Pith. sign in

REVIEW 5 major objections 6 minor 43 references

Multiscale Carbon Burden of Infrastructure in the United States

T0 review · 5 major / 6 minor · reviewed 2026-08-04 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Local land use, not metropolitan size, drives per-capita fossil CO2 emissions in US neighbourhoods.

desk verdict Genuinely new scope-2 electricity scaling result is buried under an abstract–text mismatch and a transportation attribution problem that guts the headline causal claims; worth a major revision, not publication as-is. read the letter →

arxiv 2510.08611 v2 pith:FT2JDWN3 submitted 2025-10-07 physics.soc-ph

Jason Hawkins This is my paper · ORCID
classification physics.soc-ph
keywords landuseCO2emissionscausalinferenceurbanscalingtransportationresidentialenergyinfrastructureburdendoublyrobustestimation
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper tries to establish that the relationship between land use and fossil-fuel CO2 emissions is genuinely scale-dependent: metropolitan emissions scale nearly linearly with population, but within neighbourhoods the simple density–emissions link disappears. Using a 1-km resolution US fossil CO2 inventory and census block group built-form data, it finds that local density consistently lowers per-capita residential emissions, and that local roadway design is a strong negative predictor of transportation emissions. The methodological centrepiece is treating transportation emissions as production-based 'infrastructure burden' rather than household-attributed travel demand. A sympathetic reader would care because the results imply that both neighbourhood-scale and metropolitan-scale land-use policies matter jointly, and that street design may be a more powerful transport lever than density alone.

What carries the argument

The central mechanism is the production-based allocation of transportation emissions in the Vulcan v4.0 1-km fossil-fuel CO2 dataset, combined with a doubly robust causal identification strategy that uses a non-parametric generalized propensity score and Bayesian Additive Regression Trees for the outcome response surface. Emissions are interpolated from grid cells to census block groups, and land-use treatments are the '5Ds'—density, diversity, roadway design, distance to transit, destination accessibility—measured both locally and at the metropolitan (CBSA) scale. The key conceptual move is interpreting transportation emissions as infrastructure carbon burden: emissions are attributed to th

What would settle it

Rebuild the analysis re-attributing on-road emissions to trip origins using a travel-survey-based synthetic population, and compare the roadway-design and land-use-diversity treatment effects; if the large negative roadway-design effect and the positive diversity effect disappear or reverse, the production-based allocation is the driver of those results.

Watch

Extended reading notes

Core claim

The paper claims that fossil-fuel CO2 emissions respond to land use at multiple scales, but with opposite or absent patterns depending on geography. At the metropolitan level, total emissions scale nearly linearly with population (slope 0.92), while scope-2 electricity emissions scale sublinearly (slope 0.85, R²=0.83); at the census block group level, the power-law relationship essentially vanishes (R² between 0.02 and 0.10). Using doubly robust causal estimates with generalized propensity scores and BART outcome models, the paper finds that local population density lowers per-capita emissions in residential sectors—joint-model coefficients of -1.45 for non-electric residential energy and -0

Load-bearing premise

Transportation emissions are assigned to the grid cell where fuel is burned rather than to the traveller's home neighbourhood, so the transport land-use effects measure roadway burden and through-traffic, not the travel choices of local residents.

