Comparison of UDA and Indiana
Tri-Level Causal Analyses
In this section, the UDA causal analysis results are
compared with the Indiana Tri-Level analysis results (Treat, et al, 1979).
There are several factors to be considered in reviewing comparison results:
- The focus of the Indiana Tri-Level study was identification of all
factors related to crash occurrence. In contrast, the focus of the UDA
study was identification of problem driving behaviors and identification
of situational factors/characteristics associated with these behaviors.
The more limited research objective of the UDA study was likely to result
in an underreporting of environmental and vehicle factors as compared
to the Tri-Level study or other more global studies of causation factors.
- A significant portion of the vehicle related factors in the Tri-Level
study were related to braking system deficiencies (30.8 percent). The
specific deficiencies noted in that study (e.g., gross failures, side-to-side
imbalances, premature lock-up, etc.) occur at much lower frequency levels
in the more advanced braking systems installed in vehicles manufactured
in the 1990s (most UDA study vehicles).
- The UDA study did not utilize the "certain," "probable,"
and "possible" levels to describe causal assignments. UDA
causal assignments were most directly comparable to the probable level
assignments made by the on-site teams in the Tri-Level study.
A comparison of human, environment, and vehicle causal
factors assigned in these two studies is provided in Figure 4. As was
anticipated, there was a pronounced disparity in the assigned levels of
environment and vehicle factors in the two studies. While the levels of
disparity were primarily related to the more limited research objectives
of the UDA study, improvements in vehicle system designs may have also
contributed to the very low level of vehicle factors noted in the UDA
A comparison of the six most frequently assigned human-related
causal factors in the two studies is provided in Figure 5. The UDA incidence
rates are converted from the proportion of drivers contributing to the
proportion of crashes used in the Tri-Level study. The upper portion of
Figure 5 provides a comparison of the four causal groups that were among
the six most frequently assigned causal factors in both studies. The mid
portion of the figure provides a comparison of two causal factors that
were part of the six most frequently assigned causal factors in the UDA
study, but that did not appear in the six most frequently assigned causal
factors in the Tri-Level study. Finally, the lower portion of the figure
provides a comparison of two causal factors that were part of the six
most frequently assigned causal factors in the Tri-Level study, but that
did not appear in a similar distribution for the UDA study. Major findings
may be summarized as follows:
Four Common Causal Factor Groups
- The driver inattention category, as defined in the UDA study, was
comprised of the driver inattention and driver distraction categories
as defined in the Tri-Level study. This factor was assigned to 23.0
percent of the crashes in the UDA study and 20.3 percent of the crashes
in the Tri-Level study.
- The excessive speed category was assigned to 18.9 percent of the crashes
in the UDA study and 14.7 percent of the crashes in the Tri-Level study.
- The UDA perceptual error category (15.3 percent) was directly comparable
to the Tri-Level improper lookout category (20.3 percent). Both category
labels were somewhat arbitrary in nature. It is also interesting to
note that both studies found an over-representation of older drivers
in this category.
- The UDA decision error category (10.1 percent) was directly comparable
to Tri-Level false assumption category (11.8 percent).
- In general, these four common factors demonstrated a remarkable degree
of consistency over time. Specifically, these factors were assigned
to 67.4 percent of the UDA crashes and 66.8 percent of the Tri-Level
UDA Alcohol Impairment and Incapacitation Factors
- The alcohol impairment factor was assigned to 18.4 percent of the
UDA crashes and only 6.1 percent of the Tri-Level crashes. As stated
in the Tri-Level report, that study experienced a very high incidence
rate of property damage only crashes. The report authors believed that
this large property damage incidence rate accounted for the relatively
low level of alcohol related crashes. On the other hand, the UDA study
had an overrepresentation of serious injury crashes. Other studies of
injury crashes (Terhune and Fell, 1981) show alcohol involvement at
about 20 percent.
- The UDA incapacitation category (comprised of drivers who fell asleep
or experienced a heart attack, seizure, or blackout) was assigned
to 6.5 percent of the UDA crashes and was comparable to the Tri-Level
critical non-performance category which was assigned to 1.4 percent
of the Tri-Level crashes. The UDA rate is consistent with other causal
analyses completed with NASS data. The relatively low rate reported
in the Tri-Level study may again be related to the high incidence
of property damage only crashes in that study.
Tri-Level Improper Evasive Action and Improper Maneuver
- The improper evasive action category was assigned to 10.3 percent
of the Tri-Level crashes and 2.1 percent of the UDA crashes.
- The improper maneuver category was assigned to 7.1 percent of the
Tri-Level crashes and 3.4 percent of the UDA crashes.
- The disparity level in the assignment frequencies for these categories
appeared to be associated with the classification scheme used to designate
alcohol-related crashes in the UDA study. In this effort, these behaviors
were assumed to be part of the alcohol designation. Specifically, the
only additional factors that were routinely recorded in alcohol-related
crashes in the UDA study were excessive vehicle speed and traffic control
device violations. A clinical review of a sample of UDA alcohol-related
crashes indicated that if these factors were added to the alcohol designation,
the UDA incidence rate for improper evasive action would increase by
a factor of two to three times and the incidence rate for improper maneuver
would nearly double in size.