Drivers of Household Flood Mitigation Decisions After Successive Flood Events in Waterbury, Vermont

Publication Date: 2026

Abstract

Click to View the Research Brief

Flooding has a profound global impact on human life, and the intensification of extreme weather events due to climate change will require significant adaptation, such as implementing measures to mitigate flood impacts or vacating flood zones altogether. Although decision-making strategies for flood mitigation have been studied, little attention has been given to how these decisions are made when multiple extreme events occur over a short time period. To address this gap, we interviewed 13 flood-impacted residents of Waterbury, VT—a community that has experienced four extreme weather events within 13 years—to answer the following research questions: (a) What factors led flood-vulnerable residents or business owners to take flood mitigation actions?; (b) What specific actions did interviewees perceive to be effective?; and (c) What interventions can increase the adoption of mitigation actions? Using the Protection Motivation Theory and Trust Ecology as theoretical frameworks to guide our analysis, we found that the rapid acceleration of flood event frequency in this area has accelerated individuals’ perceived vulnerability and their intent to mitigate flood damage. Accordingly, the factors that were most predictive of individuals’ decision to adopt mitigation behavior were (a) their assessment of their own self-efficacy and (b) their assessment of the efficacy of a given mitigation action. This study also contributed practitioner recommendations. State and local policymakers could improve the odds of residents adopting flood mitigation actions by establishing better communications policies, including clearer points of contact for people navigating flood recovery and developing resources that explain situationally specific mitigation ideas. State and local officials should also establish more transparency in decision-making processes around community-wide flood response projects, like floodplain restoration or buyouts.

Introduction

Flooding has an immense impact on human life on a global scale, and as climate change and associated extreme weather events increase in frequency and severity, the size and duration of floods are expected to intensify. As such, there is an increasing need for the adoption of mitigation actions at the global scale as well as at the regional and local scale. To examine adoption of mitigation actions at a local level, this report focuses on communities in the U.S. state of Vermont, which has endured reoccurring flood events in recent years. Heavy rains on July 10-11, 2023, resulted in destructive flooding that inundated floodplain communities throughout the state, including many of those along the Winooski River, which rose over 16 ft in 24 hr (Poniz, 2023¹). The flooding resulted in multiple fatalities (Banacos, 2023²), millions of dollars in damage to property and infrastructure (Giles & Dockser, 2023³), major crop loss (Sadoff & Barton, 2023⁴), increased water pollution via fuel and nutrient-laden sediment (Giles & Dockser, 2023), as well as trauma and a reduced sense of security for residents (Poniz, 2023).

In 2011, heavy rainfall from Hurricane Irene caused a 500-year flood event and resulted in the deaths of seven people and $750 million in property damage in the state (O’Connor, 2021⁵). Many of the same places that were heavily flooded during Irene were flooded again during the 2023 floods. Severe flood events also occurred in the Winooski River watershed (in the North-Central portion of the state) in December 2023 and July 2024 (Goff, n.d.⁶; Goff et al., 2024⁷). Thus, many Vermont residents have experienced three floods within two years (and four within 13 years) that approach or exceed the 100-year flood level (Federal Emergency Management Agency [FEMA], 2013⁸), putting these individuals on the front lines of climate-driven extreme weather intensification. The result has been a major increase in attention and funding from government agencies, nonprofits (Cutler, 2023⁹), and universities (Defibaugh, 2023¹⁰) across the state. However, the cumulative impact of these floods on the willingness of Vermonters to adapt and to mitigate in the face of future hazards remains unknown. This report seeks to begin addressing that knowledge gap. Its findings and implications are relevant to other communities grappling with the need to encourage mitigation behavior in response to increased flood severity and frequency, as well as to scholars interested in the factors that drive individuals’ decision-making in the face of a changing climate.

Literature Review

Flood Mitigation Actions

The question of how individuals respond to flooding is a well-researched topic, with flood mitigation literature expanding in recent years (Bockarjova et al., 2009¹¹; Bubeck et al., 2018¹²; De Koning et al., 2019¹³; Vishwanath Harish et al., 2023¹⁴; Weyrich et al., 2020¹⁵). At the household level flood mitigation refers to actions that individuals take to reduce the likelihood of future flood damage to their home or property; these mitigative actions range in complexity and cost. In the United States, the Federal Emergency Management Agency’s (FEMA) RiskMAP program (n.d.¹⁶) released plain-language resources that identify a menu of specific mitigation actions that homeowners can take both outside and inside their home. Other actions not mentioned in this document include installing floodgates or raising the home. Table 1 lists and defines these types of recommended mitigation actions. Damage mitigation is a form of adaptation to flooding, which also includes planned retreat from a flood zone (via private sale, public buyout, or relocating a home to a different part of a property) (De Koning et al., 2019).

