Abstract
People are increasingly exposed to environmental threat in the form of “slow-scale” disaster, such as the water contamination at Flint, Michigan. Little is known about the role of place attachment in determining responses to such threats. The present research tests a comprehensive model linking place attachment style to patterns of environmental threat response. Two highly powered surveys (total N = 603) test this model in the context of a water contamination scenario. Across both studies, we find that a more communal and traditionalist place inherited style predicts defensive denial of the threat and compensatory identification with spiritual powers, while a more agentic and cosmopolitan place discovered style predicts identification with responsible institutions and collective action motivation. Place relativity style—characterized by high mobility and lack of attachment—predicts scapegoating of responsible institutions, especially when the threat occurs in a location other than one’s neighborhood (Study 2).
Keywords
We live in a world of increasing exposure to environmental threats, and chief among these are “slow-scale” disasters. Cases where communities are revealed to have undergone prolonged toxic exposure—such as water contamination in Flint, Michigan—cost the government and have major psychological impact on residents. One in 10 U.S. residents have been exposed to contaminated drinking water, and one in four U.S. residents live within 4 miles of a hazardous waste site (Collins, 2010; Edelstein, 2004).
Critically, the psychological responses of citizens when they become aware of such environmental threats play a determining role in their impact. Scientists who have spent decades studying cases of water contamination and related threats (Edelstein, 2004; Rich, Edelstein, Hallman, & Wandersman, 1995) argue that the least productive responses are to defensively deny the significance of the problem, or to scapegoat government agencies whose participation is necessary to address the issue. These responses defend the individual psychologically against felt loss of control but fail to address the threat in the long term. If citizens feel sufficiently empowered and informed to exert personal control, they can work together with accountable institutions to solve environmental problems and prevent their recurrence.
Environmental psychology has the potential for major social impact by identifying psychological means of facilitating collective action when diverse communities face environmental threat. Two gaps in the literature must be filled to achieve this. First, we lack a comprehensive typology of the responses people employ to environmental threats. The present project addresses this gap using Langford’s (2002) typology. Second, to foster collective action, we need to understand not only what responses are possible but the factors that make them likely to occur. The present project addresses this second gap by investigating place attachment style (Hummon, 1992; Lewicka, 2011) as a key predictor of threat response.
A Typology of Responses to Environmental Threat
A key task for environmental psychology is to distinguish between more realistic and constructive, as opposed to more defensive or destructive, responses to environmental threat. Ian Langford (Langford, 2002; Langford, Georgiou, Bateman, Day, & Turner, 2000) put forward a comprehensive typology of threat response using data on public attitudes toward water contamination issues, food insecurity, and global climate change.
Langford (2002) identified three types of environmental threat response that he considered fairly maladaptive, insofar as they “tended to increase participant anxiety rather than alleviate it” and “can be interpreted as defensive reactions” (p. 112). The first he called a “maintenance” type, which we refer to as apathy—maintenance of equanimity in the face of environmental threats by claiming a lack of knowledge and inability to do anything about them. The second maladaptive type is the more active “avoidance” response, which we refer to as denial—fatalistic adjustment through trivialization of an environmental threat’s severity. This tactic, documented in numerous studies, involves skeptical attributions about the quality and intentions of experts, or minimization of the comparative consequences of toxic exposure (Hallman & Wandersman, 1992). The third maladaptive response Langford (2002) called “compensation,” and we refer to as scapegoating: some individuals respond to environmental threat with strong anxiety or anger and compensate by finding “meaning in ‘conspiracy theory’ and feelings of self-efficacy generated by actively ‘fighting back’” (p. 113). Scapegoating consists of the defensive attempt to attribute threat exclusively to a focal individual or group (Rothschild, Landau, Sullivan, & Keefer, 2012). Although corporations and governments often have culpability in the context of environmental threat, scapegoating is defensive when the individual blames simply to restore epistemic control (Hallman & Wandersman, 1992).
Beyond these defensive types, Langford (2002) indicated two more adaptive responses. One is the “ultimate rescuer” response, “the belief in some form of ultimate rescuer, whether it be a person, a group, a belief, an ideal, or a process. The underlying motive is the need for. . .a force or being that eternally observes, loves, and protects” (Langford, 2002, p. 105). We refer to this response as identification with external powers and institutions, either spiritual or secular (Kay, Gaucher, Napier, Callan, & Laurin, 2008). Identification involves investment in larger entities (God, government), trusting they will repair or mitigate environmental damage. Although Langford considered identifying responses to be generally more adaptive, it is nevertheless important to distinguish between spiritual and institutional forms of identification. Spiritual identification is seeking external control by investing in God to cope emotionally with the threat. Case studies of environmental threat—including research on the groundwater contamination that occurred in Legler, New Jersey, in the 1970s—suggests that although such coping can temporarily improve emotional states (Hallman & Wandersman, 1992), it routinely fails to provide long-term solace or to restore perceived control (Edelstein, 2004).
