Development and validation of an Australian Video Speed Test (AVST)
Keywords: Strategy & Theory
Submission Date: 2003
Anti-speeding educational campaigns (in television commercials, print ads, and outdoor ads, mostly) are constantly being tried but it is difficult to determine which ads are effective in reducing speed. A promising solution to this problem is to use a behavioural simulation such as the Video Speed Test, the VST (Horswill and McKenna, 1999). The driving simulation test involves getting drivers to view video excerpts of a person driving a vehicle in real driving situations. The drivers then are asked to estimate the speed that they would use in the same situations, that is, how many kilometres/hour slower or faster they would drive. This test has been shown to correlate highly with self-reported habitual speeding and also correlate highly with drivers? past involvement in speed-related accidents.
The present study develops and validates an equivalent Australian version of the driving simulation test, called the Australian Video Speed Test (AVST). A variety of roads and driving situations were filmed, including driving through residential areas, country roads, roundabouts, shopping centres, on freeways, and around bends. An initial set of 15 scenes were selected for testing, using responses from 268 drivers, with age categories ranging from 17-25 to 66+, who were recruited via mall-intercept interviews in the Illawarra region of New South Wales, Australia. Analysis of variance (ANOVA), which tests for differences between means of different populations (such as: young drivers? speed choice versus older drivers? speed choice; males? speed choice versus females? speed choice), was conducted for all demographic and driving history variables. Based on this analysis, the 15 scenes were reduced to a final set of 11 scenes, including 1 practice scene and 10 test scenes.
Overall, the AVST was validated by the fact that it correlated highly with self-reported habitual speeding (r =.52, p =.01). The 10-item AVST has high internal-consistency reliability, with an alpha coefficient of 0.83.