Editorial extensions

If this is right

  • If local density consistently lowers per-capita residential emissions (coefficients of -1.45 and -0.58), then neighbourhood-level land-use policies—zoning, infill, height limits—can meaningfully complement metropolitan-level climate policy.
  • If local roadway design has an average treatment effect of about -0.96 on transportation CO2, street design changes (connectivity, pedestrian infrastructure, speed moderation) may be at least as effective as density-based strategies for reducing transport emissions at the block-group scale.
  • If scope-2 electricity emissions scale sublinearly with metro population (slope 0.85), larger metropolitan areas gain a per-capita electricity-emissions efficiency, supporting agglomeration benefits in the electricity sector.
  • Because neighbourhood-scale power laws are essentially absent, urban scaling results at the city level cannot be extrapolated downward; policy targeting must be scale-specific.
  • Neighbouring land-use diversity exceeding local diversity implies spatial spillovers across block-group boundaries, so land-use planning should operate at corridor or district scales, not only parcel or block scales.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The production-based allocation means the large negative roadway-design effect probably reflects roadway exposure and through-traffic rather than the travel behaviour of local residents; re-attributing emissions to trip origins could substantially weaken or reverse that result.
  • The strong negative density coefficient for residential non-electric energy (-1.45) may partly capture dwelling size and building typology rather than an efficiency effect of density itself; the paper does not separate floor area from density.
  • The surprising positive effect of local land-use diversity on transportation emissions is consistent with the production-based allocation: diverse, amenity-rich neighbourhoods attract trips from outside, so a consumption-based attribution would likely flip the sign toward the conventional expectation.
  • The sublinear electricity scaling may reflect electricity grid and power-plant siting rather than household demand efficiency; the supplemental scope-2 (consumption-based) data could be used to test whether the pattern persists under consumption attribution.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

5 major / 6 minor

Summary. The paper combines Vulcan fossil-fuel CO2 grid data with EPA Smart Location Database and census data to study how land use and infrastructure shape per-capita emissions at neighborhood (CBG) and metropolitan (CBSA) scales in the United States. It reports power-law scaling relationships (notably sublinear electricity scaling, slope 0.85) and doubly robust average treatment effects for transportation, residential electricity, and residential non-electricity emissions. The authors emphasize production-based treatment of transportation emissions as an infrastructure burden and conclude that local roadway design and residential density reduce per-capita emissions, supporting coordinated multi-scale land-use policy.

Significance. If the central claims held, the paper would provide useful evidence on the relative importance of local versus metropolitan land-use features for sectoral CO2 emissions, and the sublinear metropolitan electricity scaling would be a novel descriptive result. The manuscript has constructive features: it uses publicly available high-resolution data, attempts a data-driven confounder-selection procedure, and explicitly acknowledges the transportation allocation problem. However, the abstract reports a spatial econometric framework (heteroscedasticity-robust SLX+SEM with a 10-km band) and neighbor-diversity findings that do not appear anywhere in the body; the transportation treatment effects are acknowledged to measure roadway exposure rather than residential land use; and the residential results contain direct internal contradictions. As currently written, the load-bearing empirical claims are not reproducible from the reported analysis.

