Summertime temperatures, social disadvantage, and energy rationing: Patterns of energy insecurity from population-level historical electricity data in New York State

Summertime temperatures, social disadvantage, and energy rationing: Patterns of energy insecurity from population-level historical electricity data in New York State | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Summertime temperatures, social disadvantage, and energy rationing: Patterns of energy insecurity from population-level historical electricity data in New York State Daniel Carrión, Xuezhixing Zhang, Anna Stouffer, Diana Hernández, and 3 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6184819/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Rising temperatures and energy costs have intensified energy insecurity among vulnerable populations. This study examines patterns of energy insecurity and rationing in New York State using aggregated historical electricity data (2016–2020). Using linear quantile mixed-effect models, we analyzed differences in residential energy consumption across income groups and levels of racialized economic segregation during warm seasons. Results show that higher-income households used 26.24–55.58 kWh more energy per interquartile range (IQR) increase in heat index cooling degree days compared to the lowest income group. Areas with greater economic privilege, measured by the Index of Concentration at the Extremes, consumed 161.00-234.44 kWh more energy per IQR of heat index increase than less privileged areas. These findings suggest that lower-income and minoritized communities may engage in energy rationing during high cooling demand periods, potentially exposing themselves to elevated indoor temperatures and associated health risks. This evidence highlights the need for targeted interventions to address energy insecurity in disadvantaged communities. Earth and environmental sciences/Environmental social sciences/Energy and society/Energy justice Earth and environmental sciences/Environmental social sciences/Sustainability energy insecurity energy rationing extreme heat disparities cooling demand Figures Figure 1 Figure 2 Figure 3 1. Introduction Over the past two decades, rising temperatures have driven an increase in global energy demand for cooling. 1 In New York State (NYS) in particular, the number of cooling degree days are expected to increase, representing more days when cooling is needed and a higher intensity of space-cooling demand of NYS homes relative to previous years. 2 Energy use for space cooling is critical to prevent life-threatening heat stroke, improving cognition and development, and maintaining a number of other social and public health protective factors. 3 – 6 However, as energy demand and costs rise, so does the risk for energy insecurity. 7 , 8 In order to address this growing issue, justice-oriented solutions that address the uneven distribution of temperatures and energy insecurity are needed. 9 Energy insecurity is a phenomenon that represents the interplay of the physical status of housing, household energy expenditures and finances, and related coping strategies. 10 The phenomenon disproportionately burdens low-income, ethno-racial minoritized groups, and renting individuals. Most households in the United States near or below the federal poverty line are energy insecure likely due to higher energy burden, as a greater proportion of their income is allocated towards energy costs and are more likely to live in dilapidated/older housing. 11 One hallmark of behavioral energy insecurity is the notion of energy rationing, whereby energy-insecure families employ strategies such as using appliances more sparingly in order to decrease energy cost. 10 , 12 , 13 In general this behavior is referred to as the ‘heat or eat’ dilemma, however, perhaps more fittingly for summertime temperatures, we refer to it as the ‘heat stroke or go broke’ dilemma. Previous research has primarily examined energy rationing through individual-level interviews and household surveys, less is known about how these behavioral patterns manifest at the community level. Energy rationing behavior, though enacted at the household level, may be collectively shaped by neighborhood-level social and economic conditions. Understanding these spatial patterns is crucial because they may reveal systematic disparities in how communities adapt to increasing cooling demands, which is analogous to past work on community-level wintertime energy use intensity for heating. 14 Research is needed to advance this understanding for summer months by examining evidence of energy rationing at the population level, moving beyond individual household behaviors to identify how structural disadvantage may influence collective energy use patterns across communities. There are a series of individual and structural factors that increase the likelihood of experiencing energy insecurity. Economic elements of energy insecurity can involve having low- or fixed-income, which can be compounded by physical elements of energy insecurity such as living in poor or dilapidated housing conditions, and having older appliances. 15 Additionally, renters are more likely to be energy insecure because, energy bills typically fall on renters but landlords have little incentive to replace older energy appliances; thus renters have little agency in improving the energy bills they pay, i.e. the split-incentive problem. 16 However, new appliances and energy efficiency upgrades are oftentimes still inaccessible even for low-income and minoritized homeowners. 15 On the structural level, low-income and minority residents are more likely to live in segregated neighborhoods, where patterns of social disadvantage such as redlining and concentrated poverty have prevented investment in efficient and updated housing and energy infrastructure. 17 Counties with lower average income, more minority communities, and more households headed by women of color are more likely to be energy insecure. 18 Additionally, areas with higher densities of concrete buildings and less green space experience higher temperatures because of the urban heat island effect, which can contribute to greater energy demand for low-income disadvantaged communities and greater health risks for the most socially vulnerable. 19 Most examinations of coping or behavioral energy insecurity have used individual-level interview, survey, and/or energy use data to understand the phenomenon, particularly those examining energy rationing. 10 , 20 – 23 While individual factors influence energy rationing, structural disadvantage may also lead to behavioral energy rationing being observed at the area level. Increasingly there are population-level measures of energy utilization yet, there are few studies that have examined the ability to infer collective energy rationing behavior from population-level data. In this study we examined two forms of structural disadvantage: area-level income and area-level racialized economic segregation. We first examined differences in the slopes between energy utilization and cooling degree days by area income across NYS. We hypothesized that neighborhoods earning higher incomes would use more energy on similarly hot summer days relative to neighborhoods earning lower incomes. Second, we assessed differences in the slopes between energy utilization and cooling degree days by an area-level measure of racialized economic segregation, hypothesizing that neighborhoods with higher incomes and lower percentage of minoritized racial groups would use more energy relative to neighborhoods with lower incomes and higher percentage of minority race. This study adds to the literature by providing the first analysis of energy rationing in NYS, utilizing more accurate finer-scale data relative to other pre-existing energy literature, and more broadly exploring the structural patterns that influence how vulnerable communities manage energy insecurity. 2. Results In this analysis, we utilized residential energy data from the New York State-managed Utility Energy Registry (UER) to identify energy equity gaps at the zip code tabulation area (ZCTA) level. The analysis was limited to the warm season (defined as May through September), when residents are most likely to have increased energy demand for household cooling. 24 We first extracted and processed monthly energy data as well as environmental data to construct a daily heat index which we aggregated to a monthly sum of heat index cooling degree days (HICDD) from 2016 to 2020 for 1,794 ZCTAs within New York state. Then we employed linear quantile mixed-effect models (LQMMs) to identify if the slopes of the energy use (kWh) by HICDD association varied by interaction terms for various income groups at the median ( Suppl 1 ). Furthermore, we also investigated the associations between energy use and racialized economic segregation for various ethno-racial groups by constructing an index of concentration at the extremes (ICE) for each group and fit LQMMs for each group’s ICE. 25 2.1 Descriptive analyses Figure 1 displays the average energy use per household for all four income groups {< $ 50,000; $ 50,000 to $ 69,999; $ 70,000 to $ 89,999; and ≥ $ 90,000} during the warm seasons from 2016 to 2020. In ZCTAs with higher median incomes, there was consistently greater energy consumption. An increase of energy consumption in 2020 was observed across all income groups, which might suggest systematically higher energy use during the pandemic's quarantine. After excluding observations with missing energy consumption data or income information, 1,449 ZCTAs with a total of 27,916 observations (in kWh per ZCTA-month) were included in our subsequent analyses ( Supplemental Tables 1 and 2 ). We also visualized the energy use data in relation to HICDDs by NYS climate region (Fig. 2 ). Figure 2 B demonstrates the distribution of HICDDs observations by ZCTA-months, the unit of analysis, ranging from 13.4 to 929.5 with an interquartile range (IQR) of 268.51. Figure 2 A visualizes example ZCTAs for the highest and lowest income groups. 2.2 ZCTA-Level Energy Equity Gap by Income Levels Our models revealed a significant difference in energy use between the lowest income group and the highest income groups (Table 1 ). Specifically, in the median quantile regression, an IQR increase in HICDD is associated with 9.96 (95% CI: [-8.85, 28.77]), 26.24 (95% CI: [1.73, 50.75]), and 55.58 (95% CI: [12.24, 98.91]) kWh more energy consumption per household in the “ $ 50,000 to $ 69,999”, “ $ 70,000 to $ 89,999”, and “≥ $ 90,000” income groups, compared to households with income less than $ 50,000, respectively. 