The other adaptive response Langford labels the “informed” approach, through which people find “personal meaning in being in control of themselves and their environment,” because they “believe in control and the personal responsibility for being informed” (Langford, 2002, p. 112). The informed response is an active attempt to cope with threat by engaging with science and adjusting behavior. In the present research, we focus on Langford’s reframing at the group level of the individual informed response, namely collective action (Langford et al., 2000). Collective action is an attempt to reach personal control over environmental threat via identification with and participation in relevant institutions (e.g., federal agencies). While working with and placing trust in institutions is essential to solve the complex issue of environmental threat, scapegoating leads to behavioral support for retributive punishment coupled with refusal to acknowledge personal responsibility (Langford et al., 2000), which creates an adversarial relationship between citizens and institutions detrimental to long-term mitigation efforts (Rich et al., 1995). By contrast, collective action entails commitment to high-stakes behaviors, such as joining advocacy groups or protest participation, and educating oneself about the issue of environmental threat.
Prior research (Bergstrand & Mayer, 2017) on reactions to the Deepwater Horizon oil spill shows that a strong emotional response to environmental disaster is a potent motivator for citizens to work to resolve the threat. Classic survey work on community reactions to the Three Mile Island nuclear disaster (Baum, Fleming, & Singer, 1983) demonstrated that an agentic, activist response can even buffer individuals against the psychological distress that typically results from environmental threat. Given that collective action is an adaptive response for both community members and the broader public, a key question for environmental psychology is what individual differences best predict the likelihood of such a response.
Place Attachment and Environmental Threat
Place attachment is a prominent variable reflecting individuals’ degree and nature of psychological relationship with their surrounding environment (neighborhood, community, city; Scannell & Gifford, 2010). As place attachment has been shown to have a number of stable effects on psychological outcomes (such as emotion and well-being), it would stand to reason this variable should be an important predictor of environmental threat response. Surprisingly, however, this link has not received much empirical attention, and the studies that do exist—recently reviewed by Bonaiuto, Alves, De Dominicis, and Petruccelli (2016)—reveal a complex and contradictory pattern.
On one hand, several studies show that individuals higher in place attachment are more aware of the environmental risks around them (see Bonaiuto et al., 2016). If people are more emotionally connected to their environment, they should be more informed and vigilant regarding potential threats. Somewhat surprisingly, however, a number of studies have also shown that place attachment can negatively influence intentions to adaptively cope with environmental threat. In a classic demonstration, Preston, Taylor, and Hodge (1983) found that residents of a hazardous community in Ontario were unwilling to move (despite the obvious risks) due to their strong sense of community and financial constraints. The tendency for members of a rooted community to refuse to proactively cope with threat, or even to retrospectively cope by evacuating an area that has been severely threatened, has been repeatedly observed (see Bonaiuto et al., 2016). This tendency has been theorized as an exacerbation of “spatial optimism” (Gifford et al., 2009)—people are biased to believe that environmental risks are more prevalent in distant areas and less prevalent locally. Some researchers (De Dominicis, Fornara, Ganucci, Twigger-Ross, & Bonaiuto, 2015) have suggested that highly place-attached individuals may deny local threats as a defense mechanism: because such people are emotionally attached to their environment, they distort levels of objective risk to avoid the painful process of coping.
Thus, despite the relevance of place attachment to environmental threat, the literature is unclear at best. The most obvious reason for the lack of pattern is the failure to adequately or consistently operationalize the independent (place attachment) and dependent (threat response) variables. Coping responses have typically been assessed in either gross and unidimensional or in conflicting ways, giving rise to a situation in which place attachment often has divergent directional effects on coping across studies or even within a single study (Bonaiuto et al., 2016). Place attachment is also a complex, multiply defined variable (Scannell & Gifford, 2010).
Bonaiuto et al. (2016) explicitly highlight “the lack of studies measuring individual differences in place attachment styles” (p. 49) as a problem for this area of research. Accordingly, in the present study, we link Langford’s (2002) typology of environmental threat response to the important individual difference predictor of place attachment. Building on recent environmental psychology research, we did so by theorizing connections between particular patterns of threat response and particular styles of place attachment.
The Role of Place Attachment Style
Some of the earliest work on responses to water contamination events—such as the landfill that caused groundwater contamination at a public school and community in Love Canal, New York—highlighted that divergent relationships to the environment are associated with distinct reactions to threat (Fowlkes & Miller, 1987). Building on such work, the present research highlights the potentially critical role of place attachment style (Lewicka, 2011). Based on a typology proposed by Hummon (1992), Lewicka (2011, 2013) identified and validated assessments of three distinct styles of place attachment. Place inherited style is predicted by lifelong residence and familial ancestry in a place and associated with lower socioeconomic status (SES) as well as “localist” community involvement (e.g., dense neighborhood ties; conservative values). Place discovered style is predicted by long residence in a place where one chose to settle, and associated with higher SES as well as “active” community involvement (e.g., bridging social capital; participation in civic organizations). Finally, place relativity style is predicted by a residentially mobile lifestyle and shorter residence in a place, and associated with little community involvement.
When confronted with environmental threat, response should vary as a function of place attachment style. Those with a place inherited style have lived in the environment most of their lives, cannot imagine leaving it, and might often depend on a religious community (given that they tend to have more traditionalist values). They should be motivated to adjust to the threat or turn to external agents, and will tend toward denial and spiritual identification. Suggestive evidence for these predictions can be found in case studies. For example, in a study of an Argentine shantytown with a long history of hazardous pollution, Auyero and Swistun (2009) observed that many residents seemed fatalistically resigned to their situation and often denied the extent of the risk. The working class residents living in a contaminated area near an Italian steel plant were found to display higher levels of social support coping (indicative of a more collectivist approach) as well as avoidance and denial tactics (Tartaglia, Conte, Rollero, & de Piccoli, 2018). In one of the few studies to predict disaster coping as a function of different forms of place attachment, Mishra, Mazumdar, and Suar (2010) found that in a flood-prone region of India, more economic attachments predicted greater flood preparedness, but religious place attachment was negatively associated with preparedness.
