Validation of in-car observations, a method for driver assessment

Research output: Contribution to journalArticle


An in-car observation method with human observers in the car was studied to establish whether observers could be trained to observe safety variables and register driver's behaviour in a correct and coherent way, and whether the drivers drove in their normal driving style, despite the presence of the observers. The study further discussed the observed variables from a safety perspective. First three observers were trained in the observation method and on-road observations were carried out. Their observations were then compared with a key representing a correct observation. After practising the observation method the observers showed a high correlation with the key. To establish whether the test drivers drove in a normal way during the in-car observations, comparisons of 238 spot-speed measurements were carried out. Driver's speeds when driving their own private cars were compared with their speeds during the in-car observations. The analysis showed that the drivers drove in the same way when being observed as they did normally. Most of the variables studied in the in-car observations had a well documented relevance to traffic safety. Overall, in-car observation was shown to be a reliable and valid method to observe driver behaviour, and observed changes provide relevant data on traffic safety. (C) 2003 Elsevier Ltd. All rights reserved.


Research areas and keywords

Subject classification (UKÄ) – MANDATORY

  • Infrastructure Engineering


  • method, safety assessment, driver behaviour, in-car observations, validation
Original languageEnglish
Pages (from-to)127-142
JournalTransportation Research. Part A: Policy & Practice
Issue number2
Publication statusPublished - 2004
Publication categoryResearch