2.3 ZCTA-Level Racial Disparities in Energy Equity In addition to disparities among different income groups, we also fit LQMMs using the ICE measure of racialized economic segregation (Table 2 ). The ICE score reflects the proportional difference between the most privileged group (high-income predominantly white group) and the marginalized group (low-income ethno-racially minoritized groups). A positive ICE score indicates the ZCTA area is dominated by high income white people, compared to low income minoritized groups, and vice versa. Table 2 displays estimates for each ethno-racial ICE combination at the median quantile. Specifically, in the median quantile regression, an IQR increase in HICDD resulted in 166.52 (95% CI: [95.16, 237.88]), 130.66 (95% CI: [35.98, 225.35]) and 238.39 (95% CI: [165.44, 311.35]) kWh of additional energy consumption per household for the most privileged ZCTA areas (ICE = 1) for the Black, Hispanic/Latino, and Asian groups respectively. Table 1 Model estimates for energy use by income groups. Models were fitted using a linear quantile mixed effects model with fixed effects of average household size and scaled number of households and random intercepts for combinations of county, month and year. Each row presents the model estimate for the interaction term between income groups and an interquartile range increase of heat index cooling degree days (HICDD). Each point estimate represents the additional energy consumption of ZCTA areas from the corresponding income group, compared to ZCTA areas with median income less than $ 50,000. Variable Model Estimates Main Model Sensitivity Model I Sensitivity Model II Sensitivity Model III Sensitivity Model IV HICDD: $ 50,000 to 69,999 9.96 [-8.85, 28.77] 7.26 [-14.89, 29.40] 32.94** [10.13, 55.76] 22.21*** [10.01, 34.41] 570.82 [-128.38, 1270.03] HICDD: $ 70,000 to 89,999 26.24* [1.73, 50.75] 24.22 [-2.25, 50.69] 34.69* [3.02, 0.27] 38.35*** [20.47, 56.23] 96.20 [-734.91, 927.31] HICDD:≥ $ 90,000 55.58** [12.24, 98.91] 50.72* [9.40, 92.04] 29.01 [-7.12, 65.13] 77.23*** [48.28, 106.18] 50.94 [-779.42, 881.29] Note: *: p < 0.05; **: p < 0.01; ***: p < 0.001. Main Model: Median quantile regression with the full imputed dataset (n = 27,916). Sensitivity Model I: Median quantile regression with the imputed dataset, excluding year 2020 (n = 23,547). Sensitivity Model II: Median quantile regression with the subset of ZCTA-level reported data (n = 13,645). Sensitivity Model III: Median quantile regression with the subset of community-level reported data. For demographic information, we use ACS 5-year estimates from the overlapping zip code area (n = 26,334). Sensitivity Model IV: Mean regression with the full imputed dataset (n = 27,916). Table 2 Model estimates for energy use with Index of Concentration at the Extremes (ICE) scores. Each row presents the model estimate for the interaction term between ICE score (scaled per 0 to 1 increase) and heat index cooling degree days (HICDD). Each coefficient represents the energy change of an interquartile range increase of heat index cooling degree days for various ethno-racial groups with the most privileged having an index of concentration at the extremes of 1. The privileged group defined as white people with household incomes of $ 100,000 or more and the disadvantaged group defined as Black, Hispanic/Latino, or Asian people making < $ 25,000. Model Main Model Sensitivity Model I Sensitivity Model II Sensitivity Model III Sensitivity Model IV Black 166.52*** [95.16, 237.88] 161.00*** [103.73, 218.26] 65.63 [-24.48, 155.73] 234.10*** [159.61, 308.59] 236.47 [-1218.73, 1691.66] Hispanic/ Latino 130.66** [35.98, 225.35] 198.85*** [107.34, 290.37] 71.07 [-37.61, 179.75] 249.28*** [159.40, 339.16] 254.75 [-1232.04, 1741.54] Asian 238.39*** [165.44, 311.35] 233.44*** [160.47, 306.40] 111.18 [-8.59, 230.94] 257.24*** [176.91, 337.57] 203.86 [-1450.27, 1858.00] Note: *: p < 0.05; **: p < 0.01; ***: p < 0.001. Main Model: Median quantile regression with the full imputed dataset (n = 27,916). Sensitivity Model I: Median quantile regression with the imputed dataset, excluding year 2020 (n = 23,547). Sensitivity Model II: Median quantile regression with the subset of ZCTA-level reported data (n = 13,645). Sensitivity Model III: Median quantile regression with the subset of community-level reported data. For demographic information, we use ACS 5-year estimates from the overlapping zip code area (n = 26,334). Sensitivity Model IV: Mean regression with the full imputed dataset (n = 27,916). 2.4 Sensitivity Analyses Our sensitivity analyses assessed the consistency of our models under varying conditions/assumptions. In the models assessing differences by income group, we found similar patterns for Sensitivity Model I and Model III although Sensitivity Model I showed confidence intervals that crossed 0, likely due to reduced power from fewer observations (Table 1 ). Sensitivity Model II shows a higher energy utilization for ZCTAs with median income of 50,000 to 69,999 relative to the lowest income group. As for the ICE analysis, Sensitivity Models I and III were similarly consistent with the main model (Table 2 ). Sensitivity Model II showed overall attenuated and non-significant results. In both analyses, Sensitivity Model IV was fit using a linear mixed effects regression (Tables 1 and 2 ). These regressions resulted in insignificant estimates with wide confidence intervals, indicating mean regressions are likely heavily affected by extreme outliers. 3 Discussion This study examined patterns of energy insecurity and rationing in New York State using population-level electricity data from 2016–2020. Using monthly zip code-level excess outdoor heat (HICDDs) and residential energy consumption data in the warm season, we used LQMMs to identify differences in the heat versus electricity slopes between ZCTAs based on income and ICE, showing in both cases that more advantaged/affluent zip codes exhibit greater increases in energy consumption as temperatures rose. These findings suggest that lower-income and minoritized communities may be engaging in energy rationing behaviors during periods of high cooling demand, experiencing higher indoor temperature exposures, and risking resident health. 35 – 37 Our research advances an emerging body of work on energy limiting behavior and structural drivers of temperature-based energy insecurity. 20 – 22 Like other studies, our work aims to better understand and measure energy use and insecurity among low-income and minoritized communities with regard to space cooling. Other studies have found, for example, evidence of temperature-based energy rationing or “energy equity gaps” in Arizona and northern Illinois. 20 , 21 Those studies leveraged daily person-level electricity and demographic data directly from utility providers. However, our study, identifies such patterns at the population-level, using substantially coarser data, and associations with population-level drivers of structural disadvantage. This is quite notable given the potential for aggregation bias, i.e. that the pattern remains pronounced from the individual to population level. Relatedly, there are reasons to believe that no effect would be observed or even the opposite effects at the population level, including that income and segregation are also associated with poor housing quality and older energy inefficient appliances. 38 , 39 The results of these analyses are particularly encouraging because it is unlikely that every utility provider will offer access to their datasets to researchers, but more states are planning to make aggregate energy data available. This access to energy data offers the opportunity to replicate this energy rationing analyses in other states or regions of the country. This study has several important limitations. First, our reliance on administrative and aggregated data, while allowing for the desired population-level analysis, risks ecological fallacy. Inferences about individual or household behaviors drawn from these aggregate data may not accurately reflect the diverse experiences within each ZCTA. Second, we encountered substantial missing data at the zip code level, necessitating the use of an advanced imputation approach. While this method allowed us to maintain a comprehensive dataset, it may introduce biases, particularly if data are not missing at random. This bolsters the importance of our sensitivity analyses that restrict to the true dataset. Third, our dataset completely lacked information for Long Island, a significant and populous region of New York State. This omission limits the generalizability of our findings to the entire state and may skew our understanding of statewide energy consumption patterns. These limitations underscore the need for cautious interpretation of our results and highlight areas for future research using more complete data sources. This speaks to the importance of mandated reporting and making datasets publicly available statewide. Our study offers several notable strengths that enhance its contribution to energy insecurity research. First, this study covers the majority of NYS, and thus a substantially larger and diverse population and climatological distinct region compared to past studies. Second, we employ a robust statistical methodology using LQMMs, which allows for a nuanced examination of energy consumption patterns at the median rather than the mean of the distribution. This approach revealed insights that might be overlooked by conventional regression techniques. Third, we analyzed data longitudinally from 2016 to 2020, which provided valuable heterogeneity in electricity use and temperature for insights on trends over time. Fourth, rather than using data from a local weather station, we utilized Earth science products that reconstruct daily temperature at a fine 1-km resolution, offering a more accurate representation of dynamic temperature profiles. Fifth, our study's replicability is enhanced by using publicly available data sources and a publicly available code base. Finally, we conduct multiple sensitivity analyses, which strengthen the credibility of our findings by demonstrating their robustness under different assumptions and data subsets. 4 Conclusion This study provides compelling evidence of energy rationing behavior among lower-income and minoritized and marginalized communities in NYS, revealing important patterns of energy insecurity at the population level. By leveraging advanced statistical approaches and a comprehensive dataset covering a five-year period, we have demonstrated that socioeconomic and racial disparities are strongly associated with energy consumption patterns in response to outdoor heat. These findings have profound implications for energy justice and public health, particularly as climate change intensifies the frequency and severity of extreme heat events. The observed disparities in energy consumption suggest that vulnerable populations may be exposing themselves to higher indoor temperatures, potentially increasing their risk of heat-related illnesses. This research underscores the urgent need for targeted policy interventions to address energy insecurity and improve thermal comfort in disadvantaged communities. Future energy assistance programs and cooling strategies should prioritize these vulnerable populations, considering both income levels and racial composition of neighborhoods. Moreover, this study highlights the value of using population-level data to identify and address structural inequalities in energy consumption, providing a model for similar analyses in other regions. As we navigate the challenges of climate change and energy transition, ensuring equitable access to energy for cooling must be a central consideration in policy development to protect the health and well-being of all communities. 5. Methodology 5.1 Study Population and Electricity Data Data were downloaded from the Utility Energy Registry, a project of the New York State Energy Research and Development Authority. The data are divided into energy types (i.e. electricity and natural gas) and account types (e.g. residential and commercial) and the data are made available for zip codes, counties, and communities across NYS from 2016–2020, with incomplete spatial coverage for each unit. We restricted the subset data to residential and electricity given our interest in home space cooling and chose zip codes as our unit of analysis because they are the smallest available unit that have approximations for Census data via ZCTAs. Zip codes were aggregated into ZCTAs using a crosswalk file via the U.S. Department of Health and Human Services GeoCare Navigator. 