9-8 9.6 CONCLUSIONS The analysis of the annual accident data indicated that there are a number of factors that show similar trends to the number of accidents, e.g. volume of fuel sales a:qq, UUIllber of registered vehi,~les. The anqfsis shows too great a variation to be utiliseq, llS predictors of \he nUlllber of qccidents. The resiq'\lals in historic accident data ip.~ury ~redictiQf1 ~eell to be much less tltan the targ~t set for acciq,ent, or fatality req,\lction. Tl1ere is a need for tne develoPIHl:'lllt of a moq~l tl~a t cap. predict the nump~r of accidents, fatalities , etc. to evaluf\te tlte reductipn in accidents and fatalities from the road traffic accident improvement interventions bqsed on eJ~:posure.