Severity Analysis of Pedestrian and Bike Crashes in School Buffer Zones
0
Citation
46
Reference
10
Related Paper
Abstract:
Pedestrian and bicyclist safety in school zones is critical because of the vulnerability of children and adult pedestrians to vehicle crashes. This paper explores vehicle–pedestrian/bike crash severity within a 15-min walking time buffer around schools in Detroit, Michigan, and San Jose, California—cities with high pedestrian/bike fatality rates. Using 2016–2020 crash data, we employed random-parameter multinomial logit models with heterogeneity in means and variances to understand unobserved relationships between variables. Key random parameters identified include the number of buffer zones that each crash falls into, daylight conditions, and the number of units involved in a crash, all significantly affecting injury severity. Spatial stability was investigated to see if variable effects were consistent across locations. Results revealed spatial instability across Detroit and San Jose. Factors such as Covid lockdown, dark lighting, arterial road presence, bicycle crashes, and the number of units involved showed stable effects with varying magnitudes in both cities. Network buffer zones highlighted that crash proximity to multiple schools affects injury severity. Additionally, the study found that various behavioral, roadway, weather, lighting, and school-related factors influence injury severity in school zones. These findings provide valuable insights for policymakers and planners to develop countermeasures, making school areas safer for children, adult pedestrians, and bicyclists.Keywords:
SAFER
Mixed logit
Vulnerability
For over the last thirty years the multinomial logit model has been the standard in choice modeling. Development in econometrics and computational algorithms has led to the increasing tendency to opt for more flexible models able to depict more realistically choice behavior. This study compares three discrete choice models, the standard multinomial logit, the error components logit, and the random parameters logit. Data were obtained from two choice experiments conducted to investigate consumers’ preferences for fresh pears receiving several postharvest treatments. Model comparisons consisted of in-sample and holdout sample evaluations. Results show that product characteristics hence, datasets, influence model performance. We also found that the multinomial logit model outperformed in at least one of three evaluations in both datasets. Overall, findings signal the need for further studies controlling for context and dataset to have more conclusive cues for discrete choice models capabilities.
Discrete choice
Mixed logit
Sample (material)
Multinomial distribution
Cite
Citations (0)
For over the last thirty years the multinomial logit model has been the standard in choice modeling. Development in econometrics and computational algorithms has led to the increasing tendency to opt for more flexible models able to depict more realistically choice behavior. This study compares three discrete choice models, the standard multinomial logit, the error components logit, and the random parameters logit. Data were obtained from two choice experiments conducted to investigate consumers’ preferences for fresh pears receiving several postharvest treatments. Model comparisons consisted of in-sample and holdout sample evaluations. Results show that product characteristics hence, datasets, influence model performance. We also found that the multinomial logit model outperformed in at least one of three evaluations in both datasets. Overall, findings signal the need for further studies controlling for context and dataset to have more conclusive cues for discrete choice models capabilities.
Discrete choice
Mixed logit
Sample (material)
Multinomial distribution
Cite
Citations (0)
This paper introduces the logitr R package for fast maximum likelihood estimation of multinomial logit and mixed logit models with unobserved heterogeneity across individuals, which is modeled by allowing parameters to vary randomly over individuals according to a chosen distribution. The package is faster than other similar packages such as mlogit, gmnl, mixl, and apollo, and it supports utility models specified with
Mixed logit
Multinomial distribution
Multinomial probit
Cite
Citations (12)
Mixed logit
Discrete choice
Independence
Cite
Citations (1,652)
For over the last thirty years the multinomial logit model has been the standard in choice modeling. Development in econometrics and computational algorithms has led to the increasing tendency to opt for more flexible models able to depict more realistically choice behavior. This study compares three discrete choice models, the standard multinomial logit, the error components logit, and the random parameters logit. Data were obtained from two choice experiments conducted to investigate consumers’ preferences for fresh pears receiving several postharvest treatments. Model comparisons consisted of in-sample and holdout sample evaluations. Results show that product characteristics hence, datasets, influence model performance. We also found that the multinomial logit model outperformed in at least one of three evaluations in both datasets. Overall, findings signal the need for further studies controlling for context and dataset to have more conclusive cues for discrete choice models capabilities.
