Evaluation of Trust Within a Community After Survivor Relocation Following the Great East Japan Earthquake and Tsunami

Key Points Question How is the movement of internally displaced survivors in the aftermath of a disaster associated with perceived trust towards others within a host community? Findings In this cohort study that included 3250 adults aged 65 years or older, each standard deviation increase in the influx of temporarily relocated survivors within 100 m of a resident’s home address was associated with a decrease in their trust in both people from their community and outside of it. Meaning The findings of this study suggest that opportunities for social interaction between old and new residents of host communities may be crucial for maintaining social trust.


Introduction
The long-term health sequelae of disasters range from mental health problems (eg, depression, anxiety, posttraumatic stress disorder, psychological distress) to functional decline (eg, cognitive and/or physical disability). [1][2][3][4][5] A critical feature of communities that enable survivors to remain resilient in the face of adversity is the strength of social connections, known as social capital. 4,[6][7][8] Tightly knit communities are more effective in mobilizing assistance to the neediest members of the group, before even first responders have reached the scene, and long before formal mechanisms of insurance can become activated. During the recovery phase of disasters, socially cohesive groups are more effective in giving voice to the needs of the community, and in preventing the exit of members from disaster-stricken areas. 9 Unfortunately, disasters are also commonly associated with major disruptions in social connections through the residential displacement of survivors. Such disruption of social relations affects not only the survivors who become separated from their predisaster communities, but also the host communities, which have to deal with a large influx of internally displaced persons (IDPs).
Previous studies have documented the effects of residential dislocation on the social connections of disaster victims. 10 However, few studies have focused on the impacts on communities at the receiving end of large inflows of IDPs. In the 2011 Great East Japan Earthquake and Tsunami, nearly a quarter of a million residents were forced to relocate to new communities. Japanese media at the time highlighted instances of local friction caused by the large internal flows of disaster victims into nearby communities. 11,12 However, to our knowledge, no systematic attempt has been made to document the impact of mass residential relocation on the quality of social connections in the host community.
In this study, we took advantage of a unique natural experiment stemming from the 2011 disaster. The city of Iwanuma (population 44 000) in Miyagi prefecture, located 80 km to the west of the earthquake epicenter, was one of the field sites for a nationwide cohort study of healthy aging, called Japan Gerontological Evaluation Study (JAGES). Our study specifically focused on the association of disaster with the erosion of social trust. Generalized trust-the belief that most people, including strangers, can be trusted-represents a crucial component of social capital, facilitating cooperation and collective action for mutual benefit. 13 In turn, generalized trust has been linked to better health outcomes not only in the context of disasters, but in the general population as well. 4,[14][15][16][17][18][19] Previous studies have demonstrated that generalized trust in the community is associated with factors such as residential stability and ethnic homogeneity. 20 Residential stability is recognized as an important prerequisite for developing community trust. Trust develops over time as a result of repeated social interactions (eg, the trading of mutual favors) between members of stable communities. For example, research suggests that Japanese communities reporting the highest levels of generalized trust among residents tend to be the oldest communities in continuous existence based on historical maps. 21 Residential homogeneity (eg, by ethnicity) is also correlated with levels of trust, especially in the United States, where studies consistently reported a negative association between ethnic diversity and social cohesion. 20 Japan is a notably homogeneous society (by ethnicity) owing to over 2 centuries of exclusion of people born outside of Japan (beginning with the sakoku edict of the Tokugawa era). Nevertheless, the 2011 disaster provides a pertinent natural experiment for examining outcomes associated with the mass influx of so-called outsiders into existing communities.
In the literature, a critical distinction is drawn between particularized trust (the belief that certain known others can be trusted) vs generalized trust, which includes trust of unknown groups and strangers. 20 We hypothesized that the influx of IDPs into a community would be associated with a decrease in generalized trust but not particularized trust. However, previous studies suggest that community heterogeneity has broader spillover effects in eroding generalized trust. 20  address knowledge gaps, we explore potential differences between the associations of disaster with generalized vs particularized trust within the host community.

Study Population
Data were obtained from the Iwanuma Study, a part of the JAGES nationwide cohort study In August 2010, around 7 months before the disaster, a census was conducted for every Iwanuma City resident aged 65 years or older identified through the local official residential registry.
In October 2013, approximately 2.5 years after the disaster, the town's residents were recontacted, and an in-home survey was conducted to gather information on the health status and social capital of the survivors.
Informed consent was obtained from participants at the time of data collection. This study excluded respondents who incompletely signed consent forms, and the analytic sample was restricted to the 3250 individuals (1442 men and 1808 women) who were not relocated due to the earthquake (eFigure in the Supplement

Social Trust
To measure perceived changes in trust in others in the wake of the earthquake, in the JAGES 2013 survey study participants from Iwanuma City were asked the following questions: (1) "How did your trust in people in your community change from before the earthquake?" and (2) "How did your trust in people from other communities change from before the earthquake?" Of 3250 nonrelocated participants, 3216 (1787 [55.6%] women) answered the first, and 3210 (1785 [55.6%] women) answered the second question. Possible answers included "became much stronger," "became stronger," "did not change," "became weaker," and "became much weaker."

Number of IDPs in the Host Community
The 2013 survey also asked the following question: "Did you move to a new residence because of the earthquake?" The possible answers were: (1) "No," (2) "The entire district was relocated to temporary housing" (kasetsu jutaku: similar to FEMA-style trailer homes in the US), (3) "Moved to temporary housing, but not by entire district relocation," (4) "Moved to minashi kasetsu temporary housing" (rental housing on the open market), and (5) "Purchased a new home." Because of the disaster, 131 study participants were forced to move to temporary housing, either alone (47 participants) or together with others from their district (84 participants).
Using data from the local residential registry, home address coordinates of everyone in Iwanuma city aged 65 years or older (including those who did not respond to the JAGES surveys) were verified both in 2010 and 2013, enabling us to identify a further 303 relocated residents. Using these home address coordinates for 2013, the geodetic distance between each nonrelocated and relocated participant was calculated, from which the number of IDPs within a 100 m and 250 m buffer zone of each nonrelocated resident was determined. After confirming that the interaction terms between the explanatory variables and sex were not statistically significant, data from men and women were combined in all analyses.