Individuals high in place relativity style may be less threatened due to shorter exposure to a disaster and the potential option to move away (given their generally mobile lifestyles). This possibility is supported by studies of community response to the BP Deepwater Horizon oil spill (Lee & Blanchard, 2012), which show that people with lower levels of community attachment had less negative affective reactions to the disaster. We propose that the primary concern for people higher in place relativity style is that similar threats might occur in other places where they could reside in the future. Hence, we predict that such people will be primarily motivated to re-establish a sense of epistemic control through attributions that frame environmental threats as the work of clear, identifiable forces, and therefore potentially avoidable. In other words, place relativity style should not be strongly associated with most forms of threat response (i.e., people high in relativity should be more apathetic), but to the extent that it does predict a response, that response should be scapegoating of responsible parties and institutions.
Finally, individuals high in a place discovered style are committed to the environment and also should feel efficacious enough to confront the threat. They possess social capital and are familiar with local institutions. We predict that such individuals will seek control over environmental threats via identification with institutions (governments, experts, etc.), generating a readiness for collective action. A number of studies provide preliminary evidence, such as Jacobson’s (2016) research documenting how, in a Florida community identified as a cancer cluster, citizens who already possessed a more active sense of place involvement were the most likely to adopt an activist stance and become scientifically educated regarding the cluster. In a study (Scafuto & La Barbera, 2016) of the motivation to protest contamination due to illegal waste dumping, it was found that while nonactivists were likely to rely on denial tactics, a separate activist population could be identified who felt higher in collective efficacy and had strong personal values (characteristic of a place discovered style).
Although a small body of case studies offer suggestive evidence in support of our predictions, no controlled research has yet systematically investigated the associations between all three place attachment styles and a range of potential responses to environmental threat. We performed such an initial investigation in two highly powered studies exposing participants to the hypothetical threat of water contamination in their communities.
Study 1
Method
Participants for the study were recruited in two waves or subsamples. For both subsamples, participants were recruited through Amazon’s Mechanical Turk service, and were paid US$1.50 for completing the survey. Across subsamples, participants were removed if they failed a basic attention check or if, in Subsample B, they indicated that they had already completed the study as part of Subsample A. This left a final total of Subsample A: n = 198 (86 female), and Subsample B: n = 153 (72 female).
The reason for this is that while Subsample A provided an initial test of our hypotheses, we collected additional data in Subsample B with slightly different operationalizations of the novel threat response variables to improve psychometrics. More specifically, for our first test (Subsample A), we included only one- or two-item measures of several outcomes, and we administered a validated measure of collective action motivation that differed substantially in format from the other measures (see below). After obtaining initially confirmatory results with these measures, we collected Subsample B using at least three items and the same format for each outcome. As for both subsamples our hypotheses were the same, the operationalizations only slightly different, and the patterns of results virtually identical (see below), we focus our analysis on a combined total sample to maximize power.
In both subsamples, participants first responded to basic demographics, then completed measures of place attachment style. All participants then read a short article (abridged from a New York Times article; Davey, 2015) about water contamination in Flint, Michigan. The article described Flint residents’ experience of noticing problems with the water in their homes (e.g., smells, colors) as well as unexplained ailments in their families (e.g., rashes); then reports by health experts of elevated lead levels in the water in Flint homes and schools; and initial expert and government responses, including residents being advised to install filters and avoid hot water use as well as the Mayor’s decision to change the city’s water source.
After reading the article, participants completed a series of hypothetical threat response items, which varied slightly across the two subsamples. Below we note whenever the subsamples differed in the operationalization of threat responses.
Demographics
Participants indicated their age in years, their religiosity (“How important are your religious beliefs to you?”; 1 = not very important, 9 = very important), their social class using the MacArthur subjective social status ladder measure (1 = people who are worst off, 10 = people who are best off), their length of residence in their current city in years, and the number of times they moved residences throughout their lifetime. Demographics were highly similar across the two subsamples (see Table 1).
Descriptive Statistics for Demographics Across the Subsamples (Study 1).
Place attachment style
We measured place attachment style in both samples using Lewicka’s (2011) measure, derived from Hummon’s (1992) theoretical typology. This measure assesses styles of attachment relative to the participant’s current city of residence. Six items each constitute each subfactor: place inherited (e.g., “I cannot imagine leaving my city for good”), place discovered (e.g., “I like to wander around my city and discover new places”), and place relativity (e.g., “I could equally well live here as in any other city”). Participants responded to all items on a 1 = strongly disagree, 5 = strongly agree scale. Scores for each place attachment style were calculated by averaging responses to the six items for each subfactor. Reliabilities and descriptive statistics were highly consistent across the two subsamples (see Table 2).
Descriptive Statistics for Place Attachment Styles and Threat Responses Across the Subsamples (Study 1).
The hypothetical threat response measures were derived from Langford’s (2002) typology. After reading the article describing contamination in Flint, participants were instructed to “Imagine that authorities discover the drinking water in your neighborhood is contaminated by potentially toxic levels of chemicals. For the moment, the extent of the danger and long-term effects for consumers, as well as the history of contamination in your neighborhood, is unknown.” They were then asked, “How would you react to this information?” and indicated their level of agreement (on a 5-point scale) with a series of statements. Table 2 provides a summary of descriptive statistics for the threat response measures in both subsamples.
