Filtering out confounding effects, such as regression to the mean (RTM) and trend, is crucial when using road safety data to make decisions. For example, ignoring such effects can exaggerate estimates of treatment effects in road safety scheme before-and-after studies; similarly, such effects should be properly handled when attempting to make predictions of road safety hotspots in future time periods.

The Newcastle University Statistics of Road Safety Group (NUSRG) has developed a software tool to enable road safety practitioners to make better data-informed decisions. The tool’s main dashboard supports simple data visualisation, as well as the fitting of statistical models that enable the implementation of cutting-edge methods for handling confounders such as RTM and trend.

In this presentation we will demonstrate the tool, describe the underpinning methods and give summaries of past and ongoing collaborations with road safety practitioners across the UK, including road safety partnerships in Cumbria, North Yorkshire and Gateshead; and internationally with organisations in Lisbon and New York, as well as commercially with PTV Group in Germany and LOGIT in Portugal. The aim of this presentation is to raise awareness of this work among practitioners in the hope of starting new collaborations with organisations that can use these methods to support their road safety decision making.

Lee Fawcett, Reader in Applied Statistics and Associate Dean (Global), Newcastle University

Lee Fawcett is a Reader in Applied Statistics and Associate Dean (Global) at Newcastle University, with research interests in statistical methods for road safety data analytics and prediction of environmental extremes.

Lee has 20 years’ experience working with road safety practitioners in the UK, Europe, the US and Latin America, mainly in the area of road safety scheme evaluation and predictive analytics.

Joe Matthews, Senior Lecturer in Statistical Data Science, Newcastle University

Joe Matthews is a Senior Lecturer in Statistical Data Science at Newcastle University, with research interests across a wide variety of areas of applied statistics.

Joe’s main research area of interest is applying statistical methods to road safety data analysis, particularly in the areas of collision hotspot prediction, and retrospective scheme evaluation, research into which was the topic of his PhD. As part of the road safety team at Newcastle, Joe has collaborated with local authorities across the UK and abroad, primarily through the RAPTOR suite of software tools he co-developed in order to allow road safety practitioners to analyse and interpret their data effectively.