1 Background

1.1 Level and instability of self-esteem

Self-esteem is one of the personality characteristics that predict responses after rejection (e.g., prosocial responses, withdrawal and avoidance, or antisocial responses). However, responses after rejection depend on self-esteem and wide-ranging relational and situational factors [1]. Similarly, self-esteem also predicts responses after receiving comfort. People with unstable self-esteem have greater sensitivity to cues connoting acceptance and rejection than others [2, 3]. Therefore, they may show stronger emotional reactions than others when receiving comfort. Nevertheless, the quality of the emotional response may be positive, enhancing their self-esteem, or negative, causing feelings of being incapacitated and threatened. Furthermore, the quality can differ depending on self-esteem levels and instability. Focusing on self-esteem level and instability helps explain the discrepant findings regarding these comfort effects in previous research.

Self-esteem instability (SEI) has been defined as “the magnitude of short-term fluctuations that people experience in their contextually based, immediate feelings of self-worth” [4]. People with unstable self-esteem are vulnerable to negative events. They are highly likely to be concerned about social acceptance and rejection [2]. For example, people with unstable high self-esteem are highly likely to have increased sensitivity to negative evaluations from others and experience anger and hostility. This fragile form of high self-esteem reflects vulnerable feelings of self-worth that rely upon self-deception and requires external validation [5, 6]. Nevertheless, people with stable high self-esteem have lower perceived aggression than people with unstable high self-esteem. This secure form of self-esteem reflects solid and realistic feelings of self-worth and does not require considerable external validation [7]. In recent decades, many studies have shown that unstable self-esteem moderates the effects of high self-esteem on psychological functioning [5,6,7,8,9]. Thus, investigating the level and instability of self-esteem can provide a useful perspective to understanding self-esteem.

As a developmental feature, self-esteem tends to be low from childhood to adolescence/young adulthood until around age 30 [10]. In particular, Japanese adolescents have exhibited the lowest self-esteem in 53 nations [11]. A meta-analysis covering studies of self-esteem in a Japanese sample from 1984 to 2010 revealed a decreasing trend in self-esteem [12]. While baseline levels and life span trajectories of self-esteem vary slightly across sociocultural regions [13], investigations into the developmental trajectory of self-esteem in Japanese samples have indicated low levels, with an upward trend from adulthood to old age [14]. This instability is often due to unclear self-worth and sensitivity to others’ evaluations [15]. During adolescence, people are prone to self-conflict and adaptive problems, with close friendships strongly influencing the ability to cope with developmental transitions and life stress [16, 17]. Furthermore, emotional support from friends increases during this time [18, 19]. However, the effectiveness of emotional support from friends differs depending on the recipient’s self-esteem level [20, 21]. For example, people with low self-esteem often resist emotional support from well-meaning friends after experiencing failures or rejections. SEI may moderate the influence of self-esteem levels on the effectiveness of emotional support from friends.

From a cultural perspective, it has been found that emotional support is predictive of well-being in Asian culture (including Japan), whereas it is not predictive of well-being in European and American culture [22]. Japanese people offer social support to their friends with the goal of fostering closeness and feel fulfilled when they have their needs of emotional ties met by others [23]. Based on those findings, emotional support is considered a particularly significant predictor in enhancing self-esteem for Japanese youth. As per recent statistics, 83.8% of Japanese high school students pursue tertiary education [24]. The average age of new entrants in Japanese university is 18.5 years [25], and most of them are teenagers or in their twenties [24]. Specifically, it is assumed that most Japanese university students are in a transitional phase moving from adolescence to young adulthood. During that period, many Japanese university students experience prolonged adolescence while exploring identity and self-worth and transferring closeness from parents to peers [26, 27].

1.2 Comfort as a form of emotional support

The sociometer theory describes self-esteem as an internal gauge that monitors the degree of acceptance and rejection in interpersonal relationships and a motivator for corrective actions when their quality reduces [3]. According to the theory, people with unstable self-esteem (low or high) have an unstable sociometer that responds to cues that connote acceptance and rejection.

