Measuring environmental (in)justices: Insights from a systematic literature review on methodological approaches

Meta-analysis/systematic review iScience Measuring environmental (in)justices: Insights from a systematic literature review on methodological approaches Graphical abstract Highlights • Environmental (in)justice research is expanding, yet methodologically fragmented • Review of 421 studies shows dominance of quantitative, GIS, and secondary-data methods • Participatory and qualitative, procedural, and recognition justice are underused • Research skews toward North America; future work should center on marginalized voices Authors Jacqueline Loos, Charlotte Gohr, Noelia Zafra-Calvo, Gonzalo Corte ´ s-Capano, Anna Lena Tonninger, Henrik von Wehrden Correspondence jacqueline.loo[email protected]t In brief Earth sciences; Environmental science; Social sciences Loos et al., 2025, iScience 28, 113889 December 19, 2025 © 2025 The Author(s). Published by Elsevier Inc. https://doi.org/10.1016/j.isci.2025.113889 ll iScience Meta-analysis/systematic review Measuring environmental (in)justices: Insights from a systematic literature review on methodological approaches Jacqueline Loos, 1,6, * Charlotte Gohr, 2 Noelia Zafra-Calvo, 3 Gonzalo Corte ´ s-Capano, 4,5 Anna Lena Tonninger, 1 and Henrik von Wehrden 2 1 University of Vienna, Faculty of Life Sciences, Department of Botany and Biodiversity Research, Rennweg 14, 1030 Vienna, Austria 2 Leuphana University, Center of Methods and Faculty of Sustainability, 21335 Lu¨ neburg, Germany 3 Basque Centre for Climate Change (BC3), 48940 Leioa, Spain 4 Department of Agricultural Economics and Rural Development, University of Go ¨ ttingen, Go ¨ ttingen, Germany 5 Faculty of Organic Agricultural Sciences, University of Kassel, Witzenhausen, Germany 6 Lead contact *Correspondence: [email protected] https://doi.org/10.1016/j.isci.2025.113889 SUMMARY Environmental (in)justice research uses various conceptual frameworks and methodological approaches, leading to fragmentation across contexts and disciplines. Our systematic review provides a methodological overview of how environmental (in)justice has been studied in 421 English-language scientific articles. Most studies approach environmental (in)justice from a quantitative and interdisciplinary perspective, primarily using purposive sampling, secondary data, and GIS/remote sensing tools with an emphasis on distributive justice. Although there is a notable diversification over time in data collection and analysis, there is a strong geographic bias with short-term, locally focused, and limited actor involvement, though actor diversity is growing over time. We identified eight thematic clusters with distinct methodological patterns: health, pollution, governance, climate change, collaboration, access, and green space. The lack of broadly adopted methodological approaches for evaluating environmental (in)justices largely stems from the context-specific, multi-scalar nature of cases and the philosophical and normative diversity embedded in the EJ concept itself. INTRODUCTION Environmental justice (EJ) refers to removing the barriers that cause disparities and inequalities in a given reality. 1,2 Although it started as a civil rights movement in response to disproportionate pollution and health issues beared by marginalized people, such as black communities, 3,4 over the past decades, EJ has significantly expanded its focus to a global, multifaceted matter that addresses environmental governance, pollution, land use, health, biodiversity, sustainability transitions, and climate change. 5–8 EJ has been increasingly backed up by scientific data and evidence since its origin, often through community-led science, 9 thereby unraveling systemic disadvantages of marginalized people. These disadvantages are expressed either in disproportionate exposure to environmental hazards or health risks, or in reduced access and availability of benefits that are related to environmental assets. Currently, EJ represents both a broad range of grassroot social movements and an interdisciplinary field of academic research, engaging with diverse academic traditions and disciplines (e.g., geography, sociology, environmental sciences) and activist communities around the world. 10–12 The conceptual expansion of EJ is well documented. 13,14 Existing reviews synthesize important advances and debates in relation to its conceptualization to frame and study inequality issues in climate vulnerability, green spaces, pollution, health, and governance, as well as to investigate how to support all members of society in proportion to what they initially have or what they need to tackle inequality and reach justice; a concept known as equity. 15,16 EJ is a multidimensional concept, encompassing various notions and frameworks. It is commonly understood to include distributive, procedural, recognitional, and contextual elements, 17–19 but has also been expanded to incorporate additional dimensions such as restorative or multispecies justice. 5,20–22 In addition, EJ has become an important principle in many public programs and policies in different geographies and at different scales, from local to national and global. Thus, there is a growing need for reviews and toolkits that capture and address how EJ can be studied empirically in diverse geographies and contexts. Existing reviews focus on specific analytical tools (e.g., spatial analysis, participatory mapping) or domains (e.g., health, pollution), 23–25 but there is often a lack of clear understanding and guidance on how to empirically measure EJ. Tarrant and Cordell (1999) 26 already stated early that data collection and analysis techniques in environmental justice research may lead to conflicting results. However, to our knowledge, a comprehensive methodological overview of the field is yet to come. iScience 28, 113889, December 19, 2025 © 2025 The Author(s). Published by Elsevier Inc. 1 This is an open access article under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/). ll OPEN ACCESS Empirical research on EJ is fragmented across disciplines such as geography, sociology, political ecology, and planning, with different methodological traditions, employing quantitative, qualitative, and mixed methods. 