road traffic accidents among commercial commuter drivers
Model approach
Because data replication is not feasible in most accident investigation studies, the use
of factorial design becomes difficult. The problem of degrees of freedom and the
number of runs needed for large number of predictor variables (25 in this case) would
be prohibitive (Mendenhall, 1968). The adopted approach was to use the least square
method for curve fitting the collected data.
Development of response surface equation
To investigate the extent of contributions to road traffic an accident by each identified
variable, a response surface equation of the following type is commonly adopted
(Weissten, 2007):