Traffic accidents have long been among the leading causes of deaths in industrial societies (Peltzman, 1975; Kenkel, 1993). Whether these traffic accidents are the result of drunk driving (commonly known as Driving Under Influence or DUI in America), speeding, or other unsafe driving behaviours, the injuries and fatalities that often result are devastating and can inflict tremendous costs on both the individuals involved and society.
In 1994, there were 580 fatalities and 16,600 injuries resulting from traffic accidents in New Zealand (LTSA, 2001), and the corresponding annual social cost is estimated to be NZ$3 billion (LTSA, 1996). In an effort to reduce the road toll in New Zealand, the Land Transport Safety Authority (LTSA) initiated a Supplementary Road Safety Package (SRSP) in October 1995.
The SRSP was modelled after a similar campaign in the Australian state of Victoria which contributed to reductions of around 50% of the road fatalities and 40% of the serious injuries in the three year period from 1989 to 1992 (Cameron et al., 1993).
Similarly, the LTSA had set a target for the New Zealand SRSP of a reduction of 80 fatalities, 450 serious injuries and 1600 minor injuries over a four year period. Although the complete SRSP involves the use of speed cameras, advanced laser speed detectors and compulsory breath testing (CBT) (Cameron and Vulcan, 1998), the major component (budget-wise) of the package is the use of sustained publicity to support the speed camera and CBT tools.
This publicity has taken the form of a paid advertising campaign primarily aired on television and focused on the drunk-driving and speeding themes. The advertising campaign was estimated to cost about NZ$7 million in the first year and was subsequently increased to NZ$9.8 million per year (Marketing, 1998).
It therefore constitutes a very substantial portion of the total four year SRSP budget of NZ$50.06 million (Cameron and Vulcan, 1998). The primary appeal used in the SRSP television advertising campaign has been the appeal to the emotion of fear, highlighting the consequences of unsafe driving behaviour in an extremely graphic and shocking manner.
Due to the realistic and highly graphic nature of these advertisements, it could be argued that this campaign uses a strong appeal to the emotion of fear. By implying some sort of threat or danger which evokes the emotional response of fear, advertisers and public policy makers can influence behaviour by showing individuals ways of coping with the threat and thus removing the danger (Tanner et al., 1991).
A good deal of public debate, however, has arisen regarding the effectiveness and appropriateness of an advertising campaign that uses such a strong appeal to fear (Falconer, 1996; Gregg, 1996; Karnes, 1996; Jones, 1998; O'Hanlon, 1998; Keene and Taylor, 1998; Turner, 1998).
Given the seriousness of this issue and the total cost of the campaign, it is crucial that the campaign be evaluated in terms of its effectiveness in reaching its goals. Even though a number of studies have been conducted to examine the effectiveness of the campaign, no consistent conclusion can be drawn from these studies.
For instance, Macpherson and Lewis (1996; 1998) indicated that their study provided no evidence to suggest that the LTSA advertising campaign had any substantial effects on drunk driving behaviour (drink driving offence rate) and White (2000) found no significant effects on the number of serious crashes.
Other researchers, however, found the advertising campaign to exhibit significant negative relationships with road fatalities and morbidity (Cameron and Vulcan, 1998; Tay 2001) and drink driving offence rates (Tay, 1999).The purpose of this research is to evaluate the effectiveness of the LTSA advertising campaign on fatal accident rates, related to alcohol, drugs and speed, in New Zealand.
This research differs from previous efforts in that it examines the effectiveness of the campaign on fatal accident rates among different segments of drivers in New Zealand. This approach is justified given that previous researchers have found that fear appeals are more effectively used with certain groups and not the entire population (Higbee, 1969; Burnett and Oliver 1979; Burnett and Wilkes 1980; King and Reid, 1990; Quinn et al.1992; LaTour and Rotfeld, 1997).
In addition, it utilises the Poisson regression model instead of the standard regression model used by Macpherson and Lewis (1998), Cameron and Vulcan (1998), Tay (1999, 2000) and White (2000) which assumes that the accident rates are continuous and normally distributed.
However, since accident rates are discrete variables and are more likely to have a Poisson distribution (Nicholosn, 1985; Nicholson and Wong, 1993; Michener and Tighe, 1992; Hauer, 1997), the use of standard regression models will result in inefficient, inconsistent and biased estimates (Maddala, 1983; Greene, 1997).