Driver Selection Tests and Measurement (2012)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

7 CHAPTER TWO OVERVIEW OF DRIVER INDIVIDUAL DIFFERENCES 391 at 49 CFR 391. These regulations also establish mini- mum duties of motor carriers in verifying driver qualifi- cations. An owner-operator with DOT operating authority must meet both the driver and carrier requirements. An excellent summary of driver qualifications as well as other safety-related regulations and best practices is A Motor Car- rier’s Guide to Improving Highway Safety (FMCSA 2008), available from the agency. Although there are several farming-related and other exemptions, the following are key commercial driver requirements and responsibilities: • Be at least 21 years old • Speak and read English • Be able to drive the vehicle safely • Possess a medical certificate • Have a valid Commercial Drivers License (CDL) • Provide motor carrier employer with a list of all viola- tions during the past 12 months • Pass a driver’s road test or equivalent • Know how to safety load and secure cargo. Every motor carrier must have a qualification file for each regularly employed driver, including the following: • Driver’s application for employment • 3-year driving record from state agency • Driver’s road test certificate • Annual inquiries to state agencies for driving records • Annual carrier review of the above • Annual driver’s certification of violations during pre- vious 12 months • Medical examiner’s certificate • Special certificates as applicable (e.g., drivers with lost limbs, longer combination vehicle drivers) • Driver investigation history file – Inquiry to previous employer(s) over 3 years – Drug and alcohol testing release form – Notes of responses received from above. Driver Physical Qualifications Commercial driver physical qualification standards are found in 49 CFR 391.41. These regulations prevent persons with certain specified medical conditions from operating a In the book Using Psychology: Principles of Behavior and Your Life, Holland (1975) presents following two “metaprin- ciples” of human behavior: 1. Individual Differences: Each person is physically and psychologically unique. 2. Behavioral Consistency: Each person behaves relatively consistently over time and across different situations. People have significant individual differences, and these individual differences persist over long periods because each individual behaves consistently in many ways over time. These consistent, enduring human individual differences have significant influences on the probability of crash involvement (Lancaster and Ward 2002; Knipling et al. 2004). They are a potentially fair basis for commercial driver selection because they are likely to persist on the job and affect job performance. This chapter will first define the general characteristics and qualifications that all U.S. commercial drivers must have. It will then describe and define key driver characteristics and personal dimensions with known relationships to safety- related behavior and especially to driving safety. This will lay the groundwork for the following chapter, which will address procedures, tests, and measurements to assess safety-related driver individual differences in the hiring process. GENERAL DRIVER QUALIFICATIONS This report focuses on carrier assessment of driver applicants’ crash risks during the hiring process. It does not focus on fed- eral minimum qualifications for drivers, or required carrier tasks to verify that those qualifications are met. Rather, the emphasis is on driver risk assessments beyond verification that they meet minimum requirements. Nevertheless, it is worthwhile to briefly review those qualifications, as they are a baseline for any further consideration of driver applicants, and they also reflect risk factors officially considered safety- critical. This section briefly reviews those requirements. General Federal Requirements for Drivers Commercial driver general qualification standards are found in Federal Motor Carrier Safety Regulations (FMCSR) Part

8 CMV in interstate commerce. FMCSA provides guidance to medical examiners (and motor carrier companies) in an online handbook (FMCSA n.d.), and the training specifica- tions for medical examiners who conduct the examinations are available in the core curriculum. FMCSA (2008) cites the following, from 49 CFR 391.41, as examples of its physical requirements for drivers: • Has no loss of a foot, a leg, a hand, or an arm, or has been granted a skill performance evaluation certificate pursuant to 49 CFR 391.49. • Has no impairment of a hand or finger which interferes with prehension or power grasping or has been granted a skill performance evaluation certificate pursuant to 49 CFR 391.49. • Has no impairment of an arm, foot, or leg which inter- feres with the ability to perform normal tasks associ- ated with operating a CMV or has been granted a skill performance evaluation certificate pursuant to 49 CFR 391.49. • Has no established medical history or clinical diagnosis of diabetes mellitus currently requiring insulin for con- trol or has been issued a diabetic or vision exemption. • Has no current clinical diagnosis or any disqualifying heart