Public Organization Adaptation to Extreme Events: Mediating Role of Risk Perception

Public Organization Adaptation to Extreme Events: Mediating Role of Risk Perception Abstract The study responds to the growing call for a more systematic approach to research on organizational responses to extreme events. It develops and tests an integrated framework based on the organizational adaptation and learning theory to shed light on how public organizations manage exposure and vulnerability to extreme events. The analysis uses data from a 2016 national survey of top managers in the largest fixed-route public transit agencies in the United States and from other institutional sources to test hypotheses that link exposure to extreme events, impact, risk perception, and adaptive responses. We apply a structural model to disentangle the direct effect of exposure on adaptation as well as its indirect effects through impact and risk perception. Findings underscore the critical role that organizational risk perception has in converting environmental stimuli to organizational adaptive responses and point to a perception-mediated learning model of adaptation. Introduction Public organizations are increasingly encountering extreme events that cause notable, unpredictable, and disruptive changes (Boin and Lodge 2016; Comfort et al. 2012; Tierney 2014), and have the potential to cause considerable damage. Extreme events are human or nature-induced occurrences characterized by high salience, high uncertainty, and profound impacts (Kapucu 2005). They include earthquakes, severe weather, disease outbreaks, power outages, social movements, technical break-downs and cyber-attacks (Boin and Lodge 2016; Tierney 2014). Greater complexity of technology systems, more interconnected and interdependent infrastructure (Perrow 1984; Turner and Pidgeon 1997) and ongoing changes in the Earth’s biophysical and seismic system (Folke 2006; Hegerl et al. 2007) all contribute to the increased frequency and severity of extreme events. The occurrence of extreme events is increasingly patterned (i.e., nonrandom) and the risks associated with them similar (Rosenthal and Kouzmin 1997; Roux-Dufort 2007; Taleb et al. 2009; Tierney 2014), raising the potential for detecting and understanding patterns of organizational response. Response to extreme events is typically addressed through emergency, crisis and disaster management in public administration. Governments often establish professional emergency response routines and formal contingency management structures, mainly to coordinate with first responders (Boin and Lodge 2016; Somers and Svara 2009). Yet the increase in the number, scope and intensity of disasters that repeatedly expose the limitations and weakness of the prevailing management practices (Cigler 2007; Comfort et al. 2012; Kapucu and Garayev 2011) call for new approaches “at the strategic rather than the operational level” (Boin and Lodge 2016, 291). Recurring exposure to and impacts from extreme events necessitate planned adaptation to enhance organizational ability to withstand disruption, minimize damage and maintain operations (Comfort 2002; McEntire et al. 2002). Nevertheless, some public organizations are more likely to undertake adaptive responses than others. For example, scholars have shown that while many organizations have continued their reactive strategies to manage the immediate consequences after extreme events occur, some are reevaluating and responding to vulnerability before damage and incorporating considerations about extreme events in their long-range plans, infrastructure design, asset management, and interorganizational coordination (Hill and Engle 2013; Thomson 2017; Welch et al. 2016). Why is this the case? What conditions, experiences, capacities, and perceptions lead some public organizations to adapt and not others? To date, most empirical research emerges out of case studies of organizations responding to one specific type of extreme event (Christensen et al. 2016; McGuire and Silvia 2010; Roux-Dufort 2007). Although these studies are rich in detail, it is important to quantitatively characterize responses and empirically test the determinants of organizational adaptation across multiple organizations. This study empirically tests hypotheses derived from theory linking extreme events, organizational vulnerability, risk perceptions, and adaptive capacity. It examines the interface between the organization’s environment and the learning mechanisms that enable adaptation. Specifically, it investigates how exposure to extreme events impacts public organizations, and how the occurrence and impact of extreme events are related to organizational perceptions of risk and their adaptive responses. Is organizational adaptation a spontaneous response to recurring risks or is it mediated by cognitive mechanisms that enable organization-wide sensemaking of emerging developments? Based on a unique dataset from a 2016 national survey of 892 managers in 273 US largest public transit agencies that cover 82% of the entire FTA population of transit agencies with an annual fare revenue of 1 million, the findings underscore the essential role of risk perception in channeling environmental signals to adaptive responses. The article concludes by discussing its theoretical contribution and implications for management. Organizational Adaption to Reduce Vulnerability A key premise of this research is that organizations can learn from their prior experience and adjust to their changing environment (Lawrence and Lorsch 1967; Thompson 1967), thereby reducing vulnerability from repeated perturbations and disruptions from extreme events. Vulnerability, which is the degree to which an organization is likely to experience harm due to its exposure to hazardous events (Turner et al. 2003), is recognized as an outcome of the interaction between an organization’s exposure to environmental stresses and its ability to prepare for and react to them effectively (Berkes 2007; McEntire 2008). Because organizations have limited control over the frequency, magnitude, and scope of extreme events, effective vulnerability management is a primary means by which they are able to alter the consequences of such events (Dutton and Jackson 1987; McEntire et al. 2002). Figure 1 presents the framework depicting the key determinants of vulnerability in public organizations and the organizational mechanisms through which vulnerability can be managed. From left to right, an organization is exposed to the frequency, magnitude, and scope of extreme event, but the level of harm an organization experiences is contingent on the capacity it has at time t = 0 to respond to the event (Cigler 2007; McEntire 2005). Organizations having greater capability experience less harm and manifest fewer gaps in performance (i.e., impacts) during events. Perceptible gaps in performance are called revealed vulnerability. Risk perception results from an organization’s assessment and understanding of the risk inherent in its environment (Comfort 2007; Sitkin and Pablo 1992). Depending on its experience and perceived risk, the organization builds adaptive capacity at t = 1 after the event occurs. The response of the organization at t = 1 further influences its vulnerability and response to future extreme events. The cycle from adverse impact to organization adaptive response occurs over a period of time, the length of which depends upon the size of the gap, initial organization capacity, and other factors. We discuss the theoretical rationale for this framework in the following paragraphs and in the hypothesis section. Figure 1. View largeDownload slide Organization Vulnerability and Response to Extreme Events Figure 1. View largeDownload slide Organization Vulnerability and Response to Extreme Events Extreme Events and Organizational Performance Several factors constrain public organizations’ ability to effectively deal with extreme events. Public organizations are characterized by their search for structural (i.e., lack of organizational duplication) and fiscal (i.e., frugality) efficiency with a focus on their core competencies and routines (Perrow 1984; Stark 2014). This “lean and mean” strategy (Staber and Sydow 2002), which reduces system redundancy, resilience and slack, can increase vulnerability to extreme events and complicate efforts to manage their consequences (Boin et al. 2016; Schulman 1993). As Stark (2014, 696) notes, “the pursuit of efficiency without consideration of the benefits of auxiliary resources will create systems that cannot adapt to the emergence of potentially disastrous failures.” Additionally, interconnectedness and interdependence of critical infrastructure increases organizational vulnerability and complicates adaptation. High complexity increases the potential for a single localized performance failure to cause a cascade of disruptions that result in system-wide failure (Little 2002; Perrow 1999). Crichton and colleagues (2009) aptly exemplify how failure in one electricity feeder cable during a heat wave propagated faults in another two feeders and escalated into a profound 10-week disruption in electricity supply in New Zealand. Moreover, repeated exposure to extreme events can reduce an organization’s performance well below the desired level of achievement (Comfort 2005). The compromised integrity of system components and linkages, while not necessarily precipitating catastrophe during a particular event, can increase vulnerability and widen the performance gap during future extreme events (La Porte 1996; Perrow 1994). Adaptive Capacity Strengthening adaptive capacity is a key mechanism by which public organizations can reduce vulnerability to repeated stress or perturbations (Berkes 2007; McEntire 2008; Staber and Sydow 2002). Adaptive capacity is defined not only in terms of an organization’s capability to bounce back to a state of normalcy after an extreme event, but also of its ability to absorb disruptions and reorganize while undergoing changes so as to retain the essential functions and structures (Berkes 2007; Boin and Van Eeten 2013). Increasing adaptive capacity entails deliberate efforts to make longer-term and anticipatory adjustments to fill the possible performance gaps for a wider range of observed or anticipated extreme events. It reflects an organization’s stock of resources and enables it to exploit its resources in a more productive manner (Kusumasari et al. 2010). Common measures to build adaptive capacity include improvement in material resource inputs, information and technology, infrastructure and equipment, human capital, and inter-agency coordination arrangements (Comfort and Okada 2013; Kusumasari et al. 2010; McEntire 2005). As an example, Arizona mobilized adaptive responses to severe droughts in 2008–2011 by developing an integrated water conservation system and fostering collective planning and response among water suppliers and users (Hill and Engle 2013). Development of adaptive capacity is distinguished from emergency management, which focuses predominantly on responding to the immediate impacts of extreme events (Cigler 2007; McEntire 2008; Somers and Svara 2009). Crisis management planning and preparation for extreme events establish risk-based intentions but do not constitute adaptive capacity (Christianson et al. 2009; Clarke 1999). As Perrow (1999, 152) notes, “Even ‘worst case’ scenarios usually refer only to the worst state of the environment,” giving little attention to an organization’s overall vulnerability, maintenance of infrastructure, the process of event escalation, the availability of backup resources or potential for maximum failure of the organizational management (Fischbacher-Smith 2010; McEntire 2008; Roux-Dufort 2007). Risk Perception Nevertheless, reducing vulnerability and enhancing adaptive capacity in the face of extreme events are complicated by multiple intervening factors. Vulnerability is often not obvious in the absence of significant triggers or events (Rijpma 1997; Sarewitz et al. 2003) and the evidence of capacity surfaces only after extraordinarily complex problems are solved (Kusumasari et al. 2010; Levinthal 2000). As a result, decision makers face high ambiguity, complexity and uncertain payoffs from investment and change. These challenges to building adaptive capacity, coupled with the evidence that disasters motivate adaptive planning in some organizations but not in others (Ebert et al. 2009; Haigh and Griffiths 2011), suggest a cognitive mechanism that governs how organizations respond to these external stimuli. Scholarship across disciplines has identified risk perception as a crucial factor for explaining why disasters motivate adaptation planning in organizations (Berkhout 2012; Comfort 2007; Dutton and Jackson 1987; Hoffmann et al. 2009; Somers and Svara 2009). Risk perception comprises the perceived probability of being exposed to negative impacts and the appraisal of how harmful those impacts would be on the organization (Grothmann and Patt 2005). Uncertainty and complexity associated with extreme events limit the applicability and usefulness of analytic techniques and tools (Garrett 2004; Moynihan 2008; Simon 1979), as well as the ability of existing data to capture event complexity (Fischbacher-Smith 2010). As a result, organizations rely on a messy process of sensemaking, attribution, and judgment for decision making and strategy selection (Daft and Weick 1984; Kiesler and Sproull 1982; Sitkin and Pablo 1992). Adaption is less likely to occur when the affected organizations fail to notice and attach meaning and significance to variations in their environment that pose risks (Comfort 2007; McEntire 2004; Sitkin and Pablo 1992). In contrast, perceived risk can stimulate organizations to undertake nonroutine and nonincremental action that aims to increase adaptive capacity (Comfort 2007). The theoretical framework presented in Figure 1 provides a starting point for further development of theory-based hypotheses linking exposure to extreme events, impact, risk perception, and adaptation. Following the framework, we hypothesize that organization adaptation to extreme events is affected by its vulnerability to extreme events which is manifested in the impacts it realizes during the events. We also expect that exposure and impact lead to adaptation through perception of the risk associated with extreme events. Theory And Hypotheses Exposure, Impact, and Adaptation Organizational systems are typically designed to cope with a certain range of external perturbation and disruption, the levels and scopes of which are largely determined by their historical norms and experiences. However, extreme events are invariably outliers against which the existing operating system does not provide sufficient defense (Fischbacher-Smith 2010). Repeated exposure to extreme events inevitably leads to accumulation of severe weaknesses and deficiencies until the system reaches the tipping point and loses viability and robustness (Roux-Dufort 2007). As Comfort (2002, 102) puts it, “governmental systems designed to provide security at one level of exposure may fail when they are exposed to cumulative threats of different types at different levels of operations.” The situation rapidly escalates as a result of a series of interdependent cascading failure in which failure in a single component triggers failures throughout a complex and tightly coupled system (Perrow 1984; Turner and Pidgeon 1997). Recurring extreme events are more likely to overwhelm the control and management system and lead to disastrous impacts on the affected organization. H1: Organizations that experience greater exposure to extreme events are more likely to experience greater impacts (i.e., performance gap). It is often the case that structural, political, and capacity constraints limit the ability of public organizations to pursue adaptive solutions in the face of extreme events (McGuire and Schneck 2010; Smith et al. 2009; Wise 2006). Extreme events that cause significant property damage, economic loss, human casualties, aggressive media coverage, and political pressure act as powerful catalysts for reexamining standard approaches and practices (Boin and Hart 2003; Stehr 2006). Meanwhile, the increasing inapplicability and ineffectiveness of existing routines and procedures force organizations to make greater investments in exploration and implementation of more fitting solutions (Cyert and March 1963; March 1991). Facing low control in high-risk environments, organizations can best respond by adjusting their internal processes and building adaptive capacity (Dutton and Jackson 1987; Staber and Sydow 2002). Prior work demonstrates that significant adjustments of organizational strategy and practices usually do not occur until an organization has unequivocally suffered disastrous consequences associated with extreme events (Comfort 2007; Dutton and Jackson 1987; Linnenluecke et al. 2012; McEntire 2004). Importantly, impactful extreme events open “windows of opportunity” (Kingdon 1984) for reform-minded organizations to exploit the significant damage and build support for nonincremental changes by directing attention to the flaws and deficiencies in the existing systems (Boin and Hart 2003). Such opportunities are scarce and fleeting, and organizations have to act expeditiously before the public attention and impetus for reform fade away (Birkland 2009; Dekker and Hansén 2004; Gallagher 2014; Parker et al. 2009). This suggests that organizational mobilization of support and for adaptive capacity development must be carried out within a relatively short time span. Delay or slow response can miss the opportunity. The higher the consequences of the extreme event, the more leverage an agency can apply to facilitate non-incremental changes (Birkland 1997). We thus hypothesize that higher impacts of extreme events are more likely to provide the momentum as well as the leverage for organizations to increase adaptive capacity. H2: Organizations that experience greater impacts (i.e., performance gap) from extreme events are more likely to undertake adaptive capacity building. At the same time, greater exposure to extreme events increases problem familiarity essential for organizational learning (Levitt and March 1988; Sitkin and Pablo 1992). Recurring extreme events inform organization decision makers about the rapid escalation of harm and the significant scale of failure that escalation can generate, both of which work to invalidate their assumptions about control (Fischbacher-Smith 2010). The lessons learned are reflected in their exploration and adoption of adaptive responses to the threats, independent of the outcomes from the past exposure (Sitkin and Pablo 1992). The increased exposure also allows for trail-and-error experimentation to acquire and test knowledge on how to best adapt (Berkes and Folke 2002; Levitt and March 1988). Given this discussion, it is expected that greater exposure to extreme events will be positively associated with adaptive capacity development. H3: Organizations that experience greater exposure to extreme events will be more likely to undertake adaptive capacity building. Risk Perception and Adaptation Organizations typically must experience a continuous stream of events to grasp trends and attempt to make sense of them (Daft and Weick 1984; Thomas and McDaniel 1990). Given bounded cognitions of decision makers (Simon 1979) and the magnitude of complexity and uncertainty of extreme events, the associated risks are often not easily recognized. Instead, organizations only selectively attend to emerging developments (Dutton and Jackson 1987; Weick 1977). As selectivity is governed by knowledge or ideology filters (Boin and Van Eeten 2013; Thomas and McDaniel 1990), organizations tend to resist perceptions and conclusions that challenge their prevailing routine and frames of reference (Sitkin and Pablo 1992; Staw et al. 1981). To the extent that extreme events challenge fundamental assumptions about risks and control (Boin and Lodge 2016; Fischbacher-Smith 2010; ‘t Hart 2013), they are likely to be excluded from consideration for organizational decision making. Selective filtering of stimuli is particularly strong in public organizations which are highly constrained by organizational inertia and cultures of risk denial (Cigler 2007; Ford and King 2015; Perrow 1999). Boin and Van Eeten (2013) aptly exemplify how the entrenched safety process and culture of high reliability in the National Aeronautics and Space Administration (NASA) prevented “an accurate assessment of the impending threats to the safety of the doomed shuttle” (442) and caused the well-known Challenger accident. There are many reasons for this, but at base the failure to recognize risk and act has been explained by scarcity of evidence, blindness to evidence and uncertainty in assessing the relevance of evidence (Levitt and March 1988). Therefore, before organizations engage in problem-solving behavior, they must first recognize the problem and then implement change (Daft and Weick, 1984; Sitkin and Pablo, 1992). To paraphrase Repetto (2009), just because organizations can adapt does not mean they will. The joint capacity to detect the out-of-ordinary developments and to recognize the inherent risk they pose creates a trigger that enables an organization to initiate an adaptation process (Comfort 2007; Duncan 1972; Dutton and Jackson 1987). Given this, it is relevant to ask why some organizations are capable of perceiving the risks while others not. As a first step, emerging events or trends influence perceptions because of their potential to interfere with an organization’s salient objectives (Daft and Weick 1984; Schwenk 1984). Extreme weather signals “uncontrollable” and “negative” threats (Dutton and Jackson 1987; Kahneman and Tversky 1979) to which some organizations are more sensitive than others. For example, organizations in the transportation, energy, water, and utility sectors may be particularly attuned to extreme weather signals because their operations hinge on the sound functioning of an exposed, interconnected, and dispersed lifeline infrastructure (Boin and McConnell 2007; Little 2002; McDaniels et al. 2008). The threats intensify as an organization’s experience with extreme weather events rises. As Berkhout (2012, 95) states: “… the more frequent, unambiguous and salient evidence from experience is, the greater the likelihood it will be recognized and interpreted as significant (by organizations).” An organization’s perception of risk is therefore a function of its experience with extreme events, in terms of both exposure and impact. Referring back to the discussion about the cognitive significance of risk perception, this suggests a process that links environmental stimuli to organization action through risk perception. It therefore stands to reason that risk perception positively mediates the effect of signals from extreme events on organizational adaptive responses. Based on this discussion we hypothesize H4: An organization’s risk perception positively mediates the effect of its experience with extreme events (i.e., exposure and impact) on its adaptive capacity building. Figure 2 presents the conceptual model of the theorized relationships. We hypothesize that an organization will experience greater adverse impacts from increased exposure to extreme events. Exposure to extreme events can directly affect an organization’s adaptive capacity building through a learning mechanism. The direct experience with extreme events (i.e., exposure and impacts) is also expected to trigger the organization’s perception of risks concerning ongoing environmental change and ultimately lead to a higher level of organization adaptation. Figure 2. View largeDownload slide Theoretical Model: Mediating Role of Risk Perception Figure 2. View largeDownload slide Theoretical Model: Mediating Role of Risk Perception Extreme Weather Events and Transit Agencies Extreme weather events pose a serious challenge to public agencies (Hodges 2011; Meyer et al. 2012). Not only do they disrupt operations, impair service quality, and cause additional safety threats, but they also damage infrastructure and impose strain and stress on the state of good repair. As demonstrated by Hurricane Sandy and other recent weather-related disasters, weather can have significant