25,27 Each brings distinct theories and epistemological assumptions, shaping how injustices are identified, measured, and addressed. This fragmentation produces challenges such as how to identify suitable methods and measurements aligned with the justice dimensions studied, how methods can be both context-sensitive and comparable across scales and to what extent they make visible hidden or structural forms of harm, and how they reflect underlying values, norms, and power dynamics. 28,29 The way EJ issues are constructed is deeply connected to the methods used to reveal multiple dimensions of harm across time and space. 17,30 It also depends on the capacity to critically reflect on the assumptions, limitations, and potential of these methods to make conscious decisions when using them and working with data. Likewise with the multitude of interpretations and conceptualizations of what can be considered ‘‘just,’’ there is no methodological unity in how to assess, measure, and communicate environmental (in)justice issues. 31 To our knowledge, there is also no systematic and comparative synthesis of empirical methods used to study and assess EJ across multiple dimenFigure 1. PRISMA 2020 flow diagram of the literature screening and selection process Adapted from Page MJ et al. (2021), BMJ, 372:n71, under the CC BY 4.0 license (https://creativecommons. org/licenses/by/4.0/). sions and disciplines, nor how these methods have evolved alongside conceptual developments. Such an overview could provide a clearer mapping of methodological diversity, revealing dominant approaches as well as underutilized or emerging ones. This synthesis would help identify implicit biases, neglected aspects, for example, tendencies toward certain geographic regions or methodological approaches, thereby highlighting opportunities to broaden methodological pluralism. Moreover, by tracking how methods have developed, this work could guide the integration of quantitative, qualitative, and mixed approaches, fostering more holistic and context-sensitive research. In doing so, it would support interdisciplinary collaboration, enrich future research agendas by identifying knowledge gaps and innovation spaces, and enhance the policy relevance and communication of environmental (in)justice scholarship. Ultimately, this contributes to a more inclusive and nuanced understanding of environmental (in)justice, better equipped to address plurality in environmental (in)justice research. 32 To this end, we conducted a systematic literature review of English academic publications (Figure 1) to characterize the main empirical approaches used to study environmental (in)justice across disciplines, geographies, justice dimensions, and scales. In particular we 1identified which empirical methods have been used to study environmental justice in published case studies; 2characterized which justice dimensions (e.g., contextual, distributive, procedural, recognitional, restorative, intergenerational, multispecies) they address; 3explored how these methods evolved over time and across geographies and 4which actors were considered and how they were related to which sampling, data collection and analysis methods. Overall, by providing an overview of existing methods, their uses and limitations, and how they relate to different justice claims, we aimed to support both academic research and policy applications. RESULTS Descriptive overview of methods used in environmental justice A log-linear regression model revealed a significant increase in the number of published studies over time, with an estimated annual growth rate of approximately 13% per year (β = 0.123, 2 iScience 28, 113889, December 19, 2025 iScience Meta-analysis/systematic review ll OPEN ACCESS p < 0.001, R 2 = 0.81). Articles were published in 190 different journals, with the International Journal of Environmental Research and Public Health (n = 24), Environmental Justice (n = 23), Sustainability (n = 13), Urban Planning (n = 12) and Geography (n = 10) being the five most frequented ones. Most of the studies (n = 248) were quantitative investigations, followed by mixed approaches (n = 101) and qualitative studies (n = 72). The proportion of quantitative and mixed-methods articles decreased significantly over time (p < 0.05), while the share of qualitative studies remained stable. Our set of articles entails 210 interdisciplinary and 155 disciplinary studies, with multiand transdisciplinary studies only having found their way into environmental justice research in the year 2005. Since then, however, the proportional share of transdisciplinary studies has significantly declined (estimate (− 0.132, p < 0.001). Sampling strategy Almost 50% of all articles used purposive sampling (n = 213), followed by 161 studies employing full inclusion/census sampling. While purposive sampling was associated with qualitative and mixed inter- & transdisciplinary studies, full inclusion/ census was typically related to quantitative, disciplinary studies. Stratified sampling (n = 42) and convenience sampling (n = 40) were less common, with random (n = 16), multi-stage (n = 15), and snowball sampling (n = 9) used even less frequently. Full inclusion/census was most commonly combined with purposive sampling (n = 34), and convenience sampling was also frequently paired with purposive sampling (n = 11). Overall, purposive sampling was the method most often combined with others (n = 76), followed by full inclusion/ census (n = 44). After accounting for the increasing annual number of articles, there was no significant trend in the use of single versus multiple sampling strategies over time. However, snowball sampling showed the strongest significant decline over time (estimate = − 0.233, p = 0.034), accompanied by notable decreases in basic random (estimate = − 0.104, p = 0.007) and multi-stage sampling (estimate = − 0.103, p = 0.033), while convenience sampling remained stable. Data collection techniques A total of 138 articles applied GIS and remote sensing for data collection, and 133 utilized secondary data sources, most frequently associated with a disciplinary, quantitative context. Surveys, censuses, or questionnaires were used in 119 studies, while 68 articles conducted interviews. These were mainly employed in qualitative studies. Participatory methods were applied to 41 transdisciplinary, qualitative, and mixed studies. There was a significant increase in the diversity of data collection methods over time (β = 0.47, SE = 0.04, t = 10.91, p < 0.001), with the model explaining 82% of the variance (R 2 = 0.82). Most articles (58%) relied on a single data collection method, predominantly secondary data (n = 69), GIS/remote sensing (n = 60), or surveys/census/questionnaires (n = 57). Over time, the relative use of single methods slightly decreased while the use of multiple data collection techniques increased. Among the methods analyzed, only surveys/census/questionnaires showed a significant decline (estimate = − 0.056, p = 0.003). The most frequent combinations were GIS/remote sensing with secondary data (n = 34) and GIS/remote sensing with surveys/census/questionnaires (n = 26). Data analysis methods A total of 249 articles applied only one data analysis method at a time, out of which 76 disciplinary articles solely applied univariate statistics, 69 applied spatial analysis, 23 articles used qualitative coding, and 21 used descriptive statistics, out of which a large share were of an interdisciplinary nature. 146 articles used several data analysis methods, with most of them (n = 62) combining spatial analysis with univariate statistics, while 26 combined spatial analysis with descriptive statistics. Another relatively large fraction of paired analyses methods were combined descriptive and univariate statistics (n = 21), as well as descriptive and multivariate statistics (n = 15). We detected a small but statistically significant increasing trend in the number of different methods applied over time (Kendall’s tau = 0.11, p = 0.006), indicating a gradual rise in the use of multiple data analysis methods across articles. Normalized by the increasing number of publications over time, generalized linear models revealed a significant decline in the relative use of machine learning and modeling methods (p < 0.005). Univariate statistical methods also showed a decreasing trend, though this was marginally non-significant (p = 0.0076). No other methods demonstrated significant changes in relative usage over time. Linkage between sampling, data collection, and analysis methods Pearson’s chi-squared tests revealed highly significant associations between sampling strategies and data collection methods (χ 2 = 540.49, df = 220, p < 0.001), sampling strategies and data analysis methods (χ 2 = 198.87, df = 66, p < 0.001), as well as between data collection and data analysis methods (χ 2 = 386.26, df = 120, p < 0.001), indicating strong interdependencies among these methodological choices. The results reveal clear interdependencies among sampling strategies, data collection methods, and analytical approaches. Purposive sampling was closely associated with qualitative techniques such as interviews and qualitative or mixed methods analysis. Full inclusion or census-based sampling often co-occurred with secondary data and GIS/remote sensing collection, linking further to spatial and statistical analyses. Basic random sampling was tied to structured instruments such as surveys and was frequently analyzed using machine learning or statistical techniques. Across data collection and analysis, qualitative methods such as interviews, focus groups, participatory techniques, and ethnographic approaches were strongly linked to qualitative and mixed methods analysis. In contrast, secondary data and GIS/remote sensing were associated more with statistical and spatial analyses (Figure 2). Justice dimensions addressed across studies Of the coded articles, 68% (n = 287) explicitly engaged with distributive justice, 35 articles with procedural justice, and 19 with recognition justice; 91 articles did not clearly specify a justice dimension. Other dimensions addressed included social (n = 13), epistemic (n = 6), and spatial justice (n = 6). A total of 32 articles engaged with more than one justice dimension. iScience 28, 113889, December 19, 2025 3 iScience Meta-analysis/systematic review ll OPEN ACCESS Notably, all articles published before 2005 engaged only in a single justice dimension. In 2005, one-third of the studies incorporated more than one justice dimension. Since then, the number of multidimensional justice studies has increased significantly, with an average annual rise of 0.2 articles, comprising 1–12% of annual publications. Justice dimensions are highly significantly