Table 1. Types of Flood Mitigation Actions Recommended to Property Owners

Although many studies investigate these actions, there is still debate about the factors that influence their adoption. Generally, the severity of past flood damage has been shown to be a powerful driver of damage mitigation (Vishwanath Harish et al., 2023), while social norms and relationships are also an important component of this decision-making process (Bubeck et al., 2018¹⁷). High perceptions of vulnerability have also been shown to predict avoidant adaptation measures like planned retreat (Weyrich et al., 2020), which is generally considered a more accessible option for those with greater socioeconomic resources (De Koning et al., 2019). Vulnerability, social norms, and relationships are all critical to understand, given their importance to effective communication intended to facilitate mitigation (Weyrich et al., 2020). Furthermore, while existing literature on mitigation decision-making focuses on single flood events or areas that experience regular flooding, there remains a gap in research on the perceptions, experiences, and mitigation strategies of people who have experienced multiple extreme events over a short period of time.

Theoretical Framing: Protection Motivation Theory

A theoretical framework that is often used to assess and contextualize individual behavior around flooding is Protection Motivation Theory (PMT) (Figure 1) (Babcicky & Seebauer, 2019¹⁸; Bubeck et al., 2018; De Koning et al., 2019). Proposed by R.W. Rogers (1983¹⁹), PMT designates two cognitive processes that promote behavioral intention: threat appraisal (“am I under threat?”) followed by coping appraisal (“how will I cope with this threat?”). Threat appraisal is influenced by perceived vulnerability to and severity of the threat, as well as any rewards—such as a financial benefit or praise by peers—one may receive for facing the threat. Coping appraisal is influenced by one’s assessments of the cost of action, self-efficacy, and response efficacy, which, in this case, is the perceived effectiveness of a mitigation action at minimizing the threat.

Figure 1. A Visual Representation of Protection Motivation Theory

The PMT has been used to examine risk reduction behaviors related to a variety of natural hazards, including droughts (Gebrehiwot & Van Der Veen, 2015²⁰), landslides (Mertens et al., 2018²¹), and, especially often, in studies investigating flood mitigation behaviors. Generally, these studies have found that coping appraisals are a better predictor of intent to mitigate flood damage than threat appraisals (Babcicky & Seebauer, 2019; Bubeck et al., 2018), and have used the PMT to demonstrate which factors play roles in shaping these appraisals. For example, the PMT has been used to demonstrate that social norms and networks play an important role in shaping one’s coping appraisals in some contexts (Bubeck et al., 2018), while self-efficacy and socio-economic characteristics play an important role in others (Vishwanath Harish et al., 2023). Additionally, De Koning et al. (De Koning et al., 2019) illustrated that flood damage mitigation behaviors may be primarily driven by coping appraisals, while more radical actions (like planned retreat) are predicated on high threat appraisals and fear.

A second relevant theoretical framework to mitigation decision-making is trust ecology, which depicts trust as a multidimensional and comprising four distinct forms, all of which play critical roles in supporting the resilience of a given natural resource management institution (Stern & Baird, 2015²²; Stern & Coleman, 2019²³). The four forms of trust are dispositional, rational, affinitive, and procedural (Stern & Coleman, 2018b²⁴). We describe each form below. Natural resource management institutions are defined in the trust ecology framework as any organization or system that is tasked with managing natural resource challenges, such as providing sustainable yields, provisioning ecosystem services, or building resilience to natural hazards (Stern & Baird, 2015).

Within the context of trust ecology, trust is conceptualized as a three-part relationship, where a “trustor” (Entity A) believes in a “trustee” (Entity B) to do something specific (Action C) (Hardin, 2002²⁵) (see Figure 2 below). This requires Entity A to accept some level of risk, and that risk acceptance is driven by four different types of antecedents, which determine the specific form of trust (Stern & Coleman, 2018b). Dispositional trust describes trust resulting from a general predisposition of an individual to trust others (Mayer et al., 1995²⁶; Mollering, 2006²⁷; Spector & Jones, 2004²⁸). Rational trust is based primarily upon expectations of reciprocity or perceived utility in strategic interaction (J. Coleman, 1990²⁹; Hardin, 2002; Mollering, 2006). Affinitive trust focuses more on the qualities of the potential trustee (Stern & Coleman, 2018b), and may arise through feelings of social connectedness, positive shared experiences, perceptions of shared identities, or assumptions of the similarity of salient values (Braithwaite, 1998³⁰; Cvetkovich & Winter, 2003³¹; Stern, 2008³²). Procedural trust is based on the presence of effective control systems (the context and structure in which trust-related decisions are made) (Mayer et al., 1995; Schoorman et al., 2007³³). These systems might include official rules or procedures, contracts, or other monitoring mechanisms that make behaviors more predictable and thus reduce the risk of a trust violation. Distrust, on the other hand, involves an explicit negative expectation about the potential trustee (Lewicki & Stevenson, 1997³⁴). Trust ecology is highly relevant in the context of natural disaster planning and management, such as wildland fire mitigation (K. Coleman et al., 2017³⁵; K. Coleman & Stern, 2018b, 2018a³⁶). We suggest that it is likely relevant in the context of flooding, too, in that the people or institutions an individual trusts (e.g., town officials, FEMA, neighbors) are likely to influence the decisions they make around flood mitigation behaviors.