Denial
Across both subsamples, participants responded to three items: “I would not be especially worried at first, because it is possible that the damage and risks are quite low,” “I would not expect that this event would have any major negative impact on my life,” and “Although information like this is troubling, I would not consider it any more serious than all the other hazards that exist in today’s world.”
Spiritual identification
Subsample A participants responded to a single item: “I would put faith in God or my religious community to help me through this experience.” Subsample B participants responded to three items: “I would put faith in God to help me through this experience,” “I would turn to members and leaders of my religious community for support and guidance,” and “I would engage in prayer or some other ritual to cope with the stress of this situation.”
Institutional identification
Subsample A participants responded to a single item: “I would trust local and national authorities to control and improve this issue.” Subsample B participants responded to four items: “I would seek support and information from local governmental authorities on how to handle this situation,” “I would seek support and information from experts or scientists on how to handle this situation,” “I would trust local experts and authorities to control and improve this issue,” and “I would trust national experts and authorities to control and improve this issue.”
Scapegoating
Subsample A participants responded to two items: “I would feel anger toward companies and civic authorities who might have allowed this to happen” and “I would do whatever possible to obtain financial compensation for suffering from responsible authorities.” Subsample B participants responded to these two items, as well as “I would feel outrage toward governmental officials for not preventing a situation like this from happening” and “I would support any action that might be taken to punish companies and officials who might have been involved.”
Collective action
Subsample A participants completed the “participatory actions” subscale of Alisat and Riemer’s (2015) Environmental Action scale. It is important to note that, unlike the other measures (and the collective action measure in Sample B), this measure asks participants to rate (on a 5-point scale) the frequency with which they engage in a number of pro-environmental collective action behaviors (e.g., “Financially supported an environmental cause” and “Participated in a community event which focused on environmental awareness”). Thus, this measure does not necessarily address participants’ intended responses to the specific hypothetical threat of water contamination that was presented. However, because this measure is a stronger, validated indicator of actual collective action behavior, we felt that including it in one of the subsamples would provide greater external validity. By contrast, Subsample B participants responded to three items in the context of the hypothetical water contamination threat—“I would be willing to participate in protests against companies and authorities that might have been involved,” “I would be willing to voice my opinion at a ‘town hall’ meeting organized by local politicians to discuss how they are planning to handle the situation,” and “I would be willing to attend monthly meetings of a community organization whose purpose is to discuss information and political action related to this situation.”
Results
Analysis plan
To maximize power, place attachment style and threat response scores were standardized across the two subsamples, which were combined into a single dataset for analysis (N = 351). The patterns of association and significance were nearly identical when the subsamples were examined separately (see Supplemental Table 1, in the online supplement, for the pattern of correlations separately by subsample). Indeed, controlling for subsample as an additional covariate in our primary analyses did not alter the significance nor magnitude of any of the effects (i.e., entering subsample as a predictor in the regression analyses yielded nonsignificant effects, βs < .02, ps > .68). We therefore focus on the results in the combined, total sample and do not consider the issue of subsample further.
Place attachment style and demographics
To initially examine the construct validity of our place attachment style scales, we examined their patterns of association with our demographics measures. All zero-order correlations are presented in Table 3. Summary inspection of Table 3 suggested that the scales perform in our sample in a manner very similar to the samples in which they were initially validated (Lewicka, 2011, 2013). One interesting point of divergence is that in our data, place discovered as well as place inherited style were related to length of residence.
Relationships Between Place Attachment Styles and Demographics, Study 1 (N = 351).
p < .05. **p < .001.
Place attachment style and threat responses
To test our predictions, we performed a series of hierarchical linear regressions. For each regression analysis, we entered a separate threat response as the outcome variable. In the first step, we included our demographic variables (gender, age, religiosity, social class, length of residence, and residential mobility). In the second step, we included all three place attachment styles, to examine the unique predictive effect of each (controlling for demographics). All relevant zero-order correlations and regression effects are presented in Table 4.
Predicted Relationships Between Place Attachment Style and Threat Responses, Study 1.
Note. N = 351; r values are zero-order correlations; β weights are the unique effect of place attachment style on threat response controlling for the other two styles and demographics. All represented variables were standardized. Bolded values are theoretically predicted associations.
p < .05. **p < .001.
Denial
The addition of place attachment styles in the second step of the regression contributed significantly to our ability to explain variance in denial beyond the covariates, ΔR2 = .04, F(3, 339) = 4.73, p < .01. This was driven by the predicted positive effect of place inherited style, β = .25, SE = .07, t(339) = 3.73, p < .001. Neither of the other attachment styles had significant unique effects.
Spiritual identification
The addition of place attachment styles in the second step of the regression contributed significantly to our ability to explain variance in spiritual identification beyond the other covariates, ΔR2 = .04, F(3, 340) = 5.30, p < .01. 1 This was driven by the predicted positive effect of place inherited style, β = .19, SE = .06, t(340) = 3.00, p < .01. Neither of the other attachment styles had significant unique effects.
Institutional identification
The addition of place attachment styles in the second step of the regression contributed significantly to our ability to explain variance in institutional identification, ΔR2 = .02, F(3, 339) = 2.50, p = .05. This was driven by the predicted positive effect of place discovered style, β = .13, SE = .06, t(339) = 2.19, p = .03. Neither of the other attachment styles had significant unique effects.