associated with the methodological approaches employed in the studies, including sampling strategies (χ 2 = 254.26, df = 96, p < 0.001), data collection methods (χ 2 = 703.38, df = 176, p < 0.001), and data analysis techniques (χ 2 = 437.83, df = 88, p < 0.001), indicating that the choice of methods varies systematically depending on the justice focus of the research. Articles that considered distributive justice mostly used purposive sampling (n = 141) or full inclusion/census (n = 119) as a sampling strategy. In articles looking into procedural and recognition justice, purposive sampling dominated (n = 23; n = 15, respectively). Data collection in distributive justice studies was mostly related to GIS/remote sensing (n = 109) or secondary data collection (n = 100), whereas studies with a procedural and recognition focus used interviews (n = 17; n = 14, respectively). Data analysis in distributive justice studies predominantly employed spatial analysis (n = 127) and univariate statistics (n = 122), both showing strong positive associations (std. residuals >5). Descriptive statistics (n = 58) and multivariate statistics (n = 33) were also commonly used in these studies. In contrast, procedural and recognition justice studies primarily utilized qualitative coding (procedural n = 8, recognition n = 6) and content analysis (procedural n = 6, recognition n = 6), with significant positive associations indicated by residuals above 5 and 6, respectively. Recognition studies further featured contextual analysis and machine learning/modeling methods, suggesting a broader methodological diversity within this category. Participatory methods or analyses were used in 25 articles, out of which 14 engaged with distributive justice, 6 on procedural, 2 on epistemic, and 1 on climate, recognition, and/or social justice. The conceptualization of EJ has evolved from focus on disparities to a deeper engagement toward assessing equity concerns and broadly the root causes of (in)justice. There was a significant increase over time in the use of equality-related terminology (p = 0.03), while other terminologies, such as equity, fairness, disparity, and unevenness, showed no significant change. The use of justice-related terms showed a positive but marginally non-significant upward trend (p = 0.096). Geographic and spatiotemporal patterns across studies The majority of the n = 421 reviewed articles explored environmental (in)justice cases in North America (n = 245, Figure 3A), embracing the USA (n = 227) and Canada (n = 18). Studies from Europe (n = 65) entailed cases from Spain (n = 10), Germany (n = 9), France (n = 8), and the UK (n = 8). Most studies from Asia (n = 53) came from China (n = 36). Remarkably few studies represent Oceania (n = 3), and Africa (n = 14), and 17 studies employed cases on various continents. Although relatively underrepresented, literature from South America and Africa often focuses on issues around governance, recognition, participation, livelihood, and conservation. African studies had the highest share of interdisciplinary research approaches (71%), whereas studies Figure 2. Sankey diagram showing 927 connections between data collection and analysis methods across 421 articles, reflecting multiple methods per article 4 iScience 28, 113889, December 19, 2025 iScience Meta-analysis/systematic review ll OPEN ACCESS from South America were mostly disciplinary (61%). In the case of South America, EJ research include tourism in protected area management, 33 extractivism, 34 food insecurity in indigenous communities 35 and human rights violations. 36 Methods range from geospatial tools and surveys to ethnographic and participatory methods. In the case of Africa, the focus is set on exclusion from environmental decision-making of poor and marginalized people, who remain excluded from environmental decision-making, 37 PAs impact on environmental justice and human well-being 38 and women’s vulnerability to climate change. 39 Regarding methods, Africa-based studies present in our database focus more on structured data collection and less on immersive qualitative fieldwork. Most studies investigated environmental justice issues on a local (n = 184), regional (n = 170), or national (n = 50) scale (Figure 3B). Over time, the proportion of studies conducted at local and regional scales increased slightly, while the relative share of global-scale studies declined significantly (p < 0.001). Regarding temporal coverage, the majority of studies (n = 177) focused on periods of one year or less. A total of 91 studies examined durations of 1–5 years, 95 studies focused on 5–25 years, and 34 studies addressed periods exceeding 25 years. An ordinal logistic regression showed that over time, studies tended to have shorter durations (estimate = − 0.001, p < 0.001). The temporal scale of analysis was significantly associated with the spatial scale (χ 2 = 37.90, df = 16, p = 0.0016). Short-term studies (≤1 year) primarily focused on local scales (∼50%), whereas long-term studies (>25 years) were more evenly distributed across local (41%) and broader regional to national scales (44%). Actor engagement and epistemic prioritization Most studies (n = 282) engaged with a single actor, while 139 studies engaged with several actors. Most frequently, community actors (n = 315) and vulnerable communities (n = 178) were subject to the study, with markedly less focus on institutional actors (n = 37) and governance actors (n = 28). Although there was a trend toward more actor groups being involved per study over the years, this trend was non-significant. While community actors and vulnerable people have been involved throughout the entire investigated