26 5.2 Temperature Data Cooling degree days are often used to derive a measure of the demand for space cooling, but only considers dry bulb temperature. The heat index more closely approximates the human thermoregulatory strain, so we chose to calculate a HICDD measure. HICDD data were derived using maximum and minimum temperatures and vapor pressure data from NASA’s Daymet dataset, following the National Weather Service algorithm on the Fahrenheit scale. 27 5.3 Demographic and Buildings Data ZCTA demographic data were accessed from the American Community Survey 5-year estimates. We specifically downloaded race/ethnicity, average household size (persons), the number of households, the median income, and related covariates. Buildings data were accessed from the Microsoft Buildings dataset. This dataset is the result of satellite remote sensing via Bing Imagery and machine learning to create building footprints polygons throughout the U.S. 28 5.4 Missing data In instances where zip code-level energy data were missing, we utilized an adapted dasymetric method and specifically developed an adjusted version for imputation, summarized by Fig. 3 . 29 In summary, buildings were classified as residential based on 30-m population estimates from the updatable gridded lightweight impervious (UGLI) dataset. 30 We then approximated the volume of each building using the U.S. national categorical mapping of building heights, based on building areas and height categories. 31 The average energy consumption per unit of effective volume was calculated using community-level energy data (typically geographically smaller than ZCTAs) and the total building volumes were calculated based on the formula: building area × occupied ratio × height coefficient. The occupied ratio was defined as the ratio of occupied units to all units within each block group; height coefficients represented values for each height category derived from the height data. The best height coefficients were obtained using a grid search to find values that most accurately predicted the ZCTA level energy data. In our analysis, buildings were assigned with height categories “low”, “low-medium”, “medium”, “medium-high”, “high” and “very high” based on its corresponding block group height data. The best coefficients for these categories were selected from {1.5,2.5,3.5,4.0,6.0,10.0}. Finally, we aggregated all building-level energy data into ZCTA level for imputation. Information on the number of ZCTAs imputed (and corresponding population data) can be found in Supplemental Table 1 . In areas of overlap, we calculate a mean absolute error of 547.6 MWh which, for context, is substantially lower than standard deviation of the true data ( Supplemental Tables 2 and 3 ). 5.5 Statistical Methods The data demonstrated an irregular residual structure, so our main analyses modeled the conditional median using linear quantile mixed models in order to assess different slopes for the heat-electricity use relationships by 1) median income and 2) the racialized economic segregation measure, ICE. To identify potential differences in monthly energy utilization by income levels, we fit an LQMM expressed by Eq. 1: $$\:\begin{array}{c}{Energy\:per\:Household}_{\tau\:}=\:{\beta\:}_{0}+{\beta\:}_{1}*HICDD+{\beta\:}_{2}*\:Income+\\\:{\beta\:}_{3}*\:HICDD*Income+{\beta\:}_{4}*\:Av{g}_{Househol{d}_{size}}+{b}_{Count{y}_{time}}\#\left(1\right)\end{array}$$ Where τ specifies the quantile in our regression model. We divided the median household income into four levels based on the thresholds of $ 50,000, $ 70,000 and $ 90,000. Since temperatures can fluctuate across different time and geographical positions, leading to varied energy use, we included random intercepts for the combinations of county (as defined by the ZCTA centroid), month and year to capture the spatial and temporal information for each ZCTA area. To explore the typical relationship between energy use and income levels and avoid the effect of extreme outliers, we fit the model at the median quantile. To model differences in slopes by ICE measures, we constructed ICE measures combining ethno-racial and income data [2]. Specifically, we calculated the ICE score for Black, Hispanic/Latino, and Asian households relative to Non-Hispanic white households using Eq. 2: $$\:\begin{array}{c}{ICE}_{Race\:X}\:=\frac{\:Non\:hispanic\:white\:households\:over\:US\text{\$}\text{100,000}\:-\:Race\:X\:\:alone\:households\:under\:\text{\$}\text{25,000}}{Total\:households}\:\:\:\left(2\right)\end{array}\:$$ The ICE measure is a continuous score ranging from − 1 (most deprived) to 1 (most privileged), allowing comparisons along a continuum of economic-racial polarization. We used the same LQMM modeling strategy (Eq. 1) but substituted the income categories with the ICE measures in the model expressed by Eq. 2 and fit LQMMs for each race at median. 5.6 Sensitivity Analyses We conducted a series of sensitivity analyses to examine the robustness of our findings. In the first sensitivity model, we removed 2020 from our dataset given the COVID-19 pandemic and lockdowns that likely influenced home energy utilization. Secondly, given that we imputed 51% of the analytical dataset, we reran the analysis restricting to the subset of zip code-level observations from the Utility Energy Registry ( Supplemental Table 1 ). Third, we conducted the same analyses using only the community-level data given that zip code and community observations cover relatively distinct geographies across NYS. Finally, we modeled the conditional mean using a linear mixed effects regression using the fully imputed dataset. 5.7 Computational Environment and Reproducibility All analyses were conducted in R version 4.3 and analytic code can be accessed at https://github.com/CHEE-collaborative/TEJM_Analysis . LQMM and linear mixed effects models were fit using the lqmm and lme4 packages, respectively, in R. 32 , 33 Analyses were coded using the targets package and tested across several platforms to enhance reproducibility. 34 Declarations Acknowledgements: This work was supported by a U.S. National Aeronautics and Space Administration grant 80NSSC22K1666. References Scoccimarro, E. et al. Country-level energy demand for cooling has increased over the past two decades. Commun. Earth Environ. 4 , 208 (2023). Meier, S. et al. Chapter 6: Energy. N. Y. State Clim. Impacts Assess. Interim Version Public Release (2024). Gauer, R. & Meyers, B. K. Heat-related illnesses. Am. Fam. Physician 99 , 482–489 (2019). Heyes, A. & Saberian, S. Temperature and decisions: evidence from 207,000 court cases. Am. Econ. J. Appl. Econ. 11 , 238–265 (2019). Park, R. J., Goodman, J., Hurwitz, M. & Smith, J. Heat and learning. Am. Econ. J. Econ. Policy 12 , 306–39 (2020). Parks, R. M. et al. Anomalously warm temperatures are associated with increased injury deaths. Nat. Med. 26 , 65–70 (2020). Longden, T. et al. Energy insecurity during temperature extremes in remote Australia. Nat. Energy 7 , 43–54 (2021). Carrión, D. & Hernández, D. Heatwaves and hardship: Shortcomings and solutions for enhancing the low income home energy assistance program to mitigate extreme heat and energy insecurity. Electr. J. 37 , 107440 (2024). Carrión, D., Rush, J., Colicino, E. & Just, A. C. Residential segregation and summertime air temperature across 13 northeastern U.S. states: potential implications for energy burden. Environ. Res. Lett. 19 , 084005 (2024). Hernández, D. Understanding ‘energy insecurity’ and why it matters to health. Soc. Sci. Med. 167 , 1–10 (2016). Hernández, D., Jiang, Y., Carrión, D., Phillips, D. & Aratani, Y. Housing hardship and energy insecurity among native-born and immigrant low-income families with children in the United States. J. Child. Poverty 22 , 77–92 (2016). Thomson, H., Bouzarovski, S. & Snell, C. Rethinking the measurement of energy poverty in Europe: A critical analysis of indicators and data. Indoor Built Environ. 26 , 879–901 (2017). Barrella, R. et al. The dark side of energy poverty: Who is underconsuming in Spain and why? Energy Res. Soc. Sci. 86 , 102428 (2022). Reames, T. G. Targeting energy justice: Exploring spatial, racial/ethnic and socioeconomic disparities in urban residential heating energy efficiency. Energy Policy 97 , 549–558 (2016). Brown, M. A., Soni, A., Lapsa, M. V., Southworth, K. & Cox, M. High energy burden and low-income energy affordability: conclusions from a literature review. Prog. Energy 2 , 042003 (2020). Bird, S. & Hernández, D. Policy options for the split incentive: Increasing energy efficiency for low-income renters. Energy Policy 48 , 506–514 (2012). Hernández, D. Climate Justice Starts at Home: Building Resilient Housing to Reduce Disparate Impacts From Climate Change in Residential Settings. Am. J. Public Health 112 , 66–68 (2022). Chen, C., Greig, J., Nelson, H. & Li, F. When disadvantage collides: The concentrated effects of energy insecurity and internet burdens in the United States. Energy Res. Soc. Sci. 91 , 102713 (2022). Shahmohamadi, P., Che-Ani, A., Maulud, K., Tawil, N. M. & Abdullah, N. The impact of anthropogenic heat on formation of urban heat island and energy consumption balance. Urban Stud. Res. 2011 , 497524 (2011). Cong, S., Nock, D., Qiu, Y. L. & Xing, B. Unveiling hidden energy poverty using the energy equity gap. Nat. Commun. 13 , 1–12 (2022). Huang, L., Nock, D., Cong, S. & Qiu, Y. L. Inequalities across cooling and heating in households: Energy equity gaps. Energy Policy 182 , 113748 (2023). Siegel, E. L. et al. Energy Insecurity Indicators Associated With Increased Odds Of Respiratory, Mental Health, And Cardiovascular Conditions: Study examines energy insecurity and health conditions. Health Aff. (Millwood) 43 , 260–268 (2024). Simes, M., Rahman, T. & Hernández, D. Vigilant conservation: How energy insecure households navigate cumulative and administrative burdens. Energy Res. Soc. Sci. 101 , 103092 (2023). Fikru, M. G. & Gautier, L. The impact of weather variation on energy consumption in residential houses. Appl. Energy 144 , 19–30 (2015). Justin M Feldman, Pamela D Waterman, Brent A Coull, & Nancy Krieger. Spatial social polarisation: using the Index of Concentration at the Extremes jointly for income and race/ethnicity to analyse risk of hypertension. J. Epidemiol. Community Health 69 , 1199 (2015). U.S. Health and Human Services. Health Center Program GeoCare Navigator, Zip Code to ZCTA Crosswalk. (2024). Anderson, G. B., Bell, M. L. & Peng, R. D. Methods to calculate the heat index as an exposure metric in environmental health research. Environ. Health Perspect. 121 , 1111–1119 (2013). Microsoft. Microsoft/USBuildingFootprints. (2019). Eicher, C. L. & Brewer, C. A. Dasymetric Mapping and Areal Interpolation: Implementation and Evaluation. Cartogr. Geogr. Inf. Sci. 28 , 125–138 (2001). Swanwick, R. H. et al. Dasymetric population mapping based on US census data and 30-m gridded estimates of impervious surface. Sci. Data 9 , 523 (2022). Falcone, J. A. U.S. national categorical mapping of building heights by block group from Shuttle Radar Topography Mission data. US Geological Survey - ScienceBase https://doi.org/10.5066/F7W09416 (2016). Geraci, M. & Bottai, M. Linear quantile mixed models. Stat. Comput. 24 , 461–479 (2014). Bates, D., Mächler, M., Bolker, B. & Walker, S. Fitting Linear Mixed-Effects Models Using lme4 . J. Stat. Softw. 67 , (2015). Landau, W. The targets R package: a dynamic Make-like function-oriented pipeline toolkit for reproducibility and high-performance computing. J. Open Source Softw. 6 , 2959 (2021). Harlan, S. L., Declet-Barreto, J. H., Stefanov, W. L. & Petitti, D. B. Neighborhood Effects on Heat Deaths: Social and Environmental Predictors of Vulnerability in Maricopa County, Arizona. Environ. Health Perspect. 