Figure 2. A Visual Representation of Trust Ecology Theory

Research Questions

To summarize, a robust body of research exists that seeks to understand individual decision-making in the aftermath of a single flood, including flood mitigation decisions; however, very limited research exists exploring these dynamics in settings where individuals have experienced multiple extreme hazard events. Investigating the conditions enabling or inhibiting mitigative action in these contexts is increasingly important as climate change-driven extreme weather continues to accelerate. Locally, emergency management practitioners seek strategies for increasing the adoption of mitigation actions in the communities they serve. Therefore, we seek to understand:

• What factors led flood-vulnerable residents and business owners to take flood mitigation actions on their property before Tropical Storm Irene, after Irene, and after the 2023 floods?
• What specific flood mitigation actions do residents and business owners perceive to be the most and least effective to mitigate flooding?
• What emergency management interventions are most likely to increase mitigation actions among flood-vulnerable individuals?

Research Design

Study Site and Access

Waterbury, VT—a small community of approximately 5,500 people along the Winooski River—suffered damage from four recent floods: one in 2011 during Tropical Storm Irene, two in July and December of 2023, and one in July of 2024. During the 2023 consecutive floods, damage was particularly concentrated along downtown’s Randall, Elm, and Union Streets adjacent to the Winooski River (Van Dine, 2023; Waterbury Roundabout, 2023). Smaller communities adjacent to Waterbury, such as Bolton, Duxbury, Middlesex, and Moretown, were also impacted. These streets and adjacent communities are visible in Figure 3.

Figure 3. Map of 100- and 500-Year Flood Zones in Waterbury, VT

After Tropical Storm Irene in 2011, large-scale floodplain restoration projects were proposed in three locations (Milone & MacBroom, Inc., 2013³⁷), and detailed cost-benefit analysis revealed significant benefit to the impacted municipalities (Schiff et al., 2015³⁸). One project was designed to protect a state office complex and was successfully completed by the state. The other recommended projects, a floodplain restoration project at a cornfield in Waterbury and a similar project at a racehorse training field in Duxbury, were designed to protect the village. However, each project ran into complications and neither was completed, despite urging from some impacted community members to do so (Elder-Conners, 2023³⁹; Kravitz, 2024⁴⁰). Additionally, Carmichael et al. (2020⁴¹) conducted a study in Waterbury and discussed some residents’ interest in dredging the river as a mitigation tactic. They termed this mitigation option a regional heritage narrative—a story that appeals to a subsection of a community and reveals information about the community’s social underpinnings and development. In this case, dredging is a longstanding historical practice and idea for how to manage flooding, despite concerns about the watershed-scale impact of dredging being net negative (Olson, 2000⁴²; Blair, 2024⁴³).

The flooding in 2011 also prompted the formation of Revitalize Waterbury, a community-based nonprofit that provided residents and business owners with flood recovery assistance. Between 2011 and 2023, Revitalize Waterbury continued to provide economic recovery assistance, while Community Resilience for the Waterbury Area (CReW) was formed with Revitalize Waterbury as its fiduciary agent to focus specifically on flood recovery and mitigation.

Following the 2011 flooding, one of the authors on this paper collaborated with Revitalize Waterbury on a different report to answer the research question: What were the major factors that impacted the overall response to Tropical Storm Irene in Waterbury? This mixed methods report utilized surveys and interviews of individuals involved in Waterbury’s Tropical Storm Irene response effort, finding that Waterbury’s leaders were highly collaborative and communicated effectively to constituents, despite limited preparation and pre-existing response procedures at the time. This existing relationship created a foundation on which to return to Waterbury in 2024 and assess the drivers of adopting flood mitigation practices among flood-impacted individuals within the community. In both 2011 and 2024, the goal of the research was to provide town leaders and planners with tools to understand drivers of past flood mitigation action and strategies to increase future adoption of these actions. As such, the research questions were driven largely by the needs of the community partner.