period (1997–2025), other actors were only considered later, starting with institutional actors in 2002, experts and governance actors in 2006, and policy shapers and actors from the private sector in 2010. While the number of actors considered in studies increased overall, the proportional share of each individual actor group declined except for external actors, and this was significant for governance actors (estimate − 0.016, p < 0.005), institutional actors (estimate − 0.008, p < 0.01), and policy shapers (estimate − 0.011, p < 0.01). Studies including community actors mostly engaged at the local (n = 147) and at the regional (n = 122) scale, which is a pattern that was also visible for vulnerable community groups (with n = 76 studies at the local and n = 71 at the regional scale). In contrast, institutional actors were associated with studies covering the local to global scale. Actor inclusion showed to be scale-sensitive in the scientific literature (χ 2 = 60.99, df = 32, p < 0.005). The actors investigated in the studies were as well highly significantly associated to the choices of sampling (χ 2 = 381.86, df = 104, p < 0.005), data collection (χ 2 = 431.76, df = 144, p < 0.005) and analysis methods (χ 2 = 556.86, df = 72, p < 0.005), suggesting epistemic prioritization in these studies: While experts were typically involved through purposive sampling, community actors most frequently were investigated by stratified sampling, and vulnerable actors through full inclusion/census and governance actors through snowball sampling. In terms of data collection, information from external influencers was associated with qualitative coding, whereas experts were investigated mostly through case studies and interviews, vulnerable community members mostly through surveys, census, and questionnaires, and rarely through participatory approaches. Instead, participatory data collection methods mostly occurred in relation to governance actors. Data analysis in studies that included institutional actors was primarily through content analysis, while studies considering experts, governance actors, and/or policy shapers employed contextual analysis. Actors from the private sector were included in studies using qualitative coding. Actor groups differed significantly across the considered justice dimensions (χ 2 = 162.43, df = 64, p < 0.001). Articles including community and vulnerable actors mainly addressed distributive justice, Figure 3. Detrended Correspondence Analysis Ordination showing indicator word clusters (A) Share of articles per continent, based on the 421 investigated articles. (B) Counts of published articles sorted by their spatial focus, ranging from local to global investigations. iScience 28, 113889, December 19, 2025 5 iScience Meta-analysis/systematic review ll OPEN ACCESS while those featuring governance and institutional actors engaged more evenly with procedural and recognition dimensions. Clusters/semantic map of environmental justice clusters and claims Our cluster analysis (Figure S1) and the multivariate analysis of the 409 articles used in the word analysis suggested eight different research clusters present in the environmental justice literature that we reviewed (Figure 4). Given their most abundant terms, we named these research clusters governance, pollution, climate change, health, waste, access, collaboration and green space (Table 1). The cluster comprising the largest number of articles in this review is the governance cluster (n = 73) which shows a distinct word usage profile compared to the closely overlapping clusters on pollution (n = 72), waste (n = 52), and health (n = 53); the overlapping clusters of climate change (n = 62), access (n = 43), and green space (n = 52); and the separate cluster on collaboration (n = 73) (see Figure 4). These clusters are distributed along the two DCA axes in ways that reflect broader thematic distinctions. The clusters on pollution, waste, and health center on environmental harms and the disproportionate exposure of socially disadvantaged people, often defined by race, ethnicity, or income. These articles primarily aim to identify, describe, and measure environmental injustices, emphasizing diagnosis and problem characterization. In contrast, the governance and collaboration clusters are more oriented toward institutional responses, participatory strategies, and solutions for addressing injustices. Based on this contrast, the first DCA axis can be interpreted as capturing a gradient from diagnosing harms to developing strategies for improvement. A second thematic distinction emerges between clusters focusing on environmental harms (pollution, waste) and those addressing the distribution of environmental goods and opportunities (access, green space). While the former highlights disproportionate burdens and vulnerabilities, the latter emphasize just access to environFigure 4. Eight research clusters derived from 409 empirical articles on environmental justice (12 excluded due to overlength) Colored nouns indicate cluster membership. ‘‘Accessibility’’ is prominent both in the light green and violet clusters. Proximity between clusters reflects thematic similarity. mental benefits and adaptation strategies. Accordingly, the second DCA axis may be interpreted as representing a continuum from exposure to environmental harms to access to environmental benefits. Before 2000, there were mainly studies belonging to the clusters of waste and pollution (Figure 5A). In the year 2000, access formed a new cluster in EJ research, with collaboration following in 2002, health in 2004 and governance in 2005. The climate change cluster occurred for the first time in 2012, and green space literature emerged in 2018. Studies in the cluster waste peaked in 2013 with 6 publications, and pollution studies peaked in 2024 with 12 studies. 