Skip navigation

Road Safety

Cover of Keeping People Safe When Walking: Stream 2 Pedestrian Safety Indicators and Risk Models
Keeping People Safe When Walking: Stream 2 Pedestrian Safety Indicators and Risk Models
  • Publication no: AP-R780-26
  • ISBN: 978-1-923617-78-0
  • Published: 22 October 2026

This report aims to help Austroads members measure and manage pedestrian safety performance. This report provides practical guidance, methodologies and a prototype tool that, with further development, could be used for managing pedestrian safety performance across Australia and New Zealand. It is part of the Austroads project SAG6371 Keeping People Safe When Walking program (Stream 2).

The report refines core pedestrian safety terms, concepts and frameworks from Keeping People Safe When Walking Stream 1. A pedestrian Key Performance Indicator framework is presented to track the scale, quality and effectiveness of agency action. A pedestrian Safety Performance Indicator framework was also developed, structured around Safe System pillars and walking demand.

The report describes a feasible approach to proactive pedestrian risk assessment and network screening in Australasian data environments. The approach evaluates exposure and crash risk modelling alternatives, incorporating crash prediction models and exposure proxies where count data were limited. Initial models were developed for the Melbourne CBD only. It also outlines how Empirical Bayes methods should be applied to mitigate regression-to-the-mean effects and improve the identification of high-risk locations.

A prototype tool, developed for the Melbourne CBD area, shows how model outputs can be integrated with Key Performance Indicators and Safety Performance Indicators to support prioritisation and treatment of high-risk locations.

