Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter.
Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 69
69
occupant-oriented strategies do not exist. There are also some surrogate measures that may be related to safety (e.g., behav-
other differences between highway-oriented strategies and ioral changes that may be related to crash/injury reduction).
driver-oriented strategies that need to be recognized in se- As noted earlier, if the user is considering roadway-related
lecting treatment programs and establishing crash-reduction strategies for drowsy and distracted drivers described in the re-
goals. The first relates to the data source and "ownership" of lated guide, and if the considered strategies have known
the treatment delivery system. In contrast to many of the effectiveness measures, then Procedures 1, 2A, and 2B in
highway countermeasures, most of the effectiveness measures Sections IV and V should be used. The issue here will be defin-
for these driver and vehicle occupant strategies do not relate ing the proportion of all drivers involved in lane-departure
to crash rates on sections or type of roads. Instead, the safety crashes or intersection crashes who are drowsy and distracted.
concern usually relates more to overall crash rates, perhaps Some guidance on defining such drivers will be given under
subdivided by severity. The data on which problem driver "Data Needs" below. For the remaining strategies which do
identification and effectiveness measurements are based not have known effectiveness (AMF) measures, the recom-
(traffic convictions and crashes) usually reside in DMV files. mended method for choosing and targeting strategies is a
The information on previous convictions may sometimes be modified version of Procedure 3 described in Section III. This
added to crash files. procedure could be used for any of the unsafe driving strate-
It will also be noted that some of the strategies proposed in gies found in either of the two guides.
the guides could require the enactment of legislation, de-
pending on the state in question. For example, increasing seat Procedure 3 Choosing Treatments
belt usage may require upgrading from secondary to primary and Target Subgroups Related To Unsafe
enforcement legislation. Selection of a strategy requiring leg- Driving Actions When Treatment
islation entails an assessment of the likelihood that the legis- Effectiveness in Terms of Crash/Injury
lation could be enacted in the required time-frame. Reduction Is Unknown
Another consideration is cost. An assessment of cost for
many of the proposed strategies will require subjective The assumption here is that, for the majority of the strate-
approximations. Some general information on treatment gies, there is no known level of effectiveness no defined
cost is presented for each treatment in both of the guides, CRFs or AMFs. Thus, economic analyses like those that are
and that information should be reviewed by the user. Very the basis for Procedures 1, 2A and 2B, and 4 are not possible
costly strategies should be avoided unless supported by for these treatments. Procedure 3 is aimed at helping the an-
proven effectiveness data and an estimated effect size that is alyst make an educated choice of which treatments will be
sufficient in economic terms (dollar benefits) to be cost- most effective in his or her jurisdiction, and to help the ana-
beneficial or cost-effective. Strategies that are judged to have lyst develop a targeting strategy for the treatment in cases
negligible or moderate operational costs (excluding start- where it is not to be applied jurisdiction-wide or to the "total
up) will usually be cost-beneficial if they produce statisti- problem" (e.g., where specific unsafe-driver subpopulations
cally significant annual crash reductions as small as 510 or jurisdictions are to be targeted).