JAMA Network Open | Public Health
Two sets of models were constructed to estimate the association between the number of IDPs in the proximity of nonrelocated residents within 100 and 250 m and perceived change in trust in others. Model 1 is a univariate model. Model 2 controlled for equivalent income, education, marital status, employment status, single residence, the death of a relative as a result of the disaster, the death of a friend as a result of the disaster, and depressive symptoms. SAS version 9.4 (SAS Institute Inc) was used to conduct all analyses. The data were analyzed from July 1, 2019, to January 9, 2020.
All tests were 2-tailed, with a significance level of P < .05.

Results
Of 8576 identified residents contacted to participate in 2010 who were aged 65 years or older, 5058

Discussion
Our study found that after the 2011 Tohoku Earthquake, the influx of IDPs to another community was associated with weakening of both generalized and local trust, suggesting that the concentration of IDPs within a temporary shelter village (as happened in Iwanuma) may have a particularly detrimental effect on social cohesion.
Building trust between residents of a community depends on repeated social interactions over an extended period of time, whereas exposure to outsiders or out-groups can trigger conflict and mistrust. 20 In a 2007 study, Putnam 26 found that the influx of immigrants in communities can spur perceived competition over scarce resources (eg, housing, schools), ultimately resulting in reduced community cooperation and altruism, as well as lower trust not only in people perceived as different, but also in those who are perceived as similar. This study found that internal forced migration after a disaster, even within the same city from 1 district to another, might also lead to the erosion of the trust of nonrelocated residents in people from other communities as well as in people from the same community.
Previously, we reported that relocating IDPs together as a group, as opposed to randomly housing them throughout the community, can be an effective means of preserving social connections and strengthening the resilience of disaster survivors. 10 However, the same policy may also inadvertently promote erosion of trust between older residents of the host community and newcomers.
We have therefore identified a potential dilemma in postdisaster resettlement. Our previous studies 10,27,28 have reported that the resettlement of survivors needs to take into account the preexisting social ties within a disaster-effected community in order to prevent the loss of communality associated with widespread housing destruction. In Iwanuma, the city offered 2 different means of relocation to temporary housing to survivors. People could choose between individual relocation-moving to public housing by a random lottery or seeking housing in the open rental market-or group relocation, in which whole communities would be moved together as a group into prefabricated temporary housing villages (resembling FEMA-style trailer parks in the US).
Families who wanted to escape the emergency shelters as soon as possible selected the individual option, so they could leave the shelters as soon as their number came up on the lottery. However, this mode of resettlement had the unintended consequence of disrupting existing social connections in the community and scattering the residents randomly throughout the trailer settlement. We previously found that people selecting the lottery option reported lower levels of social participation and social support. 10,28 By contrast, people selecting group resettlement were even more likely to be engaged in informal social participation 3 years after the disaster compared with before the disaster. 10 However, as the result of our present analysis suggests, the option of moving large numbers of IDPs together and concentrating them into 1 location may lead to greater friction with established residents of host communities.

Limitations
Several limitations need to be considered while interpreting the findings of this study. First, although we controlled for socioeconomic status, depressive symptoms, and personal disaster experiences, there may be residual confounders that we failed to take into account. Second, the number of nonrelocated participants reporting much weaker trust after the earthquake is quite small (12 participants), resulting in relatively wide 95% CIs around the point estimates. Therefore, the results have to be interpreted with caution. Third, because of the uneven distribution of displaced survivors in the community, we were unable to determine the precise threshold between 8 and 21 IDPs when the erosion of trust began to occur. The results suggest that the resettlement of a few scattered individuals in a community was not associated with changes in on local trust. The erosion of trust seemed to appear when larger numbers of people moved in. Fourth, we do not have information on the residential movements of people younger than 65 years, which might not be correlated with the movement of people aged 65 years or older. On the other hand, two-thirds of the population of the city of Iwanuma were aged 65 years or older before the disaster, and the age structure of IDPs was similar. Fifth, it is not clear how participants define people from their community and people from other communities. By 2013, when the question was asked, IDPs had spent approximately 2.5 years in their new environment. Thus, the respondents may have perceived the displaced population as either people from their own community or as outsiders. Hence, the 2 questions about trust might not have distinguished between particularized and generalized trust, which would also explain the similarity of the corresponding OR estimates. Also, perceived change in trust was measured based on 1 question instead of a multi-item scale, which hindered a more precise assessment of trust levels among the respondents. Sixth, the question of the generalizability of our results needs to be considered due to the relatively low response rate (59%) on the baseline survey. However, previous reports based on the JAGES study confirmed that the demographic profile of the participants is similar to the rest of the residents aged 65 years or more in Iwanuma City. 7,10 Moreover, a 59% response rate is comparable with other studies on community-dwelling respondents. 29

Conclusions
In conclusion, while a well-designed group relocation is crucial for preserving social capital and ensuring better disaster resilience among displaced survivors, a higher concentration of IDPs might lead to the erosion of trust among the residents of the host community. To avoid the erosion of social cohesion, it may be crucial to provide programs for meaningful interactions between the old and new residents of the community. 26 More research is needed to understand the short-term and longterm impact of different relocation practices on both displaced and nondisplaced survivors of disasters.