ramifications for regional mobility and functioning of economic systems. How to effectively manage the risks of extreme weather is a question that often concerns public organizations and managers. This issue has become increasingly urgent with the likely increase in the frequency and intensity of extreme weather events (Folke 2006; Hegerl et al. 2007; Parmesan and Yohe 2003), such as floods, heat waves, and severe storms. While many agencies have long experience coping with weather disruptions, they are increasingly confronted with the challenge of identifying and developing appropriate longer-term adaptation strategies to address greater risk and uncertainty associated with extreme weather (Hodges 2011; Meyer et al. 2013). Selection of transit agencies offers several advantages for the study of extreme weather and adaptation. First, because transit agencies rely on an interdependent and interconnected network of exposed and dispersed capital assets for service delivery, they are particularly sensitive and susceptible to extreme weather. In addition to the impacts on infrastructure, extreme weather also diminishes the quality and reliability of transit services in the form of increased assess time, prolonged trip duration, declined service frequency, and forced vehicle rerouting (Singhal et al. 2014). Given the interconnectivity and lack of redundancy in most current transit systems (Meyer et al. 2012), extreme weather results in substantial costs to network efficiency and mobility (Koetse and Rietveld 2009). Moreover, weather-related disruptions disproportionately affect transit-dependent populations, such as the elderly, the poor and people with disabilities (Hodges 2011). That weather impacts on the public transportation sector have received scant attention in the literature (Koetse and Rietveld 2009) contributes further to the importance of this study. Data This study matches survey data collected from US transit agencies 2016 with agency profile data obtained from the Federal Transit Administration’s National Transit Database (NTD) and demographic data taken from the US Census Bureau. The full data set is used in a structural equation model (SEM) to test our hypotheses. The NTD is the primary source for information and statistics on the US transit system. We used the entire list of agencies in NTD to identify target agencies for this study. The target population included all major US metropolitan fixed-route public transit agencies operating bus and/or rail transit services with an annual fare revenue of at least one million dollars in 2013. Agencies having a small vehicle fleet (i.e., 30 or fewer vehicles according to FTA’s definition of small systems) or those run by universities were excluded from the study because of their limited capacity to undertake adaptation. Private companies (less than 30 in number) were not included because most manage transit systems in multiple cities and states. The resulting target population included 312 public transit agencies that satisfy the selection criteria. Because different key decision makers within agencies likely have different perspectives on extreme weather events, vulnerability and risk depending upon professional background, experience, and position in the agency, the study identified the leader manager in five different departments for inclusion: maintenance, operations, engineering, service planning, and strategic planning. Names and contact information were collected using a protocol that included three methods: searching of online sources such as NTD, telephone calls to agencies, and Freedom of Information requests. Because some agencies did not respond, refused to provide information or were nonreachable after repeated attempts, we had to drop 39 agencies resulting in a final agency-level sampling frame of 273 agencies (88% of the target population). As not every agency has all five departments or functions, the final individual-level sampling frame included 892 respondents (equivalent to an average of 3.3 people per organization). The survey asked respondents about organizational experience with extreme weather events, perceptions about future weather risks, strategies employed to address extreme weather, and other questions concerning operation and management. Most questions were set within a 2-year time frame to reduce recall bias. Survey instrument development was informed by a small set of interviews of transit managers in four agencies in different regions of the United States. The instrument was coded in Sawtooth Software® and administered online first as a pretest to 20 individuals from the sample and then to the entire sample. Administration of the full survey extended from April 28 to June 11, 2016. An initial hard-copy notification letter, sent to each of the respondents informing them about the survey and requesting their participation, was followed 10 days later by an electronic invitation with web link to the live survey, username, and password. The survey was administered to an adjusted sample of 862 respondents after ineligible, retired, and noncontactable individuals were dropped. A total of 306 individuals completed the survey, yielding a survey response rate of approximately 35.5%. Of the 273 agencies surveyed, 199 provided at least one response (72.9%). Nonresponse bias analysis showed no difference by size or region (α < 0.05) between responding agencies and either the sampling frame (n = 273) or the target population (n = 312). As a result, we are confident that our study results generalize to the target population. We merged the survey response data with other organizational data taken from NTD and 2010–2014 US Census survey, including items such as organizational size and population density of the transit service area. Measurement Variables of Primary Interest: Organizational Adaptation An organization’s adaptation strategy comprises a combination of adaptation measures and the distinct strategic goals they aim to accomplish (Hoffmann et al. 2009). For conceptual focus, this article examines two critical adaptation strategies relevant for addressing the potential threats posed by extreme weather to resource-intensive public service organizations: vulnerability assessment and capital investment. Vulnerability Assessment Public organizations must decide where to make improvements and how to apply scarce resources to address extreme weather (Henstra 2010). Vulnerability assessment generates specific information necessary for making strategic decisions (Cairney et al. 2016; Somers and Svara 2009). The information is particularly useful for organizations operating in tightly coupled and complex systems prone to cascading disruptions and failures (Little 2002). Because people tend to underestimate the likelihood that natural hazards will have negative impacts on them (March and Shapira 1987; Sitkin and Weingart 1995), knowledge about system vulnerability can serve to raise vigilance and generate support for strategic changes to avert possible damage. Capital Investment Resource-intensive public organizations often find their material resources and infrastructure insufficiently designed for and severely damaged by dramatic variability in weather conditions. In the case of transit agencies, for example, excessive rainfall can flood tracks, bus ways, tunnels and stations, and hurricanes often cause power disruption, vehicle crashes and sign damage (Hodges 2011; Meyer et al. 2013). The scope and severity of the impacts depend partly on whether or not countermeasures such as redundancy are in place (Little 2002; McDaniels et al. 2008). For aged, under-designed and weather-sensitive infrastructure, capital investment is needed to reinforce, upgrade, or replace the critical assets such that they are resistant to extreme events (Fankhauser et al. 1999; Hodges 2011). Vulnerability assessment uses a composite of discrete response questionnaire items asking the respondent if his/her organization: (1) assessed the agency’s vulnerability to extreme weather; (2) estimated the costs of responding to an extreme weather event; and (3) conducted or contracted research on the risks of extreme weather events (1 = yes, 0 = no). Capital investment comprises four discrete items: (1) invested in weather-smart equipment and technologies, such as sensors that detect changes in pressure and temperature in materials; (2) invested in information and communication technologies; (3) invested in back-up power supplies and equipment; and (4) invested in weather-proof infrastructure improvement or retrofitting projects (e.g., strengthening parts of a building, improving stations or tracks). The two outcome variables—vulnerability assessment and capital investment—are constructed as two latent factors underlying the corresponding binary indicators. The variable Exposure is a composite index of a set of questionnaire items about the organization’s experience with extreme weather by event type, where extreme weather was defined in the survey as “unusually severe storms, floods, heat waves or other weather incidents that lie outside of historical norms or experience.” Respondents were asked: “During the last two years, about how many times have the following extreme weather events occurred in your transit service area?” The list of extreme weather events includes: extreme cold temperatures, extreme heat waves, floods, hurricanes/tropical storms, severe rainstorms/thunderstorms, tides/storm surges, extreme high winds, tornadoes, and extreme snow storms (Scale: 0 = never, 1 = once, 2 = two to three times, 3 = more than three times). Although the data are limited, we believe the frequency of extreme events as defined in the survey provides a reasonable measure of exposure that is sufficient to trigger responses. For each type of event experienced, respondents were subsequently asked to rate the adverse impact on their organization. The question asked: “Considering the extreme weather events that have happened in your area in the previous two years, has the level of adverse impact been catastrophic, major, moderate, minor, or no impact?” (Scale: 0 = no impact, 1 = minor, 2 = moderate, 3 = major, 4 = catastrophic). The variable impact is measured as the mean response across all events. The 2-year time span was designed in the survey to reduce recall bias and as a stable proxy measure of the longer-term location-based pattern of extreme weather the agencies experienced. Weather patterns are fundamentally localized and relatively consistent over the short term. NOAA extreme weather data confirm that while extreme weather trends are gradually increasing over the long term, they are relatively stable within a short segment of time such as 2–5 years (NOAA 2016; Smith 2017). Therefore, the 2-year snapshot provides a strong location-based measure of extreme events. We also believe that organizational adaptation—nonincremental changes from existing routines—needs to happen relatively rapidly enough so that agencies are able to leverage windows of opportunity opened by major extreme weather events and their impacts. Kimrey’s (2016) extensive literature review demonstrates that the political issue-attention for extreme events usually lasts only 2–3 years. The required timeliness of adaptive responses strengthens our confidence about using the 2-year time frame for exposure and impact to test the hypotheses. Risk perception was measured as an index of three survey questions asking about the level of agreement with the statements: (1) “Most people in my agency recognize that extreme weather events are becoming more frequent.”; 2) “My agency is increasingly concerned about the impact of extreme weather events on our transit infrastructure.”; and 3) “My agency is increasingly concerned about the impact of extreme weather events on our transit operations.” (Scale: 1 = strongly disagree, 2 = disagree, 3 = neither disagree nor agree, 4 = agree, 5 = strongly agree). Risk perception is constructed as a latent factor underlying the three ordinal response variables. We measure organizational risk perception based on responses of upper-level managers reporting about their organizations. The top managers surveyed serve in departments that are most affected by extreme events. We adopted an upper echelon perspective for conceptualizing the organizational risk perception measure because top management cognitions can causally determine organizational strategic choices and outcomes (Hambrick 2007; Hambrick and Mason 1984). In their classic work on organizational interpretation, Daft and Weick (1984) specifically stress the role of top management in synthesizing information from nested subsystems and formulating the interpretation for the organizational system as a whole. They note (1984, 285): “when one speaks of organizational interpretation one really means interpretation by a relatively small group at the top of the organizational hierarchy.” Control Variables Given the variation in complexity of transit systems, the estimations included a number of control variables. Bus only is coded as 1 if the agency operates bus services but not heavy or light rail. Infrastructure-based capacity is an index of responses to one multi-item question asking respondents to assess the quality of the transit-relevant infrastructure in their service area. Items included: bus transit services, bus station structures and shelters, commuter rail transit services, rail control systems, rail bridges, structures and tunnels, rail track, switches and track work, rail station structures and platforms, streets, roads and highways, roadway bridges, structures and tunnels, transit facility ventilation systems, transit maintenance equipment, electrification/power system, communication systems, drainage systems, and revenue/fare collection systems. Items were automatically populated in the survey depending upon the type of transit services (e.g., bus only system respondents did not receive the rail-specific questions) and respondents were given definitions of the five quality categories (Scale: 1 = poor, 2 = marginal, 3 = adequate, 4 = good, 5 = excellent; see Appendix 1 for definitions). Additional data to capture agency characteristics were collected from the NTD and the 2010–2014 US Census, matched by county code. Authority is coded 1 if the agency is an independent transit authority and 0 if the agency is affiliated to a city or state government. Organizational size, used as a proxy of an organization’s comprehensive capacity for adaptation, is measured by the natural log of an agency’s annual total funding in 2013, including funding from directly generated revenue as well as funding from local, state, and federal governments. Density is the natural log of service population normalized by service area. Commute time serves as an indicator of the demand for transit services and uses the mean travel time to work in minutes. Median household income is the natural log of median household income in the county the agency is located. Finally, a set of the dummy variables (1 = Yes, 0 = No) were used to indicate the Census region the agency is located in: Northeast, Midwest, South and West, respectively. Descriptive statistics for all variables except risk perception are displayed in Table 1. Since the mean and variance of ordinal variables have no meaning (Jöreskog 1994), we report the number of respondents and the univariate proportion for each category of the three indicators in Appendix 2. Table 1. Descriptive Statistics Variable Mean SD Min Max Size Vulnerability assessment  Vulnerability assessment 1 0.55 0.50 0 1 299  Vulnerability assessment 2 0.25 0.43 0 1 300  Vulnerability assessment 3 0.09 0.28 0 1 303 Capital investment  Capital investment 1 0.11 0.31 0 1 299  Capital investment 2 0.62 0.49 0 1 302  Capital investment 3 0.72 0.45 0 1 301  Capital investment 4 0.28 0.45 0 1 301 Exposure 9.94 4.94 0 24 304 Impacts 1.29 0.66 0 3 295 Bus only 0.74 0.44 0 1 304 Infrastructure-based capacity 3.20 0.60 1 5 304 Authority 0.64 0.48 0 1 304 Organization size 17.33 1.49 15 23 304 Density 7.50 1.29 −0.68a 12 304 Median household income (log) 10.89 0.21 10 12 304 Commute time 24.40 4.71 15 42 304 Northeast 0.13 0.33 0 1 304 Midwest 0.22 0.42 0 1 304 South 0.33 0.47 0 1 304 West 0.32 0.47 0 1 304 Variable Mean SD Min Max Size Vulnerability assessment  Vulnerability assessment 1 0.55 0.50 0 1 299  Vulnerability assessment 2 0.25 0.43 0 1 300  Vulnerability assessment 3 0.09 0.28 0 1 303 Capital investment  Capital investment 1 0.11 0.31 0 1 299  Capital investment 2 0.62 0.49 0 1 302  Capital investment 3 0.72 0.45 0 1 301  Capital investment 4 0.28 0.45 0 1 301 Exposure 9.94 4.94 0 24 304 Impacts 1.29 0.66 0 3 295 Bus only 0.74 0.44 0 1 304 Infrastructure-based capacity 3.20 0.60 1 5 304 Authority 0.64 0.48 0 1 304 Organization size 17.33 1.49 15 23 304 Density 7.50 1.29 −0.68a 12 304 Median household income (log) 10.89 0.21 10 12 304 Commute time 24.40 4.71 15 42 304 Northeast 0.13 0.33 0 1 304 Midwest 0.22 0.42 0 1 304 South 0.33 0.47 0 1 304 West 0.32 0.47 0 1 304 aThe negative log value of density comes from three observation within the same agency, with a density of about 0.50. However, the organization has normal values in other organizational attributes, so we kept the three observations in the analysis. For robust check purposes, we also run the model without the three observations and the results did not qualitatively change. View Large Table 1. Descriptive Statistics Variable Mean SD Min Max Size Vulnerability assessment  Vulnerability assessment 1 0.55 0.50 0 1 299  Vulnerability assessment 2 0.25 0.43 0 1 300  Vulnerability assessment 3 0.09 0.28 0 1 303 Capital investment  Capital investment 1 0.11 0.31 0 1 299  Capital investment 2 0.62 0.49 0 1 302  Capital investment 3 0.72 0.45 0 1 301  Capital investment 4 0.28 0.45 0 1 301 Exposure 9.94 4.94 0 24 304 Impacts 1.29 0.66 0 3 295 Bus only 0.74 0.44 0 1 304 Infrastructure-based capacity 3.20 0.60 1 5 304 Authority 0.64 0.48 0 1 304 Organization size 17.33 1.49 15 23 304 Density 7.50 1.29 −0.68a 12 304 Median household income (log) 10.89 0.21 10 12 304 Commute time 24.40 4.71 15 42 304 Northeast 0.13 0.33 0 1 304 Midwest 0.22 0.42 0 1 304 South 0.33 0.47 0 1 304 West 0.32 0.47 0 1 304 Variable Mean SD Min Max Size Vulnerability assessment  Vulnerability assessment 1 0.55 0.50 0 1 299  Vulnerability assessment 2 0.25 0.43 0 1 300  Vulnerability assessment 3 0.09 0.28 0 1 303 Capital investment  Capital investment 1 0.11 0.31 0 1 299  Capital investment 2 0.62 0.49 0 1 302  Capital investment 3 0.72 0.45 0 1 301  Capital investment 4 0.28 0.45 0 1 301 Exposure 9.94 4.94 0 24 304 Impacts 1.29 0.66 0 3 295 Bus only 0.74 0.44 0 1 304 Infrastructure-based capacity 3.20 0.60 1 5 304 Authority 0.64 0.48 0 1 304 Organization size 17.33 1.49 15 23 304 Density 7.50 1.29 −0.68a 12 304 Median household income (log) 10.89 0.21 10 12 304 Commute time 24.40 4.71 15 42 304 Northeast 0.13 0.33 0 1 304 Midwest 0.22 0.42 0 1 304 South 0.33 0.47 0 1 304 West 0.32 0.47 0 1 304 aThe negative log value of density comes from three observation within the same agency, with a density of about 0.50. However, the organization has normal values in other organizational attributes, so we kept the three observations in the analysis. For robust check purposes, we also run the model without the three observations and the results did not qualitatively change. View Large An overwhelming majority of organizations, over 95%, have experienced at least one extreme weather event and suffered the impacts during the past 2 years. Overall, organizations are relatively neutral about their perception of risks associated with extreme weather events. Approximately 25% agencies provide rail services and 64% agencies are independent transit authority organizations. Analysis And Results Method Because the model assumes multiple relationships and mixes latent and observed variables, SEM is a suitable analytic method for this study (MacKinnon 2008). SEM is preferred over regression analysis because of its capacity to estimate constructs by separating the unique variance of observed items from shared items (Kline 2015). It also has the advantage of incorporating indirect effects when mediating variables are included. Because the three adaptation measures and risk perception were constructed using categorical or ordinal variables, the relationships between the manifest variables and the latent factor are considered nonlinear. A common practice to account for the nonlinearity is to replace the observed categorical variables with their underlying latent and continuous factors (Hoyle 2012). Therefore, we analyzed the model using the weighted least square means and variance adjusted (WLSMV) estimator in Mplus 7.4 (Muthén 1993). The WLSMV estimator uses a probit function to link the underlying continuous latent variable to the observed categorical indicators, allowing analysis of relationships between the underlying latent variables. Compared to the commonly used maximum likelihood (ML) estimation, WLSMV estimation has more accurate factor loading and model fit when the number of categories is small (e.g., 2 or 3 categories) (Beauducel and Herzberg 2006). Additionally, because the transit agency is the unit of analysis and there is more than one response from some responding agencies, clustered standard errors by agency were used to account for the nested data structure (Wooldridge 2003). We included all control variables in all paths to estimate the relationships of primary interest. We also weighted the observations to ensure equal representation of each agency in the analysis. Sem Results Measurement Model A confirmatory factor analysis (CFA) on the categorical indicators shows good model fit (RMESA = 0.050 with 90% CI [0.028, 0.071], CFI = 0.984, TLI = 0.978, WRMR = 0.811). The fit indices satisfy the general cutoff points for measurement models with WLSMV estimation: RMSEA ≤ 0.05, CFI ≥ 0.96, TLI ≥ 0.95, and WRMR ≤ 1.0 (Yu 2002). The chi-square test statistic (χ2 = 56.672, df = 32, p value = .0046) indicates unsatisfactory fit, but it can be attributed to its sensitivity to sample size and is usually not considered sufficient evidence to reject the model when contrary to other fit indices (West et al. 2012). Table 2 shows the standardized factor loadings for each scale. The R-square values indicate how much variance in the indicator is explained by the underlying latent factor. The standardized factor loadings are relatively large, ranging from 0.65 to 0.96. The Wald tests for each pair of the latent factors are all highly significant with p-value <.000. Therefore, both convergent and discriminant validity are achieved for the latent constructs. Table 2. Measurement Model Factor and Indicators Estimate SE p-Value R-Square Vulnerability assessment  Vulnerability assessment 1 0.681 0.075 .000 0.464  Vulnerability assessment 2 0.783 0.066 .000 0.613  Vulnerability assessment 3 0.846 0.095 .000 0.716 Capital investment  Capital investment 1 0.649 0.096 .000 0.421  Capital investment 2 0.698 0.072 .000 0.487  Capital investment 3 0.800 0.068 .000 0.640  Capital investment 4 0.771 0.069 .000 0.594 Risk perception Risk perception 1 0.822 0.028 .000 0.676 Risk perception 2 0.704 0.036 .000 0.496 Risk perception 3 0.964 0.026 .000 0.929 Factor and Indicators Estimate SE p-Value R-Square Vulnerability assessment  Vulnerability assessment 1 0.681 0.075 .000 0.464  Vulnerability assessment 2 0.783 0.066 .000 0.613  Vulnerability assessment 3 0.846 0.095 .000 0.716 Capital investment  Capital investment 1 0.649 0.096 .000 0.421  Capital investment 2 0.698 0.072 .000 0.487  Capital investment 3 0.800 0.068 .000 0.640  Capital investment 4 0.771 0.069 .000 0.594 Risk perception Risk perception 1 0.822 0.028 .000 0.676 Risk perception 2 0.704 0.036 .000 0.496 Risk perception 3 0.964 0.026 .000 0.929 View Large Table 2. Measurement Model Factor and Indicators Estimate SE p-Value R-Square Vulnerability assessment  Vulnerability assessment 1 0.681 0.075 .000 0.464  Vulnerability assessment 2 0.783 0.066 .000 0.613  Vulnerability assessment 3 0.846 0.095 .000 0.716 Capital investment  Capital investment 1 0.649 0.096 .000 0.421  Capital investment 2 0.698 0.072 .000 0.487  Capital investment 3 0.800 0.068 .000 0.640  Capital investment 4 0.771 0.069 .000 0.594 Risk perception Risk perception 1 0.822 0.028 .000 0.676 Risk perception 2 0.704 0.036 .000 0.496 Risk perception 3 0.964 0.026 .000 0.929 Factor and Indicators Estimate SE p-Value R-Square Vulnerability assessment  Vulnerability assessment 1 0.681 0.075 .000 0.464  Vulnerability assessment 2 0.783 0.066 .000 0.613  Vulnerability assessment 3 0.846 0.095 .000 0.716 Capital investment  Capital investment 1 0.649 0.096 .000 0.421  Capital investment 2 0.698 0.072 .000 0.487  Capital investment 3 0.800 0.068 .000 0.640  Capital investment 4 0.771 0.069 .000 0.594 Risk perception Risk perception 1 0.822 0.028 .000 0.676 Risk perception 2 0.704 0.036 .000 0.496 Risk perception 3 0.964 0.026 .000 0.929 View Large Structural Model There are 306 responses (35.5% total response rate) from 199 transit agencies (72.9% of all agencies have at least one response) in our dataset. The missing values in two cases resulted in a final set of 304 observations from 199 transit agencies, which we used in the entire SEM model. The structural model achieved good model fit (χ2 = 151.263, df = 116, p-value = .0155, RMESA = 0.032 with 90% CI [0.015, 0.045], CFI = 0.966, TLI = 0.950, WRMR = 0.795). Those indices satisfy the general cutoff points for structural models using WLSMV estimation: p-value for Chi-square test ≥ 0.01, RMSEA ≤ 0.06, CFI ≥ 0.96, TLI ≥ 0.95, and WRMR ≤ 1.0 (Yu 2002). Table 3 reports the SEM results. The parameter estimates are followed by the standard errors in the parentheses. Northeast was used as the baseline for the effects of the region dummies. It is important to note that impact is positively associated with an organization’s exposure to extreme weather, yet negatively associated with its infrastructure quality. This is consistent with our theoretical framework which expects greater capacity to buffer the negative impacts of extreme events on the affected organization. Table 3. SEM Results Extreme Weather Impact Exposure 0.041 (0.009)*** Bus only −0.256 (0.099)** Infrastructure quality −0.127 (0.070)# Authority −0.111 (0.087) Organization size (log) 0.048 (0.033) Density (log) −0.047 (0.035) Median household income (log) 0.061 (0.265) Commute time −0.001 (0.012) Midwest −0.190 (0.137) South −0.175 (0.140) West −0.352 (0.141)* Risk Perception Exposure 0.034 (0.012)** Impact 0.322 (0.071)*** Bus only 0.133 (0.159) Infrastructure quality −0.043 (0.089) Authority −0.082 (0.111) Organization size (log) 0.118 (0.053)* Density (log) −0.038 (0.033) Median household income (log) 0.167 (0.311) Commute time 0.013 (0.015) Midwest −0.139 (0.195) South −0.043 (0.183) West −0.303 (0.195) Organizational Adaptation Vulnerability Assessment Capital Investment Exposure 0.033 (0.015)* 0.028 (0.012)* Impacts 0.025 (0.094) −0.036 (0.061) Risk perception 0.360 (0.075)*** 0. 