Generally, when self-esteem decreases (i.e., the sociometer goes low), people receive comfort from others and try to enhance their self-esteem. In a linguistic dictionary, comfort is defined as relief or support in mental distress or affliction, encompassing consolation, solace, and soothing [28]. Additionally, it is described as a form of emotional support, involving both verbal and nonverbal processes through which one communicates care and concern for another [29]. Based on this, the present study defines comfort as verbal or nonverbal behavior adopted when seeing someone facing some sort of difficulty, with the aim of alleviating the individual’s mental distress and restoring their psychological state. In this paper, the concepts of “consolation” and “comfort” are used interchangeably [30]. Comfort can also involve feeling sympathy or pity. For example, people feel pity, which motivates them to provide comfort when they see others requiring aid or in a negative state because of uncontrollable conditions [31]. However, comfort can be a threat to those receiving comfort. For instance, when receiving comfort, recipients may feel the consoling person perceives them as incapable of handling adverse situations and returning to a positive psychological state. Hence, a person receiving comfort could feel increasingly dejected or angry because they perceive that the other considers them lacking in some way [32]. Thus, the comfort provided to those in psychological distress is one of the important emotional support to enhance their self-esteem; nonetheless, it heightens the risk of increasingly damaging their self-esteem.

Therefore, several researchers have examined the conditions required to comfort others effectively. Supportive interactions, including comfort, are influenced by personal, relational, and situational factors [33]. For example, research reveals that the outcome of comfort varies according to comforting messages and problem severity [34, 35], psychological factors that include a recipient’s personality traits and cognitive abilities [36], and intimacy in a relationship [37, 38]. Comfort was more effective when provided by a close friend than by strangers or distant acquaintances [38]. Nevertheless, the comfort a close friend provides can also be perceived as unhelpful support by the recipient, considering several people may rely on intrapersonal self-reinforcing rather than comfort to enhance self-esteem [37].

1.3 Sex/gender differences

In accordance with the Sex and Gender Equity in Research (SAGER) guidelines [39], this study reviews previous studies about sex/gender differences in self-esteem. Men commonly tend to have higher self-esteem than women in many countries; this might reflect biological processes (such as hormonal influences) and universal sociocultural influences (such as gender roles) [40]. Research on self-esteem in a large and globally diverse cross-sectional samples has shown that females have significantly lower self-esteem than males between the ages of 10 and 30, and that the adolescent sex/gender difference is robust around age 14 in all sociocultural regions [13]. This trend has also been demonstrated in a meta-analysis of studies in Japanese samples [41].

If there are gender differences in self-esteem, the effectiveness of emotional support can differ by sex/gender. Generally, adolescent women are more likely than adolescent men to seek support from peers when stressed and adjust their emotions by receiving emotional support [42]. Additionally, women are more likely than men to show gratitude because they prioritize creating and sustaining intimate relationships [43]. Considering the above, it is necessary to take into account the influence of gender when investigating the effectiveness of comfort.

1.4 Measurement of self-esteem instability

Two general procedures may be used to assess SEI. The first is Kernis et al.’s experience sampling method [5]. In this procedure, state self-esteem is measured multiple times for several days, the standard deviation of the means is calculated, and the result is used as an index of SEI. Researchers generally administer surveys once or twice daily for one or 2 weeks. This procedure of measuring SEI has been called the “gold standard” [44] and has been used in many studies.

The experience sampling method has advantages as it assesses the variability of self-esteem in a naturalistic context. However, it may cause an error in the SEI score because it is dependent on life events experienced during the measurement period. Moreover, recent studies have raised concerns about using within-subject standard deviation as an index of SEI because of constraints on bivariate distribution [7, 45]. People with extremely low state self-esteem can only have a low standard deviation, while those with moderate state self-esteem can have a high or low standard deviation.

The second procedure is a directly administered assessment scale developed by Howard [46]: the perceived self-esteem instability (P-SEI) measure. P-SEI, defined as “one’s perception of their own variability in self-feelings” [47], measures SEI by direct assessment via a scale. P-SEI can be more trait-like than SEI measured by the gold standard because it involves self-reflection [47]. Ultimately, Roth and Altmann [48] indicate that researchers must note the risks of distorted responses due to memory and emotional biases, as a direct assessment scale requires participants to reflect on their past experiences.