Scapegoating
The addition of place attachment styles in the second step of the regression did not contribute significantly to our ability to explain variance in scapegoating, p = .06. We observed the predicted positive relationship between place relativity style and scapegoating, but it was not significant, β = .09, p = .12. We also observed a trending negative effect of place inherited style, β = −.10, p = .16.
Collective action
The addition of place attachment styles in the second step of the regression contributed significantly to our ability to explain variance in collective action, ΔR2 = .07, F(3, 339) = 10.00, p < .001. This was driven by the predicted positive effect of place discovered style, β = .27, SE = .06, t(339) = 4.90, p < .001. However, we also observed an unexpected (weaker) positive effect of place relativity style, β = .15, SE = .06, t(339) = 2.65, p < .01. This effect was particularly surprising because there was no observed zero-order relationship between place relativity style and collective action, r = .02, p = .73.
Mediation analyses
To clarify why place relativity style had an unexpected effect on collective action, we conducted post hoc mediational analyses. Specifically, we reasoned that either trust in institutions or scapegoating of institutions might conceivably lead to motivation for collective action. This was supported by the fact that collective action was positively correlated with both institutional identification (r = .25, p < .001) and scapegoating (r = .18, p < .01). We further reasoned that place discovered and place relativity style would be related to collective action primarily through one or the other of these motivational pathways.
To test these post hoc hypotheses, we performed two mediational analyses, the results of which are depicted in Figure 1. We used Preacher and Hayes’s (2008) bootstrapping approach to estimating indirect effects, with 5,000 resamples from the data. In our first analysis, we regressed collective action scores onto place discovered style (controlling for the other two attachment styles), with institutional identification and scapegoating entered as potential mediators. The 95% confidence interval (CI) for institutional identification did not contain zero [.01, .08], whereas the 95% CI for scapegoating did contain zero [–.03, .02]. This suggests that the effect of place discovered style on collective action occurs in part through institutional identification, but not through scapegoating. For our second analysis, we regressed collective action scores onto place relativity style (controlling for the other styles), with the same potential mediators. As place relativity style did not have a significant effect on scapegoating, we set a more liberal alpha level for this exploratory analysis. The 90% CI for scapegoating did not contain zero [.001, .04], whereas the 90% CI for institutional identification reached zero [–.05, .000], and a planned contrast suggested that these mediational paths were significantly different, 90% CI = [–.08, –.01]. Thus, there is suggestive evidence that the effect of place relativity style on collective action primarily occurs through scapegoating, and not through institutional identification.
Mediation models, Study 1.
Discussion
Study 1 provides compelling confirmatory support for our theorized predictions. As predicted, a place inherited style predicts denial and spiritual identification responses, and a place discovered style predicts institutional identification and collective action responses. Importantly, these effects were observed in a fairly large sample, controlling for several demographic variables that ought to independently affect threat responses and are related to place attachment. Results were mixed for place relativity style—this style is associated with scapegoating, as predicted, but not very strongly, and furthermore, this style also had some relationship with collective action, an effect that occurred indirectly through scapegoating.
While this study confirms the importance of place attachment style for predicting an array of environmental threat responses, some questions remain. We conducted a second study with three primary goals. First, we wanted to replicate the pattern of associations observed in our first study. Second, we wanted to clarify why place relativity style did not show a stronger relationship with scapegoating. We sought to do this by including an experimental manipulation. In particular, we varied whether participants imagined the hypothetical threat occurring in their own neighborhood versus in another location in the same city that was more remote. Although we did not have strong a priori hypotheses concerning this manipulation, we reasoned that it might moderate the effect of place relativity style. Individuals high in place inherited or place discovered attachment to a city might respond with equal concern regardless of where a threat occurs in their city. However, individuals high in place relativity style may respond differently to a threat depending on its location, because they have a more mobile lifestyle and would be more likely to distinguish between locations based on whether they currently reside there or might move there in the future. One possibility, suggested by our theoretical analysis (see Introduction), is that participants high in place relativity would actually be more concerned about a threat in a location other than their neighborhood, given that they might relocate to such a place in the future; whereas they might be relatively less concerned about a threat in their own neighborhood, given that they could imagine moving away to escape.
Our third goal was to assess people’s immediate emotional reaction to the hypothetical threat, something that we failed to do in Study 1. We assumed that, in general, people would experience more negative affect when imagining a threat in their own neighborhood as opposed to another location; however, we considered the possibility that place attachment style might moderate the effect of threat location on anxiety, and that the level of experienced anxiety might in turn influence the pattern of threat response. For instance, Lee and Blanchard (2012) found that people who were less attached to their community experienced less negative affect when confronted by environmental threat, suggesting that place relativity might predict less anxiety.
Study 2
Method
Participants were again recruited through Amazon Mechanical Turk and compensated US$1.50. 2 Participants who failed a basic attention check were excluded, leaving a final sample of N = 253 (117 female). Demographics were completed in a similar manner to the Study 1 subsamples, and were broadly similar—Mage = 36 years (SD = 11 years), MClass = 4.59 (SD = 1.71), MResidence = 14 years (SD = 13 years).
Procedure and materials
Participants completed the place attachment style scales as in Study 1 (αInherited = .65; αDiscovered = .86; αRelativity = .71). Next, as in Study 1, they read the same article about water contamination and institutional responses in Flint, Michigan.