121 , 197–204 (2013). Wang, C. et al. Spatial Modeling and Analysis of Heat-Related Morbidity in Maricopa County, Arizona. J. Urban Health 98 , 344–361 (2021). Putnam, H. et al. It’s not the heat, it’s the vulnerability: attribution of the 2016 spike in heat-associated deaths in Maricopa County, Arizona. Environ. Res. Lett. 13 , 094022 (2018). Hernández, D. & Bird, S. Energy burden and the need for integrated low‐income housing and energy policy. Poverty Public Policy 2 , 5–25 (2010). Drehobl, A. & Ross, L. Lifting the high energy burden in America’s largest cities: How energy efficiency can improve low income and underserved communities. (2016). Additional Declarations There is NO Competing Interest. Supplementary Files SupplementaryMaterials20250212.docx Supplementary Information Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6184819","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":446966355,"identity":"3b838afa-2106-4d0a-ac1d-5f82b0f990ff","order_by":0,"name":"Daniel Carrión","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA/klEQVRIiWNgGAWjYDCCAyBUAGfbMDCwg5hshLQYwNlpDAzMRGhhgGkBgsOEtfAdP3vwwAcDBnn+2WcfHvi443xifzPzAYYPZYdxapE8k5dwcIYBg+GMc+kGB2eeuZ044zBbAuOMc7i1GBzIMTjMY8CQwHCGjeEwb9vtxAYgl5m3DY+W828MDv8BapGHaDmXOP8w/wfmv/i03ADaAvR+ggFEy4HEDYd5GJgZ8WiRvPHG4GCPgYThRqCWgzPbko03HmYDipxLx6mF73yO8YcfFTbycmfYmD98bLOTnXe8+eGDH2XWOLVAgQSc5djAAI0sooE9SapHwSgYBaNgRAAA5oFcmCD7KYEAAAAASUVORK5CYII=","orcid":"https://orcid.org/0000-0001-6284-1508","institution":"Yale School of Public Health","correspondingAuthor":true,"prefix":"","firstName":"Daniel","middleName":"","lastName":"Carrión","suffix":""},{"id":446966356,"identity":"44b916be-4339-4302-843c-0c88509c1ff1","order_by":1,"name":"Xuezhixing Zhang","email":"","orcid":"","institution":"Duke-NUS Medical School","correspondingAuthor":false,"prefix":"","firstName":"Xuezhixing","middleName":"","lastName":"Zhang","suffix":""},{"id":446966357,"identity":"74ea311c-29fd-42df-bfdb-9c7f35a858ce","order_by":2,"name":"Anna Stouffer","email":"","orcid":"","institution":"[email protected]","correspondingAuthor":false,"prefix":"","firstName":"Anna","middleName":"","lastName":"Stouffer","suffix":""},{"id":446966358,"identity":"de124593-b23a-4665-a8d8-e2e044226b6b","order_by":3,"name":"Diana Hernández","email":"","orcid":"","institution":"Columbia University","correspondingAuthor":false,"prefix":"","firstName":"Diana","middleName":"","lastName":"Hernández","suffix":""},{"id":446966359,"identity":"bd937817-93a1-494b-82b6-6e6c5b9b721f","order_by":4,"name":"Isabel Shargo","email":"","orcid":"","institution":"Green and Healthy Homes Initiative","correspondingAuthor":false,"prefix":"","firstName":"Isabel","middleName":"","lastName":"Shargo","suffix":""},{"id":446966360,"identity":"b9053fb7-c6bf-4bb7-b051-05b49bc84133","order_by":5,"name":"Allan Just","email":"","orcid":"","institution":"Brown University","correspondingAuthor":false,"prefix":"","firstName":"Allan","middleName":"","lastName":"Just","suffix":""},{"id":446966361,"identity":"eb316c36-2c94-4a47-b046-48667b7750a3","order_by":6,"name":"Brendan Brown","email":"","orcid":"","institution":"Green and Healthy Homes Initiative","correspondingAuthor":false,"prefix":"","firstName":"Brendan","middleName":"","lastName":"Brown","suffix":""}],"badges":[],"createdAt":"2025-03-08 15:10:12","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6184819/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6184819/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":81265623,"identity":"002b6829-3457-42dc-af33-1503e6901b4a","added_by":"auto","created_at":"2025-04-24 07:26:14","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":418916,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eAverage energy use per month by income groups, 2016 to 2020. \u003c/strong\u003eAverage energy consumption per household is defined as the average ZCTA level energy consumption / total number of households within this income group.\u003c/p\u003e","description":"","filename":"Fig1.png","url":"https://assets-eu.researchsquare.com/files/rs-6184819/v1/6becc75648f02f92253c8fce.png"},{"id":81266098,"identity":"7a63ac85-afb2-40f8-abf5-09d9ed048902","added_by":"auto","created_at":"2025-04-24 07:34:14","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":280056,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eHeat index cooling degree days by climate region and average energy utilization.\u003c/strong\u003e A) Scatterplots of observations from the highest (\u0026gt;$90,000; green) and lowest (≤$50,000; red) income categories for each month of the dataset from two example ZCTAs per International Energy Conservation Code climate region, chosen by having the median value in July 2018, the center point of the dataset. Lines represent median quantile regression smooths. B) Histograms of heat index cooling days across the entire dataset (2016-2020) by climate region.\u003c/p\u003e","description":"","filename":"Fig2.png","url":"https://assets-eu.researchsquare.com/files/rs-6184819/v1/910422ac4052f4ef5d68d33e.png"},{"id":81267716,"identity":"a07924b2-0eac-404d-95be-0fcc52897ce2","added_by":"auto","created_at":"2025-04-24 07:50:14","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":141216,"visible":true,"origin":"","legend":"\u003cp\u003eFlowchart of the\u003cstrong\u003e \u003c/strong\u003eadjusted dasymetric reapportionment of electricity data from community to ZCTA.\u003c/p\u003e","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-6184819/v1/8ad893242c81b9f489f10fc2.png"},{"id":87572054,"identity":"10b8f3be-68ea-4263-b978-0fb1e1a60662","added_by":"auto","created_at":"2025-07-25 10:53:57","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1796262,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6184819/v1/884d79de-f315-40de-bef1-a7bd3fb61956.pdf"},{"id":81265622,"identity":"cd07792a-4474-4358-b731-952c302f5dbb","added_by":"auto","created_at":"2025-04-24 07:26:14","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":27363,"visible":true,"origin":"","legend":"Supplementary Information","description":"","filename":"SupplementaryMaterials20250212.docx","url":"https://assets-eu.researchsquare.com/files/rs-6184819/v1/da2f08c3a090ffcd0ea7a2c5.docx"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e Competing Interest.","formattedTitle":"Summertime temperatures, social disadvantage, and energy rationing: Patterns of energy insecurity from population-level historical electricity data in New York State","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eOver the past two decades, rising temperatures have driven an increase in global energy demand for cooling.\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e In New York State (NYS) in particular, the number of cooling degree days are expected to increase, representing more days when cooling is needed and a higher intensity of space-cooling demand of NYS homes relative to previous years.\u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e Energy use for space cooling is critical to prevent life-threatening heat stroke, improving cognition and development, and maintaining a number of other social and public health protective factors.\u003csup\u003e\u003cspan additionalcitationids=\"CR4 CR5\" citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u003c/sup\u003e However, as energy demand and costs rise, so does the risk for energy insecurity.\u003csup\u003e\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e,\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e In order to address this growing issue, justice-oriented solutions that address the uneven distribution of temperatures and energy insecurity are needed.\u003csup\u003e\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e \u003cp\u003eEnergy insecurity is a phenomenon that represents the interplay of the physical status of housing, household energy expenditures and finances, and related coping strategies.\u003csup\u003e\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u003c/sup\u003e The phenomenon disproportionately burdens low-income, ethno-racial minoritized groups, and renting individuals. Most households in the United States near or below the federal poverty line are energy insecure likely due to higher energy burden, as a greater proportion of their income is allocated towards energy costs and are more likely to live in dilapidated/older housing.\u003csup\u003e\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u003c/sup\u003e One hallmark of behavioral energy insecurity is the notion of energy rationing, whereby energy-insecure families employ strategies such as using appliances more sparingly in order to decrease energy cost.\u003csup\u003e\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e,\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e,\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e In general this behavior is referred to as the \u0026lsquo;heat or eat\u0026rsquo; dilemma, however, perhaps more fittingly for summertime temperatures, we refer to it as the \u0026lsquo;heat stroke or go broke\u0026rsquo; dilemma.\u003c/p\u003e \u003cp\u003ePrevious research has primarily examined energy rationing through individual-level interviews and household surveys, less is known about how these behavioral patterns manifest at the community level. Energy rationing behavior, though enacted at the household level, may be collectively shaped by neighborhood-level social and economic conditions. Understanding these spatial patterns is crucial because they may reveal systematic disparities in how communities adapt to increasing cooling demands, which is analogous to past work on community-level wintertime energy use intensity for heating.\u003csup\u003e\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e Research is needed to advance this understanding for summer months by examining evidence of energy rationing at the population level, moving beyond individual household behaviors to identify how structural disadvantage may influence collective energy use patterns across communities.\u003c/p\u003e \u003cp\u003eThere are a series of individual and structural factors that increase the likelihood of experiencing energy insecurity. Economic elements of energy insecurity can involve having low- or fixed-income, which can be compounded by physical elements of energy insecurity such as living in poor or dilapidated housing conditions, and having older appliances.\u003csup\u003e\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u003c/sup\u003e Additionally, renters are more likely to be energy insecure because, energy bills typically fall on renters but landlords have little incentive to replace older energy appliances; thus renters have little agency in improving the energy bills they pay, i.e. the split-incentive problem.\u003csup\u003e\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u003c/sup\u003e However, new appliances and energy efficiency upgrades are oftentimes still inaccessible even for low-income and minoritized homeowners.\u003csup\u003e\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u003c/sup\u003e On the structural level, low-income and minority residents are more likely to live in segregated neighborhoods, where patterns of social disadvantage such as redlining and concentrated poverty have prevented investment in efficient and updated housing and energy infrastructure.\u003csup\u003e\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u003c/sup\u003e Counties with lower average income, more minority communities, and more households headed by women of color are more likely to be energy insecure.