major comments (5)
  1. [Abstract vs. §4-§6] The abstract's central methodological and empirical claims are absent from the body. It states that 'the preferred heteroscedasticity-robust SLX+SEM model with a 10-km distance band' reduces spatial autocorrelation and that 'neighbouring land use diversity exceeds local diversity.' However, §4 describes only a generalized propensity score DR estimator and §6.1 describes a 'simple frequentist linear regression' joint model. No spatial lag/SLX, spatial error model, 10-km band, or neighbor-variable results appear in the text or tables. The abstract-level findings are therefore unsupported by the reported analysis.
  2. [§6.1, §7, Table 1, Fig. 6] Transportation FFCO2 is production-based in Vulcan (emissions allocated to the grid cell where fuel is burned). §6.1 concedes that CBGs with desirable 5D features also have high through-traffic and that results are 'skewed'; §7 states that the treatment effects do not correspond to the effect of land use at the home location. Since the outcome in Eqs. (5)-(7) and Table 1 is per-capita CBG emissions under this allocation, the headline ATE of -0.96 for roadway design (Fig. 6c) and the CBG density coefficient of -0.583 (Table 1) identify roadway exposure, not residential land-use effects. The abstract's policy conclusion ('local roadway design is a strong negative predictor of transportation FFCO2') is not warranted for residential mitigation. This is acknowledged as a limitation but not resolved.
  3. [§6.3, Tables 2-3] The residential non-electricity results are internally contradictory. §6.3 first states that 'population density is found to have no effect on residential energy FFCO2' and that diversity also has no effect, then immediately reports 'local (CBG) population density has a strong negative (-1.452) effect on these emissions.' The -1.452 coefficient appears in Table 3, not Table 2 (headed residential electricity, where CBG density is +0.409). The sentence attributing -1.452 to Table 2 and the 'opposite effect' comparison are not interpretable. The text and tables must be reconciled before the residential results can be evaluated.
  4. [Appendix B, §4.2] The propensity-score balancing diagnostics undermine confidence in the weights. For Transportation Density, effective sample sizes are 7%, 0%, 29%, 21%, 0%, 0%, and 78% across the listed methods, far below the stated 70% target; BART yields an effective sample of 5 observations (0%) and Super Learner 1 observation (0%). The main text does not state which balancing method produced the weights used in the DR or joint models. If weights with near-zero effective sample are used, the ATE estimates in §6 may be driven by a tiny subset of observations.
  5. [§4, Eqs. (5)-(7), Eq. (9)] The doubly robust estimator is not written in a coherent form. Eqs. (5)-(6) use T and Y(25) ambiguously: T appears as a treatment indicator and as a multiplier, while 1/f_T(X) is called an inverse propensity score even though f_T(X) in Eq. (1) is a conditional density. Eq. (9), labelled the outcome model, is a linear regression including the propensity score and CBSA fixed effects, which is inconsistent with the BART outcome model described in §4.3. These formal problems prevent replication of the reported estimates.
minor comments (6)
  1. [Abstract / §3] The abstract says Vulcan v4.0, while §3 discusses validation of v3.0; the version used should be stated consistently and the v4.0 data description should be explicit.
  2. [§3] The sector list contains 'non-road' twice and omits a clear definition of the scope-2 allocation process; please clarify which sectors are included in the three outcomes.
  3. [§4.3] The phrase 'lack of casual interpretation' should read 'causal'; similarly, 'esimate' in §7 and 'Survery' in §7 should be corrected.
  4. [Table 4 header] The table header contains 'T reatment F eature' and the word 'Minimum' appears misaligned; the table would benefit from reformatting.
  5. [§6.2-6.3] The text repeats a sentence about Table 2 across sections 6.2 and 6.3; one of the repetitions appears to be an editing artifact.
  6. [General] The paper states that results are 'available upon request' but does not provide a code or data repository; given the open-science statement in §3, a public repository would improve reproducibility.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: the headline results are fitted ATEs and scaling exponents, not quantities forced by construction.

full rationale

The paper's central outputs are empirical estimates: doubly robust ATEs computed from Eqs. (5)-(7) and (10) applied to Vulcan and EPA-SLD data, and log-log scaling slopes in Figures 4-5. No theoretical constant is derived, and no 'prediction' is generated from a fitted parameter in a way that reduces to the input. The ATE for roadway design (Fig. 6c), CBG density coefficients (Tables 1-3), and the scope-2 electricity scaling slope (0.85, R^2=0.83) are all fitted associations, not self-fulfilling constructions. The one self-citation (Hawkins and Habib, 2021) is used only to note that a similar areal interpolation approach was used before; the interpolation itself is standard and independently cited (Goodchild et al., 1993; Tsutsumi and Murakami, 2014; Comber and Zeng, 2019), so this citation is not load-bearing. The acknowledged production-based allocation of Vulcan on-road emissions (section 6.1) is a threats-to-validity issue for the transportation estimand, not circularity: the paper explicitly says treatment effects 'do not correspond to the effect of land use at the home location' and proposes a re-estimation, which confirms the authors did not rename the result as something it is not. No equation is equal to its own input by construction; no fitted parameter is relabeled as a prediction.