2 Present study

This study aims to clarify the effectiveness of emotional support (i.e., comfort) according to the level and instability of self-esteem among Japanese university students, focusing on the comfort provided by friends in the midst of negative life events. In summary, this study addresses three research gaps. First, many previous studies on self-esteem instability have focused on American samples in independent cultural contexts [e.g., 2, 5, 7, 8, 46, 47]. However, Japanese adolescents exhibit lower self-esteem compared to samples from other nations [11] as well as a heightened need for emotional ties [23]. Given this, the results from this study are expected to deepen our understanding of psychological issues faced by Japanese university students and help consider cultural commonalities and differences in the functioning of self-esteem levels and instability. Second, the results on the effectiveness of comfort can be complex and contradictory owing to the intervention of various variables. It is unclear which factors exert a stronger influence on the effectiveness of comfort as these factors have not been comprehensively addressed in previous studies. Therefore, this study also measures other variables that affect the effectiveness of comfort, such as gender, intimacy, seriousness, and controllability, and considers their influence when testing the hypothesis described later. Third, recent studies have emphasized the methodological need to study P-SEI and conventional SEI (i.e., the “gold standard” method) together because each may produce different effects [47, 49]. Thus, this study tests its hypotheses using both P-SEI and conventional SEI, and gains new insights and suggestions regarding the methods for measuring self-esteem instability.

This study examined the association between instant emotional reactions to receiving comfort and self-esteem by examining whether SEI moderates the association between self-esteem levels and the effectiveness of comfort. This study formulated the following hypotheses based on each emotional response. First, gratitude is a positive emotional reaction to receiving comfort. Studies suggest that feeling gratitude enhances self-esteem [50,51,52]. Gratitude is experienced when people receive something beneficial or when somebody does something kind or helpful [43]. The degree of gratitude when receiving comfort varies depending on whether the recipient seeks comfort or certain individual factors—individual differences in self-esteem levels and instability. According to the sociometer theory, people whose sociometer has decreased receive comfort from others and try to enhance their self-esteem [3]. Generally, people with low self-esteem feel less acceptable because they have fewer interpersonal resources to enhance their sociometer than those with high self-esteem. Therefore, people with low self-esteem strongly seek others’ comfort after rejection as compared with people with high self-esteem [53]. Additionally, the degree to which people with low self-esteem seek comfort varies with SEI. For example, for people with low self-esteem, higher instability was associated with higher comfort-seeking behavior [53]. Such people likely feel gratitude when receiving comfort because they strongly seek others’ comfort. Therefore, the following hypothesis was formulated:

Hypothesis 1: Higher instability is associated with higher gratitude when receiving comfort in people with low self-esteem.

Second, hostility is a negative emotional reaction to receiving comfort. Research shows that people with unstable high self-esteem are highly likely to have increased sensitivity to negative evaluations from others and experience anger and hostility [5]. Further, people with stable high self-esteem have lower perceived aggression than those with unstable high self-esteem [7]. For people with high self-esteem, frequent changes in feelings of self-esteem suggest that their positive views toward oneself easily fluctuate toward a negative one [46]; such fluctuation are characteristics of unstable high self-esteem, considered a form of fragile high self-esteem [54]. Thus, this study predicts that people with unstable high self-esteem have higher hostility toward receiving comfort because comfort injures their pride, and they feel they have been treated unfairly. Therefore, the following was hypothesized:

Hypothesis 2: Higher instability is associated with higher hostility when receiving comfort in people with high self-esteem.

3 Methods

3.1 Participants

The participants included 241 Japanese university students (62 men, 178 women and 1 non-response), who were provided the study materials in Japanese. The age range of the participants in the study varied from 18 to 27, with an average age of 19.64 (SDage = 1.37). The gender was collected using a self-report based on gender identity and not biological sex. The study participants were the same as Ogawa’s sample [Study 1] [55], who were additionally scored on the K6 scale [56], which is not reported in this study. The present research question differed from that in [55], as this study examined a different topic.