\u003csup\u003e\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u003c/sup\u003e Additionally, areas with higher densities of concrete buildings and less green space experience higher temperatures because of the urban heat island effect, which can contribute to greater energy demand for low-income disadvantaged communities and greater health risks for the most socially vulnerable.\u003csup\u003e\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e \u003cp\u003eMost examinations of coping or behavioral energy insecurity have used individual-level interview, survey, and/or energy use data to understand the phenomenon, particularly those examining energy rationing.\u003csup\u003e\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e,\u003cspan additionalcitationids=\"CR21 CR22\" citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u003c/sup\u003e While individual factors influence energy rationing, structural disadvantage may also lead to behavioral energy rationing being observed at the area level. Increasingly there are population-level measures of energy utilization yet, there are few studies that have examined the ability to infer collective energy rationing behavior from population-level data. In this study we examined two forms of structural disadvantage: area-level income and area-level racialized economic segregation. We first examined differences in the slopes between energy utilization and cooling degree days by area income across NYS. We hypothesized that neighborhoods earning higher incomes would use more energy on similarly hot summer days relative to neighborhoods earning lower incomes. Second, we assessed differences in the slopes between energy utilization and cooling degree days by an area-level measure of racialized economic segregation, hypothesizing that neighborhoods with higher incomes and lower percentage of minoritized racial groups would use more energy relative to neighborhoods with lower incomes and higher percentage of minority race. This study adds to the literature by providing the first analysis of energy rationing in NYS, utilizing more accurate finer-scale data relative to other pre-existing energy literature, and more broadly exploring the structural patterns that influence how vulnerable communities manage energy insecurity.\u003c/p\u003e"},{"header":"2. Results","content":"\u003cp\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003eIn this analysis, we utilized residential energy data from the New York State-managed Utility Energy Registry (UER) to identify energy equity gaps at the zip code tabulation area (ZCTA) level. The analysis was limited to the warm season (defined as May through September), when residents are most likely to have increased energy demand for household cooling.\u003csup\u003e\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e\u003c/sup\u003e We first extracted and processed monthly energy data as well as environmental data to construct a daily heat index which we aggregated to a monthly sum of heat index cooling degree days (HICDD) from 2016 to 2020 for 1,794 ZCTAs within New York state. Then we employed linear quantile mixed-effect models (LQMMs) to identify if the slopes of the energy use (kWh) by HICDD association varied by interaction terms for various income groups at the median (\u003cb\u003eSuppl 1\u003c/b\u003e). Furthermore, we also investigated the associations between energy use and racialized economic segregation for various ethno-racial groups by constructing an index of concentration at the extremes (ICE) for each group and fit LQMMs for each group\u0026rsquo;s ICE.\u003csup\u003e\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e \u003c/div\u003e \u003c/p\u003e \u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003e2.1 Descriptive analyses\u003c/h2\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e displays the average energy use per household for all four income groups {\u0026lt;\u003cspan\u003e$\u003c/span\u003e50,000; \u003cspan\u003e$\u003c/span\u003e50,000 to \u003cspan\u003e$\u003c/span\u003e69,999; \u003cspan\u003e$\u003c/span\u003e70,000 to \u003cspan\u003e$\u003c/span\u003e89,999; and \u0026ge;\u003cspan\u003e$\u003c/span\u003e90,000} during the warm seasons from 2016 to 2020. In ZCTAs with higher median incomes, there was consistently greater energy consumption. An increase of energy consumption in 2020 was observed across all income groups, which might suggest systematically higher energy use during the pandemic's quarantine. After excluding observations with missing energy consumption data or income information, 1,449 ZCTAs with a total of 27,916 observations (in kWh per ZCTA-month) were included in our subsequent analyses (\u003cb\u003eSupplemental Tables\u0026nbsp;1 and 2\u003c/b\u003e). We also visualized the energy use data in relation to HICDDs by NYS climate region (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). Figure\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eB demonstrates the distribution of HICDDs observations by ZCTA-months, the unit of analysis, ranging from 13.4 to 929.5 with an interquartile range (IQR) of 268.51. Figure\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA visualizes example ZCTAs for the highest and lowest income groups.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e2.2 ZCTA-Level Energy Equity Gap by Income Levels\u003c/h2\u003e \u003cp\u003eOur models revealed a significant difference in energy use between the lowest income group and the highest income groups (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). Specifically, in the median quantile regression, an IQR increase in HICDD is associated with 9.96 (95% CI: [-8.85, 28.77]), 26.24 (95% CI: [1.73, 50.75]), and 55.58 (95% CI: [12.24, 98.91]) kWh more energy consumption per household in the \u0026ldquo;\u003cspan\u003e$\u003c/span\u003e50,000 to \u003cspan\u003e$\u003c/span\u003e69,999\u0026rdquo;, \u0026ldquo;\u003cspan\u003e$\u003c/span\u003e70,000 to \u003cspan\u003e$\u003c/span\u003e89,999\u0026rdquo;, and \u0026ldquo;\u0026ge;\u003cspan\u003e$\u003c/span\u003e90,000\u0026rdquo; income groups, compared to households with income less than \u003cspan\u003e$\u003c/span\u003e50,000, respectively.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003e2.3 ZCTA-Level Racial Disparities in Energy Equity\u003c/h2\u003e \u003cp\u003eIn addition to disparities among different income groups, we also fit LQMMs using the ICE measure of racialized economic segregation (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). The ICE score reflects the proportional difference between the most privileged group (high-income predominantly white group) and the marginalized group (low-income ethno-racially minoritized groups). A positive ICE score indicates the ZCTA area is dominated by high income white people, compared to low income minoritized groups, and vice versa. Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e displays estimates for each ethno-racial ICE combination at the median quantile. Specifically, in the median quantile regression, an IQR increase in HICDD resulted in 166.52 (95% CI: [95.16, 237.88]), 130.66 (95% CI: [35.98, 225.35]) and 238.39 (95% CI: [165.44, 311.35]) kWh of additional energy consumption per household for the most privileged ZCTA areas (ICE\u0026thinsp;=\u0026thinsp;1) for the Black, Hispanic/Latino, and Asian groups respectively.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003e \u003cb\u003eModel estimates for energy use by income groups.\u003c/b\u003e Models were fitted using a linear quantile mixed effects model with fixed effects of average household size and scaled number of households and random intercepts for combinations of county, month and year. Each row presents the model estimate for the interaction term between income groups and an interquartile range increase of heat index cooling degree days (HICDD). Each point estimate represents the additional energy consumption of ZCTA areas from the corresponding income group, compared to ZCTA areas with median income less than \u003cspan\u003e$\u003c/span\u003e50,000.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"6\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eVariable\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"5\" nameend=\"c6\" namest=\"c2\"\u003e \u003cp\u003eModel Estimates\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMain Model\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eSensitivity Model I\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eSensitivity\u003c/p\u003e \u003cp\u003eModel II\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eSensitivity Model III\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eSensitivity Model IV\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHICDD:\u003cspan\u003e$\u003c/span\u003e50,000 to 69,999\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e9.96\u003c/p\u003e \u003cp\u003e[-8.85, 28.77]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e7.26\u003c/p\u003e \u003cp\u003e[-14.89, 29.40]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e32.94**\u003c/p\u003e \u003cp\u003e[10.13, 55.76]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e22.21***\u003c/p\u003e \u003cp\u003e[10.01, 34.41]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e570.82\u003c/p\u003e \u003cp\u003e[-128.38, 1270.03]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHICDD:\u003cspan\u003e$\u003c/span\u003e70,000 to 89,999\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e26.24*\u003c/p\u003e \u003cp\u003e[1.73, 50.75]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e24.22\u003c/p\u003e \u003cp\u003e[-2.25, 50.69]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e34.69*\u003c/p\u003e \u003cp\u003e[3.02, 0.27]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e38.35***\u003c/p\u003e \u003cp\u003e[20.47, 56.23]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e96.20\u003c/p\u003e \u003cp\u003e[-734.91, 927.31]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHICDD:\u0026ge; \u003cspan\u003e$\u003c/span\u003e90,000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e55.58**\u003c/p\u003e \u003cp\u003e[12.24, 98.91]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e50.72*\u003c/p\u003e \u003cp\u003e[9.40, 92.04]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e29.01\u003c/p\u003e \u003cp\u003e[-7.12, 65.13]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e77.23***\u003c/p\u003e \u003cp\u003e[48.28, 106.18]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e50.94\u003c/p\u003e \u003cp\u003e[-779.42, 881.29]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"6\"\u003eNote: *: p\u0026thinsp;\u0026lt;\u0026thinsp;0.05; **: p\u0026thinsp;\u0026lt;\u0026thinsp;0.01; ***: p\u0026thinsp;\u0026lt;\u0026thinsp;0.001.\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eMain Model: Median quantile regression with the full imputed dataset (n\u0026thinsp;=\u0026thinsp;27,916).\u003c/p\u003e \u003cp\u003eSensitivity Model I: Median quantile regression with the imputed dataset, excluding year 2020 (n\u0026thinsp;=\u0026thinsp;23,547).\u003c/p\u003e \u003cp\u003eSensitivity Model II: Median quantile regression with the subset of ZCTA-level reported data (n\u0026thinsp;=\u0026thinsp;13,645).\u003c/p\u003e \u003cp\u003eSensitivity Model III: Median quantile regression with the subset of community-level reported data. For demographic information, we use ACS 5-year estimates from the overlapping zip code area (n\u0026thinsp;=\u0026thinsp;26,334).\u003c/p\u003e \u003cp\u003eSensitivity Model IV: Mean regression with the full imputed dataset (n\u0026thinsp;=\u0026thinsp;27,916).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003e \u003cb\u003eModel estimates for energy use with Index of Concentration at the Extremes (ICE) scores.