disease. • Has no established medical history or clinical diagno- sis of a respiratory dysfunction. • Has no current clinical diagnosis of high blood pressure. • Has no established medical history or clinical diagno- sis of arthritis. • Has no clinical diagnosis or clinical history of epilepsy. • Has no mental, nervous, organic, or functional disease or psychiatric disorder. • Has 20/40 vision or better with or without corrective lenses. • Has distant binocular acuity of at least 20/40 in both eyes with or without corrective lenses. • Has the ability to recognize the colors (red, green, and amber) of traffic signals. • Has hearing to perceive a forced whisper voiced equal to or greater than 5 feet with or without hearing aid, or average hearing loss in the better ear equal to or less than 40 dB. • Has no history of drug use or any other substance iden- tified in Schedule 1. • Has no clinical diagnosis of alcoholism. Many detailed medical definitions and fine distinctions are applied in determining disqualifications. FMCSA’s website and the FMCSRs have specific interpretations of the qualifications and the latest changes. Changes and refinements to these requirements are continuously under consideration, and there are exemption programs for some conditions such as monocular vision. Driver physical qualifications focus primarily on major medical conditions such as cardiovascular conditions, or basic “static” psychomotor abilities such as visual acuity and color vision. Dynamic physical driving skills such as attentional focus and decision making are not assessed, nor are work-related functional requirements such as lifting and entering/exiting tractors and trailers. The nature of these skills and tests designed to assess them will be discussed later in this chapter and the next. SAFETY-RELEVANT DRIVER TRAITS AND OTHER CHARACTERISTICS A trait is a personal characteristic that differs among peo- ple and tends to be persistent over time. This report con- cerns personal traits and related characteristics that are (1) relevant to driving safety, and (2) potentially discernible through some kind of test, measurement, or other evalua- tion. Psychologists distinguish traits from states. Traits are enduring, often lifetime, characteristics, whereas states are temporary conditions (Pervin 2003). A consistent and per- sistent disposition toward anger, aggression, and/or hostility would be a trait. Temporary anger after an argument would be a state. People differ from each other in many fundamental ways. These differences may be related to heredity, developmental environments, chronic life conditions, or a combination of these. Evidence points to the following types of human traits and other characteristics as being most relevant to driv- ing safety (Lancaster and Ward 2002; Murray et al. 2003, Thiffault 2011), and thus of greatest potential interest for commercial driver assessments: • Personality • Attitudes • Psychomotor skills and cognitive functions • Medical status and conditions • Behavioral history (not a trait per se but a similar indicator) • Mental ability. These six categories are not entirely mutually exclusive. Most notably, personality is a source of attitudes, and then both are sources of behavioral history differences. Psycho- motor and cognitive skills are conceptually separable from medical conditions, but in practice the two may be conflated. In both research and practice, however, the six areas are gen- erally addressed separately. Each of these kinds of driver differences are defined and discussed here. Most can be further classified into more spe- cific categories, like different personality types and differ- ent medical conditions. Evidence for the safety-relevance of each is presented.

9 Personality This report employs a broad and simple definition of human personality: any enduring tendency or consistency in a per- son’s behavior or psychological makeup. Personality traits are consistent tendencies in emotional adjustment, interper- sonal relations, motivation, attitudes, and behavioral “style.” “Personality traits are deep individual characteristics, most often biologically rooted, that determine the broad emo- tional and behavioral orientations of the person” (Thiffault 2007). Psychological consistencies extend in two dimen- sions: consistency over time and consistency across diverse situations. Of course, such consistencies are not absolute. People change in both predictable and unpredictable ways through life, and sometimes people behave markedly dif- ferent in different situations. But there is enough individual behavioral consistency across time and across situations that it is considered a pervasive principle of psychology and a major determinate of behavior (Holland 1975; Pervin 2003). This includes commercial driver safety behavior. CTBSSP Synthesis 4 on individual differences (Knipling et al. 2004) found that both carrier safety managers and other experts considered personality dimensions like aggressive- ness and impulsivity/risk-taking to be among the top predic- tors of driver risk. Other research has also shown that these traits are associated with driver crash risk or safety-related behaviors. Other safety-relevant personality traits