Interviews

Data Collection and Sampling

We conducted semi-structured interviews with 13 flood-impacted home and business owners in the Waterbury, VT area. We did eight one-on-one interviews with residents and two group interviews with two and three residents, respectively. Altogether the interviewees had direct experience with flooding on 10 properties.

We recruited interviewees from a pool of respondents to a survey we had distributed in the Spring of 2024 with our research partner, CReW. This survey was originally a major part of our mixed-methods research design and distributed as a quantitative data collection instrument. However, the July 2024 extreme flooding event happened while we were collecting survey responses in July 2024. The event lowered response rates, as community members were addressing more pressing matters, and also rendered many of the survey questions irrelevant. As such, we adapted our data collection methods, a process typical for rapid response research (Oulahen et al., 2020⁴⁴), and converted the survey primarily into a recruiting tool. We also recruited interviewees via snowball sampling with the initial pool of participants.

Semi-structured interviews took place during September and October 2024, ranged from 40 to 80 min in length, and occurred mostly in person, with three interviews conducted virtually. Interview questions focused on participants’ general perceptions of PMT parameters, especially in the context of specific flood mitigation actions one could take at their home or business. These mitigation actions were drawn from a factsheet produced by FEMA’s RiskMAP Program (n.d.-b), with some actions removed based on input from CReW volunteers regarding their relevancy for the community. We also left space for participants to discuss other mitigation actions that they thought were relevant. All interviews were recorded.

Data Analysis

Interview data was qualitatively analyzed using the PMT and trust ecology as our theoretical lens. Qualitative analysis was driven by coding from a priori codes that were derived from the parameters of the PMT, including threat (severity, vulnerability, rewards) and coping (self-efficacy, response efficacy, and costs), as well as codes that described trust and distrust, including dispositional, rational, affinitive, and procedural antecedents. As a slight variation from the PMT, we also nested temporally-oriented subcodes under these primary codes to better understand the ways in which threat and coping appraisal—and therefore motivation to mitigate—evolved in response to successive storm events.

We used the software NVivo (Version 14) to code the data. Interviews were coded by two of the coauthors, L. Briccetti and K. Coleman. To ensure intercoder reliability, coding comparison queries were run and any instance where there was less than 90% agreement between coders prompted a discussion about how the concept should be coded. The average intercoder agreement was very high (99.38%).

Ethical Considerations and Researcher Positionality

This study was approved by the University of Vermont Institutional Review Board on February 23, 2024 (Study #00002943), and an amended version of the study was also approved on August 30, 2024. Consent procedures were implemented prior to interviews and focused on empowering participants to step away if they were uncomfortable with sharing information. Our research process emphasized reflexivity in the way that we collected information and constructed meaning from it, particularly due to the emotionally challenging nature of disaster response research. Lastly, our research agenda was driven largely by communication with our partner, CReW, who we made a conscious decision to trust as a representative of the community and involved throughout the research process.

Findings

Of the 10 flooded properties that our 13 interviewees had direct experience with, seven were homes, two were businesses, and one was the home of an adult child. As shown in Table 2, three respondents had homes or businesses that were flooded during Tropical Storm Irene. The remaining interviewees moved to their current residence after Tropical Storm Irene and were thus not impacted. Eight participants had homes or businesses that were flooded during the July 2023 flood event, while four were flooded in the December 2023 floods, and three in the July 2024 floods. Many of the respondents who did not experience severe flooding during the three most recent floods had stories about how close they came to another flood, or about more minor impacts to other areas of their property (for example, a detached garage or barn).

Table 2. Property Damage Experienced by Interviewees During Four Recent Floods in Waterbury, VT

Nine interviewees expressed that they viewed climate change as a cause of increased flood risk. Others discussed increased acknowledgment in recent years that the climate is changing (and that extreme weather is increasing) among friends and neighbors that were previously skeptical of climate change. These same interviewees also discussed community members reacting to widespread acceptance of climate change by attempting to mitigate damage or relocate altogether. Although we did not focus on FEMA buyouts, they were brought up in the majority of interviews with homeowners. Interviewees mentioned signing up to be on the buyout list or knowing neighbors who were attempting to receive buyouts. However, Vermont’s housing shortage was also discussed repeatedly as a barrier to moving. Interviewee desire to move out of their at-risk home was often prompted by the two most recent floods, which were perceived by many to clearly demonstrate a trend of increasing extreme weather.

Factors Leading Residents and Business Owners to Take Flood Mitigation Action

In general, we found that individuals were likely to have implemented at least one mitigation action and were willing to consider the majority of actions that we asked them about. Across our interviews, several key themes emerged that provide insight into the factors that drive the adoption of mitigation behavior: perceptions of vulnerability, self-efficacy, and trust (or distrust).