\u003c/b\u003e Each row presents the model estimate for the interaction term between ICE score (scaled per 0 to 1 increase) and heat index cooling degree days (HICDD). Each coefficient represents the energy change of an interquartile range increase of heat index cooling degree days for various ethno-racial groups with the most privileged having an index of concentration at the extremes of 1. The privileged group defined as white people with household incomes of \u003cspan\u003e$\u003c/span\u003e100,000 or more and the disadvantaged group defined as Black, Hispanic/Latino, or Asian people making \u0026lt; \u003cspan\u003e$\u003c/span\u003e25,000.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"6\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eModel\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMain Model\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eSensitivity\u003c/p\u003e \u003cp\u003eModel I\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eSensitivity\u003c/p\u003e \u003cp\u003eModel II\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eSensitivity\u003c/p\u003e \u003cp\u003eModel III\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eSensitivity\u003c/p\u003e \u003cp\u003eModel IV\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBlack\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e166.52***\u003c/p\u003e \u003cp\u003e[95.16, 237.88]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e161.00***\u003c/p\u003e \u003cp\u003e[103.73, 218.26]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e65.63\u003c/p\u003e \u003cp\u003e[-24.48, 155.73]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e234.10***\u003c/p\u003e \u003cp\u003e[159.61, 308.59]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e236.47\u003c/p\u003e \u003cp\u003e[-1218.73, 1691.66]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHispanic/\u003c/p\u003e \u003cp\u003eLatino\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e130.66**\u003c/p\u003e \u003cp\u003e[35.98, 225.35]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e198.85***\u003c/p\u003e \u003cp\u003e[107.34, 290.37]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e71.07\u003c/p\u003e \u003cp\u003e[-37.61, 179.75]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e249.28***\u003c/p\u003e \u003cp\u003e[159.40, 339.16]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e254.75\u003c/p\u003e \u003cp\u003e[-1232.04, 1741.54]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAsian\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e238.39***\u003c/p\u003e \u003cp\u003e[165.44, 311.35]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e233.44***\u003c/p\u003e \u003cp\u003e[160.47, 306.40]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e111.18\u003c/p\u003e \u003cp\u003e[-8.59, 230.94]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e257.24***\u003c/p\u003e \u003cp\u003e[176.91, 337.57]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e203.86\u003c/p\u003e \u003cp\u003e[-1450.27, 1858.00]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"6\"\u003eNote: *: p\u0026thinsp;\u0026lt;\u0026thinsp;0.05; **: p\u0026thinsp;\u0026lt;\u0026thinsp;0.01; ***: p\u0026thinsp;\u0026lt;\u0026thinsp;0.001.\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eMain Model: Median quantile regression with the full imputed dataset (n\u0026thinsp;=\u0026thinsp;27,916).\u003c/p\u003e \u003cp\u003eSensitivity Model I: Median quantile regression with the imputed dataset, excluding year 2020 (n\u0026thinsp;=\u0026thinsp;23,547).\u003c/p\u003e \u003cp\u003eSensitivity Model II: Median quantile regression with the subset of ZCTA-level reported data (n\u0026thinsp;=\u0026thinsp;13,645).\u003c/p\u003e \u003cp\u003eSensitivity Model III: Median quantile regression with the subset of community-level reported data. For demographic information, we use ACS 5-year estimates from the overlapping zip code area (n\u0026thinsp;=\u0026thinsp;26,334).\u003c/p\u003e \u003cp\u003eSensitivity Model IV: Mean regression with the full imputed dataset (n\u0026thinsp;=\u0026thinsp;27,916).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003e2.4 Sensitivity Analyses\u003c/h2\u003e \u003cp\u003eOur sensitivity analyses assessed the consistency of our models under varying conditions/assumptions. In the models assessing differences by income group, we found similar patterns for Sensitivity Model I and Model III although Sensitivity Model I showed confidence intervals that crossed 0, likely due to reduced power from fewer observations (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). Sensitivity Model II shows a higher energy utilization for ZCTAs with median income of 50,000 to 69,999 relative to the lowest income group. As for the ICE analysis, Sensitivity Models I and III were similarly consistent with the main model (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). Sensitivity Model II showed overall attenuated and non-significant results. In both analyses, Sensitivity Model IV was fit using a linear mixed effects regression (Tables\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e and \u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). These regressions resulted in insignificant estimates with wide confidence intervals, indicating mean regressions are likely heavily affected by extreme outliers.\u003c/p\u003e \u003c/div\u003e"},{"header":"3 Discussion","content":"\u003cp\u003eThis study examined patterns of energy insecurity and rationing in New York State using population-level electricity data from 2016\u0026ndash;2020. Using monthly zip code-level excess outdoor heat (HICDDs) and residential energy consumption data in the warm season, we used LQMMs to identify differences in the heat versus electricity slopes between ZCTAs based on income and ICE, showing in both cases that more advantaged/affluent zip codes exhibit greater increases in energy consumption as temperatures rose. These findings suggest that lower-income and minoritized communities may be engaging in energy rationing behaviors during periods of high cooling demand, experiencing higher indoor temperature exposures, and risking resident health.\u003csup\u003e\u003cspan additionalcitationids=\"CR36\" citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e \u003cp\u003eOur research advances an emerging body of work on energy limiting behavior and structural drivers of temperature-based energy insecurity.\u003csup\u003e\u003cspan additionalcitationids=\"CR21\" citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u003c/sup\u003e Like other studies, our work aims to better understand and measure energy use and insecurity among low-income and minoritized communities with regard to space cooling. Other studies have found, for example, evidence of temperature-based energy rationing or \u0026ldquo;energy equity gaps\u0026rdquo; in Arizona and northern Illinois.\u003csup\u003e\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e,\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e Those studies leveraged daily person-level electricity and demographic data directly from utility providers. However, our study, identifies such patterns at the population-level, using substantially coarser data, and associations with population-level drivers of structural disadvantage. This is quite notable given the potential for aggregation bias, i.e. that the pattern remains pronounced from the individual to population level. Relatedly, there are reasons to believe that no effect would be observed or even the opposite effects at the population level, including that income and segregation are also associated with poor housing quality and older energy inefficient appliances.\u003csup\u003e\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e,\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e\u003c/sup\u003e The results of these analyses are particularly encouraging because it is unlikely that every utility provider will offer access to their datasets to researchers, but more states are planning to make aggregate energy data available. This access to energy data offers the opportunity to replicate this energy rationing analyses in other states or regions of the country.\u003c/p\u003e \u003cp\u003eThis study has several important limitations. First, our reliance on administrative and aggregated data, while allowing for the desired population-level analysis, risks ecological fallacy. Inferences about individual or household behaviors drawn from these aggregate data may not accurately reflect the diverse experiences within each ZCTA. Second, we encountered substantial missing data at the zip code level, necessitating the use of an advanced imputation approach. While this method allowed us to maintain a comprehensive dataset, it may introduce biases, particularly if data are not missing at random. This bolsters the importance of our sensitivity analyses that restrict to the true dataset. Third, our dataset completely lacked information for Long Island, a significant and populous region of New York State. This omission limits the generalizability of our findings to the entire state and may skew our understanding of statewide energy consumption patterns. These limitations underscore the need for cautious interpretation of our results and highlight areas for future research using more complete data sources. This speaks to the importance of mandated reporting and making datasets publicly available statewide.\u003c/p\u003e \u003cp\u003eOur study offers several notable strengths that enhance its contribution to energy insecurity research. First, this study covers the majority of NYS, and thus a substantially larger and diverse population and climatological distinct region compared to past studies. Second, we employ a robust statistical methodology using LQMMs, which allows for a nuanced examination of energy consumption patterns at the median rather than the mean of the distribution. This approach revealed insights that might be overlooked by conventional regression techniques. Third, we analyzed data longitudinally from 2016 to 2020, which provided valuable heterogeneity in electricity use and temperature for insights on trends over time. Fourth, rather than using data from a local weather station, we utilized Earth science products that reconstruct daily temperature at a fine 1-km resolution, offering a more accurate representation of dynamic temperature profiles. Fifth, our study's replicability is enhanced by using publicly available data sources and a publicly available code base. Finally, we conduct multiple sensitivity analyses, which strengthen the credibility of our findings by demonstrating their robustness under different assumptions and data subsets.\u003c/p\u003e"},{"header":"4 Conclusion","content":"\u003cp\u003eThis study provides compelling evidence of energy rationing behavior among lower-income and minoritized and marginalized communities in NYS, revealing important patterns of energy insecurity at the population level. By leveraging advanced statistical approaches and a comprehensive dataset covering a five-year period, we have demonstrated that socioeconomic and racial disparities are strongly associated with energy consumption patterns in response to outdoor heat. These findings have profound implications for energy justice and public health, particularly as climate change intensifies the frequency and severity of extreme heat events. The observed disparities in energy consumption suggest that vulnerable populations may be exposing themselves to higher indoor temperatures, potentially increasing their risk of heat-related illnesses. This research underscores the urgent need for targeted policy interventions to address energy insecurity and improve thermal comfort in disadvantaged communities. Future energy assistance programs and cooling strategies should prioritize these vulnerable populations, considering both income levels and racial composition of neighborhoods. Moreover, this study highlights the value of using population-level data to identify and address structural inequalities in energy consumption, providing a model for similar analyses in other regions. As we navigate the challenges of climate change and energy transition, ensuring equitable access to energy for cooling must be a central consideration in policy development to protect the health and well-being of all communities.