International Conference on Engineering Education July 25–29, 2005, Gliwice, Poland. 1 Comparing the use of a ‘tabletop’ experiment and a collaborative desktop virtual environment for training police officers to deal with traffic accidents Authors: Ahmed BinSubaih, University of Sheffield, UK, a.binsubaih@dcs.shef.ac.uk Steve Maddock, University of Sheffield, UK, s.maddock@dcs.shef.ac.uk Daniela Romano,University of Sheffield, UK, d.romano@dcs.shef.ac.uk Abstract  The Dubai police force currently trains its accident investigators using a mixture of theoretical training and on-the-job practical training. In this paper, we investigate the suitability of a virtual environment (VE) for training and performance testing traffic accident investigators. To justify the investment for virtual training we compare trainees’ performances against the less expensive method of using a tabletop exercise. Our results show that both our proposed training methods managed to improve the overall trainees’ performances. However, overall we have not found a significant difference in performance that could be clearly attributed to one method rather than the other. There are some indications that there is an advantage for one method of training over another according to the task required at a particular investigation stage. Subjective comments from the trainees favoured the use of the virtual environment. Both environments received positive remarks from the trainers who saw the potential of using them as training and testing environments before sending the trainees to real traffic accidents. Index Terms  traffic investigators training, tabletop training, virtual environment training. INTRODUCTION The increase of computing power and its wide availability has raised interest in the use of VEs. A number of environments have been developed over the years, looking at such things as training fire-fighters [1], police officers [2] and navy personnel [3]. We have noticed that the focus of many of the environments built for the traffic accidents investigation field concerns accident reconstruction such as [4,5]. Our work differs from these environments by focusing on the accident investigation for which we have not yet found any virtual training environment. In this work we examine the suitability of such environments in training traffic accident investigators for the Dubai police force. Currently students at Dubai Police Academy receive theoretical training material on accident investigation during their course, followed by on-the-job training after the graduation under the supervision of an experienced officer. Some students also have the option of attending two specialized modules on traffic investigation. We hypothesize that there exists a large gap between theoretical and practical training which needs to be bridged by another form of training environment. We also hypothesize that the current on-the-job training on its own is not sufficient, for three reasons. First, the students get exposed to a limited range of accident types while under supervised training. According to the on-the-job training each student is assigned to work in a patrol unit under an experienced investigator, and each unit has a specific jurisdiction assigned to it in a specified area of the city. This approach for student assignment is not ideal as the accident types and frequency differs from one area to the other which means that some students might only be supervised on limited type of accidents and might never get exposed to others. Second, the evaluation of each student is done in a subjective manner by relying on the supervisor’s views. We believe that a more objective evaluation mechanism needs to be put in place to get a fairer evaluation for each individual. Third, there is no agreed period for the supervision period or the number of accidents the trainee has to investigate. The non-existence of a supervision period can be understood as it is very difficult to predict the accidents that might occur during a specified training period. However, the number of accidents can be a more accurate measure and possibly putting a minimum number on the different types of accidents that a trainee has to investigate before completing the supervised training period. Our case study was conducted over a period of two months and it was separated into two main phases: gathering knowledge and running the experiments. The first phase consisted of interviews, on-the-job observations, exercising the expert, and material analysis. The second phase involved running the experiments and the debriefing sessions that followed. The rest of the paper is organized as follows. The next section gives the reasons for the need for a practical training environment by giving real examples. This is followed by a section which describes the virtual traffic accident scenario used, the experiments design approach, the stages an investigator must complete in the scenario, and how the scoring system works. Afterwards, we present the results and discuss their implications. We also compare our results with the