2024 was also the year with the highest number of publications for all other clusters, except for access, which reached the highest number of 7 publications in 2023. Over time, the waste cluster showed a significant declining trend (estimate = − 0.119, p < 0.001), indicating reduced relative focus despite increasing total publications. The pollution, access, governance, and collaboration clusters also exhibited decreasing but statistically non-significant trends. Conversely, climate change and green space clusters showed positive yet non-significant increases, suggesting emerging interest. The health cluster remained relatively stable over time. Across all continents, research in Africa is mainly related to the governance cluster (n = 9), in Asia to green space (n = 16), in Europe to pollution (n = 28), and in North America to health (n = 50). The governance cluster also dominated in studies in Oceania, South America, and global or multiple geographic foci. There was no research on health or waste in Africa, Oceania, or South America, although these were the dominating clusters in North America. Another stark contrast is in the number of studies in the green space cluster, which dominated in Asia but was only present with 1 publication in North America, and not present in any other region. A Pearson’s Chi-Square test revealed significant associations between clusters and methodological choices, with sampling methods (χ 2 = 126.87, df = 77, p = 0.0003), data collection methods (χ 2 = 375.00, df = 147, p < 2.2e-16), and analysis methods (χ 2 = 257.52, df = 70, p < 2.2e-16) all showing nonrandom distributions across clusters. The dominating data collection and analysis methods per group are displayed in Table 1. In the governance and collaboration clusters, studies predominantly used qualitative and participatory data collection methods such as interviews and community engagement, emphasizing procedural justice and advocacy. Studies from the governance clusters were also less likely to use multiple 6 iScience 28, 113889, December 19, 2025 iScience Meta-analysis/systematic review ll OPEN ACCESS Table 1. Group characteristics ordered by size, showing article counts and the top 75% most frequent data collection and analysis methods per group by cumulative occurrence Group # Articles Five most abundant words Description Data collection method Data analysis methods 3-Governance 73 recognition, governance, livelihood, participation, conservation Most studies conducted in North America (n = 16). Cluster with the most global studies (n = 5). Most studies use a qualitative or mixed approach and interviews focusing on procedural and recognitive (in)justice. Topics evolve around conservation, capacity building, ecosystem services, and land rights. Interviews (26%); Participatory methods (12%); review (11%); GIS/remote sensing (10%); surveys, census, and questionnaires (10%) Qualitative coding (21%); Content analysis (39%); Descriptive statistics (55%); Spatial analysis (70%) 7-Pollution 72 dioxide, emission, nitrogen, concentration, particulate Most studies conducted in Europe (n = 28). The majority of articles in this cluster focus on urban areas and connect air pollution from traffic to the socio-economic status of residents. Other articles evaluate health effects of proximity to industrial sites. Most studies use a quantitative approach and spatial analysis tools focusing on distributional (in)justice. Secondary data (33%); GIS/remote sensing (23%); surveys, census, and questionnaires (18%) Univariate analysis (39%); Spatial analysis (74%) 6-Climate change 62 heat, mitigation, climate, cover, resilience Most studies were conducted in North America (n = 38). Articles in this cluster cover extreme weather events in urban and rural areas, such as heat, floods, and droughts, seismic risks. Articles develop decision support systems to reduce harm through environmental hazards via risk indices to improve community resilience. GIS/remote sensing (36%); Secondary data (25%) Spatial analysis (42%) 2-Health 53 burden, disease, exposure, poverty, tract Most studies were conducted in North America (n = 50), and three studies were from Asia. Most studies use a quantitative approach using databases and census tract information, focusing on distributional (in)justice. The analyses evolve around health risks, primarily cancer, asthma, and heart diseases, and through air pollution and soil contamination related to socioeconomic status. Secondary data (30%); surveys, census, and questionnaires (20%); GIS/remote sensing (16%); clinical trials and medical data collection (7%) Univariate statistics (55%) 1-Waste 52 minority, race, variable, facility, racism Most studies were conducted in North America (n = 45). Articles in this cluster analyze spatial inequalities from hazardous waste dump facilities, brownfields, and air toxics. Most studies use a quantitative approach and spatial analysis tools focusing on distributional (in)justice. Surveys, census, and questionnaires (29%); GIS/remote sensing (26%) Univariate statistics (46%) (Continued on next page) iScience 28, 113889, December 19, 2025 7 iScience Meta-analysis/systematic review ll OPEN ACCESS Table 1. Continued Group # Articles Five most abundant words Description Data collection method Data analysis methods 5-Access 43 accessibility, recreation, safety, activity, transportation Most studies were conducted in North America (n = 26). Studies focus on safety in urban environments, access to green spaces, and the effects of highway