\u003c/p\u003e"},{"header":"5. Methodology","content":"\u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003e5.1 Study Population and Electricity Data\u003c/h2\u003e \u003cp\u003eData were downloaded from the Utility Energy Registry, a project of the New York State Energy Research and Development Authority. The data are divided into energy types (i.e. electricity and natural gas) and account types (e.g. residential and commercial) and the data are made available for zip codes, counties, and communities across NYS from 2016\u0026ndash;2020, with incomplete spatial coverage for each unit. We restricted the subset data to residential and electricity given our interest in home space cooling and chose zip codes as our unit of analysis because they are the smallest available unit that have approximations for Census data via ZCTAs. Zip codes were aggregated into ZCTAs using a crosswalk file via the U.S. Department of Health and Human Services GeoCare Navigator.\u003csup\u003e\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003e5.2 Temperature Data\u003c/h2\u003e \u003cp\u003eCooling degree days are often used to derive a measure of the demand for space cooling, but only considers dry bulb temperature. The heat index more closely approximates the human thermoregulatory strain, so we chose to calculate a HICDD measure. HICDD data were derived using maximum and minimum temperatures and vapor pressure data from NASA\u0026rsquo;s Daymet dataset, following the National Weather Service algorithm on the Fahrenheit scale.\u003csup\u003e\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003e5.3 Demographic and Buildings Data\u003c/h2\u003e \u003cp\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003eZCTA demographic data were accessed from the American Community Survey 5-year\u003c/p\u003e \u003cp\u003eestimates. We specifically downloaded race/ethnicity, average household size (persons), the number of households, the median income, and related covariates. Buildings data were accessed from the Microsoft Buildings dataset. This dataset is the result of satellite remote sensing via Bing Imagery and machine learning to create building footprints polygons throughout the U.S.\u003csup\u003e\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e \u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003e5.4 Missing data\u003c/h2\u003e \u003cp\u003e \u003cdiv class=\"BlockQuote\"\u003e \u003cp\u003eIn instances where zip code-level energy data were missing, we utilized an adapted dasymetric method and specifically developed an adjusted version for imputation, summarized by Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e.\u003csup\u003e\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e\u003c/sup\u003e In summary, buildings were classified as residential based on 30-m population estimates from the updatable gridded lightweight impervious (UGLI) dataset.\u003csup\u003e\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e\u003c/sup\u003e We then approximated the volume of each building using the U.S. national categorical mapping of building heights, based on building areas and height categories.\u003csup\u003e\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e\u003c/sup\u003e The average energy consumption per unit of effective volume was calculated using community-level energy data (typically geographically smaller than ZCTAs) and the total building volumes were calculated based on the formula: building area \u0026times; occupied ratio \u0026times; height coefficient. The occupied ratio was defined as the ratio of occupied units to all units within each block group; height coefficients represented values for each height category derived from the height data. The best height coefficients were obtained using a grid search to find values that most accurately predicted the ZCTA level energy data. In our analysis, buildings were assigned with height categories \u0026ldquo;low\u0026rdquo;, \u0026ldquo;low-medium\u0026rdquo;, \u0026ldquo;medium\u0026rdquo;, \u0026ldquo;medium-high\u0026rdquo;, \u0026ldquo;high\u0026rdquo; and \u0026ldquo;very high\u0026rdquo; based on its corresponding block group height data. The best coefficients for these categories were selected from {1.5,2.5,3.5,4.0,6.0,10.0}. Finally, we aggregated all building-level energy data into ZCTA level for imputation. Information on the number of ZCTAs imputed (and corresponding population data) can be found in \u003cb\u003eSupplemental Table\u0026nbsp;1\u003c/b\u003e. In areas of overlap, we calculate a mean absolute error of 547.6 MWh which, for context, is substantially lower than standard deviation of the true data (\u003cb\u003eSupplemental Tables\u0026nbsp;2 and 3\u003c/b\u003e).\u003c/p\u003e \u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003e5.5 Statistical Methods\u003c/h2\u003e \u003cp\u003eThe data demonstrated an irregular residual structure, so our main analyses modeled the conditional median using linear quantile mixed models in order to assess different slopes for the heat-electricity use relationships by 1) median income and 2) the racialized economic segregation measure, ICE.\u003c/p\u003e \u003cp\u003eTo identify potential differences in monthly energy utilization by income levels, we fit an LQMM expressed by Eq.\u0026nbsp;1:\u003cdiv id=\"Equa\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equa\" name=\"EquationSource\"\u003e\n$$\\:\\begin{array}{c}{Energy\\:per\\:Household}_{\\tau\\:}=\\:{\\beta\\:}_{0}+{\\beta\\:}_{1}*HICDD+{\\beta\\:}_{2}*\\:Income+\\\\\\:{\\beta\\:}_{3}*\\:HICDD*Income+{\\beta\\:}_{4}*\\:Av{g}_{Househol{d}_{size}}+{b}_{Count{y}_{time}}\\#\\left(1\\right)\\end{array}$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eWhere τ specifies the quantile in our regression model. We divided the median household income into four levels based on the thresholds of \u003cspan\u003e$\u003c/span\u003e50,000, \u003cspan\u003e$\u003c/span\u003e70,000 and \u003cspan\u003e$\u003c/span\u003e90,000. Since temperatures can fluctuate across different time and geographical positions, leading to varied energy use, we included random intercepts for the combinations of county (as defined by the ZCTA centroid), month and year to capture the spatial and temporal information for each ZCTA area. To explore the typical relationship between energy use and income levels and avoid the effect of extreme outliers, we fit the model at the median quantile.\u003c/p\u003e \u003cp\u003eTo model differences in slopes by ICE measures, we constructed ICE measures combining ethno-racial and income data [2]. Specifically, we calculated the ICE score for Black, Hispanic/Latino, and Asian households relative to Non-Hispanic white households using Eq.\u0026nbsp;2:\u003cdiv id=\"Equb\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equb\" name=\"EquationSource\"\u003e\n$$\\:\\begin{array}{c}{ICE}_{Race\\:X}\\:=\\frac{\\:Non\\:hispanic\\:white\\:households\\:over\\:US\\text{\\$}\\text{100,000}\\:-\\:Race\\:X\\:\\:alone\\:households\\:under\\:\\text{\\$}\\text{25,000}}{Total\\:households}\\:\\:\\:\\left(2\\right)\\end{array}\\:$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eThe ICE measure is a continuous score ranging from \u0026minus;\u0026thinsp;1 (most deprived) to 1 (most privileged), allowing comparisons along a continuum of economic-racial polarization. We used the same LQMM modeling strategy (Eq.\u0026nbsp;1) but substituted the income categories with the ICE measures in the model expressed by \u003cb\u003eEq.\u0026nbsp;2\u003c/b\u003e and fit LQMMs for each race at median.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003e5.6 Sensitivity Analyses\u003c/h2\u003e \u003cp\u003eWe conducted a series of sensitivity analyses to examine the robustness of our findings. In the first sensitivity model, we removed 2020 from our dataset given the COVID-19 pandemic and lockdowns that likely influenced home energy utilization. Secondly, given that we imputed 51% of the analytical dataset, we reran the analysis restricting to the subset of zip code-level observations from the Utility Energy Registry (\u003cb\u003eSupplemental Table\u0026nbsp;1\u003c/b\u003e). Third, we conducted the same analyses using only the community-level data given that zip code and community observations cover relatively distinct geographies across NYS. Finally, we modeled the conditional mean using a linear mixed effects regression using the fully imputed dataset.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003e5.7 Computational Environment and Reproducibility\u003c/h2\u003e \u003cp\u003eAll analyses were conducted in R version 4.3 and analytic code can be accessed at \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://github.com/CHEE-collaborative/TEJM_Analysis\u003c/span\u003e\u003cspan address=\"https://github.com/CHEE-collaborative/TEJM_Analysis\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. LQMM and linear mixed effects models were fit using the \u003cem\u003elqmm\u003c/em\u003e and \u003cem\u003elme4\u003c/em\u003e packages, respectively, in R.\u003csup\u003e\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e,\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e\u003c/sup\u003e Analyses were coded using the \u003cem\u003etargets\u003c/em\u003e package and tested across several platforms to enhance reproducibility.\u003csup\u003e\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e \u003c/div\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgements:\u0026nbsp;\u003c/strong\u003eThis work was supported by a U.S. National Aeronautics and Space Administration grant 80NSSC22K1666.\u0026nbsp;\u0026nbsp;\u003c/p\u003e\n"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eScoccimarro, E. \u003cem\u003eet al.\u003c/em\u003e Country-level energy demand for cooling has increased over the past two decades. \u003cem\u003eCommun. Earth Environ.\u003c/em\u003e \u003cstrong\u003e4\u003c/strong\u003e, 208 (2023).\u003c/li\u003e\n\u003cli\u003eMeier, S. \u003cem\u003eet al.\u003c/em\u003e Chapter 6: Energy. \u003cem\u003eN. Y. State Clim. Impacts Assess. Interim Version Public Release\u003c/em\u003e (2024).\u003c/li\u003e\n\u003cli\u003eGauer, R. \u0026amp; Meyers, B. K. Heat-related illnesses. \u003cem\u003eAm. Fam. Physician\u003c/em\u003e \u003cstrong\u003e99\u003c/strong\u003e, 482\u0026ndash;489 (2019).\u003c/li\u003e\n\u003cli\u003eHeyes, A. \u0026amp; Saberian, S. Temperature and decisions: evidence from 207,000 court cases. \u003cem\u003eAm. Econ. J. Appl. Econ.\u003c/em\u003e \u003cstrong\u003e11\u003c/strong\u003e, 238\u0026ndash;265 (2019).\u003c/li\u003e\n\u003cli\u003ePark, R. J., Goodman, J., Hurwitz, M. \u0026amp; Smith, J. Heat and learning. \u003cem\u003eAm. Econ. J. Econ. Policy\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, 306\u0026ndash;39 (2020).\u003c/li\u003e\n\u003cli\u003eParks, R. M. \u003cem\u003eet al.\u003c/em\u003e Anomalously warm temperatures are associated with increased injury deaths. \u003cem\u003eNat. Med.