The datasets can be found at
ftp://ftp.nhtsa.dot.gov/nass
which can be accessed as follows: From
http://www.nhtsa.gov/people/Crash/crashstatistics, click on
NHTSA's National Center for Statistical Analysis (NCSA).
Then select "downloadable Data Files & High Interest Cases"
from the pull-down menu. Then select
CDS Annuals SAS & Flat Files
You should see Directories from 1988 through 2003 (currently).
For each year, the folder called ASCII contains a zip file; this can be downloaded an unzipped, to find SAS code to read the flat data file and produce 11 SAS datasets, of which we have used four. Another SAS program file contains labels for the levels of the variables.
The documentation for the datasets can be downloaded from the same site.
The four SAS datasets used in this article are the "occupant assessment" file, providing information about the occupant and injuries
the "general vehicle" file, providing information about the impact speeds and the vehicle model
the "vehicle exterior" dataset provides the direction of impact
the "accident description" dataset provides the "ratio inflaction factor" used for weighting the data.
These datasets are provided for each year. Individual cases are distinguished by year of accident, primary sampling unit, and case ID. The case IDs refer to the accident; within accident individual cars are identified by a "vehicle number" and within vehicles, each occupant is identified by an "occupant number". Using these identifiers, the user may combine the datasets.