include sensation-seeking, “Type A” personalities, lack of conscien- tiousness, and high stress level. Personality dimensions are not unitary physical dimen- sions like height or weight. Rather, they are constructs or explanatory labels for something that is not directly observ- able or cannot be captured by a single observation or mea- sure (DOL 2000; Pervin 2003). Personality constructs are theoretical concepts that attempt to capture a cluster of closely related personal behaviors, attitudes, or emotions. “Conscientiousness,” for example, cannot be directly seen or measured, but it exists as a human trait because its mul- tiple manifestations are visible in behavior that is considered to be morally correct. This report will also discuss construct validity, the degree to which research confirms predictions based on the construct. In other words, construct validity is the degree to which a personality label is robust and useful as an explanation of behavior. The research literature on personality includes numer- ous constructs representing personal traits and dimensions. For example, one major personality questionnaire uses mul- tiple-choice answers to classify people on five dimensions or scales: neuroticism (anxiety), extraversion, openness to experience, agreeableness, and conscientiousness (Larson and Buss 2005). These have been called the “Big Five” per- sonality dimensions, although not all are strongly related to driving safety. The major personality dimensions relevant to safety include impulsivity/risk-taking, sensation-seeking, aggres- siveness/anger/hostility, “Type A” personality, consci- entiousness, and stress level. In some cases, similar or overlapping personality dimensions are also discussed. The focus here is on the personality dimensions themselves rather than on occupational tests of them. Chapter three presents specific tests known to predict driving safety. Impulsivity/Risk-Taking An impulsive person is one who makes hasty actions and therefore is prone to error. Often, the hard-to-control impulses are related to risky behaviors or even violence. Whenever a person reacts quickly and without forethought, he or she will be at higher risk for errors. Impulsivity and risk- taking are largely inseparable as personality traits because the perception of risk is what constrains most people from hasty actions (Shinar 2007). Many motor vehicle crashes are the result of voluntary at-risk behaviors, such as excessive speed, improper following distance, and illegal maneuvers. Drivers’ perception of the risk of their actions under- lies, to a great extent, the extent to which they engage in at-risk behaviors. Risk perception is a cognitive process underlying an individual’s perceived level of risk and that determines, or strongly influences, risk-taking behaviors (Thiffault 2007). Safety belt use provides a good example. Eby (2010) reviewed the reasons why some drivers do not wear safety belts. Reasons include forgetting, discomfort, inconvenience, social motivations, and complaints about how belts are installed in some vehicles. The biggest reason, however, was the perception that there was little risk in not wearing the belt. Here, “risk” included both injury risk and traffic violation risk. Laws and company policies mandating safety belt use are effective because they “up the ante” in regard to violation risk, even though, objectively, the legal consequences of not wearing a belt are small compared with the potential injury consequences. Young male drivers have the highest rates of not wearing a belt, consistent with their high rates of risky driving behaviors. Beck et al. (2006) queried 2,030 U.S. drivers (mainly noncommercial) about their driving beliefs, attitudes, and behaviors. Of the 2,030, 305 were designated “aggressive” based on a self-report that they had driven aggressively, trav- eled 20 mph or more above the speed limit, violated a traffic sign or signal, or driven while under the influence in the past month. About 12% of the aggressive drivers admitted that they did not “always/nearly always” wear their belts, com- pared with just 2% of the remaining drivers. Not wearing a safety belt appears to be an indicator of com- mercial driver risk as well. In the LTCCS, truck crash involve- ments could be separated into three categories: single-vehicle, multivehicle where the truck/truck driver is assigned the Criti-

10 cal Reason (CR) (i.e., is “at-fault”), and multivehicle where the other vehicle/driver is assigned the CR (Knipling 2009b). Sin- gle-vehicle crashes suggest the greatest driver failure, as they generally occur as a result of a catastrophic loss of vehicle con- trol. In truck-CR multivehicle involvements, the truck driver is at fault, but the error is usually a traffic interaction mistake such as “looked but did not see” or false assumption. Truck driver error is minimal or nonexistent in multivehicle crashes where the other driver is assigned the CR. Not wearing a safety belt is strongly associated with these three crash categories and levels of driver culpability, as shown in Figure 1. Thus, not wearing a safety belt is associated not only with the risk of injuries in crashes, but also with the risk of causing crashes. FIGURE 