percent over baseline. However, unlike road user populations covered in other
The guides classify strategies into three categories: guides (e.g., older drivers or pedestrians), the choice between
alternative treatment strategies found in each of these two
1. Proven unsafe-driving guides is much less oriented to specific crash
2. Tried but not proven circumstances such as different crash types, crash location
3. Experimental not tried, effectiveness unknown (except for the roadway-based strategies), and times of crash
(except perhaps for drowsy drivers). Instead, most of the
In selecting treatment strategies, priority should be given strategies are related to improvements in programs, such as
to strategies rated as proven. However, the safety planning increasing seatbelt usage. In addition, for both the guide re-
team is encouraged to use their own judgment and to inde- lated to drowsy and distracted driving and the guide related
pendently review the evidence cited in the guides in selecting to increasing seatbelt usage, the strategies are related to the
treatments. The tried category includes those treatments that full group of such drivers and vehicle occupants who under-
have been used by agencies (in some cases used often), where take such unsafe actions. Some limited subpopulation-
there is little possibility of negative impacts on crash/injury targeting is possible for the occupant restraint strategies
frequency, and where there is an expectation (but not scien- (children vs. other occupants) and for the drowsy and dis-
tific proof) that the effect of the treatment on safety is likely tracted driver strategies (i.e., teen drivers, adult drivers, and
to be a positive one. The evidence could include poorly heavy truck drivers). Some additional targeting of a chosen
designed or executed crash/injury evaluations and indirect or strategy can occur based on jurisdiction and selected areas
OCR for page 70
70
within a jurisdiction. However, additional targeting based on Crash databases often categorize data for a given crash into
crash types and other crash data are not generally applicable up to three subfiles (1) general accident/crash variables
with these strategies. ("crash"), (2) variables for each vehicle in the crash
For these reasons, the general analysis methods presented ("vehicle"), and (3) variables for each occupant/person in the
under Procedure 3 in other sections of this manual are not as crash ("person" or "occupant"). The variables needed to de-
applicable here. For that reason, only a modified Procedure 3 termine whether a crash is a "target-population crash" are
is presented below one which continues to use relative esti- usually found in the occupant/person subfile, but could also
mates of the program effectiveness for different alternative be found in the general crash subfile (e.g., a "flag" for head-
treatments, but one that does not include further targeting on and run-off-road crashes in general) or "vehicle" subfile
steps based on crash circumstances. (e.g., driver information included with each vehicle record).
In short, crash files differ from jurisdiction to jurisdiction.
While certainly not always the case, the variables (or similar
Data Needs
variables) listed in Exhibit VIII-1 will be used in this iden-
Note that Procedure 3 is a "crash-based" procedure. It as- tification of "target-populations." Thus, the defining variables
sumes that the analyst wishes to choose among the alternative will depend on the user's definition of fatigue-related crashes
strategies and target the treatments based on crash data. It is (e.g., late-night crash involvement with no indication of DUI,
noted that an alternative way of making such choices is especially head-on or run-off-road crashes). Drivers involved
through linking crash data related to problem size to an as- in nighttime crashes, especially those who are not under the
sessment of the existing programs in a jurisdiction, and choos- influence of alcohol, are a logical subpopulation to consider
ing to implement those strategies which are either missing for fatigue involvement, although fatigue involvement can
from the current program or have the least extensive (or least also clearly occur during other time periods as well, and treat-
effective) degree of implementation. This program-deficiency ments that help drowsy drivers may also help drivers who are
procedure is described more fully in a later section. under the influence of alcohol. Research on human circadian
However, if the analyst wishes to choose and target treat- rhythms indicates that early afternoon is also a period when
ments based on crash data, the revised Procedure 3 described drowsiness is likely. Since there is no broadly accepted defi-
here basically requires crash data that will allow the analyst to nition of distracted driving crashes, defining specific crash
(1) isolate crashes involving the specific user population of types related to distracted driving may be difficult or impos-
interest (e.g., drivers involved in fatigue-related crashes or sible. Some crash files may include a variable on "distrac-
crashes involving unbelted vehicle occupants), and (2) define tion." Indeed, the MMUCC guidelines for crash variables
crash types or crash characteristics (e.g., crashes involving include such variables (i.e., P16. "Driver Distracted By").
unbelted occupants in specific age ranges) for this user pop- Narratives written by the investigating officer may include
ulation which would suggest strategies and target subgroups. driver and witness reports and the officer's own impressions
To isolate crashes involving the population of interest, the about possible distractions or fatigue. While reading narra-
analyst will need to examine the data formats/coding in their tives on every crash report can be much more time-consum-
crash file to identify variables that can be used in determining ing than simply scanning for a coded "distraction" variable,
whether or not a given crash is a "target-population crash." these statements can provide a wealth of information on the
Population Type Variable Crash Database Subfile
Drivers Involved in Fatigue- Fatigue Involvement Captured Person/Vehicle
Related Crashes under "Driver Condition" (If
Available)
Alcohol Involvement Person/Vehicle/Crash
Time of Day Crash
Violation Codes Person/Vehicle
Driver Action Prior to Crash Person (or Vehicle)
Violation Indicated Person (or Vehicle)
Contributing Circumstances Person (or Vehicle)
Distracted Drivers Driver Distracted By Person
Driver Condition Person/Vehicle
Driver Action Prior to Crash Person (or Vehicle)
Contributing Circumstances Person (or Vehicle)
Unbelted Vehicle Occupants Seatbelt Usage Person
Injury Severity Person/Vehicle
Crash Type Crash
Exhibit VIII-1. Crash variables and subfile location by population
type.