168(0.061)** Bus only −0.139 (0.164) −0.130 (0.116) Infrastructure quality 0.027 (0.107) −0.014 (0.063) Authority −0.446 (0.152)** 0.032 (0.084) Organization size (log) 0.158 (0.060)** 0.037 (0.037) Density (log) 0.066 (0.062) −0.030 (0.034) Median household income (log) 0.100 (0.365) −0.352 (0.228) Commute time −0.006 (0.017) 0.025 (0.012)* Midwest −0.056 (0.218) 0.101 (0.137) South 0.266 (0.217) 0.135 (0.128) West 0.148 (0.222) 0.189 (0.138) Extreme Weather Impact Exposure 0.041 (0.009)*** Bus only −0.256 (0.099)** Infrastructure quality −0.127 (0.070)# Authority −0.111 (0.087) Organization size (log) 0.048 (0.033) Density (log) −0.047 (0.035) Median household income (log) 0.061 (0.265) Commute time −0.001 (0.012) Midwest −0.190 (0.137) South −0.175 (0.140) West −0.352 (0.141)* Risk Perception Exposure 0.034 (0.012)** Impact 0.322 (0.071)*** Bus only 0.133 (0.159) Infrastructure quality −0.043 (0.089) Authority −0.082 (0.111) Organization size (log) 0.118 (0.053)* Density (log) −0.038 (0.033) Median household income (log) 0.167 (0.311) Commute time 0.013 (0.015) Midwest −0.139 (0.195) South −0.043 (0.183) West −0.303 (0.195) Organizational Adaptation Vulnerability Assessment Capital Investment Exposure 0.033 (0.015)* 0.028 (0.012)* Impacts 0.025 (0.094) −0.036 (0.061) Risk perception 0.360 (0.075)*** 0. 168(0.061)** Bus only −0.139 (0.164) −0.130 (0.116) Infrastructure quality 0.027 (0.107) −0.014 (0.063) Authority −0.446 (0.152)** 0.032 (0.084) Organization size (log) 0.158 (0.060)** 0.037 (0.037) Density (log) 0.066 (0.062) −0.030 (0.034) Median household income (log) 0.100 (0.365) −0.352 (0.228) Commute time −0.006 (0.017) 0.025 (0.012)* Midwest −0.056 (0.218) 0.101 (0.137) South 0.266 (0.217) 0.135 (0.128) West 0.148 (0.222) 0.189 (0.138) Unstandardized coefficients are reported. Three hundred four observations from 199 agencies; Reference category: Northeast. #p < .10, *p < .05, **p < .01, ***p < .001. View Large Table 3. SEM Results Extreme Weather Impact Exposure 0.041 (0.009)*** Bus only −0.256 (0.099)** Infrastructure quality −0.127 (0.070)# Authority −0.111 (0.087) Organization size (log) 0.048 (0.033) Density (log) −0.047 (0.035) Median household income (log) 0.061 (0.265) Commute time −0.001 (0.012) Midwest −0.190 (0.137) South −0.175 (0.140) West −0.352 (0.141)* Risk Perception Exposure 0.034 (0.012)** Impact 0.322 (0.071)*** Bus only 0.133 (0.159) Infrastructure quality −0.043 (0.089) Authority −0.082 (0.111) Organization size (log) 0.118 (0.053)* Density (log) −0.038 (0.033) Median household income (log) 0.167 (0.311) Commute time 0.013 (0.015) Midwest −0.139 (0.195) South −0.043 (0.183) West −0.303 (0.195) Organizational Adaptation Vulnerability Assessment Capital Investment Exposure 0.033 (0.015)* 0.028 (0.012)* Impacts 0.025 (0.094) −0.036 (0.061) Risk perception 0.360 (0.075)*** 0. 168(0.061)** Bus only −0.139 (0.164) −0.130 (0.116) Infrastructure quality 0.027 (0.107) −0.014 (0.063) Authority −0.446 (0.152)** 0.032 (0.084) Organization size (log) 0.158 (0.060)** 0.037 (0.037) Density (log) 0.066 (0.062) −0.030 (0.034) Median household income (log) 0.100 (0.365) −0.352 (0.228) Commute time −0.006 (0.017) 0.025 (0.012)* Midwest −0.056 (0.218) 0.101 (0.137) South 0.266 (0.217) 0.135 (0.128) West 0.148 (0.222) 0.189 (0.138) Extreme Weather Impact Exposure 0.041 (0.009)*** Bus only −0.256 (0.099)** Infrastructure quality −0.127 (0.070)# Authority −0.111 (0.087) Organization size (log) 0.048 (0.033) Density (log) −0.047 (0.035) Median household income (log) 0.061 (0.265) Commute time −0.001 (0.012) Midwest −0.190 (0.137) South −0.175 (0.140) West −0.352 (0.141)* Risk Perception Exposure 0.034 (0.012)** Impact 0.322 (0.071)*** Bus only 0.133 (0.159) Infrastructure quality −0.043 (0.089) Authority −0.082 (0.111) Organization size (log) 0.118 (0.053)* Density (log) −0.038 (0.033) Median household income (log) 0.167 (0.311) Commute time 0.013 (0.015) Midwest −0.139 (0.195) South −0.043 (0.183) West −0.303 (0.195) Organizational Adaptation Vulnerability Assessment Capital Investment Exposure 0.033 (0.015)* 0.028 (0.012)* Impacts 0.025 (0.094) −0.036 (0.061) Risk perception 0.360 (0.075)*** 0. 168(0.061)** Bus only −0.139 (0.164) −0.130 (0.116) Infrastructure quality 0.027 (0.107) −0.014 (0.063) Authority −0.446 (0.152)** 0.032 (0.084) Organization size (log) 0.158 (0.060)** 0.037 (0.037) Density (log) 0.066 (0.062) −0.030 (0.034) Median household income (log) 0.100 (0.365) −0.352 (0.228) Commute time −0.006 (0.017) 0.025 (0.012)* Midwest −0.056 (0.218) 0.101 (0.137) South 0.266 (0.217) 0.135 (0.128) West 0.148 (0.222) 0.189 (0.138) Unstandardized coefficients are reported. Three hundred four observations from 199 agencies; Reference category: Northeast. #p < .10, *p < .05, **p < .01, ***p < .001. View Large Meanwhile, among the control variables only organizational size had a statistically significant effect on risk perception such that risk perceptions are higher in larger organizations. Among the control variables, organization size is positively associated with vulnerability assessment but not associated with capital assessment. Results also show that independent organizations (transit authority agencies) are less likely to conduct vulnerability assessment compared to agencies affiliated with a city or state government, and that agencies with higher commute time are more likely to undertake capital investment. Testing Hypotheses The relationships among variables of primary interest are presented in Figure 3. Standardized coefficients are reported to facilitate comparison of the relative effect of each predictor. Figure 3. View largeDownload slide SEM Analysis Result: Mediating Role of Risk Perception Figure 3. View largeDownload slide SEM Analysis Result: Mediating Role of Risk Perception Hypothesis 1 postulates a positive relationship between an organization’s exposure to extreme events on the impacts it experienced. Our analysis supports this hypothesis, with a significant and positive effect of exposure to extreme weather on impacts. Hypothesis 2 posits that as an organization suffers greater impacts from extreme events, it is more likely to build adaptive capacity in preparation for future shocks. The results show nonsignificant effects of impacts on the two adaptive strategies, therefore suggesting no support for Hypothesis 2. These findings indicate that an organization’s increased exposure to extreme events leads to greater negative impacts, but the impacts do not directly lead to organizational adaptive responses. We interpret this finding by separating the impact from organizational learning and decision processes. Longer-term initiatives that are adaptive may require organizations to perceive a substantially different environmental context in which the harm associated with maintaining the status quo is greater than the harm associated with adaptive change. As exposure increases and causes more significant impacts, only those organizations that cognitively process a different level of risk will adapt. We also believe there is an alternative explanation for this finding: it is possible that impact can both motivate and undermine an organization’s capacity to adapt. This may be especially true when organizations focus on immediate survival and on maintaining the status quo without embracing a longer-term vision to build adaptive capacity. The “hold-the-line” practice has been widely observed as diminishing capital resources for longer-term and hampering an organization’s learning and adaptation from recurring major perturbations (Adger et al. 2011; Linnenluecke et al. 2012; Winn et al. 2011). Hypothesis 3: expects that exposure to extreme events will directly affect an organization’s adaptive capacity building due to the learning mechanisms. Exposure turns out to be a positive predictor of an organization’s capital investment and of vulnerability assessment. Thus, our results lend support to Hypothesis 3. Hypothesis 4: focuses on how risk perception mediates the effect of an organization’s exposure to and impact from extreme events on its adaptive responses. The path coefficients from exposure and impacts to risk perception are significant and positive, so are the effects of risk perception on an organization’s implementation of vulnerability assessment and capital investment. In light of the asymmetry of the distribution of the product of multiple path coefficients, we applied bootstrapping to derive the confidence interval of the mediated effects through risk perception (MacKinnon 2008). There are two paths through which exposure affects adaption via risk perception: one path from exposure to risk perception and then adaptation, and one path from exposure to impact and then risk perception and adaptation. The results indicate that an organization’s risk perception effectively mediates the effect of exposure on vulnerability assessment (standardized β = 0.103, 95% CI = [0.038, 0.181]) and capital investment (standardized β = 0.076, 95% CI = [0.010, 0.157]). It also shows that risk perception significantly mediates the effect of impact on vulnerability assessment (standardized β = 0. 097, 95% CI = [0.036, 0.164]) and capital investment (standardized β = 0.071, 95% CI = [0.011, 0.143]). Therefore, the model shows consistent support for Hypothesis 4. The findings for Hypotheses 3 and 4 reveal two mechanisms through which organizational adaptive responses occur: the direct effect of increased exposure to extreme events as well as the mediated effects of risk perception. Both mechanisms imply organizational learning through increased problem familiarity and organizational sense-making, respectively. However, the two learning mechanisms vary in their relative strength. While the relationship between exposure and capital investment is highly significant, the statistical relationship and coefficient between exposure and vulnerability assessment are weaker. Because exposure may result in physical damage that requires repairs, it is possible that capital investment decisions are affected more by actual damage. Nevertheless, both vulnerability assessment and capital investment require cognitive change and the trigger provided by heightened risk perception (based on impact and exposure). Conclusion This study sets out to examine public organizations’ experience with extreme events and to explain the variations in their adaptive responses. The conceptual framework (Figure 1) integrates organizational adaptation and learning literature to establish an initial organizational response model that captures the essence of longer-term adaptation to reduce vulnerability to recurring extreme events. The theoretical model (Figure 2) addresses key interrelated questions surrounding the reasons why some public organizations adapt to extreme events and others do not, or do so more slowly. The model posits that exposure leads to capacity-moderated impacts (i.e., performance gap), but subsequent adaptation through capacity development is mediated by organization risk perception. Our findings support three of the four hypotheses generated from the model and deeper literature: increased exposure is associated with increased impacts (i.e., performance gap). Exposure is positively associated with organizational adaptation, and risk perception mediates the effect of exposure and impact on adaptive behavior. Infrastructure-based capacity also helps buffer the negative impacts, further supporting our overall model. The study contributes to the broader literature in several ways. It captures the common patterns across extreme events and takes an organizational approach to address the challenges associated with adaptation to reduce vulnerability in public organizations. This responds to the recent literature by connecting the study on extreme events with organizational theory to enable a more systematic and generic understanding and treatment of extreme events (Boin and Van Eeten 2013; Christensen et al. 2016; Fischbacher-Smith 2010; Roux-Dufort 2007). As such, this research provides one of the few theoretically informed, quantitative studies on core topics concerning extreme events (Boin and Van Eeten 2013; Christensen et al. 2016). Our findings, in conjunction with our broader conceptual model, can be interpreted as evidence for perception-mediated learning in which adaptation requires not simply exposure to a phenomenon, but also the cognitive understanding that longer-term performance of an organization depends on some form of purposive adaptation. Both exposure and the effects of exposure on performance create opportunities for organizations to raise questions about their ability to perform. When severe events increase in frequency, and when emergency responses become more frequent, ineffective and costly, organizations begin to make sense of the pattern of the challenges and realize that longer-term investment in capacity is necessary for sustainable operation. The study may point to a cognition-based stepwise learning model in which organization commitment to change is incremental and depends on the recognition that exogenous shocks are systematic, solvable and require new investments. A baseline exists when an organization responds to extreme events using routinized or programmed emergency response actions. Increased frequency of exogenous shocks may increase familiarity with performance gaps, demonstrate fundamental limitations in capacity, and stimulate learning manifested as increased perceptions of risk. Increased perception of risk may facilitate organizational commitment to undertake some form of adaptive behavior to reduce vulnerability. The stronger linkage between exposure and capital investment combined with findings that impacts are only indirectly related to adaptation through risk perception provide further support that commitment to information gathering through vulnerability assessment may be an intermediary step, before capital investment-based adaptation. Practically, the study reveals the pitfalls in assuming that organizational adaptation to extreme events will occur spontaneously with growth in impacts from extreme events. Instead, the effect of extreme events on adaptation behavior is most likely channeled through a cognitive process wherein the risks are perceived and appropriately interpreted by the affected organization. The significant role of risk perception gives important room for management intervention. Since people are more sensitive to threats than opportunities in undertaking larger-scale internal responses (Dutton and Jackson 1987), managers can affect organization-wide risk perceptions by actively recognizing and framing threats, seeking to systematically collect and interpret new information and establishing it as part of the organizational memory. Managers can prime organizational members to better identify extreme events, recognize patterns, and develop solutions that reduce threats (Weick 1977). Managers may adopt a participatory approach to promote shared understanding and interpretation of extreme events, and pave the way for potentially adaptive solutions. The participatory efforts can be more successfully implemented by highly trusted individuals in the organization (Dutton and Jackson 1987). We acknowledge the limitations of our study. Our empirical analysis focuses on transit agencies, which may reduce generalizability of the findings of this study to other organizations. However, to some extent, the national-level approach in which agencies are surveyed across a wide range of weather conditions and extreme events and the similarity of transit agencies with many other resource intensive public agencies (i.e., utility, power, waste management) reduces this concern. Reliance on survey data for the construction of key variables in the estimation may raise concerns about common source bias. The use of survey data is valid for two reasons. First, other sources of data on organizational experience with extreme weather events, especially impacts, are difficult to collect. Additionally, because risk is fundamentally a perceptual construct, it may be best to rely on perceptual measures collected in surveys. Previous studies have suggested that self-reported data can provide valid indicators of organizational properties (Lincoln and Zeitz 1980; Moynihan and Pandey 2005; Pandey and Wright 2006). That many coefficients in our model are low and not significant can also help alleviate the common method bias concern (George and Pandey 2017). Another limitation has to do with the use of cross-sectional data to test the mediation effect. We are limited by data availability to test a three-wave longitudinal data as suggested by the literature (Cole and Maxwell 2003). However, given the scarcity of data on how public organizations adapt to extreme events, a growing and salient challenge to public organizations worldwide, we believe that the survey and our analysis still shine light on this important topic. Moreover, a major concern with longitudinal data relates to the difficulty of correctly specifying the time lag between the data collection points such that the data actually captures meaningful variation in the variable of interest, instead of stochastic changes due to time lapse (Cole and Maxwell 2003; Ployhart and Vandenberg 2010). In studies of adaptation, the time lag between extreme weather exposure, impact, and risk perception is difficult to pinpoint (Berkhout 2012; Christianson et al. 2009; Weick 1993). This makes specification of the exact duration of the time lag challenging such that longitudinal data and methods may not be necessarily superior. To the extent that the time interval is small, the use of cross-sectional data instead might be a viable representation of reality in this case (Wong and Law 1999). This study raises several new questions for further work. Are public organizations actually learning from increasingly frequent and severe events in ways that fundamentally alter investment patterns over time? This study is suggestive, but not conclusive. What other adaptation strategies are organizations undertaking besides those examined in this study? In large complex organizations, at what point do occasional shocks become recurrent; at what point do frequency, magnitude, and scope force organizational attention? Is there a step-wise learning process in which commitment is contingent on risk perception and adaptive actions are nested? Given the rise in “extreme events” that cause significant disruption (Boin and Lodge 2016; Comfort et al. 2012; Tierney 2014) future research should begin to address these issues. Acknowledgment This research was made possible through generous support by the Federal Transit Administration, US Department of Transportation. We would also like to acknowledge the anonymous reviewers for their helpful comments and suggestions. References Adger , W. Neil , Katrina Brown , Donald R. Nelson , Fikret Berkes , Hallie Eakin , Carl Folke , Kathleen Galvin , et al. 2011 . Resilience implications of policy responses to climate change: Resilience implications of policy responses to climate change . Wiley Interdisciplinary Reviews: Climate Change 2 : 757 – 66 . Google Scholar CrossRef Search ADS Beauducel , André , and Philipp Yorck Herzberg . 2006 . On the performance of maximum likelihood versus means and variance adjusted weighted least squares estimation in CFA . Structural Equation Modeling 13 : 186 – 203 . Google Scholar CrossRef Search ADS Berkes , Fikret . 2007 . Understanding uncertainty and reducing vulnerability: Lessons from resilience thinking . Natural Hazards 41 : 283 – 95 . Google Scholar CrossRef Search ADS Berkes , Fikret , and Carl Folke . 2002 . Back to the future: Ecosystem dynamics and local knowledge. In Panarchy: understanding transformation in human and natural systems, ed. Lance H. Gunderson and Crawford S. Holling, 121–146. Washington: Island Press. Berkhout , Frans . 2012 . Adaptation to climate change by organizations . Wiley Interdisciplinary Reviews: Climate Change 3 : 91 – 106 . Google Scholar CrossRef Search ADS Birkland , Thomas A . 1997 . After Disaster: Agenda Setting, Public Policy, and Focusing Events . Washington, D.C.: Georgetown Univ. Press . ———. 2009 . Disasters, lessons learned, and fantasy documents . Journal of Contingencies and Crisis Management 17 : 146 – 56 . CrossRef Search ADS Boin , Arjen , and Paul ‘t Hart . 2003 . Public leadership in times of crisis: Mission impossible ? Public Administration Review 63 : 544 – 53 . Google Scholar CrossRef Search ADS Boin , Arjen , and Martin Lodge . 2016 . Designing resilient institutions for transboundary crisis management: A time for public administration . Public Administration 94 : 289 – 98 . Google Scholar CrossRef Search ADS Boin , Arjen , and Allan McConnell . 2007 . Preparing for critical infrastructure breakdowns: The limits of crisis management and the need for resilience . Journal of Contingencies and Crisis Management 15 : 50 – 9 . Google Scholar CrossRef Search ADS Boin , Arjen , Eric Stern , and Bengt Sundelius . 2016 . The politics of crisis management: Public leadership under pressure . Cambridge, UK: Cambridge Univ. Press . Boin , Arjen , and Michel JG Van Eeten . 2013 . The resilient organization . Public Management Review 15 : 429 – 45 . Google Scholar CrossRef Search ADS Cairney , Paul , Kathryn Oliver , and Adam Wellstead . 2016 . To bridge the divide between evidence and policy: Reduce ambiguity as much as uncertainty . Public Administration Review 76 : 399 – 402 . Google Scholar CrossRef Search ADS Christensen , Tom , Per Lægreid , and Lise H. Rykkja . 