  • Summary
  • 1. Introduction
    • 1.1 Purpose
    • 1.2 Background
      • 1.2.1 Previous work (stream 1) – data, evidence and interventions
      • 1.2.2 Current stage of work (stream 2) – development of better guidance
    • 1.3 Key challenges for practitioners – pedestrian data availability, access and functionality gaps
    • 1.4 Prototype tool and guidance
    • 1.5 How this report is structured
  • 2. Pedestrian Safety and Safety Performance – Definitions, Frameworks and Data Needs
    • 2.1 Definition of key pedestrian safety concepts
    • 2.2 Review of pedestrian safety definitions
      • 2.2.1 Definition of ‘pedestrian’
        Suggested approach
      • 2.2.2 Pedestrian crash
        Commentary 1: pedestrian self-harm and death by suicide
        Commentary 2: pedestrian falls with no vehicle involvement
      • 2.2.3 Additional pedestrian safety terms
    • 2.3 Key frameworks and methodologies
      • 2.3.1 Pedestrian Safe System model
      • 2.3.2 Pedestrian Safety Data Needs Framework
      • 2.3.3 In-Depth Trauma Modelling
      • 2.3.4 Vision Zero Modelling
    • 2.4 Essential pedestrian data types for safety strategies
      • 2.4.1 Additional data required for In-Depth Trauma and Vision Zero Modelling
    • 2.5 Pedestrian safety performance – definitions and distinctions
      • 2.5.1 Distinguishing between SPIs and KPIs
      • 2.5.2 Illustrative case study: categorising 12 voluntary global road safety targets
      • 2.5.3 Illustrative case study: speed management and vehicle safety
  • 3. Pedestrian Key Performance Indicators
    • 3.1 Measuring pedestrian KPIs
    • 3.2 Further considerations
      • 3.2.1 Target setting and benchmarking
      • 3.2.2 Trauma or Vision Zero modelling
      • 3.2.3 Reporting and integration
      • 3.2.4 Governance and improvement
  • 4. Pedestrian Safety Performance Indicators
    • 4.1 Safety Performance Indicators – theory and definition
      • 4.1.1 Importance of Safety Performance Indicators
    • 4.2 Review of existing pedestrian SPI frameworks
      • 4.2.1 Swedish SPI framework
      • 4.2.2 Trendline
      • 4.2.3 Victoria and New Zealand
      • 4.2.4 Other pedestrian SPIs
      • 4.2.5 Summary of key points and what is taken forward
    • 4.3 New pedestrian SPIs
      • 4.3.1 Assessment of SPIs against ‘minimum pedestrian data requirements’
      • 4.3.2 Recommended SPIs (must have, could have)
      • 4.3.3 Target setting
  • 5. Pedestrian Risk and Exposure Assessment to Improve Pedestrian Safety
    • 5.1 Overview of pedestrian risk and exposure concepts
      • 5.1.1 Pedestrian exposure
      • 5.1.2 Pedestrian risk in crashes involving motor vehicles
      • 5.1.3 Challenges
    • 5.2 Applying pedestrian risk models in strategy development
  • 6. Methodology for Identifying, Reviewing and Assessing Pedestrian Risk Assessment Models
    • 6.1 Literature review methodology
    • 6.2 Multi-criteria assessment framework
      • 6.2.1 MCA framework for risk assessment models
      • 6.2.2 MCA framework for pedestrian exposure models
  • 7. Literature Review Findings
    • 7.1 Pedestrian risk assessment models
      • 7.1.1 Crash-based models
      • 7.1.2 Risk-based models for pedestrian crashes involving motor vehicles
      • 7.1.3 Hybrid model – Australian National Risk Assessment Model
    • 7.2 Pedestrian exposure models
      • 7.2.1 Direct demand models
      • 7.2.2 Travel demand models
      • 7.2.3 Microsimulation models
      • 7.2.4 Agent-based models
        Exposure proxy models
      • 7.2.5 Machine learning and artificial intelligence techniques
      • 7.2.6 Summary
  • 8. Results of the Multi-Criteria Assessment
    • 8.1 Pedestrian risk estimation models
      • 8.1.1 Summary
    • 8.2 Pedestrian exposure models
      • 8.2.1 Summary
  • 9. Data Availability Assessment
    • 9.1 Victoria, Australia
    • 9.2 Auckland, New Zealand
    • 9.3 Modelling options for pedestrian exposure and risk models
    • 9.4 Other relevant findings from data availability assessment
  • 10. Modelling Framework for Pedestrian Exposure and Risk 
    • 10.1 Two pathways for pedestrian risk assessment
      • 10.1.1 Pathway A – explicit exposure modelling
      • 10.1.2 Pathway B – proximal indicator approach
      • 10.1.3 Approach adopted in this study
    • 10.2 Overview of the modelling workflow
      • 10.2.1 Modelling logic and progression
    • 10.3 Data preparation and variable development
      • 10.3.1 Variable definitions
      • 10.3.2 Overview of data sources and integration
      • 10.3.3 Study area definition
      • 10.3.4 Data cleaning and quality assurance
      • 10.3.5 Spatial processing and dataset linkage
      • 10.3.6 Treatment of pedestrian exposure data