OCR for page 71
71
circumstances surrounding the crash. It must be noted that or crashes involving unbelted vehicle occupants most likely
"distraction" variables are very likely less reliable than other to be applicable in a given jurisdiction) from among the set
police collected variables since they must be based either on of all treatments presented in the applicable NCHRP Report
information provided by the driver (which can be self-serving) 500 guides. Second, where appropriate, choose the subgroups
and/or on very difficult conclusions drawn by the investigat- of users (e.g., young drivers or older drivers), highway loca-
ing police officer who was not on the scene or in the vehicle tions, or times of day to which the selected treatments should
at the time of the crash. However, this may be the only data be applied.
available, unless the user can define distraction/inattention in As described earlier in more detail, the choice of the "best
some alternative manner. Seatbelt usage for vehicle occu- treatments" from the listing of many potential user-population
pants can be based on data from the officer's investigation of treatments can be based on the following factors:
a crash. However, like distraction/inattention data, such data
will not always be as accurate as we would hope because the a) The potential treatment judged to be the most effective,
officer has to base his/her judgments on after-crash observa- even given that effectiveness is unknown
tions and occupant/witness statements. Some occupant state- b) The relative magnitude of the crash types and severity levels
ments may be untrue, particularly in states with mandatory that the treatment will affect
belt usage laws. c) The cost of the potential treatments (either jurisdiction-
As noted above, some of the strategies described in these wide or per-mile or per-location)
two guides are directed to specific driver/occupant ages and d) Other technical or policy considerations
types of drivers. The roadway-related strategies noted for
fatigued and distracted drivers can be targeted to specific These factors must be combined in some fashion to deter-
roadway location if Procedures 1, 2A, and 2B are used, but are mine which treatment to choose. While there are multiple
difficult to target under this modified Procedure 3. For this ways of making this choice, the following represents one such
limited additional targeting, the names of crash variables and procedure.
the specific codes needed to conduct these targeting analyses
will vary from jurisdiction to jurisdiction. While not all rele- 1. Prioritize the specific user-population problem(s) to be
vant crash variables are presented here, Exhibit VIII-2 addressed.
provides some guidance concerning where example variables An initial issue may be whether to treat one, two or all
related to some treatment strategies might be found. three of the groups covered in these guides drowsy driv-
ers, distracted drivers, and unbelted vehicle occupants.
This decision can be based on the frequency and severity
Procedure
of the specific types of user-population crashes occurring
As described in Section III, Procedure 3 has two basic steps. in an analyst's jurisdiction. Crashes specific to a given
First, choose the "best treatments" for the user population of user-population were defined in the table above. For each
interest (e.g., the treatments related to fatigue-related crashes user population, the analyst could begin the process by
Crash Type/Issue Variable Crash Database Subfile
Driver Age Driver Age Person/Occupant or Vehicle
Driver Date of Birth Person/Occupant or Vehicle
Occupant Age (for Child Occupant Age Person/Occupant
Restraint Strategies)
Time of Crash Light Condition Crash
Hour of Day Crash
Vehicle Type (to Identify Vehicle Type Vehicle
Large Truck Drivers) Motor Vehicle Body Type Vehicle
Category
Commercial Motor Vehicle Vehicle
Configuration
Crash Location (for Targeting County Crash
Treatments) City Crash
Route Crash
Milepost Crash
Longitude/Latitude Crash
Block Address Crash
Speed Limit (for Use in Speed Limit Crash
Developing Cost per Crash)
Exhibit VIII-2. Crash variables and subfile location by crash
type/issue.
