What does environmental justice look like in Vermont? Centering perspectives from the margins

Until 2022, Vermont was one of the few US states that did not have an Environmental Justice (EJ) policy. In 2016, the Vermont Department of Environmental Conservation (DEC) initiated a process to create an EJ policy based on an agreement with the US Environmental Protection Agency (EPA). A coalition of academics, non-profit organization leaders, legal experts, and community-based partners formed in response to the DEC’s initial approach because it lacked a robust process to center the voices of the most vulnerable Vermonters. The coalition developed a mixed-method, community-based approach to ask, “What does EJ look like in Vermont?” This article reports the door-to-door survey portion of that broader research effort. The survey of 569 Vermont residents purposively sampled sites of likely environmental harm and health concerns and sites with existing relationships with activists and community organizations engaged in ongoing EJ struggles. The survey results use logistic regression to show that non-white respondents in the sites sampled were significantly more likely to be renters, to report exposures to mold, to have trouble paying for food and electricity, to lack access to public transportation, were less likely to own a vehicle, to have a primary care doctor, and reported higher rates of Lyme disease than white respondents. Our findings contribute to EJ theory regarding the co-productive relationship between environmental privilege and environmental harms within the context of persistent characterizations of Vermont as an environmental leader with abundant environmental benefits.


Introduction
Until June of 2022, Vermont was one of the few US states that did not have an Environmental Justice (EJ) policy (see S.B. 148, Act 154 2022).Various statutes and agency regulations had addressed some inequities in the past related to legacy mining and mobile home park habitability-yet an overarching, sustained, and place-based EJ policy was elusive.In 2016, as a part of the US Environmental Protection Agency's (EPA) Environmental Justice 2020 Action Agenda (USEPA 2016), the Vermont Department of Environmental Conservation (DEC) was asked to help develop a state EJ policy as part of their performance agreement plan. 1 In early 2018, the Vermont DEC organized a meeting to seek guidance from a small group of community experts, scholars, and legal experts to initiate the process (VDEC 2017).As the meeting ensued, attendees raised concerns that the DEC's approach would likely result in simply copying policy language from a neighboring state, without engaging and incorporating the voices of those most impacted by environmental decision-making in Vermont.
In response to these concerns, a coalition formed to support a meaningful, community-informed, statewide EJ policy framework, as well as to do work beyond policy, given the many failures of EJ policies to meet the demands of vulnerable people in the context of "contemporary racial capitalism" (Pulido 2016, 525).Known as the Rural Environmental Justice Opportunities Informed by Community Expertise (REJOICE) coalition, our guiding question from the beginning was, "What does EJ look like in Vermont?"The coalition members interpreted this question broadly, emphasizing the need to recognize that many communities are already deeply embedded in struggles for health, wellbeing and justice.Some of these struggles challenge the state and the work of its agencies.Mindful of these tensions, we pooled our personnel, resources, and time to meet Vermonters where they are, compensating community members for their time and expertise, and committing to an iterative process of returning to communities to continue in a shared learning process.Our coalition is made up of academics, non-profit organization leaders, legal experts, and community-based partners, and we have woven together cooperative agreements to align our working styles, timelines, communication patterns, metrics of success, and other elements of our systems that are often too siloed for us to effectively work in coordination and collaboration.
The community-based research approaches initiated by the coalition included conducting listening sessions, interviews, focus groups, and a door-to-door survey to better define what EJ looks like in the state.The analysis in this paper reflects one facet of the broader approach we have taken to understand EJ on the ground: the self-reported environmental and health experiences of 569 Vermont residents surveyed door-to-door in sites with suspected multiple environmental and health burdens.In particular, we examine combined race, gender, and class-based environmental and health disparities and we report many EJ issues of concern identified through the survey.The results show that nonwhite respondents in the sites sampled were significantly more likely to be renters, to report exposures to mold, to have trouble paying for food and electricity, to lack access to public transportation, were less likely to own a vehicle, to have a primary care doctor, and reported higher rates of Lyme disease than white respondents.We detail results from our other related Coalition work elsewhere (Keady et al. 2021) and we detail some of our contributions to the policy process and our next steps in the conclusion.