      • 10.3.7 Identification of proximal indicators of pedestrian exposure
    • 10.4 Descriptive statistics
      • 10.4.1 Sample size and data completeness
      • 10.4.2 Intersection characteristics
      • 10.4.3 Mid-block characteristics
      • 10.4.4 Practical implications for modelling and interpretation
    • 10.5 Collinearity assessment
  • 11. Modelling Results
    • 11.1 Overview of modelling framework
    • 11.2 Exposure modelling
      • 11.2.1 Intersection exposure model
      • 11.2.2 Mid-block exposure model
    • 11.3 Pedestrian risk modelling
      • 11.3.1 Intersection risk model – FSI model
        Interpretation
      • 11.3.2 Intersection risk model – casualty model
        Interpretation
      • 11.3.3 Mid-block risk model – FSI model
        Interpretation
      • 11.3.4 Mid-block risk model – casualty model
        Interpretation
      • 11.3.5 Key findings and practitioner considerations – Melbourne CBD
      • 11.3.6 Model application and interpretation
    • 11.4 Empirical Bayes (EB) implementation
      • 11.4.1 Purpose and role
      • 11.4.2 Conceptual overview
      • 11.4.3 Overdispersion parameters for EB application
      • 11.4.4 Implementation in this study
  • 12. The Prototype Risk Visualisation Tool
    • 12.1 Map view
    • 12.2 Map and dashboard view
    • 12.3 Pedestrian safety indicator view
  • 13. Model Application, Validation and Future Improvements
    • 13.1 Improving how high-risk pedestrian locations are identified
    • 13.2 Modelling, validation and forecasting – potential Auckland and Brisbane CBD studies
    • 13.3 Improving the prototype tool
      • 13.3.1 Estimating the benefits of safety improvements
      • 13.3.2 Validation and calibration process
      • 13.3.3 Updating the base prediction using historical crash data (Empirical Bayes method)
      • 13.3.4 Expanding the prototype tool to cover other types of urban road
    • 13.4 Multi-criteria analysis
  • 14. Conclusions and Next Steps
    • 14.1 Staged roadmap for future development
      • 14.1.1 Governance and resourcing the roadmap
      • 14.1.2 Key limitations and risks
    • 14.2 Recommendations
  • References
  • Appendix A Definition of Key Road and Pedestrian Safety Concepts
  • Appendix B Development of Safety Performance Indicators
    • B.1 SPI frameworks – Victoria and New Zealand
    • B.2 Development of pedestrian SPIs
      • B.2.1 Long list of SPIs
      • B.2.2 Short list of SPIs
      • B.2.3 Consolidated short list of SPIs
    • B.3 Assessment of refined pedestrian SPIs
  • Appendix C Detailed Data Availability Assessment
  • Appendix D Detailed Variable Definitions
    • D.1 Overview of data sources
      • D.1.1 Infrastructure data
      • D.1.2 Australian Bureau of Statistics (ABS) data
      • D.1.3 Land use data
      • D.1.4 Crash data
      • D.1.5 Movement and Place data
      • D.1.6 Other datasets
    • D.2 Detailed description by dataset
      • D.2.1 Infrastructure data
      • D.2.2 ABS data
      • D.2.3 Land use data
      • D.2.4 Crash data
      • D.2.5 Movement and Place data
      • D.2.6 Other datasets
  • Appendix E Descriptive Statistics of Data – Intersections
    • E.1 Categorical characteristics of intersections
      • E.1.1 Highly urban signalised network
      • E.1.2 Strong public transport environment
      • E.1.3 Pedestrian infrastructure provision
      • E.1.4 Crosswalk design characteristics
    • E.2 Continuous variable characteristics
      • E.2.1 Injury occurrence
      • E.2.2 Pedestrian exposure
      • E.2.3 Geometric scale of intersections
      • E.2.4 Turning movement complexity
      • E.2.5 Traffic exposure
      • E.2.6 Land use intensity
      • E.2.7 Overall interpretation
    • E.3 Summary
  • Appendix F Descriptive Statistics of Data – Mid-block Data
    • F.1 Categorical characteristics of mid-blocks
      • F.1.1 Road hierarchy and functional role
      • F.1.2 Crossing provision and pedestrian exposure risk
      • F.1.3 Median and kerbside environment
      • F.1.4 Public transport context
      • F.1.5 Pedestrian protection treatments
    • F.2 Continuous variable characteristics
      • F.2.1 Injury occurrence
      • F.2.2 Pedestrian activity indicators
      • F.2.3 Geometric scale of mid-blocks
      • F.2.4 Speed environment
      • F.2.5 Traffic exposure
      • F.2.6 Land use intensity and spatial heterogeneity
    • F.3 Summary
  • Appendix G Descriptive Statistics of Crash Data
    • G.1 Intersection crashes
    • G.2 Mid-block crashes
    • G.3 Implications for modelling
  • Appendix H Collinearity Assessment Results
    • H.1 Assessment methodology
    • H.2 Intersection models – multicollinearity assessment
      • H.2.1 Geometry and lane structure variables
      • H.2.2 Population and age structure variables
