Perceived event impacts of the Tokyo 2020 Olympic Games on residents’ eudaimonic well-being: a longitudinal study of within-person changes and relationships

ABSTRACT The present study investigated 1) the changes in positive and negative event impacts and residents’ eudaimonic well-being before and after Tokyo 2020 and 2) how the social impacts are related to eudaimonic well-being. Three-wave panel data were collected from 1692 Japanese residents before and after the event. We performed linear mixed modeling (LMM) and a random-intercept cross-lagged panel model (RI-CLPM) to investigate the within-person fluctuations in the event impacts and eudaimonic well-being and the within-person relationships. The LMMs showed that only negative event impacts significantly changed after the event. The RI-CLPM revealed that positive event impacts before the event were significantly related to eudaimonic well-being right after the event, and eudaimonic well-being right after the event was significantly associated with positive event impacts two months after the event. The findings indicate that the influence of positive event impacts on residents’ eudaimonic well-being might be only temporal, but also eudaimonic well-being right after the event may have a significant role in positive evaluations of the event two months later. The results illustrated that the government and sport organizations need to pay more attention to contemplate the strategies to increase the eudaimonic well-being of residents through the power of sport to make “sport as an enabler of sustainable development”.


Introduction
The event impacts of the Olympic Games are a considerable research stream due to their significant impacts on the host country (e.g., Mair et al., 2023;Ritchie et al., 2020).Although hosting the Olympic Games requires enormous costs, whether hosting the event influences residents' long-term well-being is still a question.International Olympic Committee (IOC) launched a sustainability strategy in which one of the missions is to value the well-being of people (IOC, 2017b), closely aligned with sustainable development goals (e.g., SDG3: Ensure healthy lives and promote well-being for all at all ages (World Health Organization, 2015).Accordingly, the development of individual well-being through the power of the Olympic Games is an essential target to make "sport as an enabler of sustainable development" (IOC, 2017b, p. 17).Thus, it is essential to evaluate how hosting Olympic Games was associated with the resident's well-being.
For sport event stakeholders, how hosting a mega sport event influences residents' well-being is a significant concern (cf., Inoue et al., 2020).Specifically, due to the development of positive psychology, an understanding of residents' well-being by applying a eudaimonic approach (e.g., Ryan & Deci, 2001) rather than a hedonic approach is a growing interest in the recent event impact literature (e.g., Sato et al., 2022).In particular, a eudaimonic approach can play an essential role in sustainable policy development (cf., Kjell, 2011), and it has a significant relationship with the Olympic core values (e.g., Papaioannou, 2017).The rationales indicate the importance of understanding how the perceived impacts of Tokyo 2020 were changed over time and how those are associated with residents' eudaimonic well-being.
Scholars traditionally examined residents' perceptions before and after the events (e.g., Gibson et al., 2014) while rarely arguing the reciprocal or causal relationships between event impacts and residents' psychological outcomes (e.g., Pfitzner & Koenigstorfer, 2016).The limited evidence may obscure whether hosting a mega-event will certainly lead the host residents to greater well-being.Since longitudinal data provide rich information for statistical inference to examine reciprocal effects or causality between variables (Usami, 2021), we collected three-wave panel data with a relatively large sample size to examine the reciprocal relationship between event impacts and eudaimonic wellbeing.The present study advanced to examine the reciprocal relationships between event impacts and eudaimonic well-being at the within-person level by applying a random intercept cross-lagged panel model (RI-CLPM), allowing to infer reciprocal or causal associations at a within-person level and control for stable trait factors (e.g., individual differences).
The aims of the present study are two-fold.First, we aimed to investigate how the residents' perceptions (i.e., event impacts & eudaimonic well-being) have changed from pre-to post-event at the within-person level.More importantly, another aim of the study was to test how positive and negative event impacts and eudaimonic well-being are associated with each other at the within-person level.Consequently, the research applying a eudaimonic approach contributes to developing a eudaimonic perspective in sportevent research by understanding the relationships between the Tokyo 2020 impacts amid a social crisis and the residents' eudaimonic well-being.