Producing environmental privilege and exclusion in Vermont
Vermont's rurality and whiteness are deliberately constructed and maintained.French and English colonial military forces and settlers dispossessed Abenaki peoples of their unceded land-known as Ndakinna in Abenaki language (our land)-with the Canadian Government recognizing some tribes and the US Federal Government not recognizing those in Vermont.The State of Vermont eventually recognized four tribes in 2011, but this process was and remains contested by members of the Odanak First Nation, currently living in southern Quebec and by others (Reed et al. 2022).Migration into the state has mostly come from Canada, Europe, and other parts of the USA, especially in the late 1800s and early 1900s (Bose 2020).The National Museum of African American History and Culture (2016) notes that Vermont's declaration of independence from slavery was quite vague, possibly making it difficult for free Black people to find work, own property, or to remain in the state.Whitfield (2014) provides further evidence that Black adults and children were actively bought and sold until the early 1800s statewide.The 1771 census only noted 19 African Americans in Vermont (McReynolds 2003), which may be an underestimate as other historians show that Vermont farm families often had one or two enslaved persons to supplement their labor needs (Hardesty 2019).Other accounts of Vermont's lack of diversity throughout its history point to the theory of push and pull or lack of industrial jobs to attract communities of color (McReynolds 2003;Evancie and Sananes 2017).However, throughout the 1800s to early 1900s, Vermont's fiber mills, marble quarries, agriculture, forestry, and tourism attracted Irish, French Canadians, Italians, Poles, and Swedes into the state (McReynolds 2003, Bose 2020).Despite a decline in population due to outmigration in the 1900s and the struggles Vermont has had in attracting and retaining new residents, from 1925 to 1936, the Eugenics Survey of Vermont actively introduced discriminatory policies in an attempt to cultivate a homogenous white, able-bodied citizenry (Gallagher 1999).To increase the population density, Vermont developed a tense long-standing strategy of trying to attract wealthy tourists and second homeowners to its green mountain landscapes for leisure activities (Harrison 2005;2006).Conforti (2001) argues that in the 1920s and 1930s, in response to the growing diversity of southern New England, the idea of and the search for the "real" New England took hold in the popular geographical imagination.Vanderbeck argues (2006, 645) that Vermont Life magazine was one of the key venues that marketed, a particular careful construction of Vermont whiteness, idealized by, "white snow, white church steeples at the center of the New England 'white village,' and white faces."Vermont and its settlers were actively represented and shaped according to New England ideals and modified imaginaries of the British countryside, and purposely erased working-class industry in its messaging (Hand 1992).
Efforts to maintain a particular Vermont whiteness and character persist.As recently as 2019, Governor Phil Scott offered $10,000 through the Remote Worker Grant Program to people willing to move to Vermont while working for an out of state employer (VACCD 2020).Press coverage of the program included a picturesque red barn and the image of a white sugar maple harvester readying horses for harvesting, emphasizing the state's "unspoiled" character and landscape (see Hess 2019).Joan Goldstein, Commissioner of the Vermont Department of Economic Development, said of the program (Ibid.), "It's an acknowledgement that we need more people in this state.We need to share the tax bur-den…We are a small state with a small population and we are one of the oldest states in the nation."Vermont legislators hope that the taxes generated through the program will help pay for the initiative among other projected benefits.The program capitalizes on an increasingly common trend towards telecommuting, which has initiated a land rush to Vermont by the economically privileged who can afford to do so, especially during the COVID-19 pandemic, which has increased remote working options for many (Barry 2020).While the dominant discourse in these incentive programs emphasizes economic development and shoring up a stronger tax base in Vermont, its racialized and classed imagery and assumptions convey an economic, social, political, and environmental privilege conferred to new residents who might find co-producing iconic Vermont landscapes appealing.
Vermont is the second most rural US state, with 61.1% of the population living in rural areas (US Census Bureau 2019).It has one of the lowest state populations (623,989 residents as of 2019), the 13th lowest poverty rate with 10.2% of Vermonters living in poverty, and is one of the most racially homogenous states (89.8% of which identifies as white) (US Census Bureau 2019; WPR 2021a).Resettled refugees from Bosnia, Vietnam, Bhutan, Republic of the Congo, and Somalia began arriving in Vermont in the late 1980s, with 7956 refugees resettled largely in Burlington, Winooski, and Rutland between 1989and 2019(Bose 2014;Suozzo 2019).English is not the primary language spoken in these households.Some in these resettled groups have also moved out of the state.Latinx migrant farmworkers have increased in Vermont, especially following the passage of the North American Free Trade Agreement (NAFTA), which impacted many Central American communities who moved to the USA looking for farm work (Mares 2019).
While a USNews piece recently ranked Vermont first for the lowest levels of pollution from industrial toxins and for the lowest levels of health risk and as one of the top five forested states (Hubbard 2021;WPR 2021b), Vermont has many legacy sources of pollution such as asbestos, stone/ slate mining and nuclear power generation, and contemporary sources of drinking water pollution such as industrial dairy farming, hazardous waste generators, superfund sites, and active and capped landfills (Gribkoff 2019;Senn 2019).For example, almost 15 years after Saint-Gobain Performance Plastics closed in 2002, it was found to have significantly contaminated the neighboring communities' water with per and polyfluoroalkyl substances (PFAS), resulting in detected PFAS levels twenty-nine times higher than the state's drinking water standards (Gribkoff 2019).A recent study also found PFAS-contaminated soils in southwestern Vermont and eastern New York State, indicating dispersion via airborne emission and land deposition originating from manufacturers using Teflon-coating in their products (Schroeder et al. 2020).
Beyond industrial sources of pollution, Vermont has some of the oldest housing stock in the country, with 80% of rental units over 40 years old, and 47% of rental units over 80 years old (Scott 2019).Approximately 70% of homes in Vermont were built prior to 1978-before lead (Pb) was banned from house paint-meaning that nearly three-quarters of Vermont homes likely have lead-based paint (VDOH 2020).Additionally, many Vermont homes are heated with wood (38%), producing particulate matter air contamination (VPSD 2016).Flooding caused by tropical storm Irene in 2011 resulted in damage to 3500 homes (predominantly mobile homes), and caused many farms and businesses to lose income, threatening the safety and health of many people (Baker et al. 2014;Galford et al. 2014).
There are few public transportation options in rural Vermont, making access to healthy food, healthcare, childcare options, and employment difficult without a personal vehicle (Johnke 2017).Approximately 30% of residents of low-income towns in Vermont have to travel more than 15 minutes to get to a grocery store, with travel time being far greater for rural Vermonters (Petenko 2019).Many residents experience social isolation, with 8% of Vermonters (and 13% of Vermont residents over the age of 65) reporting never receiving social and emotional support (Dougherty and Petenko 2019).Rural healthcare providers are struggling to remain in business, with closures reducing healthcare access for those in rural areas (Faher 2019).Vermont Department of Health studies show that non-white residents are more likely to experience physical health and mental health challenges compared to white residents (VDOH 2018a).Vermont also has the highest rate of reported cases of Lyme disease in the USA (VDOH 2019).
These environmental problems and the historic and contemporary struggles of many Vermonters who live on the margins are situated within repeated portrayals of Vermont as an idyllic site for a particular kind of rural life with outdoor recreation amenities couched in a culture of whiteness and privileged socio-economic status.Below, we connect these grounded material and symbolic features of Vermont with the key theories in interdisciplinary EJ literature, which our research engages.