OCR for page 72
72
analyzing 3 to 5 years of crash data to determine the fre- volving only the applicable subpopulations must be iden-
quency of each crash population either total crashes or tified and analyzed. Here, just as in Step 1, the prioritiza-
some subset (e.g., fatal and serious-injury crashes). How- tion of subpopulations can be based on the frequency and
ever, since the severity distribution may differ between severity of the specific types of user-subpopulation crashes
some populations, and since restricting the analysis to occurring in an analyst's jurisdiction. Crashes specific to a
only fatal and serious-injury crashes will severely limit the given user-subpopulation can be defined using variables
crash sample and will omit a large component of the crash in the table above (e.g., occupant age for child restraint
problem non-serious injury and no-injury crashes a programs or vehicle type for heavy-truck driver pro-
better solution is to weight each crash for a given user pop- grams). For each user subpopulation, the analyst could
ulation by an economic cost based on its severity, and then analyze 3 to 5 years of crash data to determine the fre-
accumulate the total cost of crashes for each population. quency of each crash population. Again, either total
Information on economic cost per crash severity level can crashes or some subset (e.g., fatal and serious-injury
be found in Crash Cost Estimates by Maximum Police-Re- crashes) could be used, but the economic cost of crashes is
ported Injury Severity Within Selected Crash Geometries a better measure since crash severity may differ. The same
(22). Here, instead of using severity cost by crash type as is cost figures presented above could be used.
done in roadway-program analyses covered in earlier sec- Note that using these crash costs to develop the eco-
tions, the analyst will use the basic cost per crash catego- nomic harm of crashes involving unbelted children will
rized by police-reported severity level (i.e., K,A,B,C,O). likely result in conservative estimates of that economic
Exhibit VIII-3 below presents those costs per crash. Costs cost. These cost-per-crash estimates in Council, et al. (22)
for combinations of crash severity levels (e.g., K+A were based on standardized populations of vehicle occu-
crashes) are presented in that report (22). This analysis of pants by age and belt usage. Components of these costs re-
total crash cost will provide the analyst with overall infor- lated to lost wages and other factors would be greater for
mation on which of these three unsafe driver/occupant fatally injured children than for older populations. How-
populations is most important in his/her jurisdiction. If ever, it is felt that even though perhaps conservative, they
only one of the unsafe driver/occupant populations is are suitable for this use.
being examined, the analysis can provide useful data for 3. Identify possible treatments for use for each high-priority
public information programs concerning the economic unsafe driver group.
cost of such crashes. The analyst will then review the pertinent NCHRP Re-
2. Prioritize the specific subpopulations to be addressed. port 500 guides and list treatments that would be most ap-
Once one or more populations are identified, the sec- propriate for each of the high-priority unsafe driver
ond step involves the identification of subgroups in most groups identified in the above step. The choice should be
need of treatment. Some strategies in each of the two limited to those treatment strategies that are classified as
guides can be applied to all drivers or occupants, and thus proven or tried in the guides.
all crashes involving the population of illegal drivers are 4. Rate the possible treatments based on estimated
treatable. However, certain strategies in each of these two effectiveness.
guides are only applicable to certain user subgroups (e.g., Since this procedure deals with treatment strategies
child vs. adult restraint strategies or fatigue strategies for with unknown effectiveness, this appears to be impossible.
passenger car drivers vs. heavy truck drivers). In order to However, for a given set of possible treatments for a par-
analyze the possible benefit of these strategies, crashes in- ticular user group, it may be possible to make a judgment
concerning which treatment strategy would be expected to
Crash Severity Speed Limit Comprehensive
Category Cost/Crash*
be most effective. The judgment will be somewhat easier
Fatal (K)
< 45 mph $3,622,200 for the strategies in these two guides since there is some
> 50 mph $4,107,600 information available on estimated effectiveness for some
< 45 mph $195,700
Serious injury (A)
> 50 mph $222,300 of the strategies.
Moderate injury (B)
< 45 mph $62,200 5. Choose best treatment(s) by considering estimated
> 50 mph $91,600 effectiveness, cost and other technical and policy
< 45 mph $40,100
Minor injury (C) considerations.
> 50 mph $49,500
No injury (O)
< 45 mph $7,000 The analyst will then combine the output of the steps
> 50 mph $7,800
above with at least two other factors in making a final deci-
* Crash Cost in 2001 dollars
sion on which treatment(s) to implement the cost of the
Exhibit VIII-3. Crash cost by crash severity and treatment and other technical and policy considerations.
posted speed limit (22). Unfortunately, there are no good guidelines for how to