3.2 Research tools

3.2.1 Emotional reactions to receiving comfort from friends

This study assessed emotional reactions to a vignette involving comfort. A vignette was used because experimental vignette methodology can verify causal relationships between independent variables of self-esteem and dependent variables of emotional reactions [57]. Further, vignettes result in higher levels of confidence regarding internal validity than questionnaire surveys asking participants about their comfort experience in a complex event [57]. Negative events used vignette had selected from human relations problems that could easily be encountered by Japanese university students [38]. This vignette had been previously used with Japanese university students who confirmed that they perceived negative events in the vignette as threatening and uncontrollable, ones they cannot avoid on their own [30]. Therefore, this particular vignette was judged as worthy of receiving comfort from others and confirmed content validity. Participants read a vignette involving comfort (Table 1; see Online Resource Table S1 for original Japanese) and completed a questionnaire evaluating their emotions when a friend offered comfort. Jones and Guerrero [58] showed that a comfort’s effectiveness is complicated by the combined effects of comforting messages and nonverbal behaviors. Among them, nonverbal immediacy cues (i.e., nonverbal behaviors) significantly influence the comfort’s effectiveness because they show that the listener is attentive, cares about the person suffering mentally, and is willing to offer support. Thus, this study emphasized nonverbal behavior using a nonverbal-comfort vignette, which eliminated the effects of comfort message contents.

Table 1 A vignette involving consolation
Full size table

Participants rated gratitude (three items; e.g., “reassuring” and “glad;” α = 0.90, ω = 0.90) and hostility (two items; e.g., “irritated at friend A;” α = 0.94, ω = 0.94) using the same items as those used by Ogawa [30]. All items were rated on a six-point scale ranging from 1 (strongly disagree) to 6 (strongly agree). These items were developed based on a preliminary investigation of Japanese university students—a free description type questionnaire about actual emotional reactions to receiving comfort in the past [30]. The types of emotional reactions were classified by two researchers via discussion. Next, three university students and three post-graduate students majoring in psychology have judged the validity of these classifications. The emotional reaction items were developed according to these procedures and therefore, content validity of these items was ensured. In this study, gratitude and hostility were used for testing Hypotheses 1 and 2. The emotional reaction items used in this study (in Japanese) are provided in Online Resource Table S2.

3.2.2 Control variables in the vignette.

As effects other than self-esteem needed to be controlled, participants evaluated the following control variables in the vignette: intimacy with a friend offering comfort, seriousness, and controllability. Intimacy with a friend offering comfort (ten items; e.g., “I have trust in a friend offering comfort;” α = 0.87, ω = 0.86) was assessed using Kaneko’s psychological distance scale [59]. Seriousness was measured with two items (i.e., “I am having trouble when a friend gets angry with me” and “I get hurt when a friend gets angry with me;” α = 0.75, ω = 0.75), while controllability was measured with one item (i.e., “I was able to avoid the situation wherein a friend gets angry with me”). All items were rated on a six-point scale ranging from 1 (strongly disagree) to 6 (strongly agree). When evaluating intimacy, the participants assumed that a friend offering comfort was a same-sex friend. An itemized list of the control variables used (in Japanese) are provided in Online Resource Table S3, and the confirmatory factor analysis for intimacy are provided in Online Resource Table S4.

3.2.3 Self-esteem instability and level measured by the “gold standard” method

The participants completed a daily measure of Abe and Konno’s state self-esteem scale [60] (nine items; e.g., “I feel positive about myself now”) for up to seven consecutive days; this scale was created based on the items from Rosenberg’s self-esteem scale (RSES). These items were scored using a five-point Likert scale ranging from 1 (disagree) to 5 (agree). The participants were asked to respond to questions about how they felt at the moment rather than how they generally felt about themselves. The measures were administered using Google Forms. Over the 7 days, the mean and standard deviation of state self-esteem were calculated. The index of “self-esteem level” (GS-SEL) was used as the mean. However, recent studies have raised concerns about using within-subjects standard deviation as an index of SEI because when reliant on bounded measurements, the standard deviation measures are confounded with the mean [7, 45]. To address this concern, the relative standard deviation proposed by Mestdagh et al. as the index of “self-esteem instability” [GS-SEI; 45] was used. In this study, the scale had an internal consistency of αs = 0.86, 0.86, 0.88, 0.88, 0.90, 0.88, 0.91, ωs = 0.86, 0.86, 0.88, 0.89, 0.90, 0.89, 0.91; in order from day 1 to day 7. Items of state self-esteem (in Japanese) and confirmatory factor analysis at each day are provided in Online Resource Table S5.