Understanding environmental privilege and structural racism within an EJ approach
EJ scholars encourage an exploration of multi-scalar perspectives of difference including critical whiteness and environmental privilege and how they are entangled in the production of environmental injustice (Walker 2009;Sikor and Newell 2013;Pellow 2018).Park and Pellow (2011, 4) articulate environmental privilege as protection from ecological harms arising from, the exercise of economic, political, and cultural power that some groups enjoy, which enables them exclusive access to coveted environmental amenities such as forests, parks, mountains, rivers, coastal property, open lands, and elite neighborhoods.Environmental privilege is embodied in the fact that some groups can access spaces and resources, which are protected from the kinds of ecological harm that other groups are forced to contend with everyday [sic].These advantages include organic and pesticide free foods, neighborhoods with healthier air quality, and energy and other products siphoned from the living environments of other peoples.Park and Pellow (2011) show that environmental privilege produces and necessitates environmental inequities.They build on previous work showing that environmental privilege produces certain structural hierarchies, where racism is not simply the result of individual prejudices, but is embedded in long-standing land use, economic, and societal policies, practices, and norms (Bullard et al. 2007).Siting polluted facilities in predominantly Black, Indigenous and People of Color (BIPOC) neighborhoods, residential racial segregation, housing covenants, mass incarceration and police violence, and specific immigration policies all denote forms of structural racism (Bailey et al. 2021;Lichter et al. 2015;Rothstein 2017).Many health disparities also result from structural racism, which concentrates poverty, environmental pollution, infectious agents, and other adverse conditions among BIPOC and low-income communities (Gee and Ford 2011; Bailey et al. 2021).
Vermont did not formally engage in redlining, given its relatively small population and the small number of non-white households at the time when redlining became common practice in many cities in the USA (Watson 2021).Residential segregation was practiced, however, in the form of housing covenants forbidding a property being sold to non-white residents or members of specific religions or ethnic groups, until such covenants were outlawed in 1969 by the Fair Housing Act (Watson 2021).The extent of such covenants still in existence is not fully known and is currently under investigation (Ibid).Longstanding zoning codes and Act 250, Vermont's land use and development law outlining the review and management of development projects, employ unique criteria such as "aesthetics, scenic and natural beauty" to limit development (see VNRB 2021).These policies encourage differential access to environmental benefits and discourage multi-family housing and dense development (to retain rurality and aesthetics).
Privileging aesthetics in Vermont policymaking contextualizes the rurality of the state.Its millions of acres of protected land, few industries, and many environmental privileges continue to attract wealthy white settlers.Pellow (2016, 382) argues that the distinct structures of everyday rural lives are complexly racialized, gendered, classed, and prone to "oppositional thinking, relationships, and hierarchies that serve a particular set of interests."Rural spaces are also historic, cultural, social, and political in character and are imbued with complex and contradictory meanings, attitudes, and practices.Park and Pellow's (2011) Slums of Aspen, in some respects, mirrors how environmental privilege is actualized in Vermont, where sprawling ski resorts are serviced by marginalized communities.Rural Vermont is also a site of extraction and dumping of waste, of industrial dairy and other agricultural production.Its farmlands, woods, and mountaintops are sites of expanding energy infrastructure (natural gas pipelines and wind and solar installations), and rural sites serve as an informal refuge for individuals seeking affordable housing and labor (Sharma-Wallace 2016:175).Class is a key lens to consider within this framing of environmental privilege.Pruitt and Sobczynski (2016) show that opponents of a concentrated animal feeding operation (CAFO) hog farm in a largely white, rural, lowincome community in Arkansas were unable to frame their struggle as environmental injustice because of the insufficient attention paid to low-income white communities as targets of such injustice.Kelly-Reif and Wing (2016:350) characterize a "parasitic relationship between urban and rural communities because urban populations obtain most of their food and energy from rural areas and return their wastes to rural areas."How these practices and meanings are interpreted and made sense of also underpins the discourses and practices of inclusion, subordination, and exclusion from land, resources, policies, and governance processes ( Van Wagner 2016).
Accordingly, EJ examines not just risks of exposure to harm, but also the distribution of environmental benefits such as safe and abundant food, energy, housing, water, health, more-than-human natures, and the absence of public policies that harm people not only by race, but also by gender, sexuality, income, ability, citizenship, and language (Holifield 2001;Schlosberg 2007;Sze and London 2008;Méndez 2020).Pellow 2016 (384) argues for EJ activists and scholars to take issues of exclusion, hierarchy, and injustice seriously so as to challenge structural inequities and environmental privilege wherever they manifest, including among the poor.
Rather than center our analysis on environmental privilege and EJ in a specific city or community as in Park and Pellow's (2011) study, we argue that the "State of Vermont" is also an appropriate level of analysis for these concepts.Rural power hierarchies and relations, and ensuing inequities remain insufficiently researched and recognized by policy makers in the USA as well as in Vermont.We combine multi-scalar perspectives of difference including critical whiteness and environmental privilege and how they are entangled in the production of environmental injustice (Walker 2009;Sikor and Newell 2013;Pellow 2018).In the following sections, we detail our methods and results for the survey conducted within our broader research efforts and our place-based EJ analysis at the level of the state, where the EJ policy informed by our work was just recently passed in the Vermont State Legislature and signed by the Governor.2

