The Christmas period road toll is a highly publicized 'summary' of the number of road deaths that occur each year during New Zealand's summer holiday period. Our aims were to identify significant changes in the Christmas road toll during 1968-98, highlight the ability to detect changes and determine if focusing on the changes in the Christmas road toll provided a useful insight into trends in fatal crashes. Regression modelling of the number of fatalities was used to examine trends in the Christmas road toll in New Zealand over time. A power analysis was used to investigate the ability to detect significant reductions in the Christmas road toll. The Christmas road toll has not decreased significantly in the past 30 years and has not improved significantly in contrast to the rest of the year. The number of deaths in any given year can be expected to vary 'naturally' between 10 and 39. The statistical power to detect a hypothetical 19% reduction in the Christmas road toll is minimal. Stochastic fluctuations in the Christmas road toll make it extremely difficult to identify underlying trends, even if substantial. Little insight can be gained from yearly comparisons of road tolls considered over short periods. More focus should be placed on road tolls that are calculated over longer periods, e.g. the previous six months. Natural fluctuations arising in small counts will make it difficult to highlight real improvements in road tolls.
Samenvatting