1 Association of truck driver safety belt nonuse and crash involvement category in the LTCCS. (Source: Knipling 2009b.) Impulsivity overlaps strongly with sensation-seeking, aggressiveness/anger, “Type A” personalities, and (lack of) conscientiousness. Thus, the research findings relating to these traits generally apply qualitatively to impulsivity as well. Sensation-Seeking Sensation-seeking is the desire for varied, novel, and arous- ing experiences. It has well-established links to unsafe driving behaviors, traffic violations, and crash involvement (Schwebel et al. 2006). Sensation-seeking overlaps with other personality traits like impulsivity and aggressiveness. Sensation-seeking people tend also to be extraverts. Studies reviewed by Dewar and Olson (2002) and Knipling (2009a) link sensation-seeking with unsafe driving behaviors, traffic violations, and crash involvement. A meta-analysis by Jonah (1997) documented correlations between sensation-seeking and risky driving behaviors such as speeding, frequent lane changes, alcohol use, and failure to wear safety belts. Iverson and Rundmo (2002) also found a significant association between sensation seeking and risky driving. Rimmo (2002) found that sensation-seeking is strongly associated with violations of rules (e.g., speed lim- its and other traffic restrictions) but only weakly associated with driving mistakes not associated with rule violations, such as “looked but did not see.” Dahlen and White (2006) found sensation-seeking to be related to unsafe driving behavior, although they noted that the exact path in which it affects driv- ing is unknown. They speculated that the link was related to aggressive driving, lack of rule following (e.g., speed limits), and driver loss of concentration at critical times. Drivers who seek sensation and/or experience negative emotions while driving are more likely to be in crashes and to commit violations. Matthews et al. (1996) developed a Driver Stress Inventory (DSI) to capture emotions during driving, including aggressive feelings toward other drivers, active dislike of driving, worry over hazards, thrill-seeking, and fatigue. Subjects taking the inventory answer 48 Likert scale items which together generate scores on these various aspects of stress. DSI scale scores for both U.S. and U.K. subject groups were compared with separate questionnaire responses relating to driving behaviors, crashes, and viola- tions. Crash-involved drivers scored higher than non-crash- involved ones on feelings of thrill-seeking and aggression/ hostility while driving. Thrill-seeking and aggressive emo- tions also correlated in the +0.4 to +0.6 range with traffic violations and with self-reported speeding. Sensation-seekers appear to be generally more suscep- tible than other drivers to fatigue and drowsiness. Because 16 Driver Characteristics and Risk: Safety Manager Ratings (Arranged Highest to Lowest) 1. Aggressive/angry 2. Impatient/impulsive 3. Inattentive 4. Inexperienced (new CMV driver) 5. Unhappy with job/company 6. Young driver (e.g., less than 25) 7. Sleep apnea/other sleep disorder 8. Unhappy marriage/family problems 9. Debt or other financial problems 10. Heart or other medical condition 11. Dishonest 12. Older driver (e.g., 60 or older) 13. New to company 14. Obese/overweight 15. Introverted/unsociable 16. Did not attend truck driving school Source: Knipling et al. (2003).

11 other high-achieving individuals, but has a negative con- notation if it implies chronic anger, dissatisfaction, impa- tience, overcompetitiveness, and hostility. The “Type A” personality encompasses these characteristics. In a review of several studies, Dewar and Olson (2002) note that the Type A personality is reflected in people’s choice of vehi- cles, driving style, violation rates, crash rates, and heart attack rates. They also report a surprising association with psychomotor skills: Relative to controls, Type A individu- als have slower reaction times and generally perform worse. Type A individuals often exhibit life stress both at home and at work, are quickly irritated by other drivers, tend to dehumanize other drivers, and express anger outwardly rather than inwardly. For them, the shell of a car or truck cab can be an insulated and “safe” environment from which to project anger and hostility. Nabi et al. (2005) compared questionnaire responses on the Bortner Rating Scale for Type A Behavior Patterns (TABP) to responses relating to risky driving behaviors and past crash involvements. The subject group for this com- parison was 11,965 French national utility (electricity and gas) company employees. The researchers found a signifi- cant association between Type A behaviors and both crash rates and serious crash rates. The study controlled for annual mileage, gender, and age. Figure 2 shows driving “hazard ratios” for low, medium, and high scorers on the TABP scale. Explanations for the association include that Type A drivers engage in more risky driving behaviors (e.g., talking on cel- lular phones, eating), are less patient, and are more prone to anger in frustrating or stressful driving situations. FIGURE 2 Driving “hazard ratios” for low, medium, and high Type A questionnaire scorers in