      • H.2.3 Traffic exposure variables
      • H.2.4 Binary variable associations
      • H.2.5 Multinomial categorical associations
      • H.2.6 Overall assessment
    • H.3 Mid-block models – multicollinearity assessment
    • H.3.1 Geometry, speed and lane structure variables
      • H.3.2 Population and demographic variables
      • H.3.3 Traffic exposure variables
      • H.3.4 Binary variable associations
      • H.3.5 Nominal categorical associations
      • H.3.6 Overall assessment
  • Appendix I Detailed Modelling Results
    • I.1 Exposure modelling – modelling approach
    • I.2 Intersection exposure model
      • I.2.1 Model specification
      • I.2.2 Sample size and data coverage
      • I.2.3 Goodness-of-fit statistics
      • I.2.4 Tests of model effects
      • I.2.5 Parameter estimates and model formulation
      • I.2.6 Practical interpretation and model credibility
      • I.2.7 Modelling decision
    • I.3 Mid-block exposure model
      • I.3.1 Case processing and data sufficiency
      • I.3.2 Goodness-of-fit
      • I.3.3 Tests of model effects
      • I.3.4 Parameter estimates
      • I.3.5 Model formulation
      • I.3.6 Interpretation for practitioners
      • I.3.7 Modelling decision and implications
    • I.4 Pedestrian risk intersection model (fatalities and injuries)
      • I.4.1 Model specification
      • I.4.2 Overall model performance and goodness-of-fit
      • I.4.3 Parameter estimates and incidence rate ratios (IRRs)
      • I.4.4 Interpretation of key effects
      • I.4.5 Observed versus predicted performance
      • I.4.6 Model adequacy and limitations
    • I.5 Pedestrian risk intersection model (casualties)
      • I.5.1 Model specification
      • I.5.2 Overall model performance and goodness-of-fit
      • I.5.3 Parameter estimates and incidence rate ratios (IRRs)
      • I.5.4 Interpretation of key effects
      • I.5.5 Observed versus predicted performance
      • I.5.6 Model adequacy and limitations
    • I.6 Pedestrian risk mid-block model (FSI)
      • I.6.1 Model specification
      • I.6.2 Overall model performance and goodness-of-fit
      • I.6.3 Parameter estimates and incidence rate ratios (IRRs)
      • I.6.4 Interpretation of key effects
      • I.6.5 Observed versus predicted performance
      • I.6.6 Model adequacy and limitations
    • I.7 Pedestrian risk mid-block model (casualties)
      • I.7.1 Model specification
      • I.7.2 Overall model performance and goodness-of-fit
      • I.7.3 Parameter estimates and incidence rate ratios (IRRs)
      • I.7.4 Interpretation of key effects
      • I.7.5 Observed versus predicted performance
      • I.7.6 Model adequacy and limitations
    • I.8 Sensitivity analysis for inclusion of proxy exposure measures
  • Appendix J Empirical Bayes Methodology and Implementation
    • J.1 Purpose of the Empirical Bayes approach
    • J.2 Conceptual framework
    • J.3 Mathematical formulation
      • J.3.1 EB weight factor
    • J.4 Source of overdispersion parameters
    • J.5 Interpretation of dispersion behaviour
      • J.5.1 Intersection FSI model
      • J.5.2 Mid-block FSI model
      • J.5.3 Casualty models
    • J.6 Implementation procedure
    • J.7 Strengths of the adopted EB framework
    • J.8 Limitations and considerations
    • J.9 Concluding remarks
  • Appendix K Prototype Risk Tool User Guide
    • K.1 Map with map layer list selected (initial view)
    • K.2 Map with legend selected
    • K.3 Map and dashboard (initial view)
    • K.4 Map and dashboard (selectable SPIs from a feature)
    • K.5 Pedestrian safety indicators
      K.6 Pedestrian safety indicators (with an indicator selected)
Related publications
Austroads Pedestrian Safety Blueprint
AP-G118-26
Webinar: Managing Transport Management Centre Workplace Fatigue and Psychosocial Hazards
WEB-G116-26
Australasian Pedestrian Facility Selection Tool [V2.3.1]: User Guide
AP-R776-26
View all related publications
Latest Road Safety News
31 Aug 2026
New framework connects road design, traffic management and road safety

Road design, traffic management and road safety each play a distinct role in transport planning and delivery, yet decisions made in one discipline can significantly influence outcomes in the others. New Austroads guidance brings these fields together to support a more coordinated approach to planning, designing, operating and managing transport infrastructure.
22 Jul 2026
Updated guidance helps road agencies put Safe System principles into practice

Road authorities, local governments and road safety practitioners will be better equipped to plan safer speeds and strengthen road safety leadership through updates to two key parts of the Austroads Guide to Road Safety.
View all news