A eudaimonic approach
Since the emergence of the positive psychology movement (Seligman & Csikszentmihalyi, 2000), researchers have been inspired to examine the consumers' well-being in sporting events (cf., Inoue et al., 2020).For instance, scholars have increasingly investigated the relationship between sport event impacts and residents' well-being (e.g., Mutz, 2019;Schlegel et al., 2017).Positive psychology originated from ancient Greek philosophy, such as Aristotle's Nichomachean Ethics (Ross, 1980), claiming that well-being is associated with the highest human good (cf., Ryan & Deci, 2001).The perspective is also known as Eudaimonia (Huta & Waterman, 2014;Ryff, 1989).Eudaimonic well-being is distinguished from pursuing a pleasant psychological state or satisfying appetites, a frequently applied well-being construct (e.g., subjective wellbeing), known as Hedonia (cf., Huta & Waterman, 2014).Although a hedonic approach (e.g., Diener, 1984) is rampant in sport management, the present study applied a eudaimonic approach (e.g., Ryff, 1989) to understand the impact of Tokyo 2020 on residents' well-being.
Researchers recently argued the importance of applying a eudaimonic approach in event or tourism research (cf., Filep & Laing, 2019;Vada et al., 2020).Eudaimonism, centered on virtuous activities for the highest human good (Tiberius & Hall, 2010), aligns more with sustainable social and personal development than hedonism.As a previous longitudinal study (Joshanloo, 2019) found that eudaimonic well-being lasts longer than hedonic well-being, it fits with a research question to test the longitudinal impacts of a mega-event on residents' well-being, offering insights for sustainable policy development (Kjell, 2011).While sport event studies (e.g., Schlegel et al., 2017) have primarily been concerned with the hedonic nature of psychological experiences (e.g., positive affect, satisfaction), applying a eudaimonic approach can contribute to understanding the long-term impacts of mega-events on residents' wellbeing.
Eudaimonia also has a significant relationship with the Olympics.IOC has positioned that excellence, friendship, and respect are the core Olympic values to promote sustainable social development through sport (cf., IOC, 2017a), and researchers indicate the significant relation of the values with Aristotle's idea of eudaimonia (e.g., Loland, 2006;Papaioannou, 2017).For instance, friendship is one of the most essential virtuous attributes for reaching eudaimonia (Sherman, 1987).As the education of the values is an essential mission for the host country, Olympic values education has been implemented in the host countries (Binder, 2012).For instance, the Japanese Olympic Committee generated an Olympic value education program and offered it to all public schools in Tokyo in 2019 (Tokyo Organising Committee of the Olympic and Paralympic Committee, 2017).As consistent with the emphasis of the Olympics, the present study applied a eudaimonic approach.
In particular, the present study adopted Ryff's model of psychological well-being (Ryff, 1989) as a central theory of a eudaimonic approach.Ryff (1989) developed the model of psychological well-being to evaluate individual life from six eudaimonic dimensions: personal growth, positive relations, autonomy, environmental mastery, meaning in life, and self-acceptance.The theory of eudaimonia explains that gender, age, and social factors are significant factors for eudaimonic well-being (Huppert, 2009;Ryan & Deci, 2001;Ryan et al., 2008).For example, women showed higher eudaimonic well-being than men (Ryff & Heidrich, 1997), and age is positively associated with eudaimonic well-being (Ryff & Keyes, 1995).The present study also examined the changes in the residents' eudaimonic well-being before and after the event and the potential moderating factors on the changes in residents' eudaimonic well-being over time.

Event impacts of mega-events
Scholars have evidenced that mega-events (e.g., Olympic Games, FIFA World Cup) have variously impacted the host community and residents positively and negatively (e.g., Gibson et al., 2014;Kaplanidou et al., 2013;Kim et al., 2006).The positive impacts of the event may be offset or even outweighed by the negative event impacts, which may, in turn, influence residents' psychological outcomes negatively.Thus, it is critical to concurrently include both positive and negative event impacts in the investigation (cf., Mair et al., 2023).Regarding the conceptualization of positive and negative event impacts, the present study focused on socioeconomic aspects of event impacts due to their significant associations with well-being-related outcomes (e.g., Kaplanidou et al., 2013;Sato et al., 2022).In particular, we selected the subcomponents of positive and negative impacts from Kim et al. (2015) and Balduck et al. (2011) with some modifications to adjust the Tokyo 2020 context.
In the present study, we operationalized positive event impacts by four subcomponents consisting of 1) economic benefits, 2) cultural interest, 3) community pride, and 4) image enhancement.We eliminated the subcomponent of community development (Kim et al., 2015) and interest in a foreign culture (Balduck et al., 2011) because the items are similar to the items of cultural interest and image enhancement.Moreover, the operationalization of negative event impacts comprised 1) economic costs, 2) COVID-19 risk perception, and 3) disruption and congestion.We modified the security risks (Kim et al., 2015) to the COVID-19 risk perception to adjust the context where Tokyo 2022 did not have tourists from abroad.Also, we excluded traffic-related variables and price increases (Balduck et al., 2011) by considering the situation during the global pandemic.
Sport-event impact research often compares residents' perceptions before and after the event.Gibson et al. (2014) found that residents' social capital decreased while their psychic income increased after the 2010 FIFA World Cup.Balduck et al. (2011) examined the Tour de France's impacts, revealing that economic benefit and cultural interest did not increase, but disorder and conflict significantly decreased post-event.Furthermore, Japanese samples regarding the 2019 Rugby World Cup games found that the positive impact perceptions were relatively stable, with decreased negative impact perceptions (Oshimi et al., 2021).The previous findings could be attributed to the negativity bias, where people are generally more sensitive to negative event consequences than positive ones (Ito et al., 1998).Thus, residents tend to over-evaluate the negative impacts before the events, while the evaluations of positive impacts remain relatively stable.Moreover, adaptive forgetting (Kraemer & Golding, 1997), akin to the saying "Danger past, god forgotten", can predict that the heightened negative impact perceptions will decrease right after the Games but not positive impacts because they are not heightened before the event.Thus, we expect that positive impact perception would be constant over time, while the negative impact would decrease right after the Games, and the declined level would be maintained at the follow-up.
In addition, as Olympics core values align with some eudaimonic well-being indicators, such as excellence and friendship (cf., Huta & Waterman, 2014), we expect that the operation of the Olympic Games would positively influence the residents' eudaimonic well-being after the Games (From T1 to T2) and sustain the level at the follow-up due to the stability of eudaimonic well-being contrasts with hedonic well-being.
Nonetheless, previous research suggests some moderators, such as gender (e.g., Gibson et al., 2014), age (e.g., Kim & Morrison, 2005), geographic location (e.g., Karadakis & Kaplanidou, 2012;Sung et al., 2021), on the changes of residents' perceptions in event impact research.For instance, males showed higher psychic income than females (e.g., Gibson et al., 2014), proposing positive changes in resident perceptions would occur more in male residents as men may have more interest in sport (Pelak, 2005).Additionally, Kim and Morrison (2005) found that younger participants had positive changes in the perceived image of the 2002 FIFA World Cup after the event.The result points out that younger residents may have more cognitive flexibility (Wang et al., 2021) in their perceptions of sport event; therefore, we expect that younger residents would have positive changes in their perceptions compared to older residents.Lastly, the host city residents showed positive changes in their subjective well-being than non-host city residents (Sung et al., 2021) since the host city residents can feel more impacts of hosting the event, which can be explained by the effect of geographic proximity (e.g., Nohrstedt & Weible, 2010).Accordingly, we also examined the moderating factors (e.g., gender, age, host vs. nonhost city residents) on the changes in the residents' perception to replicate the previous findings.

