeed, these two words are not used in the most frequently used Japanese translations of the RSES (Hoshino, 1970; Sakurai, 2000; Yamamoto, Matsui, & Yamanari, 1982) . Therefore, we do not think that the results from these two negative items were sufficiently valid or reliable.
In addition, one positive item (“I feel that I’m a person of worth, at least on an equal plane with others”) indicated no change between the 1989 and 2002. This was possibly due to item wording. This item and another item showing a small change (“I am able to do things as well as most other people”) explicitly ask respondents to engage in social comparisons. People from different groups use different referents when evaluating themselves, which obscures real differences (the reference-group effect; e.g., Heine, Lehman, Peng, & Greenholtz, 2002 ). So even if the average level of self-evaluation decreased from 1989 to 2002, self-evaluations based on social comparison may show little or no difference. While people generally use social comparison to evaluate themselves in questionnaires, when social comparisons are emphasized, the reference group effect should be stronger. Thus, it is understandable to find little or no difference in these items.
Another point that needs to be addressed is that of respondent grade-level proportion differences in the two surveys, which might have affected the decrease in self- esteem. Specifically, there was a greater proportion of 2nd-year students and a lower proportion of 3rd-year students in the 2002 survey than in the 1989 survey. However, previous research has indicated that average self-esteem levels are higher in 2nd-year students than in 3rd-year students, regardless of gender (e.g., Cabinet Office Government of Japan, 2000 ). Therefore, the decrease was unlikely to be caused by these differences. Rather, it is reasonable to think that the difference would be somewhat larger if the grade-level ratio was controlled between the two surveys.
In conclusion, we suggest that self-esteem decreased between 1989 and 2002 among middle school students, which is consistent with previous findings (Ogihara et al. 2016a; Oshio et al., 2014) . Our research extends the existing literature by showing that self-esteem declined in Japan for a period of time outside that of 1999 to 2006. Because of the importance of examining long-term cultural/social changes (e.g., Grossmann & Varnum, 2015; Mesoudi, 2011 ), these findings are valuable. Given that Japanese culture has become more individualistic in some respects (Hamamura, 2012; Ogihara, 2016b; Ogihara, Fujita et al., 2015; Ogihara, Uchida, & Kusumi, 2015) , findings that indicate concurrent decreases in self-esteem are important additions to the literature.
4.2. Limitations and Future Directions
One might be skeptical about the conclusion that self-esteem decreased over this period since it is based on data from two points in time, which might be exceptional for some reason (e.g., substantial social events). This lack of additional time points is due to the difficulties of obtaining old archival data in Japan compared to in the U.S. (e.g., Ogihara, 2015; Ogihara, Fujita et al., 2015 ). However, previous research has shown that levels of self-esteem in 1989 and 2002 are not outliers compared to other years (Ogihara et al., 2016b; Oshio et al., 2014) , so our conclusion is not diminished by this limitation. Yet, it would be ideal to examine temporal changes in self-esteem with more sequential datasets.
While our evidence shows that self-esteem decreased from 1989 to 2002, it is unclear why this happened. It is necessary to investigate this issue in the future. However, identifying the decrease in self-esteem levels is an important first step.
I thank Pamela Taylor for her helpful comments on earlier versions of the manuscript. I would also like to express my appreciation to Yumi Inoue for her cooperation in obtaining archival data. This research was partially supported by Yukiko Uchida at the Kokoro Research Center, Kyoto University and the Japanese Group Dynamics Association.
1The second most frequently dropped item is “I take a positive attitude toward myself” (dropped in three studies), but this item was not used in this research. Other dropped items are “All in all, I am inclined to feel that I am a failure” (not used in this research) and “On the whole, I am satisfied with myself”, but each was dropped in only one study out of the 19 studies reporting dropped items.