3.2.4 Rosenberg self-esteem scale

The Japanese version of the Rosenberg self-esteem scale (RSES; ten items; e.g., “I wish I could have more respect for myself”) [61] translated by Mimura and Griffiths [62] was used to evaluate self-esteem level. These items were scored using a four-point Likert scale ranging from 1 (strongly disagree) to 4 (strongly agree). Five negative items were reverse-scored to compute the total scores of individual participants. In this study, the scale had an internal consistency of α = 0.85, ω = 0.85. Items of RSES (in Japanese and English) and confirmatory factor analysis are provided in Online Resource Table S6.

3.2.5 P-SEI measure

The P-SEI measure [47], which was translated into Japanese by Ogawa (eight items; e.g., “My opinion of myself often changes”) [55], was used. These items were scored using a seven-point Likert scale ranging from 1 (strongly disagree) to 7 (strongly agree). In this study, the scale had an internal consistency of α = 0.84, ω = 0.84. Items of P-SEI (in Japanese and English) and confirmatory factor analysis are provided in Online Resource Table S7.

3.2.6 Structural validity of scales

To increase the transparency of measurement practices [63], scales that comprise four or more items were provided as the supplementary results of confirmatory factor analysis in the Online Resource; psychological distance scale (Table S4), state self-esteem scale (Table S5), RSES (Table S6), and P-SEI (Table S7). Most previous research assumes that RSES (including the state self-esteem scale) is a one-factor structure, and this study also assumes the same and conducted data analysis accordingly. Nevertheless, articles considering the factor structure of RSES show lack of measurement invariance about the one factor model [64], and suggest two distinct positive and negative RSES factors [65, 66]. Indeed, RSES and state self-esteem in this study showed high internal consistency but a poor fit index on the one-factor solution (Tables S5 and S6). Therefore, results of the confirmatory factor analysis that assumed a two-factor solution is provided as supplementary information (Online Resource Tables S8 and S9).

3.3 Procedure

The survey was conducted from September 2018 to February 2019. First, the participants read the vignette and responded the control variables in the questionnaire. The participants then responded emotional reactions to RSES and P-SEI, in that order. The number of valid questionnaire responses was 200 (83.0% valid response rate). The “gold standard” method was measured from the day the survey was conducted. If the participants responded more than once a day, only the first response data was used for analysis. However, if a participant missed a day, they responded after the eighth day. Participants who did not provide 7 days of responses and those who took over 2 weeks to complete the survey were excluded from the analysis. Consequently, 116 students completed the gold standard method. Nonetheless, if GS-SEL is exactly equal to the lower or upper bound, the relative standard deviation is incalculable because each individual observation that goes into the computation is equal to the mean [45]. Hence, one student whose GS-SEL was exactly equal to the lower bound led to an incalculable GS-SEI and had to be dropped. Finally, the data of 115 students (Mage = 19.66 years, SDage = 1.50; 32 men and 83 women) were analyzed.

3.4 Data analysis

The statistical software HAD 16.01 [67] was used to analyze descriptive and inferential statistics. Moreover, the statistical software R 4.2.1 and the software package offered by Mestdagh et al. [45] were used to calculate the relative standard deviation. Initially, the mean scores, standard deviations, and correlations for all variables were calculated. Dummy variables of gender were created (1 = men, 0 = women) to test the hypotheses in the multiple regression analyses. Each self-esteem level and instability were mean-centered, then multiplied to create an interaction term. For the “gold standard” method or RSES and P-SEI, a separate regression was performed with dummy variables of gender, the control variable, self-esteem level, SEI, and interaction term predicting each emotional reaction. Diagnostic tests using the criteria (VIF < 3) from Zuur et al. [68] revealed no issues with multicollinearity. The significance level of inferential statistics was set at 0.05. The adequacy of data for performing the analysis was checked using power analysis and G*Power software [69]. The priori analysis estimation indicated that a sample size of 103 was required to obtain 80% power of detecting a medium effect size (f 2 = 0.15) at a 0.05 probability level. The post hoc analysis estimation indicated that the statistical power for a sample size of 115 was 0.85, which is above Cohen’s recommendations [70].

4 Results

Table 2 presents the mean scores, standard deviations, and correlations for all variable. Regarding the measurement of self-esteem, mean scores of GS-SEL, GS-SEI, RSES, and P-SEI had a normal distribution. GS-SEL was significantly positively correlated with RSES (r = 0.67, p < 0.01), whereas GS-SEI did not correlate with P-SEI. This result suggests that GS-SEI and P-SEI measured different qualities of SEI. Additionally, RSES was significantly negatively correlated with P-SEI (r = − 0.51, p < 0.01).