Hypothesis 2a-c:
The positive changes in event impacts (increase for positive impact but decrease for negative impact) and eudaimonic well-being would be observed more for H2a) male (vs.female), H2b) younger (vs.older), and H2c) host-city (vs.non-host city) residents.
The present study also included knowledge about COVID-19 as a potential moderator.Knowledge is an individual's cognitive resource to process information (e.g., Sinkula, 1994); thus, knowledge about COVID-19 refers to the cognitive resource to process information relating to the infectious disease in the present study.The elaboration likelihood model (Petty et al., 1983) explains that individuals tend to make their evaluations on a careful process of reasoning (central route) when they have enough cognitive resources, including knowledge.In contrast, people with insufficient cognitive resources tend to use more automatic thought processes (peripheral route) in evaluations.Whereas the model assumes that individuals with sufficient knowledge would make more stable evaluations, the individual with insufficient knowledge will be more easily affected by the information regardless of its quality.As COVID-19 was a significant factor influencing Japanese residents' supportive attitude toward Tokyo 2020 (Kyodo News, 2021), we expected that people with sufficient knowledge about COVID-19 would show more stable evaluations of the event impacts during the pandemic, while people with insufficient knowledge would demonstrate fluctuating evaluations of the event.

Hypothesis 2d (H2d):
The positive changes in positive and negative impacts and eudaimonic well-being would be observed only for those with insufficient (vs.sufficient) COVID-19 knowledge.

Mega-sport event impacts and residents' eudaimonic well-being (bottom-up theory)
The bottom-up theory of well-being posits well-being is derived from the accumulation of positive life moments, while the top-down theory postulates that individuals have a predisposition to experience events and circumstances positively or negatively (Diener, 1984).As hosting a mega event can boost host residents' well-being through bringing positive economic, sociocultural, and environmental impacts to the community (e.g., Balduck et al., 2011;Sato et al., 2022), the event impacts are significant social factors that can enhance residents' well-being.Also, social factors are theoretically influential on eudaimonic well-being since they are essential to fulfilling basic human needs (Ryan & Deci, 2001).Given that hosting the Olympic Games has the potential to facilitate social development and changes (e.g., Ritchie et al., 2020), the event impacts would spill over to the residents' eudaimonic well-being.
Economic development is traditionally linked to mental well-being (cf., Dolan et al., 2008) as a longitudinal study demonstrated the significance of economic development on eudaimonic well-being (Kaplan et al., 2008).Hosting mega-events can facilitate economic development in the host city (e.g., Lee et al., 2017), which was significantly associated with the residents' eudaimonic well-being (Sato et al., 2022).Scholars also found that the sociocultural impacts (e.g., opportunities to meet new people and cultures) had a significant relationship with residents' eudaimonic well-being (Sato et al., 2022).Community pride and image enhancement are also linked to social capital and cohesion, significant social factors relating to residents' psychological outcomes (e.g., Kaplanidou et al., 2013).In contrast, economic cost (Kim et al., 2015) and disruption and congestion (Balduck et al., 2011) have been previously adopted to examine the negative impacts on the residents' attitude toward the host-event community (e.g., Al-Emadi et al., 2017;Ouyang et al., 2019).The study included the COVID-19 risk perception as Tokyo 2020 was operated during the pandemic.Risk perception is a component of negative event impacts (e.g., Kim et al., 2015) and has been examined in the relationship with sport event participants' attitudes and a eudaimonic side of well-being (e.g., Carroll et al., 2014;Kinoshita & Matsuoka, 2023).
Scholars have applied the bottom-up spillover theory (Neal et al., 1999) to point out that the positive impacts in a specific life domain (e.g., leisure, work, family, etc.) can spill over to the overall life evaluation.Indeed, recent research found that event impacts right after the 2019 Rugby World Cup were significantly associated with residents' eudaimonic well-being two months later (Sato et al., 2022).The theoretical proposition of eudaimonia, where social factors play a significant role in influencing eudaimonic well-being (Ryff, 1989), suggests that the event impacts may spill over to the residents' psychological outcomes (e.g., Ouyang et al., 2019;Sung et al., 2021).Overall, we expect that the positive and negative event impacts would predict subsequent eudaimonic well-being.