Vulnerability

Perception of vulnerability was one of the PMT variables that changed the most throughout interviewees’ recollections of individual storm events. No respondents stated that they view a future flood as unlikely, and all agreed that flood likelihood was accelerating. Interviewees generally expressed a dramatic increase in perceived vulnerability as a result of Tropical Storm Irene, a less pronounced increase in this perception following July 2023, and dramatic increases again as a result of the December 2023 and July 2024 storms—even though a much smaller percentage of participants (and residents of the area overall) were impacted by these two recent floods. Many claimed that they originally viewed Irene as anomalous, but now perceive themselves as extremely vulnerable to future flood events. For instance, one interviewee stated:

If you asked me, when I bought the house in 2017, I said, “Oh, there will be no flood. If [a] flood comes, also there will be maybe limited damage.” But if someone asks me now, I'm sure every year there will be an event, we should be ready for that event actually.

Multiple residents mapped out the ways in which their perceptions of vulnerability changed, and the impact of that change on their decision-making. Generally, individuals expressed an increased likelihood to implement more drastic individual mitigation actions over time, like raising the home or installing floodgates. For example, moving valuables or utilities to higher floors was a mitigation action that many residents undertook after Irene, but elevating one’s home was rare at that time. It is regularly considered now. Many spoke of the accelerating emotional impact of these repeated events and near-misses, or tied an increased sense of vulnerability to climate change:

So, I think July [2023], I was like, okay, we knew this was going to happen. Bad luck. It happened really soon. December, we were like, “Oh, this is weird.” The July 2024 one, I was just pissed. I was angry that human beings suck so much and that we're just destroying our planet and no one cares. And I was just like, “This is so wrong,” and it's happening faster than anyone thinks it's happening, and unless it's happening to you, you really have no reason to be invested… Those [past floods] felt like I could write those off as flukes, but once the third one happened, it was like, I am now going to have to make some major life decisions about whether I want to stay here or not.

Self-Efficacy

Interviewees discussed how their belief in government (federal, state, or municipal) to engage in larger scale mitigation has waned as proposed mitigation projects, like the proposed floodplain restoration projects at a cornfield and racehorse training field, have gone unrealized. This perceived large-scale inaction was discussed multiple times as a barrier to individual self-efficacy and the response-efficacy of residential mitigation actions. For example, when asked about mitigation, one interviewee stated:

I think you think can do a little bit more to prepare, or to prevent it, or to mitigate it… but this has made me realize that there really would have to be some higher-level mitigation efforts in order to make it sustainable to live here at all… like the Cornfield Project or that floodplain in Duxbury, those sorts of things need to be done because history will keep repeating itself.

Low self-efficacy emerged as a key barrier to action, with interviewees discussing limits in their individual autonomy to enact change, particularly among renters and those living in higher density neighborhoods. Several individuals worried about how flood mitigation actions (like altering lot grading) might impact their neighbors. Others expressed the need for collective action to mitigate flood risks, such as community-level efforts to revegetate floodplains or to simultaneously alter lot grading. One interviewee calling mitigation in their neighborhood an “all or nothing deal.”

Higher cost and higher complexity—in terms of both physically implementing the action and the bureaucracy involved with funding and permissions—were key factors that reduced perceived self-efficacy. Regarding complexity, the lack of experienced contractors was repeatedly mentioned as a barrier. Two interview participants (a couple who shared a home) described their difficulty finding a contractor, and having to settle for delayed, unprofessional work:

Well, before we first got the flood [in July], we started looking immediately for places to stay and for people to work. We only found this one guy… they started a block of work in the cellar mid-December and then everything disappeared, nothing else went on… then 20 days go by, where are they? What's going on?... he came back in January and he finally started work. All he did was put a canvas across to get things to try to keep more rain out until they was able to go to work one or two months later.

In contrast, some interviewees expressed high self-efficacy when the mitigation options were less complex and less costly. For example, one interviewee took action by installing a sewer backflow valve at the behest of their plumber, which was a less expensive tactic than they expected:

So when I described to him [a plumber] that this is how water came in, [he said] "Well, you can stop the backflow,"… I was like "I don't really care what that costs, we're going to do that,"...It's [A backflow valve] right on the wall in our basement and it was about, maybe I think it cost us $700... It's fantastic!

Trust and Distrust

Finally, both trust and distrust had an impact on willingness to implement mitigation actions. Specifically, the concept of rational trust/distrust—forming one’s opinion on whether to trust something based on past actions or performance—was relevant in discussions about experiences with repeat flooding. Interviewees had positive and negative experiences from past floods which impacted trust in a variety of actors. People trusted those who were perceived to have helped during past crises (e.g., CReW, the town of Waterbury, some individual contractors) and linked those relationships to implementing mitigation actions. For example, one interviewee, quoted above, implemented a sewer backflow valve at the behest of a trusted plumber. Another interviewee shared an anecdote about how a positive relationship with CReW facilitated their son-in-law realizing—against his instincts—that he needed help in order to implement certain mitigation actions:

They were very helpful with telling [him] exactly what he needed to do. [Before talking to them] he [was] kind of like, “I know what I have to do and I'll just go do it,” which wasn't really in his best interest. So, he did get some advice from the CReW people.