Assumptions & free parameters 4 free parameters · 6 assumptions · 0 invented entities

The central claims rest on standard causal assumptions, on the fidelity of Vulcan and EPA-SLD data, and on two paper-specific modeling choices: population-density-based areal interpolation and the restriction of residential models to density and diversity only. The fitted scaling exponents and treatment coefficients are empirical outputs, not independent theoretical constants. No new physical or conceptual entities are postulated.

free parameters (4)
  • Sectoral scaling exponents = OLS slopes: 0.93 residential non-electricity; 1.00 commercial; 1.06 industrial; 0.92 transport; 0.85 electricity; 0.92 t
    Fitted log-log regression slopes in Figs. 4-5; used to support the power-law and scale-dependence conclusions.
  • Joint model treatment coefficients = e.g., CBG density -0.583 (transport), 0.409 (electricity), -1.451 (energy); CBG diversity 0.696, -1.139, 0.289
    Fitted multi-treatment regression coefficients in Tables 1-3, interpreted as land-use effects on per-capita FFCO2.
  • Confounder variable selection threshold = 0.05 R2 contribution
    Chosen by hand in §4.1 and 'deemed conservative and tested during propensity score balancing'; determines which confounders enter each GPS model.
  • Effective sample size target = 70% of observations
    Design criterion in §4.2 for selecting among balancing methods; affects which propensity-score estimator is treated as acceptable.
assumptions (6)
  • domain assumption Unconfoundedness and overlap hold for each continuous treatment given the selected covariates
    Invoked in §4 via the Hirano-Imbens generalized propensity score and Eqs. 2-3; not empirically testable.
  • domain assumption Vulcan v4.0 FFCO2 and EPA-SLD variables measure true emissions and land use at CBG scale without systematic error
    The data section accepts Vulcan validation against atmospheric and other inventories and uses EPA-SLD built-form measures as treatments.
  • domain assumption Areal interpolation with population-density apportionment yields unbiased CBG emissions
    §3 and Fig. 2 assign grid-cell emissions to CBGs assuming within-grid-cell homogeneity and using population density as an auxiliary variable; this can bias density-emissions relationships.
  • ad hoc to paper Only population density and land use diversity affect residential electricity and non-electricity energy outcomes
    Explicit assumption in §6.2: 'We also assume that only population density and land use diversity affect these outcomes,' excluding other 5D variables from residential models.
  • standard math The doubly robust estimator is valid if either the outcome model or the GPS model is correctly specified
    Property stated in §4; requires correctly fitted models, but the implementation details needed to verify correct specification are incomplete.
  • domain assumption Transportation emissions assigned to the production location allow interpretation as infrastructure carbon burden
    §6.1 and §7 acknowledge production-based allocation; the paper reframes it as a feature, but it is also a limitation for home-based land-use inference.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Multiscale Carbon Burden of Infrastructure in the United States." pith.science (2026). https://pith.science/paper/FT2JDWN3

@misc{pith2026251008611,
  author       = {Pith},
  title        = {Pith review of: Multiscale Carbon Burden of Infrastructure in the United States},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/FT2JDWN3}},
  note         = {Machine review of arXiv:2510.08611}
}
read the original abstract

Anthropogenic greenhouse gas (GHG) emissions vary spatially with development patterns, climate, economic structure, and energy systems. Using Vulcan v4.0 fossil-fuel CO2 (FFCO2) data for the United States at 1-km resolution, this study examines how land use and infrastructure shape emissions at local and metropolitan scales. I combine doubly robust Bayesian Additive Regression Tree estimators with multi-treatment spatial regression models to identify local and spillover effects by sector. A key methodological contribution is treating transportation emissions as production-based, capturing infrastructure carbon burden rather than household-attributed travel demand. Results show strong scale dependence and spatial interaction: in the preferred heteroscedasticity-robust SLX+SEM model with a 10-km distance band, residual spatial autocorrelation declines substantially. Local roadway design is a strong negative predictor of transportation FFCO2, while neighbouring land use diversity exceeds local diversity. In residential sectors, higher local density is consistently associated with lower per-capita emissions. These findings support coordinated multi-scale mitigation policy in the United States.