Hypothesis 3 (H3a):
A Positive event impacts would be positively associated with subsequent eudaimonic well-being.

Hypothesis 3b (H3b):
Negative event impacts would be negatively associated with subsequent eudaimonic well-being.

Research context
The Tokyo 2020 Summer Olympic Games were held from July 23rd to 8 August 2021.The Olympic Games were postponed to 2021 due to the COVID-19 pandemic and also held without public spectators because of the declaration of the emergency statement in Tokyo.Such rescheduled and no spectators Olympic Games were the first and only Olympic Games in history.The total spending was approximately $20 billion, the most expensive Olympic Games ever (Cervantes, 2021).Due to the situation, Tokyo 2020 was held despite the opposition that about 80% of residents dissented from hosting the Olympic Games (Kyodo News, 2021).

Event impacts
For the positive event impact, we adopted seven items from Kim et al. (2015) to measure economic benefits (4 items) and community pride (3 items) and six items from Balduck et al. (2011) to measure cultural interest (3 items) and image enhancement (3 items).For the negative impact perception, we adopted seven items from Kim et al. (2015) to measure economic costs (3 items) and risk perception (4 items) and three times from Balduck et al. (2011) to measure disruption and congestion (3 items).Event impacts were assessed using a 7-point Likert scale (from 1 = Strongly disagree to 7 = Strongly agree) in the present tense at T1 but were asked using past perfect tense T2 and T3.

Eudaimonic well-being
Eudaimonic well-being was measured using the 6-item psychological well-being scale from the short form of the Mental Health Continuum (Keyes, 2002).It consists of six items from each dimension of Ryff's model of psychological well-being; as such, the scale (Keyes, 2002) and the Japanese version was validated in a past study (Sato et al., 2022).The measurement applies a 6-point Likert scale (from 1 = Never to 6 = Everyday) with the stem "During the past month, how often did you feel . .." (Keyes, 2002).

COVID-19 knowledge
We adopted ten questions from Fridman et al. (2020) research for the data analysis.Original questions consist of 12 truth/false questions in their study, and we excluded two items (except items #1 & #6) for the present data collection since the nature of the items does not fit the study context.Based on Fridman's (2020) correct rate of the questions (85%), we categorized the participants with more than 90% correct rate (≤9 out of 10 questions) as sufficient knowledge and others (≥8 out of 10) as insufficient knowledge holders.As a result, 57.5 % of the participants (n = 897) were sufficient, and 42.5 % (n = 664) were insufficient knowledge holders.

Participants and procedure
A three-wave panel data (T1 = a week prior to the Games; T2 = right after the Games; T3 = two months after the Games) was collected from residents in the entire Japan using a market research firm's online panel.The time span of the data collection was decided based on a previous study (Sato et al., 2022).The authors employed a stratified sampling based on the residents' 1) age, 2) gender, and 3) geographic locations to reflect the Japanese population (Statistics Bureau of Japan, 2020).We collected 2066 (T1), 1845 (T2), and 1692 (T3) participants.The present study only included the respondents who consecutively completed the package of questionnaires from the first (T1) to the third (T3).We excluded the participants (n = 131) who responded too quickly (2 sec > per item) for further analyses; as a result, 1561 individuals (4683 observations) were used for the data analysis.
To check the difference in demographic characteristics between the attrition samples and the final samples, we conducted the chi-square tests for the categorical variables and the independent sample t-test for the continuous variable.The results showed that significant differences in gender (χ 2 = 24.17,df = 1, p < .001)and age (M attrition = 39.64,M final = 47.66;t (2064) = −13.17,p < .001)while not in the geographic location (χ 2 = 8.39, df = 7, p = .30).Therefore, females and younger populations were more likely to withdraw from the follow-up surveys.The minimum sample size for the present study was determined by the item-toresponse ratio of 1:5 (Hair et al., 2005).The study requires at least 435 cases (87 items).