constructions. Most studies use a quantitative or mixed approach and spatial analysis tools focusing on distributional (in)justice. GIS/remote sensing (30%); surveys, census, and questionnaires (28%) Spatial analysis (37%); Univariate statistics (64%) 4-Collaboration 37 partnership, advocacy, action, collaboration, team Most studies were conducted in North America (n = 36). Most studies use a qualitative approach and interviews focusing on distributive and procedural (in)justice. Articles in this cluster work with research approaches to enhance capacity building using citizen science participation, participatory research, and communityengaged research methods. Interviews (17%); participatory methods (14%); surveys, census, and questionnaires (14%); review (10%); secondary data (8%); focus groups (7%) Content analysis (24%); Spatial analysis 44%); Descriptive statistics (62%) 8-Green space 17 walking, accessibility, travel, supply, demand Most studies were conducted in Asia (n = 16), and one study was from the US. Most articles focus on accessibility to green spaces in urban areas. They use a quantitative approach and spatial analysis tools focusing on distributional (in)justice. GIS/remote sensing (40%); Secondary data (20%) Spatial analysis (39%); Univariate statistics (59%); Multivariate statistics (75%) 8 iScience 28, 113889, December 19, 2025 iScience Meta-analysis/systematic review ll OPEN ACCESS 70. Pulido, L., and De Lara, J. (2018). Reimagining ‘justice’in environmental justice: Radical ecologies, decolonial thought, and the Black Radical Tradition. Environ. Plan. E Nat. Space 1, 76–98. 71. Azevedo, F., Jost, J.T., Rothmund, T., and Sterling, J. (2019). Neoliberal ideology and the justification of inequality in capitalist societies: Why social and economic dimensions of ideology are intertwined. J. Soc. Issues 75, 49–88. 72. Vermeylen, S. (2019). Environmental Justice and Epistemic Violence (Taylor & Francis). 73. Ramcilovic-Suominen, S. (2023). Envisioning just transformations in and beyond the EU bioeconomy: inspirations from decolonial environmental justice and degrowth. Sustain. Sci. 18, 707–722. 74. A ´ lvarez, L., and Coolsaet, B. (2020). Decolonizing environmental justice studies: a Latin American perspective. Capitalism nature socialism 31, 50–69. 75. Temper, L. (2019). From boomerangs to minefields and catapults: dynamics of trans-local resistance to land-grabs. J. Peasant Stud. 46, 188–216. 76. Wong, P. (2020). Linking ‘‘local’’ to ‘‘global’’: framing Environmental Justice movements through progressive contextualization. Interface 12, 215–243. 77. Tsing, A.L. (2024). Friction: An Ethnography of Global Connection (Princeton University Press). 78. Menton, M., Larrea, C., Latorre, S., Martinez-Alier, J., Peck, M., Temper, L., and Walter, M. (2020). Environmental justice and the SDGs: from synergies to gaps and contradictions. Sustain. Sci. 15, 1621–1636. 79. Leder, S. (2024). Translocal resource governance, social relations and aspirations: Linking translocality and Feminist Political Ecology to explore farmer-managed irrigation systems and migration in Nepal. Geoforum 148, 103905. 80. Contardo, T., and Loppi, S. (2024). Assessing Environmental Justice at the Urban Scale: The Contribution of Lichen Biomonitoring for Overcoming the Dichotomy between Proximity-Based and Distribution-Based Approaches. Atmosphere 15, 275. 81. Bartlett, L., and Vavrus, F. (2016). Rethinking Case Study Research: A Comparative Approach (Routledge). 82. Thaler, G.M. (2021). Ethnography of environmental governance: Towards an organizational approach. Geoforum 120, 122–131. 83. Wilson, A.M., Polk, E., Field, C.B., and Fendorf, S. (2025). Towards environmental justice: A framework and strategic approach for implementing community based participatory research in the earth and environmental sciences. Environ. Sci. Pol. 167, 104036. 84. Tassan, M. (2022). Rethinking environmental justice in the Anthropocene: An anthropological perspective. Anthropol. Today 38, 13–16. 85. Porada, H., Boelens, R., and Vos, J. (2024). Bridging Justice Struggles: A Political Ecology of Translocal Alliance Building against Extractive Industries. Alternautas 11, 1–34. 86. Backhaus, J., and John, S. (2025). Generalization as local and translocal embedding: interrogating governance and deconstructing democratization in living labs. Sustain. Sci. Pract. Pol. 21, 2450856. 87. Sultana, F. (2022). The unbearable heaviness of climate coloniality. Polit. Geogr. 99, 102638. 88. Deivanayagam, T.A., English, S., Hickel, J., Bonifacio, J., Guinto, R.R., Hill, K.X., Huq, M., Issa, R., Mulindwa, H., Nagginda, H.P., et al. (2023). Envisioning environmental equity: climate change, health, and racial justice. Lancet 402, 64–78. 89. Massey, D.B. (2005). For space (SAGE Publications). 90. Walker, G. (2009). Globalizing environmental justice: The geography and politics of frame contextualization and evolution. Glob. Soc. Policy 9, 355–382. 91. Schlosberg, D. (1999). Environmental Justice and the New Pluralism: The Challenge of Difference for Environmentalism (Oxford University Press). 92. Turnhout, E. (2024). A better knowledge is possible: Transforming environmental science for justice and pluralism. Environ. Sci. Pol. 155, 103729. 93. Ahmed, S.K. (2024). Research Methodology Simplified. How to Choose the Right Sampling Technique and Determine the Appropriate Sample Size for Research. Oral Oncol. Rep. 12, 100662. 94. Abson, D.J., Von Wehrden, H., Baumga ¨ rtner, S., Fischer, J., Hanspach, J., Ha ¨ rdtle, W., Heinrichs, H., Klein, A.-M., Lang, D.J., Martens, P., and Walmsley, D. (2014). Ecosystem services as a boundary object for sustainability. Ecol. Econ. 103, 29–37. 