Figures

Figures reproduced from arXiv: 2510.08611 by the authors.

Figure 1
Figure 1. Annual gasoline consumption with respect to urban density (Newman and Kenworthy, 1989b). [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Grid cell-CBG spatial imputation 4 Methods of Analysis Using CBG-level observations of emissions, we are interested in the causal effect of the 5D land use metrics on FFCO2 emissions by sector. In this study, we focus on the emissions from transportation, residential electricity, and residential non-electricity sources. Among the causal identification strategies available for observations data, propensity score weig… view at source ↗
Figure 3
Figure 3. Causal graph for variable selection X no matter the value of the treatment T. Li et al. (2024) use light Gradient Boosting Machines (GBM) to train the confounder variable model. They set a threshold for CSVI to exclude variables with minimal conditional effect. We use Bayesian Additive Regression Trees (BART), described further below, for the confounder model. A simple heuristic to quantify variable importance using… view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: Metropolitan scaling effect by sector 2 4 6 8 10 ln(Population) 2.5 0.0 2.5 5.0 7.5 10.0 ln(CO2) R 2 = 0.02 Slope = 0.24 Residential Non-Electricity 2 4 6 8 10 ln(Population) 5 0 5 10 15 ln(CO2) R 2 = 0.03 Slope = 0.35 Commercial 2 4 6 8 10 ln(Population) 10 0 10 ln(CO…
Figure 5
Figure 5. Figure 5: Neighbourhood scaling effect by sector 10 [PITH_FULL_IMAGE:figures/full_fig_p010_5.png]
Figure 6
Figure 6. Figure 6: Doubly robust average treatment effects (ATE) of land use features on transportation FFCO2 [PITH_FULL_IMAGE:figures/full_fig_p012_6.png]
Figure 7
Figure 7. Figure 7: Doubly robust average treatment effects (ATE) of land use features on residential electricity [PITH_FULL_IMAGE:figures/full_fig_p013_7.png]
Figure 8
Figure 8. Figure 8: Doubly robust average treatment effects (ATE) of land use features on residential energy FFCO2 [PITH_FULL_IMAGE:figures/full_fig_p014_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

43 extracted references · 28 canonical work pages

  1. [2]

    URL:https://pnas.org/doi/full/ 10.1073/pnas.1919032117, doi:10.1073/pnas.1919032117

    Proceedings of the National Academy of Sciences 117, 13300–13307. URL:https://pnas.org/doi/full/ 10.1073/pnas.1919032117, doi:10.1073/pnas.1919032117. Bettencourt, L., Lobo, J., Helbing, D., West, G.B.,

    work page doi:10.1073/pnas.1919032117

  2. [4]

    Updated: 18 february 2021 ed., HM Treasury, London

    The economics of biodiversity: the Dasgupta review. Updated: 18 february 2021 ed., HM Treasury, London. De Luna, X., Waernbaum, I., Richardson, T.S.,

    2021

  3. [6]

    Hill, J., Weiss, C., Zhai, F.,

    doi:10.1088/1748-9326/6/1/014018. Hill, J., Weiss, C., Zhai, F.,

    work page doi:10.1088/1748-9326/6/1/014018

  4. [9]

    Journal of Planning Education and Research 38, 167–182

    Testing newman and kenworthy’s theory of density and automobile dependence. Journal of Planning Education and Research 38, 167–182. doi:10.1177/0739456X16688767. Ewing, R., Richardson, H.W., Bartholomew, K., Nelson, A.C., Bae, C.H.C.,

    work page doi:10.1177/0739456x16688767

  5. [10]

    The Annals of Applied Statistics 12, 156–177

    Covariate balancing propensity score for a continuous treatment: Application to the efficacy of political advertisements. The Annals of Applied Statistics 12, 156–177. doi:10.1214/17-AOAS1101. 15 Fragkias, M., Lobo, J., Strumsky, D., Seto, K.C.,

    work page doi:10.1214/17-aoas1101

  6. [12]