Data analyses
A confirmatory factor analysis was performed to evaluate the specified factor structure and model fit indices based on Kline's (2005) criteria, suggesting that the values greater than .90for the Comparative Fit Index (CFI) and Tucker-Lewis Index (TLI) and lower than .08 for Root Mean Square Error of Approximation (RMSEA) and Standardized Root Mean square Residual (SRMR) can be considered as adequate model fit.We also calculated the average variance extracted (AVE) and composite reliability (CR), according to Fornell and Lacker (1981): AVE > .50 and CR > .70.
Next, we tested the invariance of the measurement across three time points.The measurement model without any restrictions was applied as a baseline configural model.We compared the baseline model with three restricted models: 1) metric invariance, fixing the factor loadings to be equal across times; 2) scalar invariance, fixing the factor loadings and the intercepts to be equal across times; and 3) strict invariance, fixing the factor loadings, intercepts, and residual variances.The criteria were applied based on the following scholars' suggestions: TLI > 0.90 and ΔTLI < 0.05 (Little, 1997); ΔCFI <.005 and ΔRMSEA <.010 (Chen, 2007).
A linear mixed-effects model (LMM) was performed for each dependent variable (averaged value of four positive impact subcomponents, three negative impact subcomponents, and six eudaimonic well-being items).The models included participants' ID as a random effect on intercepts and time (T1, T2, & T3), gender (male & female), age, host city residents (host & non-host city residents), and COVID-19 knowledge (sufficient & insufficient) as the fixed effects.We performed post-hoc pairwise comparisons using Tukey adjustment with Kenward-roger degrees-offreedom calculation.
Although we hypothesized that event impact perceptions are positively associated with subsequent eudaimonic well-being, the possibility of the opposite direction, where eudaimonic well-being influences event impact perceptions, cannot be ruled out.Positive psychology theories argue that happy people would have more positive outcomes (cf., Fredrickson, 1998;Lyubomirsky et al., 2005) since positive psychological states extend and deepen awareness toward the future, while negative psychological states intensify immediate concerns.Therefore, we applied a RI-CLPM (Hamaker et al., 2015) that incorporates multiple indicators (Mulder & Hamaker, 2021) to examine the reciprocal effects of event impact perceptions and eudaimonic well-being over time.The model accounts for stable individual differences in each variable by including the autoregressive effects and cross-lagged effects among the variables at the within-person level (e.g., Selig & Little, 2012).Relative to the general cross-lagged panel model (GCLM; Zyphur et al., 2020), the RI-CLPM can clearly separate within-person fluctuations at each time from stable individual differences (i.e., stable trait factors are uncorrelated with within-person fluctuations; Usami, 2021), reducing bias from factors such as individual personality or trait-like variables.Since the test of within-person fluctuations is a significant indicator of causal inference (Hamaker, 2012), the RI-CLPM allows us to be closer to understanding the causal relationships compared to general structural modeling.R software version 4.3.0performed all statistical analyses (R Core Team, 2019).

Measurement model
The measurement model, including all 24 first-order factors and six second-order factors, demonstrated acceptable model fit indices (χ2/df = 12799.86/3543= 3.61; CFI = .93;TLI = .93;RMSEA = .04;and SRMR = .05)and the AVE and CR values for each second-order and firstorder factor (Table 2).The AVE values for each construct are greater than the squared correlations between the construct and other constructs, providing evidence of discriminant validity in the model (Fornell & Lacker, 1981).The detailed information on descriptive statistics (e.g., mean, SD, alpha) and each psychometric property is presented in Tables 1 and 2, respectively.

