In addition we find surprising results for some of the other variables that we analyze,
in particular the average vintage of vehicles on the road. Our analyses suggests that as the
average age of registered vehicles increases, fatalities are reduced. This result suggests that
changes in the type of vehicles or their relative safety may be increasing overall fatalities.
Various factors have been determined to be associated with reductions in traffic-
related fatalities. Previous research has determined that decreases in alcohol consumption,
the proportion of younger drivers, the amount of seat-belt usage, road design characteristics,
traffic speed and speed variance, and vehicle safety improvements have all been associated
with the level of traffic-related fatalities.
Much of this work has utilized macro-modelling techniques that exploit time-series
cross-sectional data sources. Recent examples of this work include Voas et al. (2000) and
Whetten-Goldstein et al. (2000) on the effect of alchohol consumption laws, McCarthy
(1999) on policies including seat belt laws and speed limits, Noland (2001a) on highway
infrastructure, and Dee (1998) on seat-belt laws. Hakim and Shefer (1991) summarize many
of the results of macro-modelling studies.
One characteristic of the majority of this work is that it generally uses US data,
primarily at the state-level. This is due partly to the easy availability of state-level data in the
US and the lengthy time-series that are becoming increasingly available. Fridstrom &
Ingebrigsten (1991) analyzed data from Norwegian counties and Karlaftis & Tarko (1998)
analyzed county data from the US state of Indiana. Some studies have also analyzed
international data using cross-sectional time-series analyses, including Noland (2001b) and
Page (2001).
We are unaware of these techniques being applied using data from Great Britain.
This study analyzes regional data based on the standard statistical regions of the UK (see