French utility company study. (Source: Nabi et al. 2005.) Note: Normed relative to “Low” hazard (1.00). Conscientiousness Conscientious people have a strong sense of right and wrong and believe in an obligation to act accordingly. Thus, they tend to be careful, scrupulous, responsible, and reliable. Unconscientious people are at the opposite extreme. Level of conscientiousness in the population may follow a skewed distribution much like that of driver risk. That is, most people are in the “hump” at the good end of the spectrum, whereas a relatively small number are in the long “tail” at the bad end. they become bored with routine tasks more easily, sensation- seekers need and seek more stimulation to keep them awake; without it (as during long, boring drives), they become vul- nerable to drowsiness. In contrast, non-sensation-seekers generate their own internal stimulation to sustain alert- ness. This finding is based on a driving simulator study by Thiffault and Bergeron (2003), as well as other studies and theories of individual differences in brain function. Aggressiveness/Anger/Hostility In the CTBSSP Synthesis 1 Safety Manager survey (Knipling et al. 2003), respondents rated “aggressive/angry” as the driver characteristic most highly associated with driver risk. The textbox shows the rank-ordered list of characteristics presented in the questionnaire. Other research corroborates this strong association. Numerous studies show relation- ships between aggression/anger and crashes and violations (Knipling et al. 2004; Schwebel et al. 2006; Thiffault 2007). “Road rage” incidents are the most extreme and highly pub- licized manifestation of driver anger, but such incidents rep- resent only the most visible part of a larger problem. Dahlen and White (2006) compared the “Big Five” per- sonality factors, sensation-seeking, and driving anger to driv- ing behaviors and history. Subjects were 312 undergraduate students who drove more than 60 miles weekly. Driving anger was measured by a 14-item questionnaire, an abridged ver- sion of a 33-item Driving Anger Scale (DAS; Deffenbacher et al. 2001). Although other measured personality traits (includ- ing sensation-seeking) showed correlations with driving risk, the trait driving anger had the clearest associations. Scores on the 14-item DAS correlated positively with close calls (+0.18), risky driving (+0.31), and aggressive driving (+0.38). The authors considered individual differences in anger while driving to be important in assessing crash risk, and recom- mended driving anger as a principal factor to include in any personality inventory to screen for risky drivers. Schwebel et al. (2006) compared personality traits with both driving behavioral history and performance on a simu- lated virtual environment task designed to assess risk-taking during driving. Traits examined included anger/hostility, sensation-seeking, and conscientiousness, all of which previ- ous studies had linked to risky driving. Anger/hostility was measured using the DAS and behavioral history using a Driver Behavior Questionnaire (DBQ; Parker et al. 1995). Subjects high in anger/hostility took more chances in the simulated driving and also had stronger histories of speeding, violations, and crashes. Findings for sensation-seeking were similar, whereas those for conscientiousness were similar but inversed. “Type A” Personality “Hard charging” is a description that may have a positive connotation when applied to successful entrepreneurs or

12 FIGURE 3 Correlations among social deviance, thoroughness, speeding, and crashes. (Source: Redrawn from West et al. 1993.) Neuroticism Neuroticism is a personality trait characterized primarily by anxiety and stress. Other characteristics include irritability, discontent, self-consciousness, and moodiness. The opposite of neuroticism is termed emotional stability. Moen (2007) found that highly anxious people had lower self-assessments of their driver skill and higher stress levels during driving. This was reflected in higher crash rates. Driving stress is also related to anger while driving. As noted earlier, Dahlen and White (2006) found that DAS scores were predictive of unsafe driving behaviors and high crash histories. Personal stress and unhappiness can be caused entirely by one’s life situation rather than by internal, constitutional factors. There is, however, clear evidence that chronic stress level can also be a true personality trait like others described above. Moreover, many people’s adverse life situations, such as family and financial problems, are long term. Thus, from the perspective of motor carriers seeking to hire low-risk drivers, applicant stress level might be seen as an enduring personal characteristic. The positive opposite is unstressed emotional stability. Individual stress level may be related to locus of control. A person with internal locus of control believes that he or she has mastery, or at least strong influence, over life events and outcomes. One with external locus of control believes that personal efforts to control events are futile. External locus of control is associated with greater stress and anxi- ety. Knipling et al. (2004) reviewed several studies indicat- ing that