Changes in event impacts of Tokyo 2020 and residents' well-being (H1 & H2)
Changes in the positive and negative event impacts of Tokyo 2020 and their eudaimonic well-being were estimated by LMMs (Table 4).In line with H1a, no effect of time was found for positive event impacts (Table 4).However, the model revealed significant effects of time on negative event impacts; Negative impacts at T2 (Estimate = −.32,p < .01)and T3 (Estimate = −.53,p < .01)showed lower values than T1.A post-hoc comparison also revealed that negative impact at T3 was not significantly different from T2 (Estimate = −.21,p = .13),indicating that the significant decline of negative impact perceptions was maintained from T2 to T3, supporting H1b.For eudaimonic well-being, we only found a marginal level of significance at T3 (Estimate = .23,p = .06).Thus, H1c was not supported.For H2, the models also tested the moderating factors on the change of the dependent variables across times.We only found significant moderating roles of gender (H2a) and COVID-19 knowledge (H2d) in the models (Table 4), partially supporting H2.The interaction between negative impact (T3) and female was marginally significant (estimate = −.09,p = .06),indicating that the rate of the decline from T1 to T3, but not from T1 to T2, was slightly larger for female residents (Figure 1a).We also found a significant relationship between eudaimonic well-being (T3) and female (estimate = −.13,p < .05),revealing significant difference from T1 to T3 for only male residents (male: estimate = −.13,p < .01;female: estimate = .01,p = 1.00).The results suggest that the eudaimonic well-being of females was stable over time, while that of males increased (Figure 1b).
Next, regarding the moderating role of knowledge about COVID-19, we found a marginal level of significant interaction between insufficient knowledge and positive event impact (T2) (estimate = .09,p = .06)and (T3) (estimate = .09,p = .06).Additionally, the post-hoc comparison showed significant increases in positive impacts from T1 to T3 (estimate = .10,p < .05)but not from T1 to T2 (estimate = .05,p = .69)and T2 to T3 (estimate = .05,p = .71)for insufficient knowledge holders (Figure 1a).The results suggest that the level of improvement from T1 to T2 and from T1 to T3 for those with insufficient knowledge was significant compared to those with sufficient knowledge.Lastly, the model revealed a marginally significant interaction between insufficient knowledge and eudaimonic well-being (T2) (estimate = −.09,p = .07),demonstrating a significant increase in eudaimonic well-being from T1 to T2 for those with sufficient COVID-19 knowledge (estimate = .09,p < .05).The findings indicate that the residents with sufficient knowledge about COVID-19 experienced a significant increase in eudaimonic well-being right after Tokyo 2020, but it was only temporal (Figure 1d).