2016 . Organizing for crisis management: Building governance capacity and legitimacy . Public Administration Review 76 : 887 – 97 . Google Scholar CrossRef Search ADS Christianson , Marlys K. , Maria T. Farkas , Kathleen M. Sutcliffe , and Karl E. Weick . 2009 . Learning through rare events: Significant interruptions at the Baltimore & Ohio Railroad Museum . Organization Science 20 : 846 – 60 . Google Scholar CrossRef Search ADS Cigler , Beverly A . 2007 . The ‘big questions’ of Katrina and the 2005 Great Flood of New Orleans . Public Administration Review 67 ( Suppl 1 ): 64 – 76 . Google Scholar CrossRef Search ADS Clarke , Lee . 1999 . Mission improbable: Using fantasy documents to tame disaster . Chicago, IL: University of Chicago Press . Cole , David A. , and Scott E. Maxwell . 2003 . Testing mediational models with longitudinal data: Questions and tips in the use of structural equation modeling . Journal of Abnormal Psychology 112 : 558–77 . Google Scholar CrossRef Search ADS PubMed Comfort , Louise K . 2002 . Rethinking security: Organizational fragility in extreme events . Public Administration Review 62 ( Suppl 1 ): 98 – 107 . Google Scholar CrossRef Search ADS ———. 2005 . Risk, security and disaster management . Annual Review of Political Science 8 : 335 – 56 . CrossRef Search ADS ———. 2007 . Crisis management in hindsight: Cognition, communication, coordination, and control . Public Administration Review 67 ( Suppl 1 ): 189 – 97 . CrossRef Search ADS Comfort , Louise K. , and Aya Okada . 2013 . Symposium-emergent leadership in extreme events: A knowledge commons for sustainable communities . International Review of Public Administration 18 : 61 – 77 . Google Scholar CrossRef Search ADS Comfort , Louise K. , William L. Waugh , and Beverly A. Cigler . 2012 . Emergency management research and practice in public administration: Emergence, evolution, expansion, and future directions . Public Administration Review 72 : 539 – 47 . Google Scholar CrossRef Search ADS Crichton , Margaret T. , Cameron G. Ramsay , and Terence Kelly . 2009 . Enhancing organizational resilience through emergency planning: Learnings from cross-sectoral lessons . Journal of Contingencies and Crisis Management 17 : 24 – 37 . Google Scholar CrossRef Search ADS Cyert , Richard M. , and James G. March . 1963 . A behavioral theory of the firm , 2nd ed.. Englewood Cliffs, NJ: Prentice Hall . Daft , Richard L. , and Karl E. Weick . 1984 . Toward a model of organizations as interpretation systems . Academy of Management Review 9 : 284 – 95 . Google Scholar CrossRef Search ADS Dekker , Sander , and Dan Hansén . 2004 . Learning under pressure: The effects of politicization on organizational learning in public bureaucracies . Journal of Public Administration Research and Theory 14 : 211 – 30 . Google Scholar CrossRef Search ADS Duncan , Robert B . 1972 . Characteristics of organizational environments and perceived environmental uncertainty . Administrative Science Quarterly 17 : 313 – 27 . Google Scholar CrossRef Search ADS Dutton , Jane E. , and Susan E. Jackson . 1987 . Categorizing strategic issues: Links to organizational action . The Academy of Management Review 12 : 76 – 90 . Google Scholar CrossRef Search ADS Ebert , Suzanne , Orieta Hulea , and David Strobel . 2009 . Floodplain restoration along the lower Danube: A climate change adaptation case study . Climate and Development 1 : 212 – 19 . Google Scholar CrossRef Search ADS Fankhauser , Samuel , Joel B. Smith , and Richard S. J. Tol . 1999 . Weathering climate change: Some simple rules to guide adaptation decisions . Ecological Economics 30 : 67 – 78 . Google Scholar CrossRef Search ADS Fischbacher-Smith , Denis . 2010 . Beyond the worst case scenario: ‘Managing’ the risks of extreme events . Risk Management 12 : 1 – 8 . Google Scholar CrossRef Search ADS Folke , Carl . 2006 . Resilience: The emergence of a perspective for social–ecological systems analyses . Global Environmental Change 16 : 253 – 67 . Google Scholar CrossRef Search ADS Ford , James D. , and Diana King . 2015 . A framework for examining adaptation readiness . Mitigation and Adaptation Strategies for Global Change 20 : 505 – 26 . Google Scholar CrossRef Search ADS Gallagher , Justin . 2014 . Learning about an infrequent event: Evidence from flood insurance take-up in the United States . American Economic Journal: Applied Economics 6 : 206 – 33 . Google Scholar CrossRef Search ADS Garrett , Terence M . 2004 . Whither Challenger, wither Columbia: Management decision making and the knowledge analytic . The American Review of Public Administration 34 : 389 – 402 . Google Scholar CrossRef Search ADS George , Bert , and Sanjay K. Pandey . 2017 . We know the Yin—but where is the Yang? Toward a balanced approach on common source bias in public administration scholarship . Review of Public Personnel Administration 37 : 245 – 70 . Google Scholar CrossRef Search ADS PubMed Grothmann , Torsten , and Anthony Patt . 2005 . Adaptive capacity and human cognition: The process of individual adaptation to climate change . Global Environmental Change 15 : 199 – 213 . Google Scholar CrossRef Search ADS Haigh , Nardia , and Andrew Griffiths . 2011 . Surprise as a catalyst for including climatic change in the strategic environment . Business & Society , 51 : 1552 – 4205 . Hambrick , D. C. , and P. A. Mason . 1984 . Upper echelons: The organization as a reflection of its top managers . Academy of Management Review 9 : 193 – 206 . Google Scholar CrossRef Search ADS Hambrick , Donald C . 2007 . Upper echelons theory: An update . Academy of Management Review 32 : 334 – 43 . Google Scholar CrossRef Search ADS Hegerl , G. C. , F. W. Zwiers , P. Braconnot , N. P. Gillett , Y. Luo , J. A. Marengo Orsini , N. Nicholls , J. E. Penner , and P. A. Stott . 2007 . Understanding and attributing climate change . In Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change , eds. S. Solomon , D. Qin , M. Manning , Z. Chen , M. Marquis , KB Averyt , M. Tignor and HL Miller , 663 – 745 . Cambridge, UK : Cambridge Univ. Press . Henstra , Daniel . 2010 . Evaluating local government emergency management programs: What framework should public managers adopt ? Public Administration Review 70 : 236 – 46 . Google Scholar CrossRef Search ADS Hill , Margot , and Nathan L. Engle . 2013 . Adaptive capacity: Tensions across scales . Environmental Policy and Governance 23 : 177 – 92 . Google Scholar CrossRef Search ADS Hodges , Tina . 2011 . Flooded bus barns and buckled rails: Public transportation and climate change adaptation . Washington, DC : Federal Transit Administration . http://www.fta.dot.gov/research (assessed July 26, 2016). Hoffmann , Volker H. , David C. Sprengel , Andreas Ziegler , Matthias Kolb , and Bruno Abegg . 2009 . Determinants of corporate adaptation to climate change in winter tourism: An econometric analysis . Global Environmental Change 19 : 256 – 64 . Google Scholar CrossRef Search ADS Hoyle , Rick H . 2012 . Handbook of structural equation modeling . New York City, NY: Guilford Press . Jöreskog , Karl G . 1994 . Structural equation modeling with ordinal variables . Multivariate Analysis and Its Applications , 24 : 297 – 310 . Google Scholar CrossRef Search ADS Kahneman , Daniel , and Amos Tversky . 1979 . Prospect theory: An analysis of decision under risk . Econometrica: Journal of the Econometric Society , 47 : 263 – 291 . Google Scholar CrossRef Search ADS Kapucu , Naim . 2005 . Interorganizational coordination in dynamic context: Networks in emergency response management . Connections 26 : 33 – 48 . Kapucu , Naim , and Vener Garayev . 2011 . Collaborative decision-making in emergency and disaster management . International Journal of Public Administration 34 : 366 – 75 . Google Scholar CrossRef Search ADS Kiesler , Sara , and Lee Sproull . 1982 . Managerial response to changing environments: Perspectives on problem sensing from social cognition . Administrative Science Quarterly , 27 : 548 – 570 . Google Scholar CrossRef Search ADS Kimrey , Christopher M . 2016 . Opportunities in crisis and catastrophe: The issue-attention cycle and political reality . Homeland Security Affairs 12 . Kingdon , John W . 1984 . Agendas, alternatives, and public policies . Vol. 45 . Little, Brown Boston . Kline , Rex B . 2015 . Principles and practice of structural equation modeling . New York City, NY: Guilford Publications . Koetse , Mark J. , and Piet Rietveld . 2009 . The impact of climate change and weather on transport: An overview of empirical findings . Transportation Research Part D: Transport and Environment 14 : 205 – 21 . Google Scholar CrossRef Search ADS Kusumasari , Bevaola , Quamrul Alam , and Kamal Siddiqui . 2010 . Resource capability for local government in managing disaster . Disaster Prevention and Management: An International Journal 19 : 438 – 51 . Google Scholar CrossRef Search ADS La Porte , Todd R . 1996 . High reliability organizations: Unlikely, demanding and at risk . Journal of Contingencies and Crisis Management 4 : 60 – 71 . Google Scholar CrossRef Search ADS Lawrence , Paul R. , and Jay W. Lorsch . 1967 . Differentiation and integration in complex organizations . Administrative Science Quarterly , 12: 1 – 47 . Google Scholar CrossRef Search ADS Levinthal , D . 2000 . Organizational capabilities in complex worlds . In The nature and dynamics of organizational capabilities, ed. Dosi, Giovanni, Nelson, Richard R., Winter, and Sidney G., 363–78). New York: Oxford University Press. Levitt , Barbara , and James G. March . 1988 . Organizational learning . Annual Review of Sociology , 14 : 319 – 40 . Google Scholar CrossRef Search ADS Lincoln , James R. , and Gerald Zeitz . 1980 . Organizational properties from aggregate data: Separating individual and structural effects . American Sociological Review , 45 : 391 – 408 . Google Scholar CrossRef Search ADS Linnenluecke , Martina K. , Andrew Griffiths , and Monika Winn . 2012 . Extreme weather events and the critical importance of anticipatory adaptation and organizational resilience in responding to impacts . Business Strategy and the Environment 21 : 17 – 32 . Google Scholar CrossRef Search ADS Little , Richard G . 2002 . Controlling cascading failure: Understanding the vulnerabilities of interconnected infrastructures . Journal of Urban Technology 9 : 109 – 23 . Google Scholar CrossRef Search ADS MacKinnon , David Peter . 2008 . Introduction to statistical mediation analysis . Abingdon, UK: Routledge . March , James G . 1991 . Exploration and exploitation in organizational learning . Organization Science 2 : 71 – 87 . Google Scholar CrossRef Search ADS March , James G. , and Zur Shapira . 1987 . Managerial perspectives on risk and risk taking . Management Science 33 : 1404 – 18 . Google Scholar CrossRef Search ADS McDaniels , Timothy , Stephanie Chang , Darren Cole , Joseph Mikawoz , and Holly Longstaff . 2008 . Fostering resilience to extreme events within infrastructure systems: Characterizing decision contexts for mitigation and adaptation . Global Environmental Change 18 : 310 – 18 . Google Scholar CrossRef Search ADS McEntire , David A . 2004 . The status of emergency management theory: Issues, barriers, and recommendations for improved scholarship . University of North Texas. Department of Public Administration. Emergency Administration and Planning . ———. 2005 . Why vulnerability matters: Exploring the merit of an inclusive disaster reduction concept . Disaster Prevention and Management: An International Journal 14 : 206 – 22 . CrossRef Search ADS ———. 2008 . A critique of emergency management policy: Recommendations to reduce disaster vulnerability . International Journal of Public Policy 3 : 302 – 10 . CrossRef Search ADS McEntire , David A. , Christopher Fuller , Chad W. Johnston , and Richard Weber . 2002 . A comparison of disaster paradigms: The search for a holistic policy guide . Public Administration Review 62 : 267 – 81 . Google Scholar CrossRef Search ADS McGuire , Michael , and Debra Schneck . 2010 . What if Hurricane Katrina hit in 2020? The need for strategic management of disasters . Public Administration Review 70 ( s1 ): s201 – 7 . Google Scholar CrossRef Search ADS McGuire , Michael , and Chris Silvia . 2010 . The effect of problem severity, managerial and organizational capacity, and agency structure on intergovernmental collaboration: Evidence from local emergency management . Public Administration Review 70 : 279 – 88 . Google Scholar CrossRef Search ADS Meyer , Michael D. , E. Rowan , C. Snow , and A. Choate . 2013 . Impacts of extreme weather on transportation: National symposium summary . Washington, DC : American Association of State Highway and Transportation Officials . Meyer , Michael D. , Emily Rowan , Michael J. Savonis , and Anne Choate . 2012 . Integrating extreme weather risk into transportation asset management . American Association of State Highway and Transportation Officials . Moynihan , Donald P . 2008 . The dynamics of performance management: Constructing information and reform . Washington, D.C.: Georgetown Univ. Press . Moynihan , Donald P. , and Sanjay K. Pandey . 2005 . Testing how management matters in an era of government by performance management . Journal of Public Administration Research and Theory 15 : 421 – 39 . Google Scholar CrossRef Search ADS Muthén , Bengt O . 1993 . Goodness of fit with categorical and other nonnormal variables . In Testing structural equation models , ed. Bollen, Kenneth A., Long, and Scott A., 205 – 243 . Newbury Park: SAGE Publications . NOAA . 2016 . Billion-dollar weather and climate disasters: Table of events . http://www.ncdc.noaa.gov/billions/events (assessed October 30, 2017). Pandey , Sanjay K. , and Bradley E. Wright . 2006 . Connecting the dots in public management: Political environment, organizational goal ambiguity, and the public manager’s role ambiguity . Journal of Public Administration Research and Theory 16(4):511–32. Parker , Charles F. , Eric K. Stern , Eric Paglia , and Christer Brown . 2009 . Preventable catastrophe? The Hurricane Katrina disaster revisited . Journal of Contingencies and Crisis Management 17 : 206 – 20 . Google Scholar CrossRef Search ADS Parmesan , Camille , and Gary Yohe . 2003 . A globally coherent fingerprint of climate change impacts across natural systems . Nature 421 : 37 – 42 . Google Scholar CrossRef Search ADS PubMed Perrow , Charles . 1984 . Normal accidents: Living with high risk systems . New York : Basic Books . ———. 1994 . The limits of safety: The enhancement of a theory of accidents . Journal of Contingencies and Crisis Management 2 : 212 – 20 . CrossRef Search ADS ———. 1999 . Organizing to reduce the vulnerabilities of complexity . Journal of Contingencies and Crisis Management 7 : 150 – 5 . CrossRef Search ADS Ployhart , Robert E. , and Robert J. Vandenberg . 2010 . Longitudinal research: The theory, design, and analysis of change . Journal of Management 36 : 94 – 120 . Google Scholar CrossRef Search ADS Repetto , Robert C . 2009 . The climate crisis and the adaptation myth. WP 13 . New Haven, CT : School of Forestry & Environmental Studies, Yale University . Rijpma , Jos A . 1997 . Complexity, tight-coupling and reliability: Connecting normal accidents theory and high reliability theory . Journal of Contingencies and Crisis Management 5 : 15 – 23 . Google Scholar CrossRef Search ADS Rosenthal , Uriel , and Alexander Kouzmin . 1997 . Crises and crisis management: Toward comprehensive government decision making . Journal of Public Administration Research and Theory 7 : 277 – 304 . Google Scholar CrossRef Search ADS Roux-Dufort , Christophe . 2007 . Is crisis management (only) a management of exceptions ? Journal of Contingencies and Crisis Management 15 : 105 – 14 . Google Scholar CrossRef Search ADS Sarewitz , Daniel , Roger Pielke , and Mojdeh Keykhah . 2003 . Vulnerability and risk: Some thoughts from a political and policy perspective . Risk Analysis 23 : 805 – 10 . Google Scholar CrossRef Search ADS PubMed Schulman , Paul R . 1993 . The negotiated order of organizational reliability . Administration & Society 25 : 353 – 72 . Google Scholar CrossRef Search ADS Schwenk , Charles R . 1984 . Cognitive simplification processes in strategic decision-making . Strategic Management Journal 5 : 111 – 28 . Google Scholar CrossRef Search ADS Simon , Herbert A . 1979 . Rational decision making in business organizations . The American Economic Review , 69 : 493 – 513 . Singhal , Abhishek , Camille Kamga , and Anil Yazici . 2014 . Impact of weather on urban transit ridership . Transportation Research Part A: Policy and Practice 69 : 379 – 91 . Google Scholar CrossRef Search ADS Sitkin , Sim B. , and Amy L. Pablo . 1992 . Reconceptualizing the determinants of risk behavior . Academy of Management Review 17 : 9 – 38 . Google Scholar CrossRef Search ADS Sitkin , Sim B. , and Laurie R. Weingart . 1995 . Determinants of risky decision-making behavior: A test of the mediating role of risk perceptions and propensity . The Academy of Management Journal 38 : 1573 – 92 . Smith , Adam B . 2017 . 2016: A historic year for billion-dollar weather and climate disasters in U.S . NOAA Climate.gov. https://www.climate.gov/news-features/blogs/beyond-data/2016-historic-year-billion-dollar-weather-and-climate-disasters-us (accessed October 23, 2017). Smith , Joel B. , Jason M. Vogel , John E. Iii , and Cromwell . 2009 . An architecture for government action on adaptation to climate change. An editorial comment . Climatic Change 95 : 53 – 61 . Google Scholar CrossRef Search ADS Somers , Scott , and James H. Svara . 2009 . Assessing and managing environmental risk: Connecting local government management with emergency management . Public Administration Review 69 : 181 – 93 . Google Scholar CrossRef Search ADS Staber , Udo , and Jorg Sydow . 2002 . Organizational adaptive capacity: A structuration perspective . Journal of Management Inquiry 11 : 408 – 24 . Google Scholar CrossRef Search ADS Stark , Alastair . 2014 . Bureaucratic values and resilience: An exploration of crisis management adaptation . Public Administration 92 : 692 – 706 . Staw , Barry M. , Lance E. Sandelands , and Jane E. Dutton . 1981 . Threat rigidity effects in organizational behavior: A multilevel analysis . Administrative Science Quarterly 26 : 501 – 24 . Google Scholar CrossRef Search ADS Stehr , Steven D . 2006 . The political economy of urban disaster assistance . Urban Affairs Review 41 : 492 – 500 . Google Scholar CrossRef Search ADS ‘t Hart , Paul . 2013 . After Fukushima: Reflections on risk and institutional learning in an era of mega-crises . Public Administration 91 : 101 – 13 . Google Scholar CrossRef Search ADS Taleb , Nassim N. , Daniel G. Goldstein , and Mark W. Spitznagel . 2009 . The six mistakes executives make in risk management . Harvard Business Review 87 : 78 – 81 . Google Scholar PubMed Thomas , James B. , and Reuben R. McDaniel . 1990 . Interpreting strategic issues: Effects of strategy and the information-processing structure of top management teams . The Academy of Management Journal 33 : 286 – 306 . Thompson , James D . 1967 . Organizations in action: Social science bases of administrative theory . New York: McGraw-Hill . Thomson , Barbara . 2017 . Managing extreme weather at bus stops . Washington, DC : Transportation Research Board . Google Scholar CrossRef Search ADS Tierney , Kathleen . 2014 . The social roots of risk: Producing disasters, promoting resilience . Palo Alto, CA: Stanford Univ. Press . Turner , Barry A. , and Nick F. Pidgeon . 1997 . Man-made disasters ., 2nd ed. Oxford, UK: Butterworth-Heinemann. Turner , Billie L. , Roger E. Kasperson , Pamela A. Matson , James J. McCarthy , Robert W. Corell , Lindsey Christensen , Noelle Eckley , et al. 2003 . A framework for vulnerability analysis in sustainability science . Proceedings of the National Academy of Sciences of the United States of America 100 : 8074 – 9 . Google Scholar CrossRef Search ADS PubMed Weick , Karl E . 1977 . Enactment processes in organizations . In New directions in organizational behavior , Staw, Barry M., and Salancik, Gerald R., 267–300. Chicago, IL: St. Clair Publications. ———. 1993 . The collapse of sensemaking in organizations: The Mann Gulch disaster . Administrative Science Quarterly , 38 : 628 – 52 . CrossRef Search ADS Welch , Eric W. , Qing Miao , Fengxiu Zhang , and P.S. Sriraj . 2016 . National survey of US public transit agency experience with and response to extreme weather events . Washington, DC : National Transportation Library . https://rosap.ntl.bts.gov/view/dot/31623 (accessed December 30, 2017). West , Stephen G. , Aaron B. Taylor , and Wei Wu . 2012 . Model fit and model selection in structural equation modeling . In Handbook of structural equation modeling , 1st ed., ed. Rick H. Hoyle, 209–231. New York City, NY: The Guilford Press. Winn , Monika , Manfred Kirchgeorg , Andrew Griffiths , Martina K. Linnenluecke , and Elmar Günther . 2011 . Impacts from climate change on organizations: A conceptual foundation . Business Strategy and the Environment 20 : 157 – 73 . Google Scholar CrossRef Search ADS Wise , Charles R . 2006 . Organizing for homeland security after Katrina: Is adaptive management what’s missing ? Public Administration Review 66 : 302 – 18 . Google Scholar CrossRef Search ADS Wong , Chi-Sum , and Kenneth S. Law . 1999 . Testing reciprocal relations by nonrecursive structural equation models using cross-sectional data. Organizational Research Methods 2(1):69–87. Wooldridge , Jeffrey M . 2003 . Cluster-sample methods in applied econometrics . American Economic Review 93 : 133 – 38 . Google Scholar CrossRef Search ADS Yu , Ching-Yun . 2002 . Evaluating cutoff criteria of model fit indices for latent variable models with binary and continuous outcomes , Vol. 30 . Los Angeles : University of California, Los Angeles . Appendix 1. Definition of the Evaluation Criteria for Infrastructure-Based Capacity The respondents were asked to rate the state-of-good-repair of their agency’s infrastructure, using the following scales: Poor—Asset is past its useful life and is in need of immediate repair or replacement; may have critically damaged component(s) Marginal—Asset reaching or just past the end of its useful life; increasing number of defective or deteriorated component(s) and increasing maintenance needs Adequate—Asset has reached its mid-life (condition 3.5); some moderately defective or deteriorated component(s) Good—Asset showing minimal signs of wear; some (slightly) defective or deteriorated component(s) Excellent—New asset; no visible defects Appendix 2. Summary Statistics on Risk Perception No. of Respondents Proportion Risk perception 1  Category 1 23 0.08  Category 2 47 0.16  Category 3 89 0.29  Category 4 111 0.37  Category 5 33 0.11 Risk perception 2  Category 1 12 0.04  Category 2 37 0.12  Category 3 133 0.45  Category 4 104 0.35  Category 5 13 0.04 Risk perception 3  Category 1 16 0.05  Category 2 49 0.16  Category 3 89 0.30  Category 4 123 0.41  Category 5 24 0.08 No. of Respondents Proportion Risk perception 1  Category 1 23 0.08  Category 2 47 0.16  Category 3 89 0.29  Category 4 111 0.37  Category 5 33 0.11 Risk perception 2  Category 1 12 0.04  Category 2 37 0.12  Category 3 133 0.45  Category 4 104 0.35  Category 5 13 0.04 Risk perception 3  Category 1 16 0.05  Category 2 49 0.16  Category 3 89 0.30  Category 4 123 0.41  Category 5 24 0.08 View Large No. of Respondents Proportion Risk perception 1  Category 1 23 0.08  Category 2 47 0.16  Category 3 89 0.29  Category 4 111 0.37  Category 5 33 0.11 Risk perception 2  Category 1 12 0.04  Category 2 37 0.12  Category 3 133 0.45  Category 4 104 0.35  Category 5 13 0.04 Risk perception 3  Category 1 16 0.05  Category 2 49 0.16  Category 3 89 0.30  Category 4 123 0.41  Category 5 24 0.08 No. of Respondents Proportion Risk perception 1  Category 1 23 0.08  Category 2 47 0.16  Category 3 89 0.29  Category 4 111 0.37  Category 5 33 0.11 Risk perception 2  Category 1 12 0.04  Category 2 37 0.12  Category 3 133 0.45  Category 4 104 0.35  Category 5 13 0.04 Risk perception 3  Category 1 16 0.05  Category 2 49 0.16  Category 3 89 0.30  Category 4 123 0.41  Category 5 24 0.08 View Large © The Author(s) 2018. Published by Oxford University Press on behalf of the Public Management Research Association. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Public Administration Research and Theory Oxford University Press

Public Organization Adaptation to Extreme Events: Mediating Role of Risk Perception

Loading next page...