Table 2 Descriptive statistics and correlations of all variables
Full size table

Table 3 presents the multiple regression analysis examining the effects of GS-SEL and GS-SEI on emotional reactions. Regarding the control variables, gender significantly predicted gratitude (β = − 0.20, p < 0.05); hence, women reported higher gratitude than men. Furthermore, intimacy significantly predicted gratitude and hostility (gratitude; β = 0.33, p < 0.01, hostility; β = − 0.37, p < 0.01). Hence, higher intimacy was associated with higher levels of gratitude and lower levels of hostility. Seriousness significantly predicted gratitude (β = 0.26, p < 0.01), and higher seriousness was associated with higher levels of gratitude. Regarding self-esteem level and instability, GS-SEL, GS-SEI, and the interaction term between them were nonsignificant in all emotional reactions.

Table 3 Regressions predicting emotional reactions for GS-SEI and GS-SEL
Full size table

Table 4 presents multiple regression analysis examining the effects of RSES and P-SEI on emotional reactions. Regarding the control variables, gender significantly predicted gratitude (β = − 0.21, p < 0.05). Women reported higher gratitude than men. Intimacy was significant for gratitude and hostility (gratitude; β = 0.36, p < 0.01, hostility; β = − 0.38, p < 0.01); higher intimacy significantly predicted higher levels of gratitude and lower levels of hostility. Seriousness significantly predicted gratitude (β = 0.18, p < 0.05); higher seriousness was associated with higher levels of gratitude. Controllability was significant for gratitude (β = 0.22, p < 0.01); higher controllability significantly predicted higher levels of gratitude. P-SEI significantly predicted gratitude (β = 0.27, p < 0.01). A higher P-SEI score was associated with higher levels of gratitude. Regarding gratitude, the interaction term between RSES and P-SEI was significant (β = − 0.19, p < 0.05). Simple-slope analysis (Fig. 1) indicated that for people who scored low on the RSES, a higher P-SEI was associated with significantly higher levels of gratitude (β = 0.46, p < 0.01). Contrarily, it was nonsignificant for people who scored high on the RSES.

Table 4 Regressions predicting emotional reactions for RSES and P-SEI
Full size table
Fig. 1
figure 1

Moderating effects of perceived self-esteem instability (P-SEI) on the relationship between the Rosenberg self-esteem scale (RSES) and gratitude. SD standard deviation

Full size image

5 Discussion

This study used P-SEI and conventional SEI to investigate whether SEI moderated the association between self-esteem levels and emotional reactions to receiving comfort. Consequently, the interaction between RSES and P-SEI in gratitude was significant. This result aligns with previous studies and shows that P-SEI analysis needs to control for self-esteem levels [46, 47]. However, other interactions, including the “gold standard” method, were nonsignificant.

For people with low self-esteem, higher SEI was associated with higher levels of gratitude. Accordingly, Hypothesis 1 was supported. This result suggests that people with low and unstable self-esteem have improved self-esteem after receiving comfort from friends. Conversely, people with low and stable self-esteem have difficulty improving it. However, comfort is not wasted on people with low, stable self-esteem because their gratitude score was high overall. This result suggests that the experience of being accepted by friends who provide comfort may improve low and stable self-esteem, better motivating such individuals than others to seek help.

Gratitude was also strongly associated with gender, intimacy, and seriousness; the effect of comfort depended on self-esteem and relational and contextual variables. For example, higher seriousness was associated with higher levels of gratitude when receiving comfort from a friend. Items of seriousness in this study referred to the degree of psychological suffering or stress resulting from a negative event. Therefore, this result suggests that comfort from friends was perceived as more beneficial when coping with high stress from serious situations. Additionally, women reported higher levels of gratitude than men. This study’s findings also suggest that peer relationship processes regarding receiving emotional support differ by gender, in line with these previous studies [42, 43].