Methods
Within the broader context of a community-centered process to elaborate what EJ looks like in Vermont, we pursued an expansive, mixed-method, and open-ended approach to answer this question while broadly centering the experiences of Vermonters living in the margins.This article reports the door-to-door survey portion of our broader research effort.Our coalition developed a door-to-door survey instrument with 58 voluntary questions arranged under the following topical headings: place connections (including local environmental risks and social concerns); water and climate change; housing (including indoor environmental risks); energy supply; food security; transportation; health; Vermont lakes and forests (outdoor recreation); safety (including sense of place); agriculture concerns; and demographic questions. 3  These categories were co-developed with coalition members and reflect a negotiation between categories of greatest concern to the different members of the coalition based on their work with communities across the state.Our coalition's structure consists of a Research Core, Community Core and Policy Core, which enables us to share different tasks according to our positionalities and strengths.Rather than treat these three categories as separate, we convene regularly with skilled facilitators to hold space, air out tensions, revise our practices, and hold each other accountable as we learn with REJOICE coalition partners and residents of the communities where we work.We included openended questions about race and gender.We categorized annual household income based on the US Department of Health and Human Services' 2019 poverty guidelines, and we asked if the respondents' household income was greater or lower than the poverty line ($25,750 per year for a fourperson household as above poverty) (Office of the Assistant Secretary for Planning and Evaluation 2019).The enumeration of "poverty" by government agencies may differ from a household's lived experience (not every person we surveyed was a member of a 4-person household).
Several steps informed our process for selecting where to administer the survey.We purposively sampled sites of likely environmental harms and health concerns.To do this, we conducted an initial spatial analysis with nine variables derived largely from publicly available datasets at the level of towns to identify areas with potentially elevated environmental and health vulnerabilities.We conducted this initial analysis using towns as the common spatial unit for all variables rather than census blocks, because community organizers work in "neighborhoods" that sit within towns but that often do not align with census block boundaries, and several of the variables were not based on census units but could be categorized or joined with their respective towns.We established a scoring mechanism similar to the one used in Washington Tracking Network's Washington Environmental Health Disparities Map (WTN 2019), with equal variable weighting to identify towns with higher concentrations of the identified 9 vulnerability variables (for details regarding our preliminary spatial analysis, please see the supplementary table (S1) published in Keady et al. 2021, an article pertaining to the energy-related aspects of the coalition's broader research). 4The lowest possible score could be 0 and the highest, 9. Our results ranged between 1 and 6.We focused on scores above 3 alongside input from community experts regarding much more specific community or neighborhood areas with known EJ struggles or that were identified as particularly underserved.We presented our initial spatial analysis to members of the coalition, which prompted critical discussions about neighborhoods and sites we would have missed without the initial screening analysis (due to not yet having strong community connections in a place), and how to move forward refining our site selection to the level of neighborhoods for administering the survey.Ultimately, our approach oversampled communities with likely multiple vulnerabilities and reflects a much broader framing of EJ than just single-source toxic exposure.
We do not disclose the chosen sites for administering the survey and we did not gather respondent addresses, due to confidentiality concerns in a state where many communities are close-knit and can easily identify people and places masked only by pseudonyms, and due to the sensitivity of the topics and the unintended impacts of identifying or labeling specific communities as vulnerable.If we merely map "data of damage" such as areas with potential EJ concerns, scholar Michelle Murphy (2017, 496) reminds us that, "Despite often antiracist intentions…such work tends to resuscitate racist, misogynist, and homophobic portraits of poor, Black, Indigenous, female, and queer lives and communities as damaged and doomed, as inhabiting irreparable states that are not just unwanted but less than fully human."Eve Tuck (2009, 422) calls researchers into "suspending damage" or refusing to do research that centers, labels or maps communities as primarily damaged.Thus, this survey should not be understood as systematically sampling all communities or spaces in Vermont with EJ concerns, and it was only one tool within a much broader suite of other community-centered approaches for elaborating EJ concerns and objectives for highlighting local expertise and demands on state and other entities for greater accountability.
Between May and August 2019, surveys were administered on weekdays between 4:00pm and 6:00pm, and on the weekends from 10:00am to 4:00pm Eastern Standard Time.Seven trained surveyors from our research team knocked on as many doors as possible within the scheduled time frames.A consent process preceded administering the survey to all participants (the institutional review board-approved process included a brief introduction of the study and its purpose, the process, risks, benefits, and confidentiality).The response rate of both answering the door and taking the survey varied by location.About 25% of people answered their doors, and approximately a third of those who answered their doors agreed to complete the survey.The survey took approximately 10 to 30 min as it was administered in a conversational manner.No identifying or address information was collected.Additionally, we developed an online version of the survey using the Lime Survey web application.Surveyors left flyers with a link to the online form at all unanswered doors.The survey link was also posted on selected local neighborhood forums.A total of 569 surveys were collected, 107 of which were gathered online, across all Vermont counties (n = 11) except Orange, Windham, and Windsor counties, which we had intended to survey in spring of 2020 and had to cancel due to restrictions on conducting in-person research early in the COVID-19 pandemic.
We first calculated descriptive statistics including crosstabulations between demographic variables (gender, race, income, and home ownership) and environmental and health variables (food, energy, transportation, recreational access, and health).Further analysis was completed using binomial logistic regression using SPSS software (IBM 2019) to determine the odds ratios (ORs) at 95% confidence intervals (CIs).Binomial logistic regression is typically used when the variables are dichotomous.Gender, race, income, and home ownership were set as covariates in SPSS and served as control variables of each other.The dependent variables included environmental and health variables.All regression results, including the statistically non-significant results, are reported in the results section following Amherin et al. (2019).Due to small sample size and missing values in the survey data, some of the predictor variables may lose significance as we add multiple controls.

