Deakin University
Browse

File(s) under permanent embargo

A Tobit model for analyzing speed limit compliance in work zones

journal contribution
posted on 2014-12-01, 00:00 authored by Ashim DebnathAshim Debnath, R Blackman, N Haworth
Poor compliance with speed limits is a serious safety concern in work zones. Most studies of work zone speeds have focused on descriptive analyses and statistical testing without systematically capturing the effects of vehicle and traffic characteristics. Consequently, little is known about how the characteristics of surrounding traffic and platoons influence speeds. This paper develops a Tobit regression technique for innovatively modeling the probability and the magnitude of non-compliance with speed limits at various locations in work zones. Speed data is transformed into two groups-continuous for non-compliant and left-censored for compliant drivers-to model in a Tobit model framework. The modeling technique is illustrated using speed data from three long-term highway work zones in Queensland, Australia. Consistent and plausible model estimates across the three work zones support the appropriateness and validity of the technique. The results show that the probability and magnitude of speeding was higher for leaders of platoons with larger front gaps, during late afternoon and early morning, when traffic volumes were higher, and when higher proportions of surrounding vehicles were non-compliant. Light vehicles and their followers were also more likely to speed than others. Speeding was more common and greater in magnitude upstream than in the activity area, with higher compliance rates close to the end of the activity area and close to stop/slow traffic controllers. The modeling technique and results have great potential to assist in deployment of appropriate countermeasures by better identifying the traffic characteristics associated with speeding and the locations of lower compliance.

History

Journal

Safety science

Volume

70

Pagination

367-377

Location

Amsterdam, The Netherlands

ISSN

0925-7535

eISSN

1879-1042

Language

eng

Publication classification

C1.1 Refereed article in a scholarly journal

Copyright notice

2014, Elsevier Ltd.

Publisher

Elsevier

Usage metrics

    Research Publications

    Categories

    No categories selected

    Exports

    RefWorks
    BibTeX
    Ref. manager
    Endnote
    DataCite
    NLM
    DC