    Journal of Urban Economics 67, 404–418

    The greenness of cities: Carbon dioxide emissions and urban develop- ment. Journal of Urban Economics 67, 404–418. URL:www.purdue.edu/eas/carbon/vulcan/index.php, doi:10.1016/j.jue.2009.11.006. Goodchild, M.F., Anselin, L., Deichmann, U.,

    work page doi:10.1016/j.jue.2009.11.006 2009

  7. [13]

    Dasgupta, P.,

    URL:https: //onlinelibrary.wiley.com/doi/10.1111/gec3.12465, doi:10.1111/gec3.12465. Dasgupta, P.,

    work page doi:10.1111/gec3.12465

  8. [18]

    URL:https://www.researchgate.net/publication/349735428_The_Influence_ of_Built_Form_on_Patterns_of_Household_Consumption_Evidence_from_American_Household_ Expenditure

    The influence of built form on patterns of household consumption: Evidence from american household expenditure, in: 100th Annual Meeting of the Transportation Research Board, Washington D.C. URL:https://www.researchgate.net/publication/349735428_The_Influence_ of_Built_Form_on_Patterns_of_Household_Consumption_Evidence_from_American_Household_ Expenditure...

    arXiv

Show all 43 references
  1. [20]

    Multivariate Behavioral Research 46, 477–513

    Challenges with propensity score strategies in a high-dimensional setting and a potential alternative. Multivariate Behavioral Research 46, 477–513. doi:10.1080/00273171.2011. 570161. Hill, J.L.,

    2011

  2. [21]

    Journal of Computational and Graphical Statistics 20, 217–240

    Bayesian nonparametric modeling for causal inference. Journal of Computational and Graphical Statistics 20, 217–240. doi:10.1198/jcgs.2010.08162. Hirano, K., Imbens, G.W.,

    2010 arXiv

  3. [23]

    household carbon footprints reveals subur- banization undermines greenhouse gas benefits of urban population density

    Spatial distribution of u.s. household carbon footprints reveals subur- banization undermines greenhouse gas benefits of urban population density. Environmental Science and Technology 48, 895–902. doi:10.1021/es4034364. Kreif, N., Grieve, R., D ´ ıaz, I., Harrison, D.,

    doi

  4. [25]

    household sector

    The influence of urban form on ghg emissions in the u.s. household sector. Energy Policy 68, 534–549. doi:10.1016/j.enpol.2014.01.024. Lee, S., Lee, B.,

    2014 doi

  5. [30]

    Regional Science and Urban Economics 70, 112–126

    Home production, expenditure, and economic geography. Regional Science and Urban Economics 70, 112–126. doi:10.1016/j.regsciurbeco.2018.03.003. Newman, P., Kenworthy, J., 1989a. Cities and automobile dependence: An international sourcebook. Alder- shot: Gower. URL:https://trid...

    2018 doi

  6. [31]

    Energy Policy 68, 232–242

    Life-cycle energy implications of different residential settings: Recognizing buildings, travel, and public infrastructure. Energy Policy 68, 232–242. URL:http: //dx.doi.org/10.1016/j.enpol.2013.12.062, doi:10.1016/j.enpol.2013.12.062. Nichols, B.G., Kockelman, K.M.,

    2013 doi

  7. [32]

    Journal of Transport and Land Use 8, 115–219

    Urban form and life-cycle energy consumption: Case studies at the city scale. Journal of Transport and Land Use 8, 115–219. doi:10.5198/jtlu.2015.598. Norman, J., MacLean, H.L., Kennedy, C.A.,

    2015 doi

  8. [34]

    Earth System Science Data 10, 87–107

    The open-source data inventory for anthropogenic co2, version 2016 (odiac2016): a global monthly fossil fuel co2 gridded emissions data product for tracer transport simu- lations and surface flux inversions. Earth System Science Data 10, 87–107. doi:10.5194/essd-10-87-2018. aD...

