Section II Rail Level Crossing Data Collection and Analysis

Understanding the Factors Influencing User Behaviour

With Contributions from:

Ashleigh Filtness, Kristie Young,

Christine Mulvihill, Casey Rampollard and Nebojsa Tomasevic


This chapter will describe Phase 1 of the research programme, which involved a series of data collection activities aimed at understanding how road users interact with existing rail level crossing systems. Road user behaviour was investigated via on-road studies, observations, interviews and surveys. Train driver experiences were explored during cab rides through urban and rural areas. Data collection methods were deliberately structured so that the data could inform Cognitive Work Analysis (Jenkins et al. 2009, Vicente 1999) and Hierarchical Task Analysis (Stanton 2006), which respectively provide formative and normative descriptions of system functioning (see Chapter 5).

The impetus for these data collection activities arose from a review of the existing literature on human factors issues in level crossing safety (Edquist et al. 2009). The most effective existing solutions to improve safety, namely, grade separation and upgrading from passive signage to boom gates, are generally considered cost prohibitive (Cairney et al. 2002, Wigglesworth and Uber 1991), particularly given there are approximately 8,838 public road level crossings in Australia and 67% of these have passive controls (i.e. signage) only (RISSB 2009). However, many lower cost interventions are not evidence based and/or have not been appropriately evaluated (Edquist et al. 2009). Although a substantive body of research exists examining road user behaviour at rail level crossings, most studies have relatively narrow scope (e.g. focussing on a single road user group) and are not been conducted in a manner consistent with systems thinking approaches (Read et al. 2013). These issues highlighted the need for a more comprehensive, indepth application of human factors methods to understand behaviour and the factors influencing behaviour at rail level crossings.

A multifaceted data collection approach was adopted to meet the requirements for systems-based analysis. Whereas most previous research has focussed on car drivers in isolation (Read et al. 2013), the current research programme was designed to incorporate perspectives from multiple system users, including car drivers, pedestrians, cyclists, motorcyclists and train drivers. It was also designed to capture road user behaviour across diverse types of rail level crossings, as there is no single design that represents a prototypical crossing (Edquist et al. 2009). Further, although there is a strong focus on novice driver safety in the broader road safety literature (Curry et al. 2011, Hatakka et al. 2002, Mayhew et al. 1998, McKnight and McKnight 2003), there was little research examining differences between novice and experienced drivers when specifically interacting with rail level crossings. Therefore, this comparison was considered within this phase of the research programme.

As a final point of note, the review of Edquist et al. (2009) highlighted the need for better data in this area, with part of the shortfall related to the methods used to study rail level crossing safety in previous research. The methods used in this research programme were selected as they provide a much greater level of detail than collected previously, particularly with respect to the cognitive processes underlying road users’ behavioural choices. For example, the use of eye tracking in instrumented vehicles provides much richer data than conventional observational studies where the behaviour of the driver is inferred by observers recording at the roadside.

The methods used to better understand end user behaviour included the following:

  • On-road instrumented vehicle studies of driver behaviour: Two on-road studies were completed: one focussing on urban active rail level crossings and the other focussing on active and passive crossings in a rural area. Participants were required to drive a pre-specified route that encompassed several rail level crossings, using an instrumented vehicle fitted with cameras and data logging equipment. Eye and head movements were recorded to measure drivers’ allocation of visual attention, and their situation awareness was measured through provision of concurrent verbal protocols (i.e. thinking aloud during the drive).
  • Cognitive task analysis interviews of driver behaviour: The on-road studies also involved a post-drive cognitive task analysis interview using the Critical Decision Method (CDM; Klein et al. 1989). CDM uses a series of structured prompts to facilitate recall of past events and probe factors that shaped decision-making. The interviews focussed on participants’ decision-making at rail level crossings (i.e. whether to proceed, slow or stop).
  • Diary study of road user behaviour: A diary study was used to capture data from multiple road user groups across diverse geographic locations. Drivers, pedestrians, cyclists and motorcyclists recorded their daily interactions at rail level crossings over a 2-week period. If participants encountered a train and/or active warnings, they were asked to recount this experience in detail. Diary questions were adapted from CDM probes to capture the factors that influenced decision-making.
  • Subject-matter expert interviews and in-cab familiarisation: Discussions were held with two train drivers and one rail subject-matter expert to gather information regarding train driver behaviour at rail level crossings and their perceptions of road users’ behaviour at rail level crossings. Researchers also participated in train cab rides through urban and rural areas to gain familiarisation with the train driving task and better understand the train driver perspective.

Observations of pedestrian and cyclist behaviour: Structured on-site observations of pedestrian and cyclist behaviour were undertaken at seven urban rail level crossing sites (see Box 2.1). The aim was to understand the range of behaviours exhibited by users of the pedestrian infrastructure at rail level crossings and identify potential factors influencing behaviour. Due to space constraints, we do not report the findings in this chapter, but further information about the observations can be found in the work of Read et al. (2014).

The remainder of this chapter summarises the methods, results and conclusions from the primary data collection activities.

< Prev   CONTENTS   Source   Next >