Accordingly, respondents did not trust those who were perceived to have done an inadequate or poor job. Some examples of those who were perceived this way included FEMA, the U.S. Small Business Administration, local officials, and some individuals. This rational distrust discouraged interviewees from working with these individuals or groups on mitigation action, and sometimes even on flood response. For example, one interviewee shared how a police officer advised her to evacuate during the July 2023 flood, leading to her car getting caught in a flash flood on the road, and stated that “if they tell me to evacuate again, I won’t.” Lastly, distrust in the ability of the river’s flood gauge system to provide accurate information about water levels engendered animosity. Sentiments towards the flood gauges stood out, particularly in reference to the July 2024 flood, where flood peaks were much higher than had been forecasted. However, with the flood gauges, this distrust did not seem to work against mitigation action, and one interviewee even cited the gauges as a motivator to mitigate: “I trust that when it says it's going to be bad, to just assume it's going to be way worse than what they think it's going to be.”

Perceived Effectiveness of Mitigation Actions

Overall, interview respondents projected a strong desire to mitigate flood damage, and a willingness to consider most flood mitigation actions recommended for their properties. However, belief in the effectiveness of actions varied, with most respondents expressing some level of doubt at the efficacy of many actions given the scale of past flooding, and a perception that flooding will only get worse. Discussions about individual mitigation strategies often veered toward the need for large-scale municipal or regional mitigation strategies for the area. The large-scale actions mentioned ranged from completing the floodplain restoration projects proposed after Tropical Storm Irene to dredging the river near the village. Some respondents also mused about the utility of mitigating over leaving altogether, from both an emotional and financial standpoint. One of the interviewees discussed weighing the cost of mitigating versus the cost of finding new housing amidst skyrocketing Vermont housing prices:

Yeah, if you back up 10, 15 years ago… you look at what we have and you think, well, we've got enough money to survive. We're doing pretty good here. And all of a sudden the inflation hits and the weather changes hit and all of a sudden what we had isn't worth shit. Despite this overall sense of doubt regarding individual mitigation strategies, the perceived effectiveness of these actions still varied relative to each other. Participant perspectives on individual mitigation strategies are documented in Table 3 below.

Table 3. Willingness to Adopt Individual Mitigation Strategies Among Interviewees

The actions that were perceived as effective by respondents were a mix of both high and low barrier-of-entry mitigation strategies. The lowest cost strategy—moving valuables to upper floors—was acknowledged by all respondents as a good idea or something they had done, whether permanently or in the days preceding an expected flood. Raising the home (a high-cost strategy) and installing flood vents (a much lower-cost strategy) emerged as mitigation actions that were perceived as effective, and interviewees were aware of, despite not being one of the actions that we asked about. Raising the home was cited as being “astonishingly expensive” but was repeatedly mentioned as the best option in a high flood-risk area, with another interviewee going as far as saying: “a lot of them are trying to, they're raising the houses… that’s really the only thing you can do.” Installing floodgates also came up as effective in multiple conversations, with an interviewee expressing confidence that it would eliminate the need for increasingly costly flood insurance:

When we first got flood insurance, when we moved here in 2012… it was $500 a month, and now it's almost a thousand. So, we're going to cancel it and invest in floodgates for the doors 'cause we don't have a basement.

Additionally, some mitigation actions were mentioned as being effective only in specific situations—or as being generally effective, but that many individuals lacked the autonomy to implement the action. Sewer backflow valves were an example of an action that was generally considered to be unhelpful by respondents, as they perceived from their flood experiences that a sewer backflow was not important when river floodwaters were filling up their home anyway. However, one interviewee described a backflow valve as being particularly important in their situation, where river floodwaters did not reach their home in July 2023, but a sewer backflow caused flooding. Similarly, moving appliances to upper floors and lot grading were mostly considered to be effective, but only situationally applicable.

All interviewees that expressed high self-efficacy for the ability to move appliances to upper floors had already done so, with the exception of one person who was in the process of doing so. However, some respondents expressed low self-efficacy for this action, usually because of a sense that their home did not have the requisite space on upper floors for a boiler or water tank, or that rerouting electrical wiring to accommodate the raising of an electrical panel would be nearly impossible logistically:

Our furnace, hot water tank, washer, dryer, extra refrigerator, all that stuff is on the floor. So, one thing I didn't mention describing our house is 1200 square feet. It's a very small house, so I don't have a lot of extra room to move stuff up to the second floor. It would require a remodel that I just don't know we have room for.