Discussion
The present study investigated 1) the changes in event impacts of Tokyo 2020 and eudaimonic well-being (H1 & H2) and 2) how the positive and negative event impacts perceptions and eudaimonic well-being are associated across three time points (H3).The results showed that residents' positive impacts and eudaimonic well-being were relatively stable across time (H1a), while the negative impacts significantly decreased over time (H1b).Also, gender (H2a) and COVID-19 knowledge (H2b) significantly moderated changes in  residents' perceptions.In addition, only the positive impacts before the event predicted eudaimonic well-being right after the Games (H3a), while eudaimonic well-being right after the event was associated with positive event impacts two months after the event.As a eudaimonic approach can provide valuable insight into understanding sustainable personal and social development through the Olympic Games (Kjell, 2011;Loland, 2006;Papaioannou, 2017), this study contributes to the mega-event literature by elucidating how the impacts of Tokyo 2020 were perceived and their relations to residents' well-being.
The findings provide significant theoretical contributions.Consistent with H1a, LMMs revealed that positive event impacts remained stable across times, while negative event impacts declined from T1 to T2.Notably, the significant decline in negative impact perceptions was maintained at the follow-up (from T2 to T3).Residents often feel a sense that something "important" is happening before mega-events (Chalip, 2006), making them sensitive to the impacts of prior events, especially negative ones as explained by negativity bias (Ito et al., 1998).The pre-event pandemic situation may have intensified the residents' attention to the negative impacts, and the adaptive forgetting (Kraemer & Golding, 1997) gradually decreased the heightened negative impact perceptions over time.Thus, the results support that the residents may have over-evaluated their negative, but not positive, event impacts only at T1 (Balduck et al., 2011;Oshimi et al., 2021).
The results testing H1c found that eudaimonic well-being has not changed over time.However, the p-value for the change from T1 to T3 was close to the significant level (p = .06).It is noteworthy that there was a 0.23-point increase on the 6-point Likert scale from T1 to T3.Moreover, a supplemental analysis ruled out the influence a pandemic-related policy (i.e., the declaration of emergency statement) on the changes in eudaimonic well-being (Appendix B).Since eudaimonic well-being is theoretically a relatively stable global life evaluation compared to hedonic wellbeing (e.g., Joshanloo, 2019;Tiberius & Hall, 2010), the time span of the present study (about three months from T1 to T3) might not be long enough to examine the changes.Nevertheless, the findings suggest the difficulty of improving residents' eudaimonic well-being solely through the power of hosting a mega-sport event during a crisis period.
Second, the examination of moderating factors on the residents' perception changes over time revealed significant gender moderation effects on the changes in negative impacts and eudaimonic well-being over time (H2a).Contrary to our hypothesis, the results showed a slightly larger decline in negative event impacts for women (Figure 1a), which can be attributed to the evidence that women may exhibit a stronger negativity bias (Ito & Cacioppo, 2005).They tend to have a less favorable attitude toward negative messages but a more favorable attitude toward positive messages (Putrevu, 2010).Thus, women might have been more influenced by negatively framed messages before the event but affected by positively framed messages during the Games, as the media frequently underlies the positive side of Tokyo 2020 during the event (Peaslee & Berggreen, 2012).Moreover, while eudaimonic well-being was relatively stable over time among women, it increased among men from T1 to T3 (Figure 1b), supporting our hypothesis.The results indicate that men's eudaimonic well-being T2 might be more affected by Tokyo 2020 than women since men generally showed more interest in sport (Pelak, 2005), However, inconsistent with H2b and H2c, the present results found that age and host or non-host city residents were not significant moderators on the residents' perception changes, which were anticipated based on the cognitive flexibility of younger people (Wang et al., 2021) and geographic proximity effect (Nohrstedt & Weible, 2010).The current results may be found because Tokyo 2020 was held on a unique occasion, such as no spectators and no tourists due to COVID-19.The findings suggest that the unusual event operations during a crisis period may have buffered the influence of cognitive flexibility and proximity effects on the changes in residents' perceptions.
Regarding the moderating role of COVID-19 knowledge (H2d), we found a significant interaction with the changes in positive impact perceptions.(Figure 1c).The results demonstrated significant increases from T1 to T3 for insufficient knowledge holders, while it was stable across times for those with sufficient knowledge.Knowledge is a cognitive resource that influences individual attitudes and behaviors since it is associated with information processing (Sinkula, 1994).For instance, people with sufficient knowledge are related to adhering to social distancing (Fridman et al., 2020).According to the elaboration likelihood model (Petty et al., 1983), consumers with enough cognitive resources often implement a central route for information processing, resulting in more stable decision-making and attitude.In other words, the individuals are less affected by additional information (e.g., new knowledge).Hence, the results support our hypothesis.
However, contrary to our hypothesis, individuals with sufficient knowledge showed temporal improvement (from T1 to T2) in eudaimonic well-being, while the eudaimonic well-being of those with insufficient knowledge was stable across time (Figure 1d).While we expected that sufficient knowledge about COVID-19 is related to the more stable residents' perceptions, the effect may be only applicable to the evaluation of the Olympic Games.A recent study revealed that education level is a positive antecedent of eudaimonic well-being during the pandemic because it reduces financial stress (Jiang et al., 2022).Accordingly, as individuals with sufficient knowledge about COVID-19 might feel less stress during the event, it might increase their eudaimonic well-being.
Third, for H3a and H3b, a RI-CLPM revealed that positive event impacts (T1), but not negative event impacts (T1), were significantly associated with residents' eudaimonic well-being (T2).In contrast, both positive and negative event impacts (T2) were not associated with eudaimonic well-being (T3).Therefore, the residents who had perceived higher positive impacts right before Tokyo 2020 (T1) were positively associated with their eudaimonic well-being only right after Tokyo 2020 (T2).The positive relationship is somewhat consistent with the bottom-up theory of well-being, although the relationship was only temporal.This finding suggests that the influence of event impacts of Tokyo 2020 on residents' eudaimonic well-being was not strong.
The temporal relationship between positive impacts and eudaimonic well-being aligns with the perspective proposed by Gutman (1982), indicating that the event impacts would positively influence residents' evaluation when the positive perceptions were improved at the end of the event.Although positive impact perceptions did not significantly improve, the subcomponents (cultural interest, community pride, and image enhancement) have increased after Tokyo 2020, except for economic benefits.However, because the three subcomponents of positive impacts were stagnant from T2 to T3 (Appendix A), the positive impacts (T2) might not be associated with eudaimonic well-being (T3).Moreover, the results showed relatively lower positive impacts and higher negative impacts compared to previous research (e.g., Balduck et al., 2011;Oshimi et al., 2021), possibly due to dissent among Japanese residents regarding hosting the event (Kyodo News, 2021).This suggests ease in perceiving negative impacts but challenges in perceiving positive impacts.Consequently, those who perceived higher positive impacts before Tokyo 2020 May have high sport involvement (e.g., sport fans).The facts also support that positive impacts (T1) were positively associated with eudaimonic well-being (T2).
Lastly, eudaimonic well-being (T2) was significantly associated with positive event impacts two months after Tokyo 2020 (T3).The results confirm that happy people get more positive outcomes (cf., Fredrickson, 1998).In other words, people with greater eudaimonic well-being right after the Games, a concept focusing on virtuous activities for the highest human good, evaluated the event impacts more positively two months after the Olympics.As we conceptualized well-being only from a positive psychological perspective (e.g., lack of negative well-being-related variables such as anxiety), negative event impacts might not be influenced by eudaimonic well-being.Hence, the results suggest that eudaimonic well-being may be an important construct to be associated with positive event impacts.
In summary, our study makes significant theoretical contributions to the literature.First, we explored how hosting a mega-sport event influences residents' well-being by applying a eudaimonic approach, an aspect rarely examined in sport management literature.Specifically, our findings indicate that the bottom-up spillover effect of Tokyo 2020 on residents' eudaimonic well-being was temporal, while the level of eudaimonic well-being is an important predictor of positive event perceptions.The results suggest that not only hedonic well-being (i.e., positive emotions; Fredrickson, 1998) but also eudaimonic well-being can foster individuals' attention to the positive aspects of events.Second, the study employs relatively rigorous methodologies and a unique data source, contributing to the theoretical development in the field since a rigorous methodology plays an essential role in developing theoretical knowledge.Specifically, many studies on the topic do not apply longitudinal data collection, a key solution for addressing common method bias (Rindfleisch et al., 2008).Another significant issue is that previous research often neglects individual fluctuations in the analysis, such as relying on traditional repeated-measure ANOVA.It can be a significant risk of inflating the risk of type-1 errors because repeated-measure ANOVA does not consider the correlations between the repeatedly measured variables (Quene & van den Bergh, 2008).Lastly, the study suggests COVID-19 knowledge plays a significant role in improving residents' eudaimonic wellbeing.As previous studies indicate the significance of education level on eudaimonic well-being (Jiang et al., 2022), appropriate knowledge can enhance coping strategies for stressors.The findings also suggest that future research could examine how other knowledge (e.g., Olympic values) is associated with the residents' eudaimonic well-being.Based on the findings, we also suggest several practical implications.