 
/lp/ou_press/public-organization-adaptation-to-extreme-events-mediating-role-of-V1pzGv5zVd
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of the Public Management Research Association. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.
ISSN
1053-1858
eISSN
1477-9803
D.O.I.
10.1093/jopart/muy004
Publisher site
See Article on Publisher Site

Abstract

Abstract The study responds to the growing call for a more systematic approach to research on organizational responses to extreme events. It develops and tests an integrated framework based on the organizational adaptation and learning theory to shed light on how public organizations manage exposure and vulnerability to extreme events. The analysis uses data from a 2016 national survey of top managers in the largest fixed-route public transit agencies in the United States and from other institutional sources to test hypotheses that link exposure to extreme events, impact, risk perception, and adaptive responses. We apply a structural model to disentangle the direct effect of exposure on adaptation as well as its indirect effects through impact and risk perception. Findings underscore the critical role that organizational risk perception has in converting environmental stimuli to organizational adaptive responses and point to a perception-mediated learning model of adaptation. Introduction Public organizations are increasingly encountering extreme events that cause notable, unpredictable, and disruptive changes (Boin and Lodge 2016; Comfort et al. 2012; Tierney 2014), and have the potential to cause considerable damage. Extreme events are human or nature-induced occurrences characterized by high salience, high uncertainty, and profound impacts (Kapucu 2005). They include earthquakes, severe weather, disease outbreaks, power outages, social movements, technical break-downs and cyber-attacks (Boin and Lodge 2016; Tierney 2014). Greater complexity of technology systems, more interconnected and interdependent infrastructure (Perrow 1984; Turner and Pidgeon 1997) and ongoing changes in the Earth’s biophysical and seismic system (Folke 2006; Hegerl et al. 2007) all contribute to the increased frequency and severity of extreme events. The occurrence of extreme events is increasingly patterned (i.e., nonrandom) and the risks associated with them similar (Rosenthal and Kouzmin 1997; Roux-Dufort 2007; Taleb et al. 2009; Tierney 2014), raising the potential for detecting and understanding patterns of organizational response. Response to extreme events is typically addressed through emergency, crisis and disaster management in public administration. Governments often establish professional emergency response routines and formal contingency management structures, mainly to coordinate with first responders (Boin and Lodge 2016; Somers and Svara 2009). Yet the increase in the number, scope and intensity of disasters that repeatedly expose the limitations and weakness of the prevailing management practices (Cigler 2007; Comfort et al. 2012; Kapucu and Garayev 2011) call for new approaches “at the strategic rather than the operational level” (Boin and Lodge 2016, 291). Recurring exposure to and impacts from extreme events necessitate planned adaptation to enhance organizational ability to withstand disruption, minimize damage and maintain operations (Comfort 2002; McEntire et al. 2002). Nevertheless, some public organizations are more likely to undertake adaptive responses than others. For example, scholars have shown that while many organizations have continued their reactive strategies to manage the immediate consequences after extreme events occur, some are reevaluating and responding to vulnerability before damage and incorporating considerations about extreme events in their long-range plans, infrastructure design, asset management, and interorganizational coordination (Hill and Engle 2013; Thomson 2017; Welch et al. 2016). Why is this the case? What conditions, experiences, capacities, and perceptions lead some public organizations to adapt and not others? To date, most empirical research emerges out of case studies of organizations responding to one specific type of extreme event (Christensen et al. 2016; McGuire and Silvia 2010; Roux-Dufort 2007). Although these studies are rich in detail, it is important to quantitatively characterize responses and empirically test the determinants of organizational adaptation across multiple organizations. This study empirically tests hypotheses derived from theory linking extreme events, organizational vulnerability, risk perceptions, and adaptive capacity. It examines the interface between the organization’s environment and the learning mechanisms that enable adaptation. Specifically, it investigates how exposure to extreme events impacts public organizations, and how the occurrence and impact of extreme events are related to organizational perceptions of risk and their adaptive responses. Is organizational adaptation a spontaneous response to recurring risks or is it mediated by cognitive mechanisms that enable organization-wide sensemaking of emerging developments? Based on a unique dataset from a 2016 national survey of 892 managers in 273 US largest public transit agencies that cover 82% of the entire FTA population of transit agencies with an annual fare revenue of 1 million, the findings underscore the essential role of risk perception in channeling environmental signals to adaptive responses. The article concludes by discussing its theoretical contribution and implications for management. Organizational Adaption to Reduce Vulnerability A key premise of this research is that organizations can learn from their prior experience and adjust to their changing environment (Lawrence and Lorsch 1967; Thompson 1967), thereby reducing vulnerability from repeated perturbations and disruptions from extreme events. Vulnerability, which is the degree to which an organization is likely to experience harm due to its exposure to hazardous events (Turner et al. 2003), is recognized as an outcome of the interaction between an organization’s exposure to environmental stresses and its ability to prepare for and react to them effectively (Berkes 2007; McEntire 2008). Because organizations have limited control over the frequency, magnitude, and scope of extreme events, effective vulnerability management is a primary means by which they are able to alter the consequences of such events (Dutton and Jackson 1987; McEntire et al. 2002). Figure 1 presents the framework depicting the key determinants of vulnerability in public organizations and the organizational mechanisms through which vulnerability can be managed. From left to right, an organization is exposed to the frequency, magnitude, and scope of extreme event, but the level of harm an organization experiences is contingent on the capacity it has at time t = 0 to respond to the event (Cigler 2007; McEntire 2005). Organizations having greater capability experience less harm and manifest fewer gaps in performance (i.e., impacts) during events. Perceptible gaps in performance are called revealed vulnerability. Risk perception results from an organization’s assessment and understanding of the risk inherent in its environment (Comfort 2007; Sitkin and Pablo 1992). Depending on its experience and perceived risk, the organization builds adaptive capacity at t = 1 after the event occurs. The response of the organization at t = 1 further influences its vulnerability and response to future extreme events. The cycle from adverse impact to organization adaptive response occurs over a period of time, the length of which depends upon the size of the gap, initial organization capacity, and other factors. We discuss the theoretical rationale for this framework in the following paragraphs and in the hypothesis section. Figure 1. View largeDownload slide Organization Vulnerability and Response to Extreme Events Figure 1. View largeDownload slide Organization Vulnerability and Response to Extreme Events Extreme Events and Organizational Performance Several factors constrain public organizations’ ability to effectively deal with extreme events. Public organizations are characterized by their search for structural (i.e., lack of organizational duplication) and fiscal (i.e., frugality) efficiency with a focus on their core competencies and routines (Perrow 1984; Stark 2014). This “lean and mean” strategy (Staber and Sydow 2002), which reduces system redundancy, resilience and slack, can increase vulnerability to extreme events and complicate efforts to manage their consequences (Boin et al. 2016; Schulman 1993). As Stark (2014, 696) notes, “the pursuit of efficiency without consideration of the benefits of auxiliary resources will create systems that cannot adapt to the emergence of potentially disastrous failures.” Additionally, interconnectedness and interdependence of critical infrastructure increases organizational vulnerability and complicates adaptation. High complexity increases the potential for a single localized performance failure to cause a cascade of disruptions that result in system-wide failure (Little 2002; Perrow 1999). Crichton and colleagues (2009) aptly exemplify how failure in one electricity feeder cable during a heat wave propagated faults in another two feeders and escalated into a profound 10-week disruption in electricity supply in New Zealand. Moreover, repeated exposure to extreme events can reduce an organization’s performance well below the desired level of achievement (Comfort 2005). The compromised integrity of system components and linkages, while not necessarily precipitating catastrophe during a particular event, can increase vulnerability and widen the performance gap during future extreme events (La Porte 1996; Perrow 1994). Adaptive Capacity Strengthening adaptive capacity is a key mechanism by which public organizations can reduce vulnerability to repeated stress or perturbations (Berkes 2007; McEntire 2008; Staber and Sydow 2002). Adaptive capacity is defined not only in terms of an organization’s capability to bounce back to a state of normalcy after an extreme event, but also of its ability to absorb disruptions and reorganize while undergoing changes so as to retain the essential functions and structures (Berkes 2007; Boin and Van Eeten 2013). Increasing adaptive capacity entails deliberate efforts to make longer-term and anticipatory adjustments to fill the possible performance gaps for a wider range of observed or anticipated extreme events. It reflects an organization’s stock of resources and enables it to exploit its resources in a more productive manner (Kusumasari et al. 2010). Common measures to build adaptive capacity include improvement in material resource inputs, information and technology, infrastructure and equipment, human capital, and inter-agency coordination arrangements (Comfort and Okada 2013; Kusumasari et al. 2010; McEntire 2005). As an example, Arizona mobilized adaptive responses to severe droughts in 2008–2011 by developing an integrated water conservation system and fostering collective planning and response among water suppliers and users (Hill and Engle 2013). Development of adaptive capacity is distinguished from emergency management, which focuses predominantly on responding to the immediate impacts of extreme events (Cigler 2007; McEntire 2008; Somers and Svara 2009). Crisis management planning and preparation for extreme events establish risk-based intentions but do not constitute adaptive capacity (Christianson et al. 2009; Clarke 1999). As Perrow (1999, 152) notes, “Even ‘worst case’ scenarios usually refer only to the worst state of the environment,” giving little attention to an organization’s overall vulnerability, maintenance of infrastructure, the process of event escalation, the availability of backup resources or potential for maximum failure of the organizational management (Fischbacher-Smith 2010; McEntire 2008; Roux-Dufort 2007). Risk Perception Nevertheless, reducing vulnerability and enhancing adaptive capacity in the face of extreme events are complicated by multiple intervening factors. Vulnerability is often not obvious in the absence of significant triggers or events (Rijpma 1997; Sarewitz et al. 2003) and the evidence of capacity surfaces only after extraordinarily complex problems are solved (Kusumasari et al. 2010; Levinthal 2000). As a result, decision makers face high ambiguity, complexity and uncertain payoffs from investment and change. These challenges to building adaptive capacity, coupled with the evidence that disasters motivate adaptive planning in some organizations but not in others (Ebert et al. 2009; Haigh and Griffiths 2011), suggest a cognitive mechanism that governs how organizations respond to these external stimuli. Scholarship across disciplines has identified risk perception as a crucial factor for explaining why disasters motivate adaptation planning in organizations (Berkhout 2012; Comfort 2007; Dutton and Jackson 1987; Hoffmann et al. 2009; Somers and Svara 2009). Risk perception comprises the perceived probability of being exposed to negative impacts and the appraisal of how harmful those impacts would be on the organization (Grothmann and Patt 2005). Uncertainty and complexity associated with extreme events limit the applicability and usefulness of analytic techniques and tools (Garrett 2004; Moynihan 2008; Simon 1979), as well as the ability of existing data to capture event complexity (Fischbacher-Smith 2010). As a result, organizations rely on a messy process of sensemaking, attribution, and judgment for decision making and strategy selection (Daft and Weick 1984; Kiesler and Sproull 1982; Sitkin and Pablo 1992). Adaption is less likely to occur when the affected organizations fail to notice and attach meaning and significance to variations in their environment that pose risks (Comfort 2007; McEntire 2004; Sitkin and Pablo 1992). In contrast, perceived risk can stimulate organizations to undertake nonroutine and nonincremental action that aims to increase adaptive capacity (Comfort 2007). The theoretical framework presented in Figure 1 provides a starting point for further development of theory-based hypotheses linking exposure to extreme events, impact, risk perception, and adaptation. Following the framework, we hypothesize that organization adaptation to extreme events is affected by its vulnerability to extreme events which is manifested in the impacts it realizes during the events. We also expect that exposure and impact lead to adaptation through perception of the risk associated with extreme events. Theory And Hypotheses Exposure, Impact, and Adaptation Organizational systems are typically designed to cope with a certain range of external perturbation and disruption, the levels and scopes of which are largely determined by their historical norms and experiences. However, extreme events are invariably outliers against which the existing operating system does not provide sufficient defense (Fischbacher-Smith 2010). Repeated exposure to extreme events inevitably leads to accumulation of severe weaknesses and deficiencies until the system reaches the tipping point and loses viability and robustness (Roux-Dufort 2007). As Comfort (2002, 102) puts it, “governmental systems designed to provide security at one level of exposure may fail when they are exposed to cumulative threats of different types at different levels of operations.” The situation rapidly escalates as a result of a series of interdependent cascading failure in which failure in a single component triggers failures throughout a complex and tightly coupled system (Perrow 1984; Turner and Pidgeon 1997). Recurring extreme events are more likely to overwhelm the control and management system and lead to disastrous impacts on the affected organization. H1: Organizations that experience greater exposure to extreme events are more likely to experience greater impacts (i.e., performance gap). It is often the case that structural, political, and capacity constraints limit the ability of public organizations to pursue adaptive solutions in the face of extreme events (McGuire and Schneck 2010; Smith et al. 2009; Wise 2006). Extreme events that cause significant property damage, economic loss, human casualties, aggressive media coverage, and political pressure act as powerful catalysts for reexamining standard approaches and practices (Boin and Hart 2003; Stehr 2006). Meanwhile, the increasing inapplicability and ineffectiveness of existing routines and procedures force organizations to make greater investments in exploration and implementation of more fitting solutions (Cyert and March 1963; March 1991). Facing low control in high-risk environments, organizations can best respond by adjusting their internal processes and building adaptive capacity (Dutton and Jackson 1987; Staber and Sydow 2002). Prior work demonstrates that significant adjustments of organizational strategy and practices usually do not occur until an organization has unequivocally suffered disastrous consequences associated with extreme events (Comfort 2007; Dutton and Jackson 1987; Linnenluecke et al. 2012; McEntire 2004). Importantly, impactful extreme events open “windows of opportunity” (Kingdon 1984) for reform-minded organizations to exploit the significant damage and build support for nonincremental changes by directing attention to the flaws and deficiencies in the existing systems (Boin and Hart 2003). Such opportunities are scarce and fleeting, and organizations have to act expeditiously before the public attention and impetus for reform fade away (Birkland 2009; Dekker and Hansén 2004; Gallagher 2014; Parker et al. 2009). This suggests that organizational mobilization of support and for adaptive capacity development must be carried out within a relatively short time span. Delay or slow response can miss the opportunity. The higher the consequences of the extreme event, the more leverage an agency can apply to facilitate non-incremental changes (Birkland 1997). We thus hypothesize that higher impacts of extreme events are more likely to provide the momentum as well as the leverage for organizations to increase adaptive capacity. H2: Organizations that experience greater impacts (i.e., performance gap) from extreme events are more likely to undertake adaptive capacity building. At the same time, greater exposure to extreme events increases problem familiarity essential for organizational learning (Levitt and March 1988; Sitkin and Pablo 1992). Recurring extreme events inform organization decision makers about the rapid escalation of harm and the significant scale of failure that escalation can generate, both of which work to invalidate their assumptions about control (Fischbacher-Smith 2010). The lessons learned are reflected in their exploration and adoption of adaptive responses to the threats, independent of the outcomes from the past exposure (Sitkin and Pablo 1992). The increased exposure also allows for trail-and-error experimentation to acquire and test knowledge on how to best adapt (Berkes and Folke 2002; Levitt and March 1988). Given this discussion, it is expected that greater exposure to extreme events will be positively associated with adaptive capacity development. H3: Organizations that experience greater exposure to extreme events will be more likely to undertake adaptive capacity building. Risk Perception and Adaptation Organizations typically must experience a continuous stream of events to grasp trends and attempt to make sense of them (Daft and Weick 1984; Thomas and McDaniel 1990). Given bounded cognitions of decision makers (Simon 1979) and the magnitude of complexity and uncertainty of extreme events, the associated risks are often not easily recognized. Instead, organizations only selectively attend to emerging developments (Dutton and Jackson 1987; Weick 1977). As selectivity is governed by knowledge or ideology filters (Boin and Van Eeten 2013; Thomas and McDaniel 1990), organizations tend to resist perceptions and conclusions that challenge their prevailing routine and frames of reference (Sitkin and Pablo 1992; Staw et al. 1981). To the extent that extreme events challenge fundamental assumptions about risks and control (Boin and Lodge 2016; Fischbacher-Smith 2010; ‘t Hart 2013), they are likely to be excluded from consideration for organizational decision making. Selective filtering of stimuli is particularly strong in public organizations which are highly constrained by organizational inertia and cultures of risk denial (Cigler 2007; Ford and King 2015; Perrow 1999). Boin and Van Eeten (2013) aptly exemplify how the entrenched safety process and culture of high reliability in the National Aeronautics and Space Administration (NASA) prevented “an accurate assessment of the impending threats to the safety of the doomed shuttle” (442) and caused the well-known Challenger accident. There are many reasons for this, but at base the failure to recognize risk and act has been explained by scarcity of evidence, blindness to evidence and uncertainty in assessing the relevance of evidence (Levitt and March 1988). Therefore, before organizations engage in problem-solving behavior, they must first recognize the problem and then implement change (Daft and Weick, 1984; Sitkin and Pablo, 1992). To paraphrase Repetto (2009), just because organizations can adapt does not mean they will. The joint capacity to detect the out-of-ordinary developments and to recognize the inherent risk they pose creates a trigger that enables an organization to initiate an adaptation process (Comfort 2007; Duncan 1972; Dutton and Jackson 1987). Given this, it is relevant to ask why some organizations are capable of perceiving the risks while others not. As a first step, emerging events or trends influence perceptions because of their potential to interfere with an organization’s salient objectives (Daft and Weick 1984; Schwenk 1984). Extreme weather signals “uncontrollable” and “negative” threats (Dutton and Jackson 1987; Kahneman and Tversky 1979) to which some organizations are more sensitive than others. For example, organizations in the transportation, energy, water, and utility sectors may be particularly attuned to extreme weather signals because their operations hinge on the sound functioning of an exposed, interconnected, and dispersed lifeline infrastructure (Boin and McConnell 2007; Little 2002; McDaniels et al. 2008). The threats intensify as an organization’s experience with extreme weather events rises. As Berkhout (2012, 95) states: “… the more frequent, unambiguous and salient evidence from experience is, the greater the likelihood it will be recognized and interpreted as significant (by organizations).” An organization’s perception of risk is therefore a function of its experience with extreme events, in terms of both exposure and impact. Referring back to the discussion about the cognitive significance of risk perception, this suggests a process that links environmental stimuli to organization action through risk perception. It therefore stands to reason that risk perception positively mediates the effect of signals from extreme events on organizational adaptive responses. Based on this discussion we hypothesize H4: An organization’s risk perception positively mediates the effect of its experience with extreme events (i.e., exposure and impact) on its adaptive capacity building. Figure 2 presents the conceptual model of the theorized relationships. We hypothesize that an organization will experience greater adverse impacts from increased exposure to extreme events. Exposure to extreme events can directly affect an organization’s adaptive capacity building through a learning mechanism. The direct experience with extreme events (i.e., exposure and impacts) is also expected to trigger the organization’s perception of risks concerning ongoing environmental change and ultimately lead to a higher level of organization adaptation. Figure 2. View largeDownload slide Theoretical Model: Mediating Role of Risk Perception Figure 2. View largeDownload slide Theoretical Model: Mediating Role of Risk Perception Extreme Weather Events and Transit Agencies Extreme weather events pose a serious challenge to public agencies (Hodges 2011; Meyer et al. 2012). Not only do they disrupt operations, impair service quality, and cause additional safety threats, but they also damage infrastructure and impose strain and stress on