Results
We characterize the differential experiences of survey respondents by race, gender, income, and residence type.The demographics of the survey respondents broadly reflect the demographics of the residents of Vermont.However, our survey used a convenience sampling method, prioritizing sites with high preliminary environmental and health risk scores in our initial spatial analysis.Therefore, our results are not generalizable to all sites in Vermont, as we are centering voices likely in the margins.More women participated in the survey than men.As an EJ study, this survey sought to represent more BIPOC individuals and residents with an income below the federal poverty line, as seen in Table 1, in order to better characterize their experiences.Just over 14% of survey respondents were non-white compared to the state average of 6.6% non-white residents, and over 31% of the respondents had a household income below $25,750 (referred to as below the poverty line from here forward), compared to the statewide level of around 11%.More than half of the survey respondents were renters (54.1%), while 70.7% of housing units in the state between 2014 and 2018 were owner occupied (US Census Bureau 2019).We did not seek to represent more renters, yet these statistics reflect the characteristics of the surveyed population by race and income.
Women (42%) were slightly more likely to live below the federal poverty line than men (38%), but this observation was not significant (see Fig. 1).According to the 2020 US Census, the poverty rate for male householders was 11.4% versus 23.4% among female householders.However, women were slightly more likely (45%) to own their home compared to men (40%) in our survey.About 49% of non-white respondents identified as women.BIPOC respondents on average were poorer and less likely to own a house than the white respondents.Approximately 46% of the non-white respondents were living under the poverty line compared to 40% of the white respondents.Only 17% of the nonwhite respondents owned their home compared to 45% of the white respondents in the survey.While there is some overlap between race and income in this study, there was no significant relationship between income and race.However, the regression results significantly confirmed that white respondents were three times more likely to be homeowners compared to those who identified as non-white (OR: 3.592, CI: 1.745-7.395).Those living below the poverty line were eight times more likely to be renters than those living above it (OR: 8.919 CI: 5.225-15.224).

General environmental concerns
Our questions framed environmental concerns broadly to include plastics, air pollution, water pollution, and climate change.Both non-white respondents and respondents living below the poverty line were twice as likely to be less concerned about broader environmental issues like plastics, air pollution, and climate change compared to white respondents and respondents living above the poverty line (Tables 2  and 3).This could be because there are more specific issues in the community or home that they are concerned about than more general topics as framed.We also know that social issues and environmental issues cannot be easily separated,

Exposures at home
We examined exposures to mold, lead, arsenic, asbestos, radon, pests, and pesticide uses at home by gender, race, income, and home ownership.Very few respondents indicated that they had been exposed to arsenic, asbestos, radon, or pests at home, so these variables were excluded from the regression analysis.Respondents who identified as non-white (OR=2.297CI 1.161-4.547)and renters (OR: 2.567 CI: 1.166-5.651)were twice as likely to indicate problems with mold at home compared to respondents who identified as white and as homeowners.Mold is one of the most common complaints made by homeowners and renters in Vermont, and it can result in upper and lower respiratory problems (Weinhold 2007;Dawson 2018).Survey respondents who live below the poverty line were more than twice as likely to be concerned about pesticide use (12%; OR=2.649,CI 0.808-8.684)than those who live above the poverty line (3%), though this observation was not significant (see Tables 2 and 3).