external locus of control is associated with higher crash risk. For example, Jones and Foreman (1984) classified bus driver applicants with two or more moving violations as high-risk and those with no moving violations as low-risk. On a personality profile, 79% of the high-risk group scored high on external locus of control, versus only 31% of the low-risk group. A nondriving study of 283 hospital workers compared individual “safety locus of control” to their on-job accidents (Jones and Wuebker 1993). Thirty-eight percent of the low safety consciousness group was involved in one or more Arthur and Graziano (1996) administered question- naires on conscientiousness and five other personality traits to nearly 500 subjects, including both college students and workers. Conscientious individuals were those who charac- terized themselves as self-disciplined, responsible, reliable, and dependable. Of the six traits measured, conscientious- ness was found to have the strongest relation (in this case, an inverse relation) to crash involvement for both students and workers. The authors noted that “conscientious individuals may be especially sensitive to social responsibility norms,” making them less likely to engage in dangerous activities. Controlling for other factors, the correlation between consci- entiousness and number of at-fault crashes was −0.22, which may be considered a moderate correlation given the various measurement difficulties and confounding factors affecting such a study. Extreme lack of conscience is seen in antisocial per- sonalities. Individuals with this personality disorder have been called “sociopaths” or “psychopaths.” “Antisocial” in this context does not mean introverted, but rather that the individual has little social regard for others. These individu- als tend to be sensation-seekers who do not appreciate the potential consequences of their actions for themselves or others. The antisocial personality type is often seen among criminals and among individuals with a history of traffic vio- lations and crashes. Thus, hiring individuals with criminal backgrounds poses safety and security concerns (Knipling 2009a). In a large poll of more than 700 general population driv- ers, West et al. (1993) related “social deviance” to self- reports of speeding while driving and crashes. Socially deviant individuals were characterized as being selfish, focused on immediate gratification, and having a disregard for the law and for other people. Questionnaires were used to assess socially deviant attitudes and behaviors. Ques- tionnaires were also used to assess subjects’ “thorough- ness” and their driving histories and behaviors. Figure 3 summarizes correlations seen among these personal char- acteristics and histories. Negative correlations indicate an inverse relationship. Although none of the correlations was particularly high (probably reflecting the difficulty of precisely measuring these traits), the highest seen were between social deviance and speeding, and between social deviance and crashes. Several project interviews mentioned an “attitude of compliance” as an important safety-related characteristic of good commercial drivers. One bus company safety director believed that drivers who were “passive” and nonassertive in traffic were the safest drivers because they avoided con- flicts with other vehicles. A truck company safety director regarded ex-service members as a good bet for success as commercial drivers because they were used to complying with rules and orders.

13 which in turn become behaviors. Perceived behavioral con- trol is related to a person’s expectations of rewards or pun- ishments associated with the behavior, and the degree to which they control those consequences. Figure 4 shows this schematically. FIGURE 4 Simple schematic of the Theory of Planned Behavior. (Source: Ajzen 1991.) The TPB is a theoretical framework for studies of individ- ual factors in driving safety. Numerous studies have related attitudes, subjective norms, or perceived behavioral control to intentions and to behavior. All three have been shown to be predictors of dangerous driving behaviors (Parker et al. 1998; Thiffault 2007, 2011; Poulter et al. 2008). Chapter three will discuss how “slack” driver attitudes toward rule violations (e.g., speeding) are related to both a relative lack of concern about crashing and to violation frequency (Ma et al. 2010). Several of the personality traits discussed in the previous section are intertwined with safety attitudes. Conscientious individuals, for example, value morality and safety highly, are strongly influenced by safety-related social norms, or perceive controls on their behavior to be strong. An “attitude of compliance” appears to characterize many of the most conscientious and reliable commercial drivers. In Britain, Poulter et al. (2008) tested the application of the TPB to truck driver safety. Based on past studies, they identified two principal driver factors associated with crash involvement: (1) driving behavior and (2) driver compli- ance with driver- and vehicle-related regulations. Driver behavior is reflected by moving traffic violations, whereas driver compliance is reflected by driver-related (e.g., hours