the state of good repair. As demonstrated by Hurricane Sandy and other recent weather-related disasters, weather can have significant ramifications for regional mobility and functioning of economic systems. How to effectively manage the risks of extreme weather is a question that often concerns public organizations and managers. This issue has become increasingly urgent with the likely increase in the frequency and intensity of extreme weather events (Folke 2006; Hegerl et al. 2007; Parmesan and Yohe 2003), such as floods, heat waves, and severe storms. While many agencies have long experience coping with weather disruptions, they are increasingly confronted with the challenge of identifying and developing appropriate longer-term adaptation strategies to address greater risk and uncertainty associated with extreme weather (Hodges 2011; Meyer et al. 2013). Selection of transit agencies offers several advantages for the study of extreme weather and adaptation. First, because transit agencies rely on an interdependent and interconnected network of exposed and dispersed capital assets for service delivery, they are particularly sensitive and susceptible to extreme weather. In addition to the impacts on infrastructure, extreme weather also diminishes the quality and reliability of transit services in the form of increased assess time, prolonged trip duration, declined service frequency, and forced vehicle rerouting (Singhal et al. 2014). Given the interconnectivity and lack of redundancy in most current transit systems (Meyer et al. 2012), extreme weather results in substantial costs to network efficiency and mobility (Koetse and Rietveld 2009). Moreover, weather-related disruptions disproportionately affect transit-dependent populations, such as the elderly, the poor and people with disabilities (Hodges 2011). That weather impacts on the public transportation sector have received scant attention in the literature (Koetse and Rietveld 2009) contributes further to the importance of this study. Data This study matches survey data collected from US transit agencies 2016 with agency profile data obtained from the Federal Transit Administration’s National Transit Database (NTD) and demographic data taken from the US Census Bureau. The full data set is used in a structural equation model (SEM) to test our hypotheses. The NTD is the primary source for information and statistics on the US transit system. We used the entire list of agencies in NTD to identify target agencies for this study. The target population included all major US metropolitan fixed-route public transit agencies operating bus and/or rail transit services with an annual fare revenue of at least one million dollars in 2013. Agencies having a small vehicle fleet (i.e., 30 or fewer vehicles according to FTA’s definition of small systems) or those run by universities were excluded from the study because of their limited capacity to undertake adaptation. Private companies (less than 30 in number) were not included because most manage transit systems in multiple cities and states. The resulting target population included 312 public transit agencies that satisfy the selection criteria. Because different key decision makers within agencies likely have different perspectives on extreme weather events, vulnerability and risk depending upon professional background, experience, and position in the agency, the study identified the leader manager in five different departments for inclusion: maintenance, operations, engineering, service planning, and strategic planning. Names and contact information were collected using a protocol that included three methods: searching of online sources such as NTD, telephone calls to agencies, and Freedom of Information requests. Because some agencies did not respond, refused to provide information or were nonreachable after repeated attempts, we had to drop 39 agencies resulting in a final agency-level sampling frame of 273 agencies (88% of the target population). As not every agency has all five departments or functions, the final individual-level sampling frame included 892 respondents (equivalent to an average of 3.3 people per organization). The survey asked respondents about organizational experience with extreme weather events, perceptions about future weather risks, strategies employed to address extreme weather, and other questions concerning operation and management. Most questions were set within a 2-year time frame to reduce recall bias. Survey instrument development was informed by a small set of interviews of transit managers in four agencies in different regions of the United States. The instrument was coded in Sawtooth Software® and administered online first as a pretest to 20 individuals from the sample and then to the entire sample. Administration of the full survey extended from April 28 to June 11, 2016. An initial hard-copy notification letter, sent to each of the respondents informing them about the survey and requesting their participation, was followed 10 days later by an electronic invitation with web link to the live survey, username, and password. The survey was administered to an adjusted sample of 862 respondents after ineligible, retired, and noncontactable individuals were dropped. A total of 306 individuals completed the survey, yielding a survey response rate of approximately 35.5%. Of the 273 agencies surveyed, 199 provided at least one response (72.9%). Nonresponse bias analysis showed no difference by size or region (α < 0.05) between responding agencies and either the sampling frame (n = 273) or the target population (n = 312). As a result, we are confident that our study results generalize to the target population. We merged the survey response data with other organizational data taken from NTD and 2010–2014 US Census survey, including items such as organizational size and population density of the transit service area. Measurement Variables of Primary Interest: Organizational Adaptation An organization’s adaptation strategy comprises a combination of adaptation measures and the distinct strategic goals they aim to accomplish (Hoffmann et al. 2009). For conceptual focus, this article examines two critical adaptation strategies relevant for addressing the potential threats posed by extreme weather to resource-intensive public service organizations: vulnerability assessment and capital investment. Vulnerability Assessment Public organizations must decide where to make improvements and how to apply scarce resources to address extreme weather (Henstra 2010). Vulnerability assessment generates specific information necessary for making strategic decisions (Cairney et al. 2016; Somers and Svara 2009). The information is particularly useful for organizations operating in tightly coupled and complex systems prone to cascading disruptions and failures (Little 2002). Because people tend to underestimate the likelihood that natural hazards will have negative impacts on them (March and Shapira 1987; Sitkin and Weingart 1995), knowledge about system vulnerability can serve to raise vigilance and generate support for strategic changes to avert possible damage. Capital Investment Resource-intensive public organizations often find their material resources and infrastructure insufficiently designed for and severely damaged by dramatic variability in weather conditions. In the case of transit agencies, for example, excessive rainfall can flood tracks, bus ways, tunnels and stations, and hurricanes often cause power disruption, vehicle crashes and sign damage (Hodges 2011; Meyer et al. 2013). The scope and severity of the impacts depend partly on whether or not countermeasures such as redundancy are in place (Little 2002; McDaniels et al. 2008). For aged, under-designed and weather-sensitive infrastructure, capital investment is needed to reinforce, upgrade, or replace the critical assets such that they are resistant to extreme events (Fankhauser et al. 1999; Hodges 2011). Vulnerability assessment uses a composite of discrete response questionnaire items asking the respondent if his/her organization: (1) assessed the agency’s vulnerability to extreme weather; (2) estimated the costs of responding to an extreme weather event; and (3) conducted or contracted research on the risks of extreme weather events (1 = yes, 0 = no). Capital investment comprises four discrete items: (1) invested in weather-smart equipment and technologies, such as sensors that detect changes in pressure and temperature in materials; (2) invested in information and communication technologies; (3) invested in back-up power supplies and equipment; and (4) invested in weather-proof infrastructure improvement or retrofitting projects (e.g., strengthening parts of a building, improving stations or tracks). The two outcome variables—vulnerability assessment and capital investment—are constructed as two latent factors underlying the corresponding binary indicators. The variable Exposure is a composite index of a set of questionnaire items about the organization’s experience with extreme weather by event type, where extreme weather was defined in the survey as “unusually severe storms, floods, heat waves or other weather incidents that lie outside of historical norms or experience.” Respondents were asked: “During the last two years, about how many times have the following extreme weather events occurred in your transit service area?” The list of extreme weather events includes: extreme cold temperatures, extreme heat waves, floods, hurricanes/tropical storms, severe rainstorms/thunderstorms, tides/storm surges, extreme high winds, tornadoes, and extreme snow storms (Scale: 0 = never, 1 = once, 2 = two to three times, 3 = more than three times). Although the data are limited, we believe the frequency of extreme events as defined in the survey provides a reasonable measure of exposure that is sufficient to trigger responses. For each type of event experienced, respondents were subsequently asked to rate the adverse impact on their organization. The question asked: “Considering the extreme weather events that have happened in your area in the previous two years, has the level of adverse impact been catastrophic, major, moderate, minor, or no impact?” (Scale: 0 = no impact, 1 = minor, 2 = moderate, 3 = major, 4 = catastrophic). The variable impact is measured as the mean response across all events. The 2-year time span was designed in the survey to reduce recall bias and as a stable proxy measure of the longer-term location-based pattern of extreme weather the agencies experienced. Weather patterns are fundamentally localized and relatively consistent over the short term. NOAA extreme weather data confirm that while extreme weather trends are gradually increasing over the long term, they are relatively stable within a short segment of time such as 2–5 years (NOAA 2016; Smith 2017). Therefore, the 2-year snapshot provides a strong location-based measure of extreme events. We also believe that organizational adaptation—nonincremental changes from existing routines—needs to happen relatively rapidly enough so that agencies are able to leverage windows of opportunity opened by major extreme weather events and their impacts. Kimrey’s (2016) extensive literature review demonstrates that the political issue-attention for extreme events usually lasts only 2–3 years. The required timeliness of adaptive responses strengthens our confidence about using the 2-year time frame for exposure and impact to test the hypotheses. Risk perception was measured as an index of three survey questions asking about the level of agreement with the statements: (1) “Most people in my agency recognize that extreme weather events are becoming more frequent.”; 2) “My agency is increasingly concerned about the impact of extreme weather events on our transit infrastructure.”; and 3) “My agency is increasingly concerned about the impact of extreme weather events on our transit operations.” (Scale: 1 = strongly disagree, 2 = disagree, 3 = neither disagree nor agree, 4 = agree, 5 = strongly agree). Risk perception is constructed as a latent factor underlying the three ordinal response variables. We measure organizational risk perception based on responses of upper-level managers reporting about their organizations. The top managers surveyed serve in departments that are most affected by extreme events. We adopted an upper echelon perspective for conceptualizing the organizational risk perception measure because top management cognitions can causally determine organizational strategic choices and outcomes (Hambrick 2007; Hambrick and Mason 1984). In their classic work on organizational interpretation, Daft and Weick (1984) specifically stress the role of top management in synthesizing information from nested subsystems and formulating the interpretation for the organizational system as a whole. They note (1984, 285): “when one speaks of organizational interpretation one really means interpretation by a relatively small group at the top of the organizational hierarchy.” Control Variables Given the variation in complexity of transit systems, the estimations included a number of control variables. Bus only is coded as 1 if the agency operates bus services but not heavy or light rail. Infrastructure-based capacity is an index of responses to one multi-item question asking respondents to assess the quality of the transit-relevant infrastructure in their service area. Items included: bus transit services, bus station structures and shelters, commuter rail transit services, rail control systems, rail bridges, structures and tunnels, rail track, switches and track work, rail station structures and platforms, streets, roads and highways, roadway bridges, structures and tunnels, transit facility ventilation systems, transit maintenance equipment, electrification/power system, communication systems, drainage systems, and revenue/fare collection systems. Items were automatically populated in the survey depending upon the type of transit services (e.g., bus only system respondents did not receive the rail-specific questions) and respondents were given definitions of the five quality categories (Scale: 1 = poor, 2 = marginal, 3 = adequate, 4 = good, 5 = excellent; see Appendix 1 for definitions). Additional data to capture agency characteristics were collected from the NTD and the 2010–2014 US Census, matched by county code. Authority is coded 1 if the agency is an independent transit authority and 0 if the agency is affiliated to a city or state government. Organizational size, used as a proxy of an organization’s comprehensive capacity for adaptation, is measured by the natural log of an agency’s annual total funding in 2013, including funding from directly generated revenue as well as funding from local, state, and federal governments. Density is the natural log of service population normalized by service area. Commute time serves as an indicator of the demand for transit services and uses the mean travel time to work in minutes. Median household income is the natural log of median household income in the county the agency is located. Finally, a set of the dummy variables (1 = Yes, 0 = No) were used to indicate the Census region the agency is located in: Northeast, Midwest, South and West, respectively. Descriptive statistics for all variables except risk perception are displayed in Table 1. Since the mean and variance of ordinal variables have no meaning (Jöreskog 1994), we report the number of respondents and the univariate proportion for each category of the three indicators in Appendix 2. Table 1. Descriptive Statistics Variable Mean SD Min Max Size Vulnerability assessment  Vulnerability assessment 1 0.55 0.50 0 1 299  Vulnerability assessment 2 0.25 0.43 0 1 300  Vulnerability assessment 3 0.09 0.28 0 1 303 Capital investment  Capital investment 1 0.11 0.31 0 1 299  Capital investment 2 0.62 0.49 0 1 302  Capital investment 3 0.72 0.45 0 1 301  Capital investment 4 0.28 0.45 0 1 301 Exposure 9.94 4.94 0 24 304 Impacts 1.29 0.66 0 3 295 Bus only 0.74 0.44 0 1 304 Infrastructure-based capacity 3.20 0.60 1 5 304 Authority 0.64 0.48 0 1 304 Organization size 17.33 1.49 15 23 304 Density 7.50 1.29 −0.68a 12 304 Median household income (log) 10.89 0.21 10 12 304 Commute time 24.40 4.71 15 42 304 Northeast 0.13 0.33 0 1 304 Midwest 0.22 0.42 0 1 304 South 0.33 0.47 0 1 304 West 0.32 0.47 0 1 304 Variable Mean SD Min Max Size Vulnerability assessment  Vulnerability assessment 1 0.55 0.50 0 1 299  Vulnerability assessment 2 0.25 0.43 0 1 300  Vulnerability assessment 3 0.09 0.28 0 1 303 Capital investment  Capital investment 1 0.11 0.31 0 1 299  Capital investment 2 0.62 0.49 0 1 302  Capital investment 3 0.72 0.45 0 1 301  Capital investment 4 0.28 0.45 0 1 301 Exposure 9.94 4.94 0 24 304 Impacts 1.29 0.66 0 3 295 Bus only 0.74 0.44 0 1 304 Infrastructure-based capacity 3.20 0.60 1 5 304 Authority 0.64 0.48 0 1 304 Organization size 17.33 1.49 15 23 304 Density 7.50 1.29 −0.68a 12 304 Median household income (log) 10.89 0.21 10 12 304 Commute time 24.40 4.71 15 42 304 Northeast 0.13 0.33 0 1 304 Midwest 0.22 0.42 0 1 304 South 0.33 0.47 0 1 304 West 0.32 0.47 0 1 304 aThe negative log value of density comes from three observation within the same agency, with a density of about 0.50. However, the organization has normal values in other organizational attributes, so we kept the three observations in the analysis. For robust check purposes, we also run the model without the three observations and the results did not qualitatively change. View Large Table 1. Descriptive Statistics Variable Mean SD Min Max Size Vulnerability assessment  Vulnerability assessment 1 0.55 0.50 0 1 299  Vulnerability assessment 2 0.25 0.43 0 1 300  Vulnerability assessment 3 0.09 0.28 0 1 303 Capital investment  Capital investment 1 0.11 0.31 0 1 299  Capital investment 2 0.62 0.49 0 1 302  Capital investment 3 0.72 0.45 0 1 301  Capital investment 4 0.28 0.45 0 1 301 Exposure 9.94 4.94 0 24 304 Impacts 1.29 0.66 0 3 295 Bus only 0.74 0.44 0 1 304 Infrastructure-based capacity 3.20 0.60 1 5 304 Authority 0.64 0.48 0 1 304 Organization size 17.33 1.49 15 23 304 Density 7.50 1.29 −0.68a 12 304 Median household income (log) 10.89 0.21 10 12 304 Commute time 24.40 4.71 15 42 304 Northeast 0.13 0.33 0 1 304 Midwest 0.22 0.42 0 1 304 South 0.33 0.47 0 1 304 West 0.32 0.47 0 1 304 Variable Mean SD Min Max Size Vulnerability assessment  Vulnerability assessment 1 0.55 0.50 0 1 299  Vulnerability assessment 2 0.25 0.43 0 1 300  Vulnerability assessment 3 0.09 0.28 0 1 303 Capital investment  Capital investment 1 0.11 0.31 0 1 299  Capital investment 2 0.62 0.49 0 1 302  Capital investment 3 0.72 0.45 0 1 301  Capital investment 4 0.28 0.45 0 1 301 Exposure 9.94 4.94 0 24 304 Impacts 1.29 0.66 0 3 295 Bus only 0.74 0.44 0 1 304 Infrastructure-based capacity 3.20 0.60 1 5 304 Authority 0.64 0.48 0 1 304 Organization size 17.33 1.49 15 23 304 Density 7.50 1.29 −0.68a 12 304 Median household income (log) 10.89 0.21 10 12 304 Commute time 24.40 4.71 15 42 304 Northeast 0.13 0.33 0 1 304 Midwest 0.22 0.42 0 1 304 South 0.33 0.47 0 1 304 West 0.32 0.47 0 1 304 aThe negative log value of density comes from three observation within the same agency, with a density of about 0.50. However, the organization has normal values in other organizational attributes, so we kept the three observations in the analysis. For robust check purposes, we also run the model without the three observations and the results did not qualitatively change. View Large An overwhelming majority of organizations, over 95%, have experienced at least one extreme weather event and suffered the impacts during the past 2 years. Overall, organizations are relatively neutral about their perception of risks associated with extreme weather events. Approximately 25% agencies provide rail services and 64% agencies are independent transit authority organizations. Analysis And Results Method Because the model assumes multiple relationships and mixes latent and observed variables, SEM is a suitable analytic method for this study (MacKinnon 2008). SEM is preferred over regression analysis because of its capacity to estimate constructs by separating the unique variance of observed items from shared items (Kline 2015). It also has the advantage of incorporating indirect effects when mediating variables are included. Because the three adaptation measures and risk perception were constructed using categorical or ordinal variables, the relationships between the manifest variables and the latent factor are considered nonlinear. A common practice to account for the nonlinearity is to replace the observed categorical variables with their underlying latent and continuous factors (Hoyle 2012). Therefore, we analyzed the model using the weighted least square means and variance adjusted (WLSMV) estimator in Mplus 7.4 (Muthén 1993). The WLSMV estimator uses a probit function to link the underlying continuous latent variable to the observed categorical indicators, allowing analysis of relationships between the underlying latent variables. Compared to the commonly used maximum likelihood (ML) estimation, WLSMV estimation has more accurate factor loading and model fit when the number of categories is small (e.g., 2 or 3 categories) (Beauducel and Herzberg 2006). Additionally, because the transit agency is the unit of analysis and there is more than one response from some responding agencies, clustered standard errors by agency were used to account for the nested data structure (Wooldridge 2003). We included all control variables in all paths to estimate the relationships of primary interest. We also weighted the observations to ensure equal representation of each agency in the analysis. Sem Results Measurement Model A confirmatory factor analysis (CFA) on the categorical indicators shows good model fit (RMESA = 0.050 with 90% CI [0.028, 0.071], CFI = 0.984, TLI = 0.978, WRMR = 0.811). The fit indices satisfy the general cutoff points for measurement models with WLSMV estimation: RMSEA ≤ 0.05, CFI ≥ 0.96, TLI ≥ 0.95, and WRMR ≤ 1.0 (Yu 2002). The chi-square test statistic (χ2 = 56.672, df = 32, p value = .0046) indicates unsatisfactory fit, but it can be attributed to its sensitivity to sample size and is usually not considered sufficient evidence to reject the model when contrary to other fit indices (West et al. 2012). Table 2 shows the standardized factor loadings for each scale. The R-square values indicate how much variance in the indicator is explained by the underlying latent factor. The standardized factor loadings are relatively large, ranging from 0.65 to 0.96. The Wald tests for each pair of the latent factors are all highly significant with p-value <.000. Therefore, both convergent and discriminant validity are achieved for the latent constructs. Table 2. Measurement Model Factor and Indicators Estimate SE p-Value R-Square Vulnerability assessment  Vulnerability assessment 1 0.681 0.075 .000 0.464  Vulnerability assessment 2 0.783 0.066 .000 0.613  Vulnerability assessment 3 0.846 0.095 .000 0.716 Capital investment  Capital investment 1 0.649 0.096 .000 0.421  Capital investment 2 0.698 0.072 .000 0.487  Capital investment 3 0.800 0.068 .000 0.640  Capital investment 4 0.771 0.069 .000 0.594 Risk perception Risk perception 1 0.822 0.028 .000 0.676 Risk perception 2 0.704 0.036 .000 0.496 Risk perception 3 0.964 0.026 .000 0.929 Factor and Indicators Estimate SE p-Value R-Square Vulnerability assessment  Vulnerability assessment 1 0.681 0.075 .000 0.464  Vulnerability assessment 2 0.783 0.066 .000 0.613  Vulnerability assessment 3 0.846 0.095 .000 0.716 Capital investment  Capital investment 1 0.649 0.096 .000 0.421  Capital investment 2 0.698 0.072 .000 0.487  Capital investment 3 0.800 0.068 .000 0.640  Capital investment 4 0.771 0.069 .000 0.594 Risk perception Risk perception 1 0.822 0.028 .000 0.676 Risk perception 2 0.704 0.036 .000 0.496 Risk perception 3 0.964 0.026 .000 0.929 View Large Table 2. Measurement Model Factor and Indicators Estimate SE p-Value R-Square Vulnerability assessment  Vulnerability assessment 1 0.681 0.075 .000 0.464  Vulnerability assessment 2 0.783 0.066 .000 0.613  Vulnerability assessment 3 0.846 0.095 .000 0.716 Capital investment  Capital investment 1 0.649 0.096 .000 0.421  Capital investment 2 0.698 0.072 .000 0.487  Capital investment 3 0.800 0.068 .000 0.640  Capital investment 4 0.771 0.069 .000 0.594 Risk perception Risk perception 1 0.822 0.028 .000 0.676 Risk perception 2 0.704 0.036 .000 0.496 Risk perception 3 0.964 0.026 .000 0.929 Factor and Indicators Estimate SE p-Value R-Square Vulnerability assessment  Vulnerability assessment 1 0.681 0.075 .000 0.464  Vulnerability assessment 2 0.783 0.066 .000 0.613  Vulnerability assessment 3 0.846 0.095 .000 0.716 Capital investment  Capital investment 1 0.649 0.096 .000 0.421  Capital investment 2 0.698 0.072 .000 0.487  Capital investment 3 0.800 0.068 .000 0.640  Capital investment 4 0.771 0.069 .000 0.594 Risk perception Risk perception 1 0.822 0.028 .000 0.676 Risk perception 2 0.704 0.036 .000 0.496 Risk perception 3 0.964 0.026 .000 0.929 View Large Structural Model There are 306 responses (35.5% total response rate) from 199 transit agencies (72.9% of all agencies have at least one response) in our dataset. The missing values in two cases resulted in a final set of 304 observations from 199 transit agencies, which we used in the entire SEM model. The structural model achieved good model fit (χ2 = 151.263, df = 116, p-value = .0155, RMESA = 0.032 with 90% CI [0.015, 0.045], CFI = 0.966, TLI = 0.950, WRMR = 0.795). Those indices satisfy the general cutoff points for structural models using WLSMV estimation: p-value for Chi-square test ≥ 0.01, RMSEA ≤ 0.06, CFI ≥ 0.96, TLI ≥ 0.95, and WRMR ≤ 1.0 (Yu 2002). Table 3 reports the SEM results. The parameter estimates are followed by the standard errors in the parentheses. Northeast was used as the baseline for the effects of the region dummies. It is important to note that impact is positively associated with an organization’s exposure to extreme weather, yet negatively associated with its infrastructure quality. This is consistent with our theoretical framework which expects greater capacity to buffer the negative impacts of extreme events on the affected organization. Table 3. SEM Results Extreme Weather Impact Exposure 0.041 (0.009)*** Bus only −0.256 (0.099)** Infrastructure quality −0.127 (0.070)# Authority −0.111 (0.087) Organization size (log) 0.048 (0.033) Density (log) −0.047 (0.035) Median household income (log) 0.061 (0.265) Commute time −0.001 (0.012) Midwest −0.190 (0.137) South −0.175 (0.140) West −0.352 (0.141)* Risk Perception Exposure 0.034 (0.012)** Impact 0.322 (0.071)*** Bus only 0.133 (0.159) Infrastructure quality −0.043 (0.089) Authority −0.082 (0.111) Organization size (log) 0.118 (0.053)* Density (log) −0.038 (0.033) Median household income (log) 0.167 (0.311) Commute time 0.013 (0.015) Midwest −0.139 (0.195) South −0.043 (0.183) West −0.303 (0.195) Organizational Adaptation Vulnerability Assessment Capital Investment Exposure 0.033 (0.015)* 0.028 (0.012)* Impacts 0.025 (0.094) −0.036 (0.061) Risk perception 0.360 (0.075)*** 0. 