Discrete choice
Mixed logit
Sample (material)
Multinomial distribution
Cite
Citations (0)
Abstract. The examination of car driver behavior deciding which parking space to choose. The application of various logit models has led to an insight of selecting between the available alternatives: free on-street parking, paid on-street parking and parking in an underground car park. Several logit models allowing for correlation between random taste parameters calculate coefficients using stated choice data. The main purpose of this paper is to extend a Mixed Multinomial Logit (M-MNL) model to similar models which can also implement the correlation between random parameters. This leads to the following: Nested Logit (NL), Nested Generalized Extreme Value (NGEV), Cross-Nested Logit (CNL) and Mixed-Mixed Multinomial Logit (MM-MNL) models to approach modeling parking choice models. The estimated coefficients are used to compute subjective-value of time (SVT) when looking for a parking space.
Mixed logit
Discrete choice
Nested logit
Value of time
Parking space
Value (mathematics)
Cite
Citations (1)
mixmixlogit is a Stata command that implements the mixed-mixed multinomial logit model (MM-MNL) for binary dependent variable data. It was first proposed in Keane and Wasi (2013) and Greene and Hensher (2013), and applied recently in Keane et al. (2020). It generalises both 'mixed logit' and 'latent class logit' by allowing for multiple latent types in the underlying data that are each characterised by a distribution of random parameters (as opposed to latent class logit, which assumes a homogeneous coefficient vector for each latent type, and mixed logit that allows for a distribution of random parameters for a single type of consumer or agent).
Mixed logit
Multinomial distribution
Mixed model
Multinomial probit
Discrete choice
Cite
Citations (0)
The paper investigates the influence of different model specifications for interpreting the results of discrete choice experiments when investigating heterogeneous public landscape preferences. Comparing model specifications based on the Mixed Multinomial Logit and the Generalized Multinomial Logit Model reveals that the parameter estimates appear qualitatively comparable. Still, a more in-depth investigation of the conditional estimate distributions of the sample show that parameter interactions in the Generalized Multinomial Logit Model lead to different interpretations compared to the Mixed Multinomial Logit Model. This highlights the potential impact of common model specifications in the results in landscape preference studies.
Mixed logit
Multinomial distribution
Discrete choice
Multinomial probit
Cite
Citations (0)
Mixed logit
Discrete choice
Choice set
Data set
Multinomial probit
Cite
Citations (17)
사회과학에서 널리 쓰이는 범주형 데이터, 특히 범주형 종속변수의 분석에 있어 로짓모델은 지난 수 십년간 매우 널리 사용되어왔다. 특히, 다항로짓(multinomial logit)과 조건부로짓(conditional logit)은 종속변수가 3항이상의 명목형 범주로 이루어져 있는 경우 매우 유용한 것이 사실이다. 하지만 이 모델들은 널리 알려진 한계가 있다. 첫째, 개인들 간의 선호도의 동질성의 가정, 둘째, 무관한 대안으로 부터의 독립성(independence of irrelevant alternatives)의 가정, 셋째, 시간별, 개인별 무상관성(uncorrelatedness across time and individual)의 원칙의 가정이 그것이다. 시뮬레이션에 기반한 기법이 확산되기 시작한 이래 범주형 데이터 연구에서 새롭게 활발히 개발, 적용중인 방법중 하나인 혼합로짓(mixed logit)은 개인간 불균질한 선호도(heterogeneous preferences), 교체패턴(substitution patterns)의 자유로운 설정, 시간 및 개인간 상관성 등을 범주형 모델링 과정에 포함시킬 수 있는 가능성을 제시함으로써 전통적으로 사용되던 범주형 데이터 분석의 한계의 극복 뿐만아니라 보다 심화된 데이터 분석을 제시 할 수 있는 방법을 제시한다. 본 논문의 목적은 혼합 로짓을 소개하는 것이다.
Mixed logit
Independence
Multinomial probit
Substitution (logic)
Cite
Citations (1)