Lot grading was also praised as effective, especially by landowners who were able to implement it, including some who credited it with mitigating damage from recent floods. However, the majority of interviewees did not perceive the autonomy to make such a change, because they were not the property’s landowner or because they lived in dense neighborhoods where grading could impact surrounding homes. An example of this line of thinking came from an interviewee who stated in regard to lot grading that “anything that we would do at this point will be to the detriment of our neighbors.” Similar thinking arose in reference to the reduction of impervious surfaces, with all respondents either perceiving that they already had a low surface area of impervious surfaces, or that reducing impervious surfaces was not something that they had the power to do.

Lastly, some actions were considered minimally effective in a severe flood, such as sump pumps/French drains, clearing gutters/downspouts, sealing basements, flood alert systems, and flood vents. The overriding sentiment for respondents who experienced river inundation flooding was that installing permanent pumping mechanisms or keeping downspouts clear lacked utility because “If you're flooded, you're going to pump it back into the flood? It doesn't make any sense.” In other words, the pumped-out water would end up right back in the waters flooding them or draining into the home of a neighbor. Additionally, sealing basements was derided by respondents as the least effective and most dangerous of the mitigation techniques that we discussed, with one interviewee stating:

I think probably the least impact would be sealing the basements, because usually the water comes above the sill plate… I think the problem is if you try to keep the water out of your basement and you got a lot of water around it, the force of the water pushing in on it can basically knock the walls over. It's really better to let the water in and equalize the pressure up through.

When discussing flood vents, most respondents said their home was not susceptible to this type of pressure buildup from floodwaters or it would be cheaper to break their basement windows in order to relieve this pressure. Lastly, flood alert systems were deemed unnecessary by most interviewees, with reasons ranging from preferring to check security cameras instead to preferring to check public flood gauges or rely on neighbors to alert them. One interviewee did have an alert system and found it to be effective. Other respondents were simply confused by the concept—accordingly, this mitigation strategy was the one that respondents seemed least aware of.

Discussion

This study provides a valuable example of decision-making for flood mitigation action in a context where residents have experienced multiple extreme flooding events in quick succession. As supported by previous flood mitigation research, both threat and coping appraisals played an important role in mitigation decision-making (Bubeck et al., 2018; Vishwanath Harish et al., 2023; Weyrich et al., 2020), but the relative importance of these appraisals appeared to shift with successive storm events. More specifically, perceived vulnerability has gained salience as flood frequency has increased, particularly following the two most recent floods. When asked to recall their perceived vulnerability to flooding over time, interview participants generally expressed that Tropical Storm Irene was shocking but could be dismissed as an anomaly. They also didn’t perceive the first 2023 flood as a shock due to previous experience with Irene. However, two subsequent floods within the span of a year rendered Waterbury’s vulnerabilities undeniable, resulting in statements like: “I’m sure every year there will be a flood event.” This perception of accelerating vulnerability may be driving participants to take mitigative action, as all interview participants had taken some action and expressed willingness to implement others, which is not consistent with mitigation literature in contexts where communities experienced just one event (Weyrich et al., 2020). Therefore, a high intention to engage in mitigation in this context may reveal how individuals impacted by disasters respond to a rapid increase in event frequency. This particular series of flood events appears to have exceeded the residents’ vulnerability threshold, prompting them to shift their attitudes and behavior.

Despite generally high intent to mitigate, a common theme in interviews was that both self-efficacy and response-efficacy were important for intent and for actual implementation, which is supported by prior research (Babcicky & Seebauer, 2019; Weyrich et al., 2020). However, participants in our interviews were likely to consider any action that they perceived as effective, unlike previous literature in contexts without multiple recent extreme events (Vishwanath Harish et al., 2023; Weyrich et al., 2020). In conjunction with the perceptions of accelerating vulnerability, this finding suggests that threat appraisal was generally high, leaving assessments of one’s self-efficacy and an action’s response efficacy as the primary antecedents to mitigation. Stated differently, motivation to mitigate was not an issue in this context. Rather, in Waterbury, the drivers of mitigation behavior were resource availability, autonomy, and belief that a specific action was effective.