    2016 doi

  9. [35]

    American Journal of Epidemiology 174, 1223–1227

    Invited commentary: Understanding bias amplification. American Journal of Epidemiology 174, 1223–1227. doi:10.1093/aje/kwr352. Pushkar, A., Hollingworth, B., Miller, E.,

    doi

  10. [37]

    Lessons from Sustainable Transit-Oriented Cities. p. 49–94. doi:10.1596/9780821397459_CH02. 17 Tang, D., Kong, D., Pan, W., Wang, L.,

    doi

  11. [39]

    Interna- tional Regional Science Review 37, 273–297

    New spatial econometrics–based areal interpolation method. Interna- tional Regional Science Review 37, 273–297. doi:10.1177/0160017612463233. T¨ ubbicke, S.,

    doi

  12. [41]

    doi:10.1016/j

    Energy requirements of consumption: Urban form, climatic and socio-economic factors, rebounds and their policy implications 63, 696–707. doi:10.1016/j. enpol.2013.07.035. Zheng, S., Wang, R., Glaeser, E.L., Kahn, M.E.,

    2013 doi

  13. [42]

    Journal of Economic Geography 11, 761–792

    The greenness of china: household carbon dioxide emissions and urban development. Journal of Economic Geography 11, 761–792. doi:10.1093/jeg/lbq031. Zhu, Y., Coffman, D.L., Ghosh, D.,

    doi

  14. [43]

    Journal of Causal Inference 3, 25–40

    A boosting algorithm for estimating generalized propensity scores with continuous treatments. Journal of Causal Inference 3, 25–40. doi:10.1515/jci-2014-0022. A Propensity Score Confounding Variable Models p_highschool p_owner pct_ao0 avg_hh_size med_dwelling_age totpop_cbsa s...

    2014 doi

  15. [84]

    16 Li, S., Pu, Z., Cui, Z., Lee, S., Guo, X., Ngoduy, D.,

    doi:10.1016/j.jtrangeo.2020.102694. 16 Li, S., Pu, Z., Cui, Z., Lee, S., Guo, X., Ngoduy, D.,

    2020

  16. [160]

    2024.104537

    doi:10.1016/j.trc. 2024.104537. Liu, Y., Bi, J., Lv, J., Ma, Z., Wang, C.,

    2024

  17. [553]

    Hawkins, J., Habib, K.N.,

    doi:10.1038/s41467-020-20871-0. Hawkins, J., Habib, K.N.,

    doi

  18. [1344]

    Gurney, K.R., Liang, J., Patarasuk, R., Song, Y., Huang, J., Roest, G.,

    doi:10.1038/s41597-024-04180-5. Gurney, K.R., Liang, J., Patarasuk, R., Song, Y., Huang, J., Roest, G.,

    doi

  19. [1993]

    Environment and Planning A: Economy and Space 25, 383–397

    A framework for the areal interpolation of socioeconomic data. Environment and Planning A: Economy and Space 25, 383–397. doi:10.1068/a250383. Graham, D.J.,

    doi

  20. [2004]

    Transportation Research Part A: Policy and Practice 38, 143–162

    Urban density and energy consumption: A new look at old statistics. Transportation Research Part A: Policy and Practice 38, 143–162. doi:10.1016/j.tra.2003. 10.004. Murphy, D.,

    2003 doi

  21. [2005]

    Wiley-Blackwell

    The Propensity Score with Continuous Treatments. Wiley-Blackwell. p. 73–84. URL:http://www.scopus.com/inward/record.url?scp=84987736345&partnerID=8YFLogxK, doi:10.1002/0470090456.ch7. Jones, C., Kammen, D.M.,

    doi

  22. [2006]

    Journal of Urban Planning and Development 132, 10–21

    Comparing High and Low Residential Density: Life-Cycle Analysis of Energy Use and Greenhouse Gas Emissions. Journal of Urban Planning and Development 132, 10–21. doi:10.1061/(asce)0733-9488(2006)132:1(10). Oda, T., Maksyutov, S., Andres, R.J.,