Transportation access
Public transportation access and use, and ownership of vehicles varied significantly by gender, race, income, and home ownership.Women respondents (OR: 1.964 CI: 1.113-3.403)were 1.9 times more likely and non-white respondents (OR: 2.169 CI: 1.091-4.313)were twice as likely to report a lack of access to public transportation compared to men and white respondents.Non-white respondents (OR: 3.837 CI: 1.760-8.365),renters (OR: 3.460 CI: 1.105-10.833),and those below the poverty line (OR: 2.993 CI: 1.401-6.396)were three times more likely to report using public transportation daily or weekly than white respondents, those above the poverty line, and homeowners.Correspondingly, non-white respondents were over 2.5 times significantly less likely to own a personal vehicle than white respondents (OR: 0.378 CI: 0.1176-0.809)and those below the federal poverty line were over five times less likely to own a personal vehicle compared to respondents living above the poverty line (OR: 0.169 CI: 0.081-0.353).

Food access and affordability
Overall, gender, race, and income level were correlated with access to fresh fruits and vegetables.Women (OR: 0.336 CI:0.133-0.848)and those below the federal poverty line (OR: 0.377 CI:0.147-0.969)were significantly over 2.5 times less likely to report access to fresh food than men and those above the poverty line.Women (OR: 2.827 CI: 1.300-6.147)and non-white respondents (OR: 2.445 CI: 1.057-5.662)were significantly over twice as likely to report being unable to afford fresh produce than men and non-white respondents.Those below the federal poverty line and renters were also twice as likely to be unable to afford fresh produce, although this observation was not significant (Tables 2  and 3).Significantly, respondents below the poverty line were six times less likely to report eating fresh fruits and vegetables weekly than those above the poverty line (OR: 0.189 CI: 0.181-0.653).Non-white respondents, interestingly, were four times more likely than white respondents to report consuming fresh vegetables and fruits on a weekly basis, though this observation was not significant (OR:4.667

Utilities affordability
Renters were, significantly, more than two times more likely (OR: 2.741 CI: 1.065-7.049)to report having trouble paying for heat compared to homeowners.Non-white respondents (OR: 3.153 CI: 1.423-6.986)and respondents living below the poverty line (OR: 2.966 CI: 1.352-6.511)were three times more likely to have trouble paying for electricity compared to the white respondents and those above the poverty line.Renters were four times more likely to report having trouble paying their rent/mortgages than owners (OR: 4.480 CI: 1.714-11.711).Respondents with a household income below the poverty line did not report similar concerns for paying rent.Eleven percent of white respondents had trouble paying for rent or their mortgage compared to 20% of nonwhite respondents.Sixteen percent of respondents below the poverty line reported having trouble paying their rent/ mortgage compared to 12% of those above the poverty line.

Health access and outcomes
Non-white respondents were significantly less likely to report having a primary care doctor than white respondents (OR: 0.343 CI: 0.144-0.815).We also explored asthma, allergies, autoimmune disorders, cardiovascular disease, cancer, Lyme disease, and participation in outdoor physical activities in relation to gender, race, income, and home ownership (Tables 2 and 3).Respondents living below the poverty line (OR=1.839,CI 0.913-3.703)were 1.8 times more likely to report asthma than those above the poverty line.Women were, significantly, twice as likely as men to have allergies (OR=2.453,CI: 1.473-4.084),and renters were significantly 1.8 times as likely to report allergies than homeowners (OR=1.864,CI: 1.016-3.419).Non-white respondents were almost twice as likely to report autoimmune disorders than white respondents, but this observation was not significant (OR=1.923,CI: 0.739-5.006).Cardiovascular disease in the household was significantly twice as likely to be reported by those below the poverty line compared to those above the poverty line (OR=2.272,CI 1.118-4.618).Respondents below the poverty line were also twice as likely to report cancer and to engage in less physical activity outdoors than those above the poverty line, but this observation was not significant.Additionally, nonwhite respondents were three times more likely than white respondents to have reported Lyme disease in the household (OR=3.351,CI 1. 267-8.867).