Participants were then randomly assigned to one of two experimental conditions. In the neighborhood location condition, participants were instructed to complete the hypothetical threat response measures with the same text used in Study 1 (e.g., “Imagine authorities discover that the drinking water in your neighborhood is contaminated. . .”). By contrast, participants assigned to the other location condition were instructed to complete the hypothetical threat response measures with text that referenced a location other than their own neighborhood (e.g., “Imagine authorities discover that the drinking water in a part of your city where you have never lived is contaminated. . .”).
Participants first completed a manipulation check of anxiety, because we reasoned that anxious responses to the hypothetical threat would be higher in the neighborhood compared with the other location condition. On a 5-point scale, participants responded to three items assessing their emotional response to the threat: “I would be concerned for my personal well-being,” “I would be concerned for the purity of the natural environment in my city,” and “I would feel anxiety for my and my family’s future” (α = .75).
Finally, participants responded to all of the threat response measures as they were operationalized in Study 1, Subsample B (αDenial = .85; αSpiritual = .94; αInstitutional = .73; αScapegoating = .83; αAction = .90). The only modification to these measures was that our scapegoating measure omitted one of the prior items (“I would do whatever possible to obtain financial compensation for suffering from responsible authorities”).
Results
Place attachment style and threat responses
Our first goal for Study 2 was to replicate the pattern of results observed in Study 1 irrespective of how the location of the threat was experimentally framed. As Study 1 established the unique effects of place attachment beyond demographics, in these analyses, we only controlled for the most relevant factors in Step 1 (age, social class, and length of residence), and focused on comparing the attachment style effects observed in Step 2. The pattern of relationships can be observed in Table 5.
Predicted Relationships Between Place Attachment Style and Threat Responses, Study 2.
Note. N = 253; r values are zero-order correlations; β weights are the unique effect of place attachment style on threat response controlling for the other two styles and demographics. Data are collapsed across the two experimental conditions. Bolded values are theoretically predicted.
p < .05. **p < .001.
Denial
Replicating Study 1 and as predicted, we observed a positive effect of place inherited style on denial, β = .17, SE = .09, t(245) = 2.15, p = .03. Unexpectedly, we also observed a positive effect of place relativity style, β = .19, SE = .09, t(245) = 2.85, p = .01.
Spiritual identification
Replicating Study 1 and as predicted, we observed a positive effect of place inherited style on spiritual identification, β = .27, SE = .12, t(245) = 3.58, p < .001. Neither of the other attachment styles had unique effects.
Institutional identification
Replicating Study 1 and as predicted, we observed a positive effect of place discovered style on institutional identification, β = .28, SE = .06, t(245) = 3.90, p < .001. Neither of the other attachment styles had unique effects.
Scapegoating
Replicating Study 1, and also contrary to our original prediction, we did not observe an effect of place relativity style on scapegoating, β = .08, p = .22. Unexpectedly, we did observe an effect of place discovered style on scapegoating, β = .16, SE = .06, t(245) = 2.19, p = .03. This effect was somewhat surprising because there was little evidence for a zero-order association between these variables, r = .07, p = .26.
Collective action
Replicating Study 1 and as predicted, we observed a positive effect of place discovered style on collective action, β = .38, SE = .08, t(245) = 5.43, p < .001. Neither of the other attachment styles had unique effects.
Mediation analyses for place discovered style
As in Study 1, we conducted post hoc, exploratory mediation analyses to better understand the processes behind some of the observed effects. In particular, we were interested in why place discovered style had unique effects on both collective action and scapegoating. We reasoned that while institutional identification might be driving the relationship between discovered style and collective action (as found in Study 1), a more emotional response to the threat could be motivating individuals high in discovered style to also engage in scapegoating. We thus performed two mediational analyses (using the Preacher & Hayes, 2008, bootstrapping method), in which we regressed first scapegoating and then collective action onto place discovered style (controlling for the other two styles), with both institutional identification and anxiety entered as potential mediators.
The results of these analyses are presented in Figure 2. The effect of place discovered style on scapegoating occurred at least partly as a result of felt anxiety (95% CI = [.002, .09]), but clearly did not occur via institutional identification (95% CI = [–.01, .06]). The findings for collective action were more mixed: distinct, significant mediational paths were observed for both anxiety (95% CI = [.003, .11]) and institutional identification (95% CI = [.001, .10]). A bootstrapped contrast did not distinguish between the two indirect effects (95% CI = [–.06, .08]). To summarize, it is clear that threat-related anxiety, rather than institutional identification, drives the effect of place discovered style on scapegoating, but there seem to be two potential motivational paths from a discovered style to collective action, one through institutional identification (also observed in Study 1) and the other through anxiety (untested in Study 1).
Mediation models for place discovered, Study 2.
Effects of experimental condition
Although we replicated most of the associations observed in Study 1 when collapsing across our experimental conditions, we were also interested in testing whether our threat location manipulation had any effect on threat response tendencies. In particular, we wanted to determine whether the effect of place relativity style on threat response would be moderated by the proximity of the hypothetical threat.
Anxiety manipulation check
We found that our manipulation had a significant effect on reported anxiety, t(251) = −4.95, p < .001, d = .63, such that participants who imagined the threat in their own neighborhood were more anxious (M = 4.57, SD = 0.55) compared with participants who imagined the threat in another location (M = 4.11, SD = 0.88).