Conclusions

Implications for Policy and Practice

Collective Mitigation Actions at the Municipal or County Level

One theme that emerged was that interviewees perceived mitigation actions at the individual level to be hampered by the lack of large-scale mitigative action at the collective level—like floodplain restoration—on the part of the state or municipality. Previous research conducted in this area of Vermont coined the term heritage narratives to describe longstanding traditional ideas about how to reduce flooding (Carmichael et al., 2020). A pertinent example of a heritage narrative from this study would be dredging the river in selective locations to protect the village, an action which came up occasionally in our interviews and is an important historical idea for how to prevent flooding in the community. Some interviewees believed that there is no point in taking individual action if the river was not dredged by local authorities. Dredging, however, would likely increase damage downstream (Olson, 2000) and flood practitioners have sought to discourage the practice in the area (Blair, 2024). At the same time, ignoring this heritage narrative (especially without other meaningful mitigation action) serves to undermine trust and reduce individual response efficacy. Accordingly, as response efficacy is an important antecedent to individual mitigation action, a strategy for increasing mitigation among flood-vulnerable populations in this community would be to make progress towards meaningful large-scale mitigation.

Improve Communication System to Inform Residents About Flood Mitigation

In terms of self-efficacy, perceived low complexity of an action often drove its uptake, while high complexity was often cited as a reason not to implement an action, a concept which is illustrated by some theories of behavior change like Diffusion of Innovations Theory (Rogers, 1983). Unlike cost—the other major barrier we saw to self-efficacy—the complexity of implementing a mitigation action can be reduced by the actions of state and local assistance organizations. For instance, multiple interviewees expressed confusion at the number of different people and organizations offering help in the aftermath of the flood, and that it was difficult to know who was the right person to talk to for a given problem. These interviewees also discussed how this confusion could have been ameliorated if they had a central person or organization they could trust as a first point of contact to help direct them through the logistical hurdles inherent to flood recovery and mitigation. Therefore, decreasing the complexity of communication channels, along with not overpromising organizational capacity to assist—while certainly not easy goals—are actionable and would have a notable impact on perceived self-efficacy.

Additionally, the specific flood actions that had the least barriers were often the ones that were only situationally effective, such as installing sewer backflow valves on the edge of the flood zone, improving lot grade in less dense neighborhoods, or reducing impervious surfaces in the few areas where they were high. Therefore, a worthwhile strategy for increasing the adoption of mitigation actions could be identifying the people through which the situationally effective strategies are relevant and tailoring communication to them.

Build Trust With Residents Through Consistent Communication and Clear Expectations

Finally, procedural and rational distrust emerged across all our interviews as important factors influencing individual decision-making around mitigation behaviors. Accordingly, local, state, and federal officials may be well served to take specific steps to engender and preserve rational and procedural trust to promote the adoption of mitigation behaviors. Specific recommendations for building procedural trust include using clear and consistent centralized communication and hosting meetings, field days, or other media opportunities designed to clarify decision-making processes around state and federal projects, such as buy-outs. To build rational trust, we recommend that FEMA staff, state emergency response crews, and town planners host presentations to demonstrate professional competence. These presentations could include updates on regular progress of large-scale mitigation projects, such as floodplain restoration projects.

We heard from interviewees that officials routinely over-promised and under-delivered; these experiences almost certainly eroded both rational and procedural trust. Thus, officials should take care not to over-commit, and to provide consistent and clear information to the individuals who are depending on them. These recommendations can help to bolster existing affinitive trusting relationships or build a foundation where affinitive trust is low.

Limitations

To more fully explore the themes that we discuss here, additional sampling is needed to bolster this limited dataset. These interviews were largely exploratory, with a wide breadth of questions. A more focused set of interviews, with recruitment derived from additional snowball sampling, could be used to strengthen support for our theoretical conclusions regarding the impact of successive flood events on flood mitigation decision-making. An alternative strategy would be to use these interviews as a starting point for an exploratory mixed methods design, which would involve the re-implementation of a more focused survey instrument. This design should involve expansion to a different and wider geographical area, both to facilitate a higher survey response rate and to ameliorate the geographical limitations of this study.

Future Research Directions

The findings of this research are well-suited to a follow-up study that utilizes behavioral nudges in outreach messaging. Future research could use threat-based and coping-based outreach communication by practitioners in an area that has experienced rapidly accelerating disaster frequency, which would test which strategy generates greater interest in mitigation action, similar to the experimental design of Byerly et al.’s (2019⁴⁵) on conservation messaging to maple syrup producers.

This study also explores which mitigation actions are broadly considered to be effective by residents, as well as which actions are perceived as ineffective or only situationally effective, and the relationship between effectiveness and cost, complexity, and autonomy. By partnering with a community-based organization, CReW, in Waterbury, these recommendations will be heard by practitioners with considerable local influence, and the ability to benefit the individuals living and working in the area.

Acknowledgments. Special thanks to Liz Schlegel and other volunteers at CReW for their assistance within the community of Waterbury, and to the Gund Institute for Environment at the University of Vermont for supporting this project.

References