168(0.061)** Bus only −0.139 (0.164) −0.130 (0.116) Infrastructure quality 0.027 (0.107) −0.014 (0.063) Authority −0.446 (0.152)** 0.032 (0.084) Organization size (log) 0.158 (0.060)** 0.037 (0.037) Density (log) 0.066 (0.062) −0.030 (0.034) Median household income (log) 0.100 (0.365) −0.352 (0.228) Commute time −0.006 (0.017) 0.025 (0.012)* Midwest −0.056 (0.218) 0.101 (0.137) South 0.266 (0.217) 0.135 (0.128) West 0.148 (0.222) 0.189 (0.138) Extreme Weather Impact Exposure 0.041 (0.009)*** Bus only −0.256 (0.099)** Infrastructure quality −0.127 (0.070)# Authority −0.111 (0.087) Organization size (log) 0.048 (0.033) Density (log) −0.047 (0.035) Median household income (log) 0.061 (0.265) Commute time −0.001 (0.012) Midwest −0.190 (0.137) South −0.175 (0.140) West −0.352 (0.141)* Risk Perception Exposure 0.034 (0.012)** Impact 0.322 (0.071)*** Bus only 0.133 (0.159) Infrastructure quality −0.043 (0.089) Authority −0.082 (0.111) Organization size (log) 0.118 (0.053)* Density (log) −0.038 (0.033) Median household income (log) 0.167 (0.311) Commute time 0.013 (0.015) Midwest −0.139 (0.195) South −0.043 (0.183) West −0.303 (0.195) Organizational Adaptation Vulnerability Assessment Capital Investment Exposure 0.033 (0.015)* 0.028 (0.012)* Impacts 0.025 (0.094) −0.036 (0.061) Risk perception 0.360 (0.075)*** 0. 168(0.061)** Bus only −0.139 (0.164) −0.130 (0.116) Infrastructure quality 0.027 (0.107) −0.014 (0.063) Authority −0.446 (0.152)** 0.032 (0.084) Organization size (log) 0.158 (0.060)** 0.037 (0.037) Density (log) 0.066 (0.062) −0.030 (0.034) Median household income (log) 0.100 (0.365) −0.352 (0.228) Commute time −0.006 (0.017) 0.025 (0.012)* Midwest −0.056 (0.218) 0.101 (0.137) South 0.266 (0.217) 0.135 (0.128) West 0.148 (0.222) 0.189 (0.138) Unstandardized coefficients are reported. Three hundred four observations from 199 agencies; Reference category: Northeast. #p < .10, *p < .05, **p < .01, ***p < .001. View Large Table 3. SEM Results Extreme Weather Impact Exposure 0.041 (0.009)*** Bus only −0.256 (0.099)** Infrastructure quality −0.127 (0.070)# Authority −0.111 (0.087) Organization size (log) 0.048 (0.033) Density (log) −0.047 (0.035) Median household income (log) 0.061 (0.265) Commute time −0.001 (0.012) Midwest −0.190 (0.137) South −0.175 (0.140) West −0.352 (0.141)* Risk Perception Exposure 0.034 (0.012)** Impact 0.322 (0.071)*** Bus only 0.133 (0.159) Infrastructure quality −0.043 (0.089) Authority −0.082 (0.111) Organization size (log) 0.118 (0.053)* Density (log) −0.038 (0.033) Median household income (log) 0.167 (0.311) Commute time 0.013 (0.015) Midwest −0.139 (0.195) South −0.043 (0.183) West −0.303 (0.195) Organizational Adaptation Vulnerability Assessment Capital Investment Exposure 0.033 (0.015)* 0.028 (0.012)* Impacts 0.025 (0.094) −0.036 (0.061) Risk perception 0.360 (0.075)*** 0. 168(0.061)** Bus only −0.139 (0.164) −0.130 (0.116) Infrastructure quality 0.027 (0.107) −0.014 (0.063) Authority −0.446 (0.152)** 0.032 (0.084) Organization size (log) 0.158 (0.060)** 0.037 (0.037) Density (log) 0.066 (0.062) −0.030 (0.034) Median household income (log) 0.100 (0.365) −0.352 (0.228) Commute time −0.006 (0.017) 0.025 (0.012)* Midwest −0.056 (0.218) 0.101 (0.137) South 0.266 (0.217) 0.135 (0.128) West 0.148 (0.222) 0.189 (0.138) Extreme Weather Impact Exposure 0.041 (0.009)*** Bus only −0.256 (0.099)** Infrastructure quality −0.127 (0.070)# Authority −0.111 (0.087) Organization size (log) 0.048 (0.033) Density (log) −0.047 (0.035) Median household income (log) 0.061 (0.265) Commute time −0.001 (0.012) Midwest −0.190 (0.137) South −0.175 (0.140) West −0.352 (0.141)* Risk Perception Exposure 0.034 (0.012)** Impact 0.322 (0.071)*** Bus only 0.133 (0.159) Infrastructure quality −0.043 (0.089) Authority −0.082 (0.111) Organization size (log) 0.118 (0.053)* Density (log) −0.038 (0.033) Median household income (log) 0.167 (0.311) Commute time 0.013 (0.015) Midwest −0.139 (0.195) South −0.043 (0.183) West −0.303 (0.195) Organizational Adaptation Vulnerability Assessment Capital Investment Exposure 0.033 (0.015)* 0.028 (0.012)* Impacts 0.025 (0.094) −0.036 (0.061) Risk perception 0.360 (0.075)*** 0. 168(0.061)** Bus only −0.139 (0.164) −0.130 (0.116) Infrastructure quality 0.027 (0.107) −0.014 (0.063) Authority −0.446 (0.152)** 0.032 (0.084) Organization size (log) 0.158 (0.060)** 0.037 (0.037) Density (log) 0.066 (0.062) −0.030 (0.034) Median household income (log) 0.100 (0.365) −0.352 (0.228) Commute time −0.006 (0.017) 0.025 (0.012)* Midwest −0.056 (0.218) 0.101 (0.137) South 0.266 (0.217) 0.135 (0.128) West 0.148 (0.222) 0.189 (0.138) Unstandardized coefficients are reported. Three hundred four observations from 199 agencies; Reference category: Northeast. #p < .10, *p < .05, **p < .01, ***p < .001. View Large Meanwhile, among the control variables only organizational size had a statistically significant effect on risk perception such that risk perceptions are higher in larger organizations. Among the control variables, organization size is positively associated with vulnerability assessment but not associated with capital assessment. Results also show that independent organizations (transit authority agencies) are less likely to conduct vulnerability assessment compared to agencies affiliated with a city or state government, and that agencies with higher commute time are more likely to undertake capital investment. Testing Hypotheses The relationships among variables of primary interest are presented in Figure 3. Standardized coefficients are reported to facilitate comparison of the relative effect of each predictor. Figure 3. View largeDownload slide SEM Analysis Result: Mediating Role of Risk Perception Figure 3. View largeDownload slide SEM Analysis Result: Mediating Role of Risk Perception Hypothesis 1 postulates a positive relationship between an organization’s exposure to extreme events on the impacts it experienced. Our analysis supports this hypothesis, with a significant and positive effect of exposure to extreme weather on impacts. Hypothesis 2 posits that as an organization suffers greater impacts from extreme events, it is more likely to build adaptive capacity in preparation for future shocks. The results show nonsignificant effects of impacts on the two adaptive strategies, therefore suggesting no support for Hypothesis 2. These findings indicate that an organization’s increased exposure to extreme events leads to greater negative impacts, but the impacts do not directly lead to organizational adaptive responses. We interpret this finding by separating the impact from organizational learning and decision processes. Longer-term initiatives that are adaptive may require organizations to perceive a substantially different environmental context in which the harm associated with maintaining the status quo is greater than the harm associated with adaptive change. As exposure increases and causes more significant impacts, only those organizations that cognitively process a different level of risk will adapt. We also believe there is an alternative explanation for this finding: it is possible that impact can both motivate and undermine an organization’s capacity to adapt. This may be especially true when organizations focus on immediate survival and on maintaining the status quo without embracing a longer-term vision to build adaptive capacity. The “hold-the-line” practice has been widely observed as diminishing capital resources for longer-term and hampering an organization’s learning and adaptation from recurring major perturbations (Adger et al. 2011; Linnenluecke et al. 2012; Winn et al. 2011). Hypothesis 3: expects that exposure to extreme events will directly affect an organization’s adaptive capacity building due to the learning mechanisms. Exposure turns out to be a positive predictor of an organization’s capital investment and of vulnerability assessment. Thus, our results lend support to Hypothesis 3. Hypothesis 4: focuses on how risk perception mediates the effect of an organization’s exposure to and impact from extreme events on its adaptive responses. The path coefficients from exposure and impacts to risk perception are significant and positive, so are the effects of risk perception on an organization’s implementation of vulnerability assessment and capital investment. In light of the asymmetry of the distribution of the product of multiple path coefficients, we applied bootstrapping to derive the confidence interval of the mediated effects through risk perception (MacKinnon 2008). There are two paths through which exposure affects adaption via risk perception: one path from exposure to risk perception and then adaptation, and one path from exposure to impact and then risk perception and adaptation. The results indicate that an organization’s risk perception effectively mediates the effect of exposure on vulnerability assessment (standardized β = 0.103, 95% CI = [0.038, 0.181]) and capital investment (standardized β = 0.076, 95% CI = [0.010, 0.157]). It also shows that risk perception significantly mediates the effect of impact on vulnerability assessment (standardized β = 0. 097, 95% CI = [0.036, 0.164]) and capital investment (standardized β = 0.071, 95% CI = [0.011, 0.143]). Therefore, the model shows consistent support for Hypothesis 4. The findings for Hypotheses 3 and 4 reveal two mechanisms through which organizational adaptive responses occur: the direct effect of increased exposure to extreme events as well as the mediated effects of risk perception. Both mechanisms imply organizational learning through increased problem familiarity and organizational sense-making, respectively. However, the two learning mechanisms vary in their relative strength. While the relationship between exposure and capital investment is highly significant, the statistical relationship and coefficient between exposure and vulnerability assessment are weaker. Because exposure may result in physical damage that requires repairs, it is possible that capital investment decisions are affected more by actual damage. Nevertheless, both vulnerability assessment and capital investment require cognitive change and the trigger provided by heightened risk perception (based on impact and exposure). Conclusion This study sets out to examine public organizations’ experience with extreme events and to explain the variations in their adaptive responses. The conceptual framework (Figure 1) integrates organizational adaptation and learning literature to establish an initial organizational response model that captures the essence of longer-term adaptation to reduce vulnerability to recurring extreme events. The theoretical model (Figure 2) addresses key interrelated questions surrounding the reasons why some public organizations adapt to extreme events and others do not, or do so more slowly. The model posits that exposure leads to capacity-moderated impacts (i.e., performance gap), but subsequent adaptation through capacity development is mediated by organization risk perception. Our findings support three of the four hypotheses generated from the model and deeper literature: increased exposure is associated with increased impacts (i.e., performance gap). Exposure is positively associated with organizational adaptation, and risk perception mediates the effect of exposure and impact on adaptive behavior. Infrastructure-based capacity also helps buffer the negative impacts, further supporting our overall model. The study contributes to the broader literature in several ways. It captures the common patterns across extreme events and takes an organizational approach to address the challenges associated with adaptation to reduce vulnerability in public organizations. This responds to the recent literature by connecting the study on extreme events with organizational theory to enable a more systematic and generic understanding and treatment of extreme events (Boin and Van Eeten 2013; Christensen et al. 2016; Fischbacher-Smith 2010; Roux-Dufort 2007). As such, this research provides one of the few theoretically informed, quantitative studies on core topics concerning extreme events (Boin and Van Eeten 2013; Christensen et al. 2016). Our findings, in conjunction with our broader conceptual model, can be interpreted as evidence for perception-mediated learning in which adaptation requires not simply exposure to a phenomenon, but also the cognitive understanding that longer-term performance of an organization depends on some form of purposive adaptation. Both exposure and the effects of exposure on performance create opportunities for organizations to raise questions about their ability to perform. When severe events increase in frequency, and when emergency responses become more frequent, ineffective and costly, organizations begin to make sense of the pattern of the challenges and realize that longer-term investment in capacity is necessary for sustainable operation. The study may point to a cognition-based stepwise learning model in which organization commitment to change is incremental and depends on the recognition that exogenous shocks are systematic, solvable and require new investments. A baseline exists when an organization responds to extreme events using routinized or programmed emergency response actions. Increased frequency of exogenous shocks may increase familiarity with performance gaps, demonstrate fundamental limitations in capacity, and stimulate learning manifested as increased perceptions of risk. Increased perception of risk may facilitate organizational commitment to undertake some form of adaptive behavior to reduce vulnerability. The stronger linkage between exposure and capital investment combined with findings that impacts are only indirectly related to adaptation through risk perception provide further support that commitment to information gathering through vulnerability assessment may be an intermediary step, before capital investment-based adaptation. Practically, the study reveals the pitfalls in assuming that organizational adaptation to extreme events will occur spontaneously with growth in impacts from extreme events. Instead, the effect of extreme events on adaptation behavior is most likely channeled through a cognitive process wherein the risks are perceived and appropriately interpreted by the affected organization. The significant role of risk perception gives important room for management intervention. Since people are more sensitive to threats than opportunities in undertaking larger-scale internal responses (Dutton and Jackson 1987), managers can affect organization-wide risk perceptions by actively recognizing and framing threats, seeking to systematically collect and interpret new information and establishing it as part of the organizational memory. Managers can prime organizational members to better identify extreme events, recognize patterns, and develop solutions that reduce threats (Weick 1977). Managers may adopt a participatory approach to promote shared understanding and interpretation of extreme events, and pave the way for potentially adaptive solutions. The participatory efforts can be more successfully implemented by highly trusted individuals in the organization (Dutton and Jackson 1987). We acknowledge the limitations of our study. Our empirical analysis focuses on transit agencies, which may reduce generalizability of the findings of this study to other organizations. However, to some extent, the national-level approach in which agencies are surveyed across a wide range of weather conditions and extreme events and the similarity of transit agencies with many other resource intensive public agencies (i.e., utility, power, waste management) reduces this concern. Reliance on survey data for the construction of key variables in the estimation may raise concerns about common source bias. The use of survey data is valid for two reasons. First, other sources of data on organizational experience with extreme weather events, especially impacts, are difficult to collect. Additionally, because risk is fundamentally a perceptual construct, it may be best to rely on perceptual measures collected in surveys. Previous studies have suggested that self-reported data can provide valid indicators of organizational properties (Lincoln and Zeitz 1980; Moynihan and Pandey 2005; Pandey and Wright 2006). That many coefficients in our model are low and not significant can also help alleviate the common method bias concern (George and Pandey 2017). Another limitation has to do with the use of cross-sectional data to test the mediation effect. We are limited by data availability to test a three-wave longitudinal data as suggested by the literature (Cole and Maxwell 2003). However, given the scarcity of data on how public organizations adapt to extreme events, a growing and salient challenge to public organizations worldwide, we believe that the survey and our analysis still shine light on this important topic. Moreover, a major concern with longitudinal data relates to the difficulty of correctly specifying the time lag between the data collection points such that the data actually captures meaningful variation in the variable of interest, instead of stochastic changes due to time lapse (Cole and Maxwell 2003; Ployhart and Vandenberg 2010). In studies of adaptation, the time lag between extreme weather exposure, impact, and risk perception is difficult to pinpoint (Berkhout 2012; Christianson et al. 2009; Weick 1993). This makes specification of the exact duration of the time lag challenging such that longitudinal data and methods may not be necessarily superior. To the extent that the time interval is small, the use of cross-sectional data instead might be a viable representation of reality in this case (Wong and Law 1999). This study raises several new questions for further work. Are public organizations actually learning from increasingly frequent and severe events in ways that fundamentally alter investment patterns over time? This study is suggestive, but not conclusive. What other adaptation strategies are organizations undertaking besides those examined in this study? In large complex organizations, at what point do occasional shocks become recurrent; at what point do frequency, magnitude, and scope force organizational attention? Is there a step-wise learning process in which commitment is contingent on risk perception and adaptive actions are nested? Given the rise in “extreme events” that cause significant disruption (Boin and Lodge 2016; Comfort et al. 2012; Tierney 2014) future research should begin to address these issues. Acknowledgment This research was made possible through generous support by the Federal Transit Administration, US Department of Transportation. We would also like to acknowledge the anonymous reviewers for their helpful comments and suggestions. References Adger , W. Neil , Katrina Brown , Donald R. Nelson , Fikret Berkes , Hallie Eakin , Carl Folke , Kathleen Galvin , et al. 2011 . Resilience implications of policy responses to climate change: Resilience implications of policy responses to climate change . Wiley Interdisciplinary Reviews: Climate Change 2 : 757 – 66 . Google Scholar CrossRef Search ADS Beauducel , André , and Philipp Yorck Herzberg . 2006 . On the performance of maximum likelihood versus means and variance adjusted weighted least squares estimation in CFA . Structural Equation Modeling 13 : 186 – 203 . Google Scholar CrossRef Search ADS Berkes , Fikret . 2007 . Understanding uncertainty and reducing vulnerability: Lessons from resilience thinking . Natural Hazards 41 : 283 – 95 . Google Scholar CrossRef Search ADS Berkes , Fikret , and Carl Folke . 2002 . Back to the future: Ecosystem dynamics and local knowledge. In Panarchy: understanding transformation in human and natural systems, ed. Lance H. Gunderson and Crawford S. Holling, 121–146. Washington: Island Press. Berkhout , Frans . 2012 . Adaptation to climate change by organizations . Wiley Interdisciplinary Reviews: Climate Change 3 : 91 – 106 . Google Scholar CrossRef Search ADS Birkland , Thomas A . 1997 . After Disaster: Agenda Setting, Public Policy, and Focusing Events . Washington, D.C.: Georgetown Univ. Press . ———. 2009 . Disasters, lessons learned, and fantasy documents . Journal of Contingencies and Crisis Management 17 : 146 – 56 . CrossRef Search ADS Boin , Arjen , and Paul ‘t Hart . 2003 . Public leadership in times of crisis: Mission impossible ? Public Administration Review 63 : 544 – 53 . Google Scholar CrossRef Search ADS Boin , Arjen , and Martin Lodge . 2016 . Designing resilient institutions for transboundary crisis management: A time for public administration . Public Administration 94 : 289 – 98 . Google Scholar CrossRef Search ADS Boin , Arjen , and Allan McConnell . 2007 . Preparing for critical infrastructure breakdowns: The limits of crisis management and the need for resilience . Journal of Contingencies and Crisis Management 15 : 50 – 9 . Google Scholar CrossRef Search ADS Boin , Arjen , Eric Stern , and Bengt Sundelius . 2016 . The politics of crisis management: Public leadership under pressure . Cambridge, UK: Cambridge Univ. Press . Boin , Arjen , and Michel JG Van Eeten . 2013 . The resilient organization . Public Management Review 15 : 429 – 45 . Google Scholar CrossRef Search ADS Cairney , Paul , Kathryn Oliver , and Adam Wellstead . 2016 . To bridge the divide between evidence and policy: Reduce ambiguity as much as uncertainty . Public Administration Review 76 : 399 – 402 . Google Scholar CrossRef Search ADS Christensen , Tom , Per Lægreid , and Lise H. Rykkja . 