Discussion
Overall, our results suggest that environmental and health issues are experienced differently across gender, race, income, and renter/owner status categories.Communities of color, those below the federal poverty line for a four-person household, and renters in our survey were disproportionally impacted by environmental health concerns.The survey results represent typically hard-to-reach and underrepresented populations in Vermont who are typically not wellrepresented in most studies and databases (VDOH 2010; O'Hare 2019).Our analysis explores patterns of social and environmental inequality and shows that people's lives are interconnected and mutually dependent even if such connections remain largely invisible in broader datasets and policy discussions (Hankivsky 2012).We found overlapping vulnerabilities by race, income, and residence.Approximately 46% of non-white respondents were living below the poverty line, and 80% of nonwhite respondents were likely to be renters (the relationship between race and income and home ownership was significant).While we controlled these variables in our regression analysis, there might be multiple overlapping issues that influence the results in this study.For instance, in the case of food insecurity, while non-white respondents reported having more fresh fruits and vegetables as part of their diet compared to white respondents and respondents below the poverty line, those who were non-white and below the poverty line reported the highest odds of going hungry.This suggests that non-white respondents strive to eat fresh, healthy food when they can, even if they have less secure food access and are more likely to report going hungry from time to time.In comparison, respondents below the poverty line were also less likely to include fresh produce as part of their weekly diet.They were also more likely to report less outdoor activity and have higher reports of cardiovascular problems and asthma.
These statistics broadly align with national trends: 21.2% of Black, non-Hispanic households and 16.2% of Hispanic households are food insecure, compared to the national average of 11.1% (USDA 2019).National trends also show 29.1% of households below the federal poverty line facing food insecurity compared to the national average of 11.1%, as observed in this study (USDA 2019).Petenko (2019) reports that food deserts are a problem in Vermont, and that income (not race) is likely a major factor in the ability of a population to access fresh and healthy food.However, our results demonstrate that accessibility, affordability, and hunger are significant concerns for non-white respondents as much as low-income populations in Vermont.These results are reflected in the 2018 Vermont Department of Health Behavioral Risk Factor Surveillance System (BRFSS) data, where BIPOC individuals and those below the poverty line were 9-12% more likely to be food insecure (VDOH 2018b).This database also showed slightly higher rates of cardiovascular disease among those living below the poverty line (14%) than BIPOC individuals (8%), consistent with our analysis (Ibid.).
An intersectional perspective is also required to understand access to transportation.Discriminatory city and town planning and insufficient funding allocated to rural areas make it increasingly difficult to access transportation (Ward 2009;McKenzie 2013;Rowangould et al. 2016).Our analysis demonstrates that BIPOC communities and households below the poverty line are significantly more likely to use public transportation, bicycle, and walk than communities with higher incomes and white populations.Transportation issues were not part of the VDOH BRFSS database, but our results also indicate that access to environmental and health services are closely linked to transportation.For example, food security and health access issues-distance to food and healthcare access points, and limited availability of affordable and culturally familiar foods-might also be related to lack of access to personal transportation.These factors may conspire to prevent physical, social, and economic access to sufficient, safe, and nutritious food that meets the food preferences and dietary needs of particular populations.The data suggests that a portion of the study population not only relies on public transportation but also that many do not own personal vehicles.Further analysis shows that those who own a personal vehicle were, significantly, twice as likely to report better access to fresh produce (OR: 2.398 CI: 1.128-5.101)and to have a primary care doctor (OR:2.654CI:1.350-5.215).State databases should incorporate the impact of transportation on health and food insecurity in the state, as the rurality of the state compounds lack of access to these essential services.

3
Housing is also a key factor related to environmental justice and health.The rental housing stock in Vermont is old, often poorly maintained, and has severe issues with heating, plumbing, electricity, and safely functioning household appliances.The renters in this study reported higher exposures to mold, difficulty paying for heat and rent, and were significantly more likely to report allergies.The people who are already economically marginalized are also made further vulnerable when health emergencies such as COVID-19 strike, as they face the greatest risk for unemployment and evictions.A recent survey found that up to a third of all renters in the USA were not able to pay rent in April 2020 and were experiencing continued housing instability due to the COVID-19 pandemic (Greene and McCargo 2020).Not having stable housing is also tied to other problems, such as food insecurity and increased stress, which can trigger long-term physical and mental health issues.
Accessibility of healthcare was low for both non-white respondents and respondents below the poverty line relative to others in our study.While non-white respondents reported better access to healthcare compared to those below the poverty line, non-white respondents were 2.5 times more likely to not have a primary care doctor than white respondents.If one does not have a primary care doctor, it is likely that they do not have health insurance, a variable that our study did not explore.The BRFSS (VDOH 2018b) database shows that, as a group, non-white Vermont residents were less likely to have medical health plan coverage, a medical healthcare access provider, and medical healthcare access due to cost than all state residents living below the poverty line.These results suggest that income and race both play a big role in access to healthcare.These results are consistent with other studies that have shown health disparities by race and income (Panikkar et al. 2012;Panikkar et al. 2014;Lopez and Gadsden 2016;Weinstein et al. 2017;Mariotto et al. 2018).Non-white respondents in the study were twice as likely to report autoimmune disorders, but this observation was not significant.The lack of a primary care provider among the non-white population may also be connected to higher reports of Lyme disease among non-white communities.BRFSS databases (VDOH 2018b) did not include information on these variables.
In 2017, Vermont had the highest rate of Lyme disease nationally, with over 500 cases being reported each year since 2011 (VDOH 2019).Lyme disease is currently most prominent in the Northeast of the USA; however, with climate change impacts, the areas with Lyme disease carrying ticks are shifting (APHA 2020).Our study showed that non-white respondents (71%) were as likely as white respondents (75%) to spend time outdoors in Vermont, but that non-white respondents were three times more likely to report Lyme disease (this contrasts with other regions of the USA based on a much older study) (Fix et al. 2000).With timely treatment, it is possible to manage Lyme disease.In combination with the survey finding of lower rates of nonwhite access to health insurance and care, it could be that some segments of the population are not accessing timely treatment and intervention, generating increased concerns for long-haul Lyme disease health impacts in this group.Another concern is that non-white individuals are often taken less seriously by doctors and their complaints are not always considered, resulting in some non-white individuals not going to visit the doctor, in part out of fear of not being heard (Anwar 2019).Issues with preferred language access might be another reason for reduced health access among non-white individuals, as well as the limited reach of the state's public health outreach regarding Lyme disease, especially in multiple languages.
While the data indicates that environmental and health disparities are present in Vermont, it is likely that the level of data analysis completed does not fully explain all of the nuances within the data.In this study, much of the analysis is based on binary categorization; further categorization was minimized to ensure robust sample size.Additionally, the number of non-white respondents in the study was low (n=82), which is a key limitation in the study.In addition, there are likely key differences among the represented BIPOC population, as other studies have noted (Panikkar et al. 2014).This limitation points to the difficulties in conducting an extensive study of hard-to-reach populations, and shows that more concerted effort and funding are needed to understand and serve the needs of underserved populations.