\u003c/em\u003e \u003cstrong\u003e26\u003c/strong\u003e, 65\u0026ndash;70 (2020).\u003c/li\u003e\n\u003cli\u003eLongden, T. \u003cem\u003eet al.\u003c/em\u003e Energy insecurity during temperature extremes in remote Australia. \u003cem\u003eNat. Energy\u003c/em\u003e \u003cstrong\u003e7\u003c/strong\u003e, 43\u0026ndash;54 (2021).\u003c/li\u003e\n\u003cli\u003eCarri\u0026oacute;n, D. \u0026amp; Hern\u0026aacute;ndez, D. Heatwaves and hardship: Shortcomings and solutions for enhancing the low income home energy assistance program to mitigate extreme heat and energy insecurity. \u003cem\u003eElectr. J.\u003c/em\u003e \u003cstrong\u003e37\u003c/strong\u003e, 107440 (2024).\u003c/li\u003e\n\u003cli\u003eCarri\u0026oacute;n, D., Rush, J., Colicino, E. \u0026amp; Just, A. C. Residential segregation and summertime air temperature across 13 northeastern U.S. states: potential implications for energy burden. \u003cem\u003eEnviron. Res. Lett.\u003c/em\u003e \u003cstrong\u003e19\u003c/strong\u003e, 084005 (2024).\u003c/li\u003e\n\u003cli\u003eHern\u0026aacute;ndez, D. Understanding \u0026lsquo;energy insecurity\u0026rsquo; and why it matters to health. \u003cem\u003eSoc. Sci. Med.\u003c/em\u003e \u003cstrong\u003e167\u003c/strong\u003e, 1\u0026ndash;10 (2016).\u003c/li\u003e\n\u003cli\u003eHern\u0026aacute;ndez, D., Jiang, Y., Carri\u0026oacute;n, D., Phillips, D. \u0026amp; Aratani, Y. Housing hardship and energy insecurity among native-born and immigrant low-income families with children in the United States. \u003cem\u003eJ. Child. Poverty\u003c/em\u003e \u003cstrong\u003e22\u003c/strong\u003e, 77\u0026ndash;92 (2016).\u003c/li\u003e\n\u003cli\u003eThomson, H., Bouzarovski, S. \u0026amp; Snell, C. Rethinking the measurement of energy poverty in Europe: A critical analysis of indicators and data. \u003cem\u003eIndoor Built Environ.\u003c/em\u003e \u003cstrong\u003e26\u003c/strong\u003e, 879\u0026ndash;901 (2017).\u003c/li\u003e\n\u003cli\u003eBarrella, R. \u003cem\u003eet al.\u003c/em\u003e The dark side of energy poverty: Who is underconsuming in Spain and why? \u003cem\u003eEnergy Res. Soc. Sci.\u003c/em\u003e \u003cstrong\u003e86\u003c/strong\u003e, 102428 (2022).\u003c/li\u003e\n\u003cli\u003eReames, T. G. Targeting energy justice: Exploring spatial, racial/ethnic and socioeconomic disparities in urban residential heating energy efficiency. \u003cem\u003eEnergy Policy\u003c/em\u003e \u003cstrong\u003e97\u003c/strong\u003e, 549\u0026ndash;558 (2016).\u003c/li\u003e\n\u003cli\u003eBrown, M. A., Soni, A., Lapsa, M. V., Southworth, K. \u0026amp; Cox, M. High energy burden and low-income energy affordability: conclusions from a literature review. \u003cem\u003eProg. Energy\u003c/em\u003e \u003cstrong\u003e2\u003c/strong\u003e, 042003 (2020).\u003c/li\u003e\n\u003cli\u003eBird, S. \u0026amp; Hern\u0026aacute;ndez, D. Policy options for the split incentive: Increasing energy efficiency for low-income renters. \u003cem\u003eEnergy Policy\u003c/em\u003e \u003cstrong\u003e48\u003c/strong\u003e, 506\u0026ndash;514 (2012).\u003c/li\u003e\n\u003cli\u003eHern\u0026aacute;ndez, D. Climate Justice Starts at Home: Building Resilient Housing to Reduce Disparate Impacts From Climate Change in Residential Settings. \u003cem\u003eAm. J. Public Health\u003c/em\u003e \u003cstrong\u003e112\u003c/strong\u003e, 66\u0026ndash;68 (2022).\u003c/li\u003e\n\u003cli\u003eChen, C., Greig, J., Nelson, H. \u0026amp; Li, F. When disadvantage collides: The concentrated effects of energy insecurity and internet burdens in the United States. \u003cem\u003eEnergy Res. Soc. Sci.\u003c/em\u003e \u003cstrong\u003e91\u003c/strong\u003e, 102713 (2022).\u003c/li\u003e\n\u003cli\u003eShahmohamadi, P., Che-Ani, A., Maulud, K., Tawil, N. M. \u0026amp; Abdullah, N. The impact of anthropogenic heat on formation of urban heat island and energy consumption balance. \u003cem\u003eUrban Stud. Res.\u003c/em\u003e \u003cstrong\u003e2011\u003c/strong\u003e, 497524 (2011).\u003c/li\u003e\n\u003cli\u003eCong, S., Nock, D., Qiu, Y. L. \u0026amp; Xing, B. Unveiling hidden energy poverty using the energy equity gap. \u003cem\u003eNat. Commun.\u003c/em\u003e \u003cstrong\u003e13\u003c/strong\u003e, 1\u0026ndash;12 (2022).\u003c/li\u003e\n\u003cli\u003eHuang, L., Nock, D., Cong, S. \u0026amp; Qiu, Y. L. Inequalities across cooling and heating in households: Energy equity gaps. \u003cem\u003eEnergy Policy\u003c/em\u003e \u003cstrong\u003e182\u003c/strong\u003e, 113748 (2023).\u003c/li\u003e\n\u003cli\u003eSiegel, E. L. \u003cem\u003eet al.\u003c/em\u003e Energy Insecurity Indicators Associated With Increased Odds Of Respiratory, Mental Health, And Cardiovascular Conditions: Study examines energy insecurity and health conditions. \u003cem\u003eHealth Aff. (Millwood)\u003c/em\u003e \u003cstrong\u003e43\u003c/strong\u003e, 260\u0026ndash;268 (2024).\u003c/li\u003e\n\u003cli\u003eSimes, M., Rahman, T. \u0026amp; Hern\u0026aacute;ndez, D. Vigilant conservation: How energy insecure households navigate cumulative and administrative burdens. \u003cem\u003eEnergy Res. Soc. Sci.\u003c/em\u003e \u003cstrong\u003e101\u003c/strong\u003e, 103092 (2023).\u003c/li\u003e\n\u003cli\u003eFikru, M. G. \u0026amp; Gautier, L. The impact of weather variation on energy consumption in residential houses. \u003cem\u003eAppl. Energy\u003c/em\u003e \u003cstrong\u003e144\u003c/strong\u003e, 19\u0026ndash;30 (2015).\u003c/li\u003e\n\u003cli\u003eJustin M Feldman, Pamela D Waterman, Brent A Coull, \u0026amp; Nancy Krieger. Spatial social polarisation: using the Index of Concentration at the Extremes jointly for income and race/ethnicity to analyse risk of hypertension. \u003cem\u003eJ. Epidemiol. Community Health\u003c/em\u003e \u003cstrong\u003e69\u003c/strong\u003e, 1199 (2015).\u003c/li\u003e\n\u003cli\u003eU.S. Health and Human Services. Health Center Program GeoCare Navigator, Zip Code to ZCTA Crosswalk. (2024).\u003c/li\u003e\n\u003cli\u003eAnderson, G. B., Bell, M. L. \u0026amp; Peng, R. D. Methods to calculate the heat index as an exposure metric in environmental health research. \u003cem\u003eEnviron. Health Perspect.\u003c/em\u003e \u003cstrong\u003e121\u003c/strong\u003e, 1111\u0026ndash;1119 (2013).\u003c/li\u003e\n\u003cli\u003eMicrosoft. Microsoft/USBuildingFootprints. (2019).\u003c/li\u003e\n\u003cli\u003eEicher, C. L. \u0026amp; Brewer, C. A. Dasymetric Mapping and Areal Interpolation: Implementation and Evaluation. \u003cem\u003eCartogr. Geogr. Inf. Sci.\u003c/em\u003e \u003cstrong\u003e28\u003c/strong\u003e, 125\u0026ndash;138 (2001).\u003c/li\u003e\n\u003cli\u003eSwanwick, R. H. \u003cem\u003eet al.\u003c/em\u003e Dasymetric population mapping based on US census data and 30-m gridded estimates of impervious surface. \u003cem\u003eSci. Data\u003c/em\u003e \u003cstrong\u003e9\u003c/strong\u003e, 523 (2022).\u003c/li\u003e\n\u003cli\u003eFalcone, J. A. U.S. national categorical mapping of building heights by block group from Shuttle Radar Topography Mission data. US Geological Survey - ScienceBase https://doi.org/10.5066/F7W09416 (2016).\u003c/li\u003e\n\u003cli\u003eGeraci, M. \u0026amp; Bottai, M. Linear quantile mixed models. \u003cem\u003eStat. Comput.\u003c/em\u003e \u003cstrong\u003e24\u003c/strong\u003e, 461\u0026ndash;479 (2014).\u003c/li\u003e\n\u003cli\u003eBates, D., M\u0026auml;chler, M., Bolker, B. \u0026amp; Walker, S. Fitting Linear Mixed-Effects Models Using \u003cstrong\u003elme4\u003c/strong\u003e. \u003cem\u003eJ. Stat. Softw.\u003c/em\u003e \u003cstrong\u003e67\u003c/strong\u003e, (2015).\u003c/li\u003e\n\u003cli\u003eLandau, W. The targets R package: a dynamic Make-like function-oriented pipeline toolkit for reproducibility and high-performance computing. \u003cem\u003eJ. Open Source Softw.\u003c/em\u003e \u003cstrong\u003e6\u003c/strong\u003e, 2959 (2021).\u003c/li\u003e\n\u003cli\u003eHarlan, S. L., Declet-Barreto, J. H., Stefanov, W. L. \u0026amp; Petitti, D. B. Neighborhood Effects on Heat Deaths: Social and Environmental Predictors of Vulnerability in Maricopa County, Arizona. \u003cem\u003eEnviron. Health Perspect.\u003c/em\u003e \u003cstrong\u003e121\u003c/strong\u003e, 197\u0026ndash;204 (2013).\u003c/li\u003e\n\u003cli\u003eWang, C. \u003cem\u003eet al.\u003c/em\u003e Spatial Modeling and Analysis of Heat-Related Morbidity in Maricopa County, Arizona. \u003cem\u003eJ. Urban Health\u003c/em\u003e \u003cstrong\u003e98\u003c/strong\u003e, 344\u0026ndash;361 (2021).\u003c/li\u003e\n\u003cli\u003ePutnam, H. \u003cem\u003eet al.\u003c/em\u003e It\u0026rsquo;s not the heat, it\u0026rsquo;s the vulnerability: attribution of the 2016 spike in heat-associated deaths in Maricopa County, Arizona. \u003cem\u003eEnviron. Res. Lett.\u003c/em\u003e \u003cstrong\u003e13\u003c/strong\u003e, 094022 (2018).\u003c/li\u003e\n\u003cli\u003eHern\u0026aacute;ndez, D. \u0026amp; Bird, S. Energy burden and the need for integrated low‐income housing and energy policy. \u003cem\u003ePoverty Public Policy\u003c/em\u003e \u003cstrong\u003e2\u003c/strong\u003e, 5\u0026ndash;25 (2010).\u003c/li\u003e\n\u003cli\u003eDrehobl, A. \u0026amp; Ross, L. Lifting the high energy burden in America\u0026rsquo;s largest cities: How energy efficiency can improve low income and underserved communities. (2016).\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"energy insecurity, energy rationing, extreme heat, disparities, cooling demand","lastPublishedDoi":"10.21203/rs.3.rs-6184819/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6184819/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eRising temperatures and energy costs have intensified energy insecurity among vulnerable populations. This study examines patterns of energy insecurity and rationing in New York State using aggregated historical electricity data (2016\u0026ndash;2020). Using linear quantile mixed-effect models, we analyzed differences in residential energy consumption across income groups and levels of racialized economic segregation during warm seasons. Results show that higher-income households used 26.24\u0026ndash;55.58 kWh more energy per interquartile range (IQR) increase in heat index cooling degree days compared to the lowest income group. Areas with greater economic privilege, measured by the Index of Concentration at the Extremes, consumed 161.00-234.44 kWh more energy per IQR of heat index increase than less privileged areas. These findings suggest that lower-income and minoritized communities may engage in energy rationing during high cooling demand periods, potentially exposing themselves to elevated indoor temperatures and associated health risks. This evidence highlights the need for targeted interventions to address energy insecurity in disadvantaged communities.\u003c/p\u003e","manuscriptTitle":"Summertime temperatures, social disadvantage, and energy rationing: Patterns of energy insecurity from population-level historical electricity data in New York State","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-04-24 07:26:09","doi":"10.21203/rs.3.rs-6184819/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"591e3d23-7d1d-4bf1-b1c0-53a37a245960","owner":[],"postedDate":"April 24th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":47574834,"name":"Earth and environmental sciences/Environmental social sciences/Energy and society/Energy justice"},{"id":47574835,"name":"Earth and environmental sciences/Environmental social sciences/Sustainability"}],"tags":[],"updatedAt":"2025-07-25T10:45:48+00:00","versionOfRecord":[],"versionCreatedAt":"2025-04-24 07:26:09","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-6184819","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6184819","identity":"rs-6184819","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}