Practical implications
The findings can provide significant implications.To begin with, eudaimonic perspectives are more influential on individual mental health (Brandel et al., 2017;Kinoshita et al., 2022); thus, enhancement of eudaimonic well-being through the power of hosting sport events would help sustainable personal and social development.Further, the moderating role of gender on the changes in residents' perceptions suggests that targeting women in event promotion, considering their generally lower interest in sports (Pelak, 2005), may mitigate the negative impact perceptions.Also, initiatives such as pre-event documentaries focusing on female athletes could attract women, potentially enhancing positive impact perceptions before the event, which eventually improves their eudaimonic wellbeing after the event.The results also suggest that providing sufficient crisis-related knowledge may positively influence positive impact perceptions.Thus, event organizers should consider providing knowledge applicable to cope with residents' stressors and anxiety related to hosting mega-events.
The results further indicated that residents with higher positive event impacts before the event increased their eudaimonic well-being right after the Games.Thus, the event impacts before the Games are more important to residents' eudaimonic well-being than the impacts after the Games.In other words, regarding the impacts of hosting a mega-sport event on residents' eudaimonic well-being, management strategies to increase positive event impacts before the event may be essential while the media in Japan has broadcasted a number of negative news before the Games (Kyodo News, 2021).For instance, the Olympic values education was mostly implemented for children, but such Olympic-related knowledge or the positive impacts of hosting the Games should be more emphasized through media.As a result, the important information will reach a more general population, not only children, which may result in higher positive impact perceptions and greater eudaimonic well-being after the event.
Lastly, a RI-CLPM revealed that eudaimonic well-being (T2) was significantly associated with positive event impacts (T3).The results suggest that residents' eudaimonic well-being might be a significant antecedent of the evaluation of mega-events.The government and IOC should not overlook the reciprocal effects because eudaimonic well-being is a relatively stable construct (i.e.lasts longer), and it is significantly associated with the Olympic core values (Friendship, Excellence, and Respect).Therefore, the residents' eudaimonic well-being, rather than hedonic well-being, should be an important target in the host country to maintain the positive evaluation of hosting the event.This, in turn, can lead to additional positive impacts on the residents' eudaimonic well-being.To make "sport as an enabler of sustainable development" (IOC, 2017b, p. 17), the government and sport organizations need to prioritize strategies that enhance residents' eudaimonic well-being through the power of sport.

Limitations and future research
First, the present results should be carefully interpreted as Tokyo 2020 is an unprecedented case due to the pandemic.Furthermore, the time span of the study was limited.By adopting a longer time span (e.g. a year) for the pre-and post-event data collection, the change in the perceptions would be more deeply understood.The research adopted statistical procedures to account for individual differences (unobserved heterogeneity) by focusing on withinperson fluctuations.However, the supplemental analysis revealed that the declaration of the emergency statement had an impact on changes in positive impact perceptions (Appendix B).Therefore, we could not completely eliminate the influence of other external factors.It is possibly because the impact of the pandemic has overshadowed that of Tokyo 2020.Also, other COVID-19-related policies or governances as a significant social factor may have explained a greater proportion of the residents' well-being.Future research may need to include such external factors to examine how those are associated with the variables of researchers' interests.
In conclusion, we found that only negative event impacts significantly changed from preto post-Olympic Games, while gender and knowledge about COVID-19 moderated the residents' perception changes.Finally, positive impact perceptions before the Games positively affected eudaimonic well-being right after the event, but the impacts were not maintained.However, well-being right after the event played a significant role in positive evaluations of the event two months later.The present findings suggest how the event impacts are associated with residents' eudaimonic well-being.

Figure 1 .
Figure 1.Visualized line charts for the significant moderating effects in the results of linear mixed modeling.

Figure 2 .
Figure 2. The results for the multiple-indicator version of the random intercept cross-lagged panel model.

Table 1 .
Descriptive statistics and pearson correlations.

Table 3 .
The results for measurement invariance test.

Table 4 .
The results for the linear mixed modeling: fixed effects parameters estimates.