of driving) and vehicle-related (e.g., overloading, mechani- cal problems) roadside violations. The researchers recruited 232 truck drivers from several companies and other sources to complete a questionnaire assessing attitudes toward both specific driving behaviors and specific regulatory violations. In subject comparisons, they found positive interrelation- major accidents during the study period, compared with 28% of the medium safety and 21% of the high safety con- sciousness groups. Attitudes Interwoven with the concept of human personality is the con- cept of attitude. An attitude is an individual’s positive or nega- tive evaluation of a particular thing, where “thing” can be any object of thought. Attitudes toward particular driving behav- iors, including both positive behaviors (e.g., safety belt use) and negative behaviors (e.g., speeding), are of greatest interest. Attitudes have two internal components: cognitive (knowledge and beliefs) and emotional (Dewar and Olson, 2002). Attitudes are revealed in individual statements and, most important, in behavior. Extreme behaviors like aggres- sive driving strongly reflect negative attitudes, such as a general hostility toward society and rules. Less extreme behaviors also reflect attitudes, although situational factors also affect such behaviors. For example, a driver with a nega- tive attitude toward safety belt use may still wear one if there is a strong company belt-use policy and clear negative con- sequences for non-use. People tend to attribute their own behavior to external circumstances (e.g., “I didn’t have time to react to the sig- nal change.”) while attributing the behavior of others more to their character or personality (“red-light runner”). This difference in how individuals view their own behavior ver- sus that of others is called the attribution bias (Dewar and Olson 2002). The truth lies somewhere in between. People do have persistently different personalities and attitudes, and thus these are “fair ground” in driver selection. However, the environment can change specific behaviors and even specific attitudes. For example, individuals forced to comply with a rule (e.g., safety belt policy) will often develop more positive attitudes toward the rule and required behavior over time. Social norms are an important part of the human environ- ment; a company driver will be more likely to buckle up if he or she believes that all the other drivers are doing so. One can conceptualize a loose causal relationship con- necting individual personality, attitudes, intentions, and behaviors. This is illustrated as: The Theory of Planned Behavior (TPB; Ajzen 1991) has been formulated to explain how attitudes and other factors combine to become behavioral intentions, and then behav- ior. At any given time, an individual’s attitudes (the positive or negative value of a behavior) combine and interact with subjective norms (social norms as perceived by the person) and perceived behavioral control to determine intentions,

14 ships among all of the TPB elements shown in Figure 4. The intention to observe traffic laws had the greatest association with driving behaviors, whereas behavioral control had the greatest association with regulatory compliance. A key point for driver selection is that well-constructed questionnaires can assess persistent individual differences in safety attitudes and that such attitudes can be predic- tive of driving behaviors. However, one should not view all safety-related attitudes as fixed. They may change based on new knowledge, experience, and maturation. Although they are outside of the realm of driver selection, Behavior-Based Safety programs (e.g., Hickman et al. 2007) often result in positive changes in driver safety attitudes even though their focus is on specific behaviors. Psychomotor Skills and Cognitive Functions Dynamic Skills in Driving Perceptual: Static Visual Acuity Dynamic Visual Acuity Visual Contrast Sensitivity Peripheral Vision/Field-of-View Detection of Objects in a Visual Field Depth Perception Cognitive (Mental): Information Processing/Thinking Decision Making Selective Attention Attention Sharing (multitasking) Psychomotor Coordination: Reaction Time Multilimb Coordination Precision Control Tracking (following a target or path) Range-of-Motion Adapted from Llaneras et al. 1995. Driving is a demanding sensory-motor task that requires keen perception, quick thinking and decisions, and precise execution of responses. In some respects, it is like a com- puter or video game, and indeed many such games involve driving or similar maneuvering. Sensorimotor and cognitive (mental) skills are of paramount importance for high per- formance in video games but not generally for safe driving. If they were, then teenagers and young adults would be the best drivers, and safe performance would decline in later adulthood along with sensorimotor and quick reaction skills. Instead, middle-age and “young old” drivers up to their late 60s or even older are generally the safest drivers (Knipling 2009a). Driving involves many dynamic skills. The Trucking Research Institute (Llaneras et al. 1995) analyzed the dynamic perceptual, cognitive, and psychomotor (sensorimotor) skills