2016 . Organizing for crisis management: Building governance capacity and legitimacy . Public Administration Review 76 : 887 – 97 . Google Scholar CrossRef Search ADS Christianson , Marlys K. , Maria T. Farkas , Kathleen M. Sutcliffe , and Karl E. Weick . 2009 . Learning through rare events: Significant interruptions at the Baltimore & Ohio Railroad Museum . Organization Science 20 : 846 – 60 . Google Scholar CrossRef Search ADS Cigler , Beverly A . 2007 . The ‘big questions’ of Katrina and the 2005 Great Flood of New Orleans . Public Administration Review 67 ( Suppl 1 ): 64 – 76 . Google Scholar CrossRef Search ADS Clarke , Lee . 1999 . Mission improbable: Using fantasy documents to tame disaster . Chicago, IL: University of Chicago Press . Cole , David A. , and Scott E. Maxwell . 2003 . Testing mediational models with longitudinal data: Questions and tips in the use of structural equation modeling . Journal of Abnormal Psychology 112 : 558–77 . Google Scholar CrossRef Search ADS PubMed Comfort , Louise K . 2002 . Rethinking security: Organizational fragility in extreme events . Public Administration Review 62 ( Suppl 1 ): 98 – 107 . Google Scholar CrossRef Search ADS ———. 2005 . Risk, security and disaster management . Annual Review of Political Science 8 : 335 – 56 . CrossRef Search ADS ———. 2007 . Crisis management in hindsight: Cognition, communication, coordination, and control . Public Administration Review 67 ( Suppl 1 ): 189 – 97 . CrossRef Search ADS Comfort , Louise K. , and Aya Okada . 2013 . Symposium-emergent leadership in extreme events: A knowledge commons for sustainable communities . International Review of Public Administration 18 : 61 – 77 . Google Scholar CrossRef Search ADS Comfort , Louise K. , William L. Waugh , and Beverly A. Cigler . 2012 . Emergency management research and practice in public administration: Emergence, evolution, expansion, and future directions . Public Administration Review 72 : 539 – 47 . Google Scholar CrossRef Search ADS Crichton , Margaret T. , Cameron G. Ramsay , and Terence Kelly . 2009 . Enhancing organizational resilience through emergency planning: Learnings from cross-sectoral lessons . Journal of Contingencies and Crisis Management 17 : 24 – 37 . Google Scholar CrossRef Search ADS Cyert , Richard M. , and James G. March . 1963 . A behavioral theory of the firm , 2nd ed.. Englewood Cliffs, NJ: Prentice Hall . Daft , Richard L. , and Karl E. Weick . 1984 . Toward a model of organizations as interpretation systems . Academy of Management Review 9 : 284 – 95 . Google Scholar CrossRef Search ADS Dekker , Sander , and Dan Hansén . 2004 . Learning under pressure: The effects of politicization on organizational learning in public bureaucracies . Journal of Public Administration Research and Theory 14 : 211 – 30 . Google Scholar CrossRef Search ADS Duncan , Robert B . 1972 . Characteristics of organizational environments and perceived environmental uncertainty . Administrative Science Quarterly 17 : 313 – 27 . Google Scholar CrossRef Search ADS Dutton , Jane E. , and Susan E. Jackson . 1987 . Categorizing strategic issues: Links to organizational action . The Academy of Management Review 12 : 76 – 90 . Google Scholar CrossRef Search ADS Ebert , Suzanne , Orieta Hulea , and David Strobel . 2009 . Floodplain restoration along the lower Danube: A climate change adaptation case study . Climate and Development 1 : 212 – 19 . Google Scholar CrossRef Search ADS Fankhauser , Samuel , Joel B. Smith , and Richard S. J. Tol . 1999 . Weathering climate change: Some simple rules to guide adaptation decisions . Ecological Economics 30 : 67 – 78 . Google Scholar CrossRef Search ADS Fischbacher-Smith , Denis . 2010 . Beyond the worst case scenario: ‘Managing’ the risks of extreme events . Risk Management 12 : 1 – 8 . Google Scholar CrossRef Search ADS Folke , Carl . 2006 . Resilience: The emergence of a perspective for social–ecological systems analyses . Global Environmental Change 16 : 253 – 67 . Google Scholar CrossRef Search ADS Ford , James D. , and Diana King . 2015 . A framework for examining adaptation readiness . Mitigation and Adaptation Strategies for Global Change 20 : 505 – 26 . Google Scholar CrossRef Search ADS Gallagher , Justin . 2014 . Learning about an infrequent event: Evidence from flood insurance take-up in the United States . American Economic Journal: Applied Economics 6 : 206 – 33 . Google Scholar CrossRef Search ADS Garrett , Terence M . 2004 . Whither Challenger, wither Columbia: Management decision making and the knowledge analytic . The American Review of Public Administration 34 : 389 – 402 . Google Scholar CrossRef Search ADS George , Bert , and Sanjay K. Pandey . 2017 . We know the Yin—but where is the Yang? Toward a balanced approach on common source bias in public administration scholarship . Review of Public Personnel Administration 37 : 245 – 70 . Google Scholar CrossRef Search ADS PubMed Grothmann , Torsten , and Anthony Patt . 2005 . Adaptive capacity and human cognition: The process of individual adaptation to climate change . Global Environmental Change 15 : 199 – 213 . Google Scholar CrossRef Search ADS Haigh , Nardia , and Andrew Griffiths . 2011 . Surprise as a catalyst for including climatic change in the strategic environment . Business & Society , 51 : 1552 – 4205 . Hambrick , D. C. , and P. A. Mason . 1984 . Upper echelons: The organization as a reflection of its top managers . Academy of Management Review 9 : 193 – 206 . Google Scholar CrossRef Search ADS Hambrick , Donald C . 2007 . Upper echelons theory: An update . Academy of Management Review 32 : 334 – 43 . Google Scholar CrossRef Search ADS Hegerl , G. C. , F. W. Zwiers , P. Braconnot , N. P. Gillett , Y. Luo , J. A. Marengo Orsini , N. Nicholls , J. E. Penner , and P. A. Stott . 2007 . Understanding and attributing climate change . In Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change , eds. S. Solomon , D. Qin , M. Manning , Z. Chen , M. Marquis , KB Averyt , M. Tignor and HL Miller , 663 – 745 . Cambridge, UK : Cambridge Univ. Press . Henstra , Daniel . 2010 . Evaluating local government emergency management programs: What framework should public managers adopt ? Public Administration Review 70 : 236 – 46 . Google Scholar CrossRef Search ADS Hill , Margot , and Nathan L. Engle . 2013 . Adaptive capacity: Tensions across scales . Environmental Policy and Governance 23 : 177 – 92 . Google Scholar CrossRef Search ADS Hodges , Tina . 2011 . Flooded bus barns and buckled rails: Public transportation and climate change adaptation . Washington, DC : Federal Transit Administration . http://www.fta.dot.gov/research (assessed July 26, 2016). Hoffmann , Volker H. , David C. Sprengel , Andreas Ziegler , Matthias Kolb , and Bruno Abegg . 2009 . Determinants of corporate adaptation to climate change in winter tourism: An econometric analysis . Global Environmental Change 19 : 256 – 64 . Google Scholar CrossRef Search ADS Hoyle , Rick H . 2012 . Handbook of structural equation modeling . New York City, NY: Guilford Press . Jöreskog , Karl G . 1994 . Structural equation modeling with ordinal variables . Multivariate Analysis and Its Applications , 24 : 297 – 310 . Google Scholar CrossRef Search ADS Kahneman , Daniel , and Amos Tversky . 1979 . Prospect theory: An analysis of decision under risk . Econometrica: Journal of the Econometric Society , 47 : 263 – 291 . Google Scholar CrossRef Search ADS Kapucu , Naim . 2005 . Interorganizational coordination in dynamic context: Networks in emergency response management . Connections 26 : 33 – 48 . Kapucu , Naim , and Vener Garayev . 2011 . Collaborative decision-making in emergency and disaster management . International Journal of Public Administration 34 : 366 – 75 . Google Scholar CrossRef Search ADS Kiesler , Sara , and Lee Sproull . 1982 . Managerial response to changing environments: Perspectives on problem sensing from social cognition . Administrative Science Quarterly , 27 : 548 – 570 . Google Scholar CrossRef Search ADS Kimrey , Christopher M . 2016 . Opportunities in crisis and catastrophe: The issue-attention cycle and political reality . Homeland Security Affairs 12 . Kingdon , John W . 1984 . Agendas, alternatives, and public policies . Vol. 45 . Little, Brown Boston . Kline , Rex B . 2015 . Principles and practice of structural equation modeling . New York City, NY: Guilford Publications . Koetse , Mark J. , and Piet Rietveld . 2009 . The impact of climate change and weather on transport: An overview of empirical findings . Transportation Research Part D: Transport and Environment 14 : 205 – 21 . Google Scholar CrossRef Search ADS Kusumasari , Bevaola , Quamrul Alam , and Kamal Siddiqui . 2010 . Resource capability for local government in managing disaster . Disaster Prevention and Management: An International Journal 19 : 438 – 51 . Google Scholar CrossRef Search ADS La Porte , Todd R . 1996 . High reliability organizations: Unlikely, demanding and at risk . Journal of Contingencies and Crisis Management 4 : 60 – 71 . Google Scholar CrossRef Search ADS Lawrence , Paul R. , and Jay W. Lorsch . 1967 . Differentiation and integration in complex organizations . Administrative Science Quarterly , 12: 1 – 47 . Google Scholar CrossRef Search ADS Levinthal , D . 2000 . Organizational capabilities in complex worlds . In The nature and dynamics of organizational capabilities, ed. Dosi, Giovanni, Nelson, Richard R., Winter, and Sidney G., 363–78). New York: Oxford University Press. Levitt , Barbara , and James G. March . 1988 . Organizational learning . Annual Review of Sociology , 14 : 319 – 40 . Google Scholar CrossRef Search ADS Lincoln , James R. , and Gerald Zeitz . 1980 . Organizational properties from aggregate data: Separating individual and structural effects . American Sociological Review , 45 : 391 – 408 . Google Scholar CrossRef Search ADS Linnenluecke , Martina K. , Andrew Griffiths , and Monika Winn . 2012 . Extreme weather events and the critical importance of anticipatory adaptation and organizational resilience in responding to impacts . Business Strategy and the Environment 21 : 17 – 32 . Google Scholar CrossRef Search ADS Little , Richard G . 2002 . Controlling cascading failure: Understanding the vulnerabilities of interconnected infrastructures . Journal of Urban Technology 9 : 109 – 23 . Google Scholar CrossRef Search ADS MacKinnon , David Peter . 2008 . Introduction to statistical mediation analysis . Abingdon, UK: Routledge . March , James G . 1991 . Exploration and exploitation in organizational learning . Organization Science 2 : 71 – 87 . Google Scholar CrossRef Search ADS March , James G. , and Zur Shapira . 1987 . Managerial perspectives on risk and risk taking . Management Science 33 : 1404 – 18 . Google Scholar CrossRef Search ADS McDaniels , Timothy , Stephanie Chang , Darren Cole , Joseph Mikawoz , and Holly Longstaff . 2008 . Fostering resilience to extreme events within infrastructure systems: Characterizing decision contexts for mitigation and adaptation . Global Environmental Change 18 : 310 – 18 . Google Scholar CrossRef Search ADS McEntire , David A . 2004 . The status of emergency management theory: Issues, barriers, and recommendations for improved scholarship . University of North Texas. Department of Public Administration. Emergency Administration and Planning . ———. 2005 . Why vulnerability matters: Exploring the merit of an inclusive disaster reduction concept . Disaster Prevention and Management: An International Journal 14 : 206 – 22 . CrossRef Search ADS ———. 2008 . A critique of emergency management policy: Recommendations to reduce disaster vulnerability . International Journal of Public Policy 3 : 302 – 10 . CrossRef Search ADS McEntire , David A. , Christopher Fuller , Chad W. Johnston , and Richard Weber . 2002 . A comparison of disaster paradigms: The search for a holistic policy guide . Public Administration Review 62 : 267 – 81 . Google Scholar CrossRef Search ADS McGuire , Michael , and Debra Schneck . 2010 . What if Hurricane Katrina hit in 2020? The need for strategic management of disasters . Public Administration Review 70 ( s1 ): s201 – 7 . Google Scholar CrossRef Search ADS McGuire , Michael , and Chris Silvia . 2010 . The effect of problem severity, managerial and organizational capacity, and agency structure on intergovernmental collaboration: Evidence from local emergency management . Public Administration Review 70 : 279 – 88 . Google Scholar CrossRef Search ADS Meyer , Michael D. , E. Rowan , C. Snow , and A. Choate . 2013 . Impacts of extreme weather on transportation: National symposium summary . Washington, DC : American Association of State Highway and Transportation Officials . Meyer , Michael D. , Emily Rowan , Michael J. Savonis , and Anne Choate . 2012 . Integrating extreme weather risk into transportation asset management . American Association of State Highway and Transportation Officials . Moynihan , Donald P . 2008 . The dynamics of performance management: Constructing information and reform . Washington, D.C.: Georgetown Univ. Press . Moynihan , Donald P. , and Sanjay K. Pandey . 2005 . Testing how management matters in an era of government by performance management . Journal of Public Administration Research and Theory 15 : 421 – 39 . Google Scholar CrossRef Search ADS Muthén , Bengt O . 1993 . Goodness of fit with categorical and other nonnormal variables . In Testing structural equation models , ed. Bollen, Kenneth A., Long, and Scott A., 205 – 243 . Newbury Park: SAGE Publications . NOAA . 2016 . Billion-dollar weather and climate disasters: Table of events . http://www.ncdc.noaa.gov/billions/events (assessed October 30, 2017). Pandey , Sanjay K. , and Bradley E. Wright . 2006 . Connecting the dots in public management: Political environment, organizational goal ambiguity, and the public manager’s role ambiguity . Journal of Public Administration Research and Theory 16(4):511–32. Parker , Charles F. , Eric K. Stern , Eric Paglia , and Christer Brown . 2009 . Preventable catastrophe? The Hurricane Katrina disaster revisited . Journal of Contingencies and Crisis Management 17 : 206 – 20 . Google Scholar CrossRef Search ADS Parmesan , Camille , and Gary Yohe . 2003 . A globally coherent fingerprint of climate change impacts across natural systems . Nature 421 : 37 – 42 . Google Scholar CrossRef Search ADS PubMed Perrow , Charles . 1984 . Normal accidents: Living with high risk systems . New York : Basic Books . ———. 1994 . The limits of safety: The enhancement of a theory of accidents . Journal of Contingencies and Crisis Management 2 : 212 – 20 . CrossRef Search ADS ———. 1999 . Organizing to reduce the vulnerabilities of complexity . Journal of Contingencies and Crisis Management 7 : 150 – 5 . CrossRef Search ADS Ployhart , Robert E. , and Robert J. Vandenberg . 2010 . Longitudinal research: The theory, design, and analysis of change . Journal of Management 36 : 94 – 120 . Google Scholar CrossRef Search ADS Repetto , Robert C . 2009 . The climate crisis and the adaptation myth. WP 13 . New Haven, CT : School of Forestry & Environmental Studies, Yale University . Rijpma , Jos A . 1997 . Complexity, tight-coupling and reliability: Connecting normal accidents theory and high reliability theory . Journal of Contingencies and Crisis Management 5 : 15 – 23 . Google Scholar CrossRef Search ADS Rosenthal , Uriel , and Alexander Kouzmin . 1997 . Crises and crisis management: Toward comprehensive government decision making . Journal of Public Administration Research and Theory 7 : 277 – 304 . Google Scholar CrossRef Search ADS Roux-Dufort , Christophe . 2007 . Is crisis management (only) a management of exceptions ? Journal of Contingencies and Crisis Management 15 : 105 – 14 . Google Scholar CrossRef Search ADS Sarewitz , Daniel , Roger Pielke , and Mojdeh Keykhah . 2003 . Vulnerability and risk: Some thoughts from a political and policy perspective . Risk Analysis 23 : 805 – 10 . Google Scholar CrossRef Search ADS PubMed Schulman , Paul R . 1993 . The negotiated order of organizational reliability . Administration & Society 25 : 353 – 72 . Google Scholar CrossRef Search ADS Schwenk , Charles R . 1984 . Cognitive simplification processes in strategic decision-making . Strategic Management Journal 5 : 111 – 28 . Google Scholar CrossRef Search ADS Simon , Herbert A . 1979 . Rational decision making in business organizations . The American Economic Review , 69 : 493 – 513 . Singhal , Abhishek , Camille Kamga , and Anil Yazici . 2014 . Impact of weather on urban transit ridership . Transportation Research Part A: Policy and Practice 69 : 379 – 91 . Google Scholar CrossRef Search ADS Sitkin , Sim B. , and Amy L. Pablo . 1992 . Reconceptualizing the determinants of risk behavior . Academy of Management Review 17 : 9 – 38 . Google Scholar CrossRef Search ADS Sitkin , Sim B. , and Laurie R. Weingart . 1995 . Determinants of risky decision-making behavior: A test of the mediating role of risk perceptions and propensity . The Academy of Management Journal 38 : 1573 – 92 . Smith , Adam B . 2017 . 2016: A historic year for billion-dollar weather and climate disasters in U.S . NOAA Climate.gov. https://www.climate.gov/news-features/blogs/beyond-data/2016-historic-year-billion-dollar-weather-and-climate-disasters-us (accessed October 23, 2017). Smith , Joel B. , Jason M. Vogel , John E. Iii , and Cromwell . 2009 . An architecture for government action on adaptation to climate change. An editorial comment . Climatic Change 95 : 53 – 61 . Google Scholar CrossRef Search ADS Somers , Scott , and James H. Svara . 2009 . Assessing and managing environmental risk: Connecting local government management with emergency management . Public Administration Review 69 : 181 – 93 . Google Scholar CrossRef Search ADS Staber , Udo , and Jorg Sydow . 2002 . Organizational adaptive capacity: A structuration perspective . Journal of Management Inquiry 11 : 408 – 24 . Google Scholar CrossRef Search ADS Stark , Alastair . 2014 . Bureaucratic values and resilience: An exploration of crisis management adaptation . Public Administration 92 : 692 – 706 . Staw , Barry M. , Lance E. Sandelands , and Jane E. Dutton . 1981 . Threat rigidity effects in organizational behavior: A multilevel analysis . Administrative Science Quarterly 26 : 501 – 24 . Google Scholar CrossRef Search ADS Stehr , Steven D . 2006 . The political economy of urban disaster assistance . Urban Affairs Review 41 : 492 – 500 . Google Scholar CrossRef Search ADS ‘t Hart , Paul . 2013 . After Fukushima: Reflections on risk and institutional learning in an era of mega-crises . Public Administration 91 : 101 – 13 . Google Scholar CrossRef Search ADS Taleb , Nassim N. , Daniel G. Goldstein , and Mark W. Spitznagel . 2009 . The six mistakes executives make in risk management . Harvard Business Review 87 : 78 – 81 . Google Scholar PubMed Thomas , James B. , and Reuben R. McDaniel . 1990 . Interpreting strategic issues: Effects of strategy and the information-processing structure of top management teams . The Academy of Management Journal 33 : 286 – 306 . Thompson , James D . 1967 . Organizations in action: Social science bases of administrative theory . New York: McGraw-Hill . Thomson , Barbara . 2017 . Managing extreme weather at bus stops . Washington, DC : Transportation Research Board . Google Scholar CrossRef Search ADS Tierney , Kathleen . 2014 . The social roots of risk: Producing disasters, promoting resilience . Palo Alto, CA: Stanford Univ. Press . Turner , Barry A. , and Nick F. Pidgeon . 1997 . Man-made disasters ., 2nd ed. Oxford, UK: Butterworth-Heinemann. Turner , Billie L. , Roger E. Kasperson , Pamela A. Matson , James J. McCarthy , Robert W. Corell , Lindsey Christensen , Noelle Eckley , et al. 2003 . A framework for vulnerability analysis in sustainability science . Proceedings of the National Academy of Sciences of the United States of America 100 : 8074 – 9 . Google Scholar CrossRef Search ADS PubMed Weick , Karl E . 1977 . Enactment processes in organizations . In New directions in organizational behavior , Staw, Barry M., and Salancik, Gerald R., 267–300. Chicago, IL: St. Clair Publications. ———. 1993 . The collapse of sensemaking in organizations: The Mann Gulch disaster . Administrative Science Quarterly , 38 : 628 – 52 . CrossRef Search ADS Welch , Eric W. , Qing Miao , Fengxiu Zhang , and P.S. Sriraj . 2016 . National survey of US public transit agency experience with and response to extreme weather events . Washington, DC : National Transportation Library . https://rosap.ntl.bts.gov/view/dot/31623 (accessed December 30, 2017). West , Stephen G. , Aaron B. Taylor , and Wei Wu . 2012 . Model fit and model selection in structural equation modeling . In Handbook of structural equation modeling , 1st ed., ed. Rick H. Hoyle, 209–231. New York City, NY: The Guilford Press. Winn , Monika , Manfred Kirchgeorg , Andrew Griffiths , Martina K. Linnenluecke , and Elmar Günther . 2011 . Impacts from climate change on organizations: A conceptual foundation . Business Strategy and the Environment 20 : 157 – 73 . Google Scholar CrossRef Search ADS Wise , Charles R . 2006 . Organizing for homeland security after Katrina: Is adaptive management what’s missing ? Public Administration Review 66 : 302 – 18 . Google Scholar CrossRef Search ADS Wong , Chi-Sum , and Kenneth S. Law . 1999 . Testing reciprocal relations by nonrecursive structural equation models using cross-sectional data. Organizational Research Methods 2(1):69–87. Wooldridge , Jeffrey M . 2003 . Cluster-sample methods in applied econometrics . American Economic Review 93 : 133 – 38 . Google Scholar CrossRef Search ADS Yu , Ching-Yun . 2002 . Evaluating cutoff criteria of model fit indices for latent variable models with binary and continuous outcomes , Vol. 30 . Los Angeles : University of California, Los Angeles . Appendix 1. Definition of the Evaluation Criteria for Infrastructure-Based Capacity The respondents were asked to rate the state-of-good-repair of their agency’s infrastructure, using the following scales: Poor—Asset is past its useful life and is in need of immediate repair or replacement; may have critically damaged component(s) Marginal—Asset reaching or just past the end of its useful life; increasing number of defective or deteriorated component(s) and increasing maintenance needs Adequate—Asset has reached its mid-life (condition 3.5); some moderately defective or deteriorated component(s) Good—Asset showing minimal signs of wear; some (slightly) defective or deteriorated component(s) Excellent—New asset; no visible defects Appendix 2. Summary Statistics on Risk Perception No. of Respondents Proportion Risk perception 1  Category 1 23 0.08  Category 2 47 0.16  Category 3 89 0.29  Category 4 111 0.37  Category 5 33 0.11 Risk perception 2  Category 1 12 0.04  Category 2 37 0.12  Category 3 133 0.45  Category 4 104 0.35  Category 5 13 0.04 Risk perception 3  Category 1 16 0.05  Category 2 49 0.16  Category 3 89 0.30  Category 4 123 0.41  Category 5 24 0.08 No. of Respondents Proportion Risk perception 1  Category 1 23 0.08  Category 2 47 0.16  Category 3 89 0.29  Category 4 111 0.37  Category 5 33 0.11 Risk perception 2  Category 1 12 0.04  Category 2 37 0.12  Category 3 133 0.45  Category 4 104 0.35  Category 5 13 0.04 Risk perception 3  Category 1 16 0.05  Category 2 49 0.16  Category 3 89 0.30  Category 4 123 0.41  Category 5 24 0.08 View Large No. of Respondents Proportion Risk perception 1  Category 1 23 0.08  Category 2 47 0.16  Category 3 89 0.29  Category 4 111 0.37  Category 5 33 0.11 Risk perception 2  Category 1 12 0.04  Category 2 37 0.12  Category 3 133 0.45  Category 4 104 0.35  Category 5 13 0.04 Risk perception 3  Category 1 16 0.05  Category 2 49 0.16  Category 3 89 0.30  Category 4 123 0.41  Category 5 24 0.08 No. of Respondents Proportion Risk perception 1  Category 1 23 0.08  Category 2 47 0.16  Category 3 89 0.29  Category 4 111 0.37  Category 5 33 0.11 Risk perception 2  Category 1 12 0.04  Category 2 37 0.12  Category 3 133 0.45  Category 4 104 0.35  Category 5 13 0.04 Risk perception 3  Category 1 16 0.05  Category 2 49 0.16  Category 3 89 0.30  Category 4 123 0.41  Category 5 24 0.08 View Large © The Author(s) 2018. Published by Oxford University Press on behalf of the Public Management Research Association. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com. This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

Journal

Journal of Public Administration Research and TheoryOxford University Press

Published: Feb 27, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off