Finally, higher controllability was associated with higher levels of gratitude only in the multiple regression analysis examining the effects of RSES and P-SEI. This result suggests that it is probably the comfort from friends that boosted motivation for problem-solving in situations perceived as controllable and led to higher gratitude. However, caution is warranted when considering this result. Generally, it is considered that emotional support (and emotion-focused coping) is more fitting when the situation is less controllable [71]. Nevertheless, there is also a study which shows that emotional support is perceived as highly appropriate and desirable by a recipient regardless of controllability [72]. A recent study points out that the results examining effects of controllability are quite mixed because of measurement, analysis, and study design issues [73]. Furthermore, this study revealed different results across two multiple regression analysis. This may be because the effects of variables change depending on the combination of explanatory variables that are input in the multiple regression analysis [74].

Conversely, SEI was not associated with hostility for people with high self-esteem, thereby rejecting Hypothesis 2. Moreover, hostility was strongly associated with intimacy, corresponding with Ogawa’s findings [38]. Additionally, most of the study participants did not feel hostility (M = 1.57) because “intimacy with a friend offering comfort” received high scores. Therefore, the interaction between self-esteem levels and instability regarding hostility may not have been detected. This suggests that hostility that occurs when comforted is strongly affected by intimacy in a relationship and less affected by the personality characteristic of self-esteem.

Finally, Hypothesis 1 was supported by a direct assessment via P-SEI, but not by the “gold standard” method (i.e., conventional SEI). It is possible that SEI measured using the “gold standard method” had a greater measurement error than a direct assessment because of the life events experienced during the measurement period. Therefore, the P-SEI might be a better index for predicting emotional responses than the “gold standard” method. Additionally, the differences of those results also may suggest that GS-SEI and P-SEI measured different qualities of SEI; as Howard points out, P-SEI is higher trait-like than GS-SEI [47].

6 Limitations and future directions

Notably, the generalizability of the results is limited. The hypotheses in this study were tested using a vignette that required self-perception about emotional reactions. P-SEI with self-reflection might be strongly associated with emotional reactions measured by a vignette, including imagining and thinking about a situation. The dynamics of comforting suggest that face-to-face interactions may be much more complex than vignettes [58]. Further, this study only used a vignette involving nonverbal comfort and trouble with friendships. The effectiveness of comfort differs with types of negative life events, such as an illness, an academic failure, or personal problems [38]. Therefore, future research should examine the relationship between P-SEI and emotional reactions using face-to-face interactions and content involving various troubles. It must also be admitted that the vignettes used in this study may not be entirely for experimental manipulation. This is because a significant portion of the study participants did not experienced hostility as a result of increased intimacy with a friend. In future research, the experimental manipulation’s validity should be examined in a pilot study in accordance with best practices for validating experimental manipulations revealed by Chester and Lasko [75]. Moreover, the effectiveness of comfort can differ with culture-specific influences (such as cultural construal of self, gender roles). This study concluded that comfort provided by close friends generally induces positive emotions, but this may be limited to a sample of Japanese college students in interdependent cultural contexts. Additionally, Japan still has a traditional view of gender roles and greater gender inequality [76, 77]. Thus, the effect size of gender differences may be greater in Japanese samples than in samples from other cultures.

Finally, in the GS-SEI using gold standard method, more than half of the participants dropped out because they were unable to complete repeated measurements, which also happened in previous research [78, 79]. The gold standard method has many problems considering that GS-SEI scores regarding respondents’ feelings and situation at the time requires them to spend a considerable amount of time and effort logging daily measures repeatedly. Therefore, P-SEI is also better than GS-SEI in this regard.

7 Conclusion

The present study provides evidence that the effectiveness of comfort differs by the level and instability of self-esteem. Although people with stable low self-esteem experience less comfort than others, this study showed that comfort provided by friends with high intimacy generally induces positive emotions. In particular, this trend was more pronounced in women than in men. This result is consistent with previous studies emphasizing that close friendship plays a key role in arousing positive emotional experiences [80, 81]. Additionally, gratefulness motivates prosocial responses, strengthens relationship bonds, and enhances feelings of closeness within relationships [81]. The accumulation of these positive emotions may promote the maintenance of close friendships and a high sociometer (i.e., high self-esteem), developing high and stable self-esteem among Japanese students experiencing extended adolescence. This study contributes to the field of social and personality psychology promoting the understanding of the effectiveness of emotional support according to the level and instability of self-esteem among Japanese university students. Additionally, the methodological contributions include encouraging further research using both P-SEI and the conventional gold standard method.