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 19
19 respondents required a minimum number of years of com- tions and other unsafe driving acts are direct reflections of mercial driving experience and that managers rated this age driving behavior. hiring criterion as the 4th most effective safety management Past crash involvement was not included in the list of the technique of the 28 presented. 16 driver risk factors in the research project survey, but the A contrasting view is that experience leads to complacency management practices sections did ask safety managers if they and unsafe practices for some drivers. One respondent claimed checked the MVR of crashes and violations for driver appli- that "CMV drivers appear to become complacent after about cants. All 178 safety manager respondents (100%) indicated 4 or 5 years on the job (regardless of age) and are less safe." that they check the MVR as part of their hiring procedures, While this may be true for some drivers, no evidence was and both safety managers and other experts rated this driver found indicating that this is a general trend. hiring practice as No.1 in effectiveness among the eight listed. Similarly, 99% of the safety manager respondents reported that they continuously track their fleet drivers' involvement in 4.2.2 Longevity with Company crashes, violations, and incidents, and this performance eval- uation practice was rated by both respondent groups as the In the research project survey, "new to company" was not No.1 evaluation practice of the four presented. rated by either safety managers or other experts as a strong cor- relate of crash risk. Safety managers rated "new to company" as 13th of 16 factors, and other experts rated it 11th of 16. Stud- Example of Commercial Driver Risk Prediction ies have shown a relationship between employment longevity Using Violation and Incident Data with a company and crash involvement. One safety manager respondent commented that, "The problem drivers are usually A major insurance carrier for the CMV industry has devel- the bottom 1020% who are a constant turnover in the indus- oped a powerful methodology for relating past driver viola- try. Often other employers do not give all information [about tions and incidents to current crash risk. This insurer has them]. . . ." made significant efforts to assist their clients in identifying In a study of CMV driver retention and safety, Staplin et al. high-risk drivers. These efforts involved a three-part approach, (2002) found that drivers with frequent job changes (three or which built a crash predictive index, assembled a comprehen- more different carriers per year for 2 years or more) were sive safety history, and included a ranking system comparing more than twice as likely to be involved in a crash as the at- drivers' safety history with that of their peers. fault driver than drivers with less frequent job changes. One may ask whether job changes increase a driver's risk, or Predictive Index. First, the insurer teamed with several whether poor driving results in dismissal or other management insured clients to determine the ability of various safety actions resulting in job changes. infractions to predict crash and incident risk. For example, the insurer partnered with a major national motor carrier of 4.2.3 Crashes, Violations, and Incidents bulk liquids and chemicals to monitor the roadside inspec- tion results and crash/incident history of 1,682 drivers who A driver's history of involvement in crashes, violations, and were continuously employed over a 37-month period. For other incidents is known to be a strong predictor of future crash these 1,682 individuals, the insurer tracked the number and involvement and the driver's role (e.g., at-fault vs. not-at-fault). type of "risk-indicative" driver violations. Example viola- For example, using a sample of more than 200,000 (mostly tion types included the following: non-commercial) drivers in Kentucky, Chandraratna and Stamatiadis (2004) were able to use past data on driver crash · Traffic Enforcement: Speeding, following too close, and involvement and violations to predict future crash involve- improper lane change. ment as the at-fault driver with 88% accuracy using a logis- · Log Out-of-Service: Exceeding hours-of-service limits tic regression model. One input to the predictive model was and log falsification. the time gap between successive crashes. If a driver was · General Log: Log not current, failure to retain previous recently involved at-fault in a crash, he or she is more likely 7 days' logs. to be at-fault in a second crash event. · Illegal Use of Radar Detector. Miller and Schuster (1983) followed 2,283 drivers in Cali- · Driving While Disqualified: Cited for driving with sus- fornia and Iowa for 10 to 18 years. They found that traffic vio- pended or revoked license. lations were a better predictor (than were crashes) of future · Driver Violation Rate: Total number of risk-indicative crashes. Their results indicate that violation behavior is more violations per inspection. stable over time than crashes, thereby making them a better predictor of future crashes than crashes themselves. This find- The violation data were gleaned from FMCSA carrier ing may reflect that crashes are relatively rare and are the result profiles. Carrier profile reports, which can be obtained from of both at-risk driving behavior and chance. In contrast, viola- FMCSA (www.fmcsa.dot.gov), contain tabulations of all
OCR for page 20
20 state-reported crashes and roadside inspections. For each averaged 0.64 crashes/incidents and drivers with four violations roadside inspection, these reports provide identifying infor- averaged 0.93 crashes/incidents. mation (e.g., time, date, and driver name) as well as a tabula- The predictive ability of violation types was also calcu- tion of cited violations. lated. Results indicated that "Driving While Disqualified" Next, the insurer tracked these drivers' involvement in such had nearly twice the predictive ability of the other violations. events as crashes, spills, and mis-deliveries. These data were This was followed by "Traffic Enforcement" and "General extracted from both the carrier's internal accident/incident reg- Log" violations. ister and the insurer's claim management system. The com- A similar process was used across a range of other safety bined files yielded information regarding driver name, violations, such as motorist complaints (e.g., "1-800-Hows- crash/incident type (e.g., preventable rear-end collision, side- My-Driving") and MVR convictions (e.g., speeding, failure to swipe crashes, and chemical spills), and incident cost. yield right-of-way, and failure to obey traffic warning device). During the 37-month tracking period, the 1,682 drivers were involved in 4,662 roadside inspections, 1,102 risk- Comprehensive History. Next, the insurer assisted its indicative violations and 836 claim incidents. Drivers were clients in developing a comprehensive safety history. Typi- involved in an average of 0.50 crashes/incidents each. How- cally, this history comes from multiple files, similar to those ever, involvement rates ranged from 0 to 13 incidents per shown in Figure 7. driver. Violation involvement rates ranged from 0 to 7 per The files are often stored in vastly different data formats. driver. Interestingly, 20% of the drivers were involved in Driver rosters and accident registers, for example, are often 79% of all crashes/incidents and 76% of all violations. maintained in a spreadsheet (e.g., Microsoft® Excel) format. The insured carrier next examined the statistical relationship MVR reports and motorist complaints are frequently in "hard between violations and incident involvement. This deter- copy" format and are kept within a driver's qualification file. mines the influence of the various factors (e.g., number of traf- Other information, such as customer complaints or observed fic enforcement actions, log out-of-service, general log, radar incidents, is frequently maintained within operations or driver detector, and total driver violations) in the number of incidents. dispatching software programs. The findings indicated that the total number of driver violations was a very strong predictor of crash/incident involvement. For Peer Ranking System. With the comprehensive safety example, drivers with no violations averaged 0.41 crash/ history assembled, the insured assisted their clients in devel- incident events each. In contrast, drivers with two violations oping a total "Risk Index" score for each driver. This was DRIVER MVR REPORTS ROSTER Source: Internal company roster Data Source: State MVR Reports O THER COMPANY ACCIDENT DATA REGISTER Comprehensive Safety File Source: Various internal or external files Source: Internal company file CARRIER MOTORIST PROFILE COMPLAINTS Source: Federal Roadside Inspections Source: "How's My Driving" Reports Figure 7. Assembling comprehensive safety history.