    2006 doi

  23. [2010]

    Journal of the American Planning Associ- ation 76, 265–294

    Travel and the built environment. Journal of the American Planning Associ- ation 76, 265–294. doi:10.1080/01944361003766766. Ewing, R., Hamidi, S.,

    doi

  24. [2011]

    Biometrika 98, 861–875

    Covariate selection for the nonparametric estimation of an average treatment effect. Biometrika 98, 861–875. doi:10.1093/biomet/asr041. Duranton, G., Turner, M.A.,

    doi

  25. [2013]

    urban areas

    Does size matter? scaling of co2 emissions and u.s. urban areas. PLoS ONE 8, e64727. doi:10.1371/journal.pone.0064727. Glaeser, E.L., Kahn, M.E.,

    doi

  26. [2014]

    Ewing, R., Hamidi, S., Tian, G., Proffitt, D., Tonin, S., Fregolent, L.,

    URL:https://smartgrowthamerica.org/ resources/measuring-sprawl-2014/. Ewing, R., Hamidi, S., Tian, G., Proffitt, D., Tonin, S., Fregolent, L.,

    2014

  27. [2015]

    Health Economics 24, 1213–1228

    Evaluation of the effect of a continuous treatment: A machine learning approach with an application to treatment for traumatic brain injury. Health Economics 24, 1213–1228. doi:10.1002/hec.3189. Lee, S., Lee, B.,

    doi

  28. [2017]

    doi:10.1080/01944363.2016.1240044

    Does compact development make people drive less? Journal of the American Planning Association 83, 7–18. doi:10.1080/01944363.2016.1240044. Suzuki, H., Cervero, R., Iuchi, K.,

    2016

  29. [2018]

    Journal of Urban Economics 108, 170–191

    Urban form and driving: Evidence from us cities. Journal of Urban Economics 108, 170–191. doi:10.1016/j.jue.2018.10.003. Ewing, R., Bartholomew, K., Winkelman, S., Walters, J., Chen, D.,

    2018 doi

  30. [2020]

    Journal of Geophysical Research: Atmospheres 125, e2020JD032974

    The vulcan version 3.0 high- resolution fossil fuel co2 emissions for the united states. Journal of Geophysical Research: Atmospheres 125, e2020JD032974. doi:10.1029/2020JD032974. Gurney, K.R., Liang, J., Roest, G., Song, Y., Mueller, K., Lauvaux, T.,

    doi

  31. [2021]

    Applied Energy 304, 117761

    A physics-informed data reconciliation framework for real-time electricity and emissions tracking. Applied Energy 304, 117761. doi:10.1016/j.apenergy.2021.117761. Chapman, J., Fox, E.H., Bachman, W., Frank, L.D., Thomas, J., Rourk Reyes, A.,

    2021

  32. [2022]

    Journal of Econometric Methods 11, 71–89

    Entropy balancing for continuous treatments. Journal of Econometric Methods 11, 71–89. doi:10.1515/jem-2021-0002. Vegetabile, B.G., Griffin, B.A., Coffman, D.L., Cefalu, M., Robbins, M.W., McCaffrey, D.F.,

    2021 doi

  33. [2023]

    Biometrics 79, 903–914

    Ultra-high dimensional variable selection for doubly robust causal inference. Biometrics 79, 903–914. doi:10.1111/biom.13625. Tsutsumi, M., Murakami, D.,

    doi

  34. [2025]

    Transportation Research Part A: Policy and Practice 192, 104324

    Causal inference for transport research. Transportation Research Part A: Policy and Practice 192, 104324. doi:10.1016/j.tra.2024.104324. Greifer, N.H.,

    2024

  35. [9486]

    Mindali, O., Raveh, A., Salomon, I.,

    doi:10.1038/s41598-017-09863-1. Mindali, O., Raveh, A., Salomon, I.,

    doi

Pith tools

Reviewed August 4, 2026 · model on record in the stance chip above.