Mediation analysis for place relativity style
We also found that experimental condition interacted with place relativity style to predict anxiety, β = −.71, SE = .13, t(249) = −2.64, p < .01. This interaction occurred because relativity style was positively related to anxiety in the other location condition, r(125) = .15, p = .10, but negatively related to anxiety in the own neighborhood condition, r(128) = −.20, p = .02. This pattern is interpretable because people who are more habitually mobile may not be overly concerned when a threat occurs in their own community (as they can easily consider evacuating), but may indeed be relatively concerned when a threat occurs elsewhere (as they might possibly move to the threatened area). Although we did not make this prediction a priori, we reasoned that the different patterns of anxiety for high-relativity participants across conditions might impact their threat responses. Indeed, we observed that while relativity style and scapegoating were completely unrelated in the own neighborhood condition, r(128) = .03, p = .74, they were related as predicted in the other location condition, r(125) = .17, p = .05. We performed a moderated mediation analysis, depicted in Figure 3, to determine whether there was a significant indirect effect of the Place Relativity Style × Condition interaction on scapegoating via anxiety. The 95% CI for the indirect effect did not contain zero [–.24, –.04], suggesting that only in the other location condition did higher place relativity scores predict greater anxiety and consequently greater scapegoating.
Mediation model for Place Relativity × Condition, Study 2.
Other effects of condition
We did not observe interactions between experimental condition and the other two place attachment styles on anxiety. We also did not observe main effects of condition on our threat response measures, ts < .60, ps > .50. 3
Discussion
Study 2 broadly replicated the predicted pattern of associations between place attachment style and threat response. As in Study 1, there was one predicted association that was unobserved—that between place relativity style and scapegoating. Our follow-up analyses on the effect of threat location partly explain the lack of a significant relationship between these variables. It appears that place relativity style is associated with a fairly apathetic response when the threat is more proximate; if anything, participants higher in relativity seem to deny the threat under these conditions (hence the significant effect of relativity on denial in these data). Interestingly, when the threat is more distal, place relativity predicts a more anxious reaction, compensated for by greater levels of scapegoating. These exploratory findings are broadly consistent with our theoretical account: place relativity style predicts a lack of concern with local threats, presumably because high-relativity participants would simply elect to leave the area, but it predicts greater concern with remote threats, presumably because high-relativity participants can imagine moving into other contaminated areas and are hence motivated to attribute environmental threats to clearly identifiable agents (i.e., to scapegoat).
The only other observed relationship that was not predicted a priori was the unique effect of place discovered style on scapegoating. Follow-up mediation analyses revealed that while a discovered style predicts scapegoating via elevated anxiety, both anxiety and institutional identification have independent indirect effects on collective action. Although these findings were somewhat surprising, it should be noted that (a) place discovered style did not have an effect on scapegoating in Study 1; (b) there was no zero-order relationship between discovered style and scapegoating in Study 2; and (c) the effect of discovered style on collective action in Study 2 is much larger. We therefore conclude that the predicted association between place discovered style and collective action is more robust (indeed, this association represented the largest effects across the two studies).
General Discussion
Across two highly powered studies, we obtained general support for our predictions. In both studies, place inherited style was uniquely related to responses of denial and spiritual identification, while place discovered style was strongly related to institutional identification and collective action. Although, as predicted, place relativity style was related to a scapegoating response, this effect was not as strong. An experimental manipulation of threat location (Study 2) revealed that the reason for this weaker effect is that, among participants high in place relativity style, a threat in their own neighborhood is less anxiety-provoking and evoked denial or apathy; whereas a threat in another location actually elicits more anxiety and a stronger scapegoating response. Presumably, this is because individuals high in place relativity style imagine moving away from a threat in their own neighborhood, whereas they are more concerned with threats in other locations, given that they might relocate to contaminated locations in the future.
Limitations
The primary limitation of these data is their hypothetical nature. To address this, in each of our studies, we included a single-item measure of exposure to environmental threat: “Have you, or any of your friends or family, actually experienced or are currently experiencing a water contamination incident?” Using this item, we conducted supplementary, exploratory analyses to examine any differences in the pattern of results based on actual exposure. When we combined both datasets, standardizing all variables, approximately 8% of the total sample (n = 47) reported having personal experience with contamination. Note that this number is only slightly lower than the national average for exposure to drinking water reported by Collins (2010), and it stands to reason that we would obtain a lower number, given that race and SES are primary predictors of exposure in the United States (and our samples are not very racially diverse). Thus, we may have some confidence that this is a fairly accurate exposure rate. Correlations within the exposed and unexposed subsamples are presented in Table 6.
Comparison of Relationships Within the Exposed and Unexposed Subsamples.
Note. Results from the exposed subsample (N = 47) are presented below, from the unexposed subsample (N = 556) above the diagonal.
p < .05. **p < .001.
Although the sample sizes are dramatically different, a few preliminary observations can be drawn. Among the small group of participants with personal threat exposure, the relationships between place inherited style, denial, and identification become even more pronounced, as does the relationship between place relativity style and scapegoating; however, the relationship between place discovered style and collective action is weakened. This could be a function of the fact that exposure to environmental contamination in the United States is strongly linked to lower socioeconomic and minority racial status (Taylor, 2014), which also positively predict place inherited and place relativity styles (Lewicka, 2011). Future research is needed, however, to compare the effects of place attachment style across more diverse samples and across groups that both have and have not been exposed to threat. For now, our supplementary analysis suggests that our predictions largely hold among individuals who have personal threat exposure.