Conclusions
Our objective was to describe the broader coalition research effort leading to the passage of Vermont's first EJ policy and to highlight the quantitative research results from surveys administered in that broader process.The survey focused on gaining greater understanding of access to environmental benefits and existing health disparities in a state that is largely rural with many environmental privileges.The survey results alone are not surprising given that environmental benefits are not equitably distributed, with many environmental benefits specifically lower among communities of color and low-income populations.Our results, in conjunction with historical analysis of typical EJ concerns in the state suggest that as the state privileges rurality, and certain kinds of environmental amenities to attract wealthy white settlers, this coproduces vulnerabilities in communities of color and low-income communities.
Vermont has one of the smallest BIPOC populations in the USA and greater environmental privileges than many other states.This may lead to false assumptions that EJ and health disparities do not exist, thereby obfuscating the experiences of BIPOC individuals in Vermont.This is compounded by the fact that the largest health database in Vermont does not require the inclusion of race and ethnicity data (see GMCB 2021).Our findings confirm EJ and health disparities by income, race, and home ownership in Vermont when we center voices from sites of likely environmental harm and health concerns.These differential experiences cannot be explained merely by residential segregation.The identification of disparities in this study by itself is not surprising.While we cannot identify the cause of these disparities based on the data presented here, the historical policies adapted by the state may contribute towards ongoing structural racism in Vermont beyond a focus on income alone.
Beyond material disparities, structural racism can also be entrenched through white-coded ideological framings within institutions.Lack of conversation on EJ concerns as they relate to many state policies for so long points to how Vermont centers whiteness implicitly within its environmental decision-making.We argue that issues of exclusion, hierarchy, power, and injustice should be integrally incorporated in environmental analysis and decision-making in Vermont to decenter whiteness within decision-making.Prioritizing economic investments in public services, people, and communities that need them the most are critical steps to redressing these disparities.Consistent with the arc of the EJ movement, our findings suggest that EJ issues and environmental privilege persist, co-exist, and are at times co-productive in rural contexts.We call for effective policy approaches, proper investments, and community-centered programs to improve egregious environmental and health disparities.
The coalition's legal expert members drafted "S.B. 148 An act relating to environmental justice in Vermont" in spring of 2021 in collaboration with members of Conservation Law Foundation, vetting versions of the bill when feasible with community partners.Members of the research core of the REJOICE coalition presented our research findings more than 13 times to community, academic, and other groups, and they were invited to legislative sessions to testify on behalf of the policy, some of which was covered by local press.The final version of the policy was signed by Governor Scott on May 31st, 2022.Members of the coalition are aware of the limits of engaging in such policy formation processes, echoing the critical perspectives and debates within broader EJ literature (see Pulido et al. 2016 andPellow 2018 on the limits of engaging in formal EJ policy formation efforts).However, they still see such efforts as necessary and part of a broader and multi-faceted strategy (see Harrison 2022 regarding the "diversity of tactics EJ activists use" (15)).The coalition is currently transitioning to co-create a broader EJ network in the state.Some coalition members were recently awarded EPA funding to support more robust participatory engagement with communities and individuals previously excluded from state environmental interventions.Our community-centered findings, including qualitative analyses reported elsewhere, combined with these policy-and community-focused efforts represent a necessary and multi-layered strategic approach to understanding EJ in Vermont.
Funding This research was supported by a Lintilhac Foundation Grant, the James M. Jeffords Fund Grant Program for Policy Studies, and an Ian A. Worley Award through the Environmental Program at the University of Vermont.

Fig. 1
Fig. 1 Cross-tabulations of gender and race by income and home ownership

Table 1
Descriptive statistics of survey respondents *Missing variables represent respondents who did not respond to the question.Household income and home ownership numbers are from 2019 (US Census Bureau 2021) making their separation in the survey limiting.Women were significantly more concerned about water pollution than men (OR: 1.852 CI: 1.159-2.959).

Table 2
Cross-tabs of environmental and health concerns by gender, race, income, and residence

Table 3
Odds OR(CI) of environmental and health outcomes by gender, race, income, and residence (Bold numeric entries denote statistically significant associations.)