95. Dufre ˆ ne, M., and Legendre, P. (1997). Species assemblages and indicator species: the need for a flexible asymmetrical approach. Ecol. Monogr. 67, 345–366. iScience 28, 113889, December 19, 2025 15 iScience Meta-analysis/systematic review ll OPEN ACCESS STAR★METHODS KEY RESOURCES TABLE METHOD DETAILS We conducted a systematic literature review using the Scopus database on November 29, 2024, applying the following search string: (TITLE-ABS(‘‘environmental equity’’ OR ‘‘environmental inequity’’ OR ‘‘environmental justice’’ OR ‘‘environmental injustice’’ OR ‘‘environmental disparities’’ OR ‘‘environmental inequality’’ OR ‘‘environmental equality’’) AND TITLE-ABS(‘‘case study’’ OR ‘‘area’’ OR ‘‘study region’’) AND TITLE-ABS(‘‘method’’ OR ‘‘assessment*’’ OR ‘‘tool*’’)). We limited the search to English-language, peer-reviewed articles, resulting in 730 initial records. Screening occurred in three stages: 1. Title screening excluded 256 articles that were clearly unrelated to environmental (in)justice or lacked an empirical focus. 2. Abstract screening removed an additional 26 articles that did not meet our inclusion criteria: empirical focus and engagement with environmental (in)justice. 3. Full-text screening excluded 27 more articles due to insufficient relevance or lack of methodological detail. EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAIL The final dataset included 421 articles for full analysis (Figure 1). We classified data collection temporal scales into five categories: short (<1 year), medium (1–5 years), long (>5 years), and very long (>30 years). Studies lacking temporal information were coded as missing. Data types were categorized as quantitative, mixed, or qualitative. Sampling strategies were classified as probability-based (e.g., simple random, stratified, cluster, systematic, multi-stage) or non-probability-based (e.g., quota, snowball, purposive, convenience) following Ahmed (2024). 93 Full inclusion or census approaches were also noted when entire datasets were analyzed. Data collection methods were grouped into the following broad categories based on methodological traits: Interviews, surveys/ census/questionnaires, GIS/remote sensing, clinical trials or medial data collection, simulation & modelling, statistical designs, instrumental or laboratory measurements, field sampling, reviews, participatory methods, observation, focus groups, case studies, ethnographic methods, socio-legal sources or secondary data sources. Data analysis methods were coded into univariate statistics, multivariate statistics, descriptive statistics, spatial analysis, machine learning & modelling, contextual analysis, participatory methods or analysis, qualitative coding or content analysis. Finally, we classified studies by their disciplinary orientation (disciplinary, interdisciplinary, multidisciplinary or transdisciplinary), and the dimensions of justice addressed. We conducted two rounds of intercoder reliability checks during the title and abstract screening phases to ensure consistency. For full-text coding, variable definitions were collaboratively developed and refined through team discussions. Each article was coded by one researcher and reviewed or supplemented by a second. Although we did not calculate formal intercoder agreement scores, collaborative review aimed to minimize inconsistencies. QUANTIFICATION AND STATISTICAL ANALYSIS We analyzed our dataset at two levels. We performed descriptive statistical analysis on all 421 articles. This included univariate analyses to explore associations between variables using Chi-square tests, generalized linear models (GLMs), log-linear models to investigate proportions, Fisher’s exact tests, and Kendall’s tau rank correlations, depending on the type and distribution of the data. Following Abson et al. (2014), 94 we conducted a multivariate full-text word analysis based on nouns extracted from the 409 REAGENT or RESOURCE SOURCE IDENTIFIER Software and algorithms Python (v3.12) Python Software Foundation https://python.org R (v4.4.2) R project https://www.r-project.org Other Literature database for systematic review Scopus https://www.scopus.com Analytical procedures: Chi-square, GLM, log-linear models, Fisher’s exact test, Kendall’s tau, hierarchical cluster analysis, indicator species analysis, DCA Analytical procedures Abson et al. 93 ; This paper Systematic review reporting guideline PRISMA 2020 https://prisma-statement.org e1 iScience 28, 113889, December 19, 2025 iScience Meta-analysis/systematic review ll OPEN ACCESS article pdfs. Assuming that research communities tend to use distinct vocabularies, we applied hierarchical cluster analysis to group articles by similarity in word use. Clustering was based on minimizing within-group variance and maximizing between-group differences in word frequency distributions. To characterize each cluster, we applied an indicator species analysis, 95 treating words as ‘‘species’’ and clusters as ‘‘habitats’’. This allowed us to identify statistically significant indicator words, i.e., terms that are disproportionately associated with specific clusters, forming the semantic core of each research cluster. Finally, we employed a detrended correspondence analysis (DCA) to reduce dimensionality and visualize the primary axes of variation in word use across articles. The resulting ordination plot shows indicator words along two primary axes, enabling interpretation of the relationships between thematic clusters in the environmental justice literature. All analyses were conducted in R,Version 4.4.2 (R Core Team 2021). iScience 28, 113889, December 19, 2025 e2 iScience Meta-analysis/systematic review ll OPEN ACCESS