Implications for Place Attachment and Environmental Threat
The present research clearly supports an important role for place attachment style in response to environmental threat, thereby shedding some light on current uncertainties. Although their review focused primarily on studies of natural (as opposed to technological) disaster, recall that Bonaiuto et al. (2016) concluded that place attachment tends to be associated with greater environmental risk perception, but less environmental risk coping. Indeed, in some studies, place attachment is actually negatively related to risk perception as well. Our findings suggest a key variable—place attachment style—as a moderator of these processes. The conflicting results in the prior literature may be due to the fact that those high in place inherited style tend to deny the significance of a threat (which would result in lower risk perception and less coping), whereas those high in a place discovered style have the motivation to engage in collective action and are therefore willing to acknowledge the presence of risk.
While it is important to distinguish between these two styles of positive place attachment, future research should also assess negative or nonattachment as well. Indeed, Hummon (1992) originally identified five types of place attachment and distinguished between three “nonattached” forms (place relativity, placelessness, and alienation), which in the present research were reduced to the single place relativity style. Lewicka (2011) was able to use a variety of measures and a cluster analytic approach to distinguish these five forms, and such a method might be employed in future work on environmental threat response. The distinction between place relativity and alienation seems particularly important given the experimental moderation in Study 2; while those high in place relativity may actually be motivated to take defensive action concerning threats in other locations, alienated individuals may be relatively indifferent and apathetic toward threats occurring in their own or any other environment.
Water Contamination and Other Forms of Environmental Threat
Research shows that threat responses vary dramatically as a function of affective and cognitive processes associated with different kinds of threat. For instance, Slovic (2010) has conducted a program of research showing that people’s initial emotional response to a threat, as well as factors such as the number of lives lost or endangered and the visibility of hazard, have a major impact on risk perceptions (cf. Lindell, 1994). Although we kept these processes largely constant in our studies by consistently using the same hypothetical threat of water contamination, we still saw some evidence of variation in risk perception: for example, in Study 2, individuals higher in place relativity style were less likely to feel anxious about a threat in their own neighborhood.
Our focus on water contamination could be viewed as a limitation insofar as the question remains of how generalizable these patterns would be to other threats. Comparing studies of community response with different forms of contamination and with natural disasters, Edelstein (2004) proposed responses vary as a function of perceived causality, consequences, and controllability. Research covering a variety of environmental threats (Lindell & Perry, 2004) shows that the outcomes of spiritual and institutional identification are relatively common, and the present work adds that they will be differentially determined by place attachment style. However, attributions of causality and consequences may significantly shape the likelihood of scapegoating and denial as viable responses. Outright denial is less common in the case of highly visible and destructive natural disasters (although it can occur with respect to the perceived likelihood of future disasters), and scapegoating is far more common when there is a clear human or corporate agent to which blame can be assigned (Kroll-Smith & Couch, 1990; Wolfenstein, 1977). Where controllability of the threat is concerned, place attachment style may be expected to interact with the nature of the threat. If a threat is truly low in controllability—for instance, a once-in-a-lifetime storm or eruption—than individuals high in place inherited style may actually cope best with the threat, because they are able to draw on spiritual resources to adjust to the uncontrollable. However, for threats that require lifestyle changes to successfully cope—such as avoiding a contaminated groundwater source—a place discovered style, with its characteristic aspects of agentic coping and collective action readiness, may prove most advantageous.
Although in our research we focused on place attachment as a predictor of environmental threat response, this relationship is most likely bidirectional. Studies show that undergoing environmental threat can lead to diminished feelings of place attachment (Ruiz & Hernandez, 2014). Longitudinal research with an exposed sample may reveal both positive and negative cycles of relationship between initial place attachment style, responses to threat, and subsequent changes in place attachment. It is also quite possible that the form of environmental threat could play an important moderating role. While natural disasters may sometimes strengthen a community and increase place attachment, technological disasters and contamination events often seem to have the opposite effect over time (Tartaglia et al., 2018).
The Prospect of Intervention
Importantly, this model not only identifies the types of individuals who will be less likely to show an adaptive response to environmental threat, but crucially suggests distinct motivational paths for intervention among these individuals. Those with a place inherited style are committed to their local environment, but require an enhanced sense of identification with and trust in institutions to engage in collective action. By contrast, those with a place relativity style need to feel more strongly attached to their local environment before they will be motivated to take collective action. In addition to different aims, it might also be important to consider different modalities for discrete interventions. Given their more communalistic orientation (Lewicka, 2013), individuals high in place inherited style may respond more to “grassroots” interventions involving word-of-mouth and neighborhood participation. On the contrary, high-relativity individuals might be more responsive to interventions that target them personally (e.g., personalized letters or emails). Future studies should develop and test such motivational interventions.
Supplemental Material
Online_Supplement – Supplemental material for Place Attachment Style as a Predictor of Responses to the Environmental Threat of Water Contamination
Supplemental material, Online_Supplement for Place Attachment Style as a Predictor of Responses to the Environmental Threat of Water Contamination by Daniel Sullivan and Isaac F. Young in Environment and Behavior
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
Notes
Supplemental Material
Supplemental material for this article is available online.
Author Biographies
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