involved in driving. The text box lists these skills. Tests on a group of commercial drivers compared these dynamic skills to performance on an interactive truck driving simula- tor. Dynamic or “neurocognitive” skills most predictive of simulator performance included depth perception, peripheral vision/field-of-view, field independence/dependence, atten- tion sharing, and range of motion. The tests showed that many dynamic skills generally declined with age, but that age alone was not a good predictor of performance. Even if age did reli- ably predict dynamic skills, it appears “that behavior usually trumps performance in driving safety” (Knipling 2009a). Although dynamic performance generally declines for older commercial drivers in their 50s and 60s, these drivers are among the best when it comes to crash rates and likelihood of being at fault in crashes. Dynamic skill tests are not likely to be highly predic- tive of crash rates across the wide range of drivers, but they may be useful to identify those with significant deficits. This might include the assessment of some serious medical con- ditions or impairments from drug or alcohol use (Llaneras et al. 1995). They might also be useful to provide baseline performance measures for later comparisons should drivers undergo significant health changes or show other signs of possible increased risk. Most perceptual information in driving is visual. A com- mon estimate is that 90% or more of the information a driver receives is visual, though this estimate is not based on rigor- ous studies (Dewar and Olson 2002). Driver licensing tests to measure visual acuity screen out most of those with bad vision, but otherwise they are not known to be predictive of driving safety. The visual skill apparently most related to safe driv- ing is not a static skill but rather a dynamic one related to peripheral vision. It is called Useful Field-of-View (UFOV) and has been studied mostly in older drivers. UFOV can be described as an “occupational visual field” test, in contrast to a clinical visual field test using flashing peripheral lights in an ophthalmologic setting. Young adult fixed-head field- of-view is about 180°, but this generally declines by age 70 to about 140° (Dewar and Olson 2002). Head and eye move-

15 ments allow a wider field, of course. The UFOV test flashes peripheral lights while a subject focuses on a center target. Subjects’ ability to see and react to the peripheral lights determines their UFOV. The UFOV test is different from the standard ophthalmological vision tests because it measures the central processing speed at which visual information is analyzed. It includes subtests that evaluate speed of informa- tion processing, ability to divide attention, and susceptibility to distraction. The test expresses the patient’s UFOV as a percentage reduction from the ideal (Crabb et al. 2004). Studies among older drivers and those with known atten- tion or mental impairments find that UFOV is predictive of crash rates, especially for intersection crashes. The UFOV is much less predictive for younger and unimpaired drivers (Dewar and Olson 2002). In driving, UFOV varies inversely with speed; that is, the higher the speed, the less the angle of the useful visual field. This may be one reason why many drivers drive slower as they age. Clay et al. (2005) completed a cumulative meta-analy- sis on the relationship of UFOV and driving performance in older adults. A meta-analysis combines previous stud- ies’ results as data to analyze the same research questions. Among older drivers, the UFOV correlation with safety is robust across multiple indices of driving performance and several research laboratories. This convergence of evidence from numerous studies using different methodologies con- firms the importance of the UFOV assessment as a valid index of driving competence and safety (Clay et al. 2005). Sumer et al. (2005) administered computer-based cogni- tive and psychomotor tests to 716 professional and nonprofes- sional drivers. Tests included traffic monotonous attention, selective attention, visual pursuit/tracking, eye-hand coordi- nation, reaction time, and peripheral perception. Scores on these tests were compared with self-reported driving behav- iors, skills, violations, and inattention errors. The peripheral perception test, similar to the UFOV test, was found to have the strongest positive correlations with driving and safety skills, as well as the strongest negative correlations with driving violations and inattention errors. The Trail-making test and WayPoint are two similar psy- chomotor tests of visual attention and task switching. The Trail-making test task requires a subject to “connect the dots” of 25 consecutive targets using paper and pencil or a computer screen. Scoring is based primarily on speed but also on errors. Two versions are available. Version A is sim- pler: The targets are simply numbered (1, 2, 3, etc). Version B requires the subject to switch from numbers to letters (1, A, 2, B, 3, C, etc.). WayPoint is similar to Trail-making Ver- sion B, but adds distractors in some parts to make the task more difficult. Trail-making is used primarily to diagnose brain damage (Corrigan and Hinkