|
Paper Title:
Implementation of Victoria's
new Hazard Perception Test
View
full paper
Authors:
John Catchpole, Corinne
Leadbeatter, Peter Congdon
Abstract:
In 1999, the Registration
and Licensing Department of VicRoads commissioned a team led by
ARRB Transport Research Ltd (ARRB TR) to update and extend the Hazard
Perception Test (HPT). This paper describes the methods and results
of a major trial of the newly developed test items and final construction
of the test based on the results of the trial.
The sample of drivers
who participated in the trial included current learner permit holders,
probationary licence holders and a small group of highly experienced
drivers. Item responses from 363 participants recruited from metropolitan
and country areas were available for analysis. 114 new test items
were trialled. Analysis of the response data collected during the
trial established the relative difficulty of all items and assessed
the extent to which the items work together to represent a single
underlying trait, in this case a candidate’s driving hazard
perception ability. Some items were dropped from the pool due to
poor fit with the underlying trait measured by the majority of items.
This procedure resulted in a final pool of 90 items, from which
four items were set aside for exclusive use as practice items.
Fifteen fixed test forms
were constructed from the final item pool, each containing 28 test
items and four practice items. All forms are matched in difficulty
and in the number of items addressing each type of accident. Each
form contains a range of correct response window locations (early,
middle, late or no correct response window). Thus candidates cannot
expect to pass the test by systematically responding in the same
way to every item. The trial revealed that responses made by touching
the screen were systematically delayed when compared with responses
to the same items made by clicking a mouse button, but there was
little or no difference in response variability and there was no
interaction between items and response modes. The test is expected
to function equally well using either response mode. The reliability
of the updated test to separate the trial population based on ability
is 0.68, which compares favourably with the reliability value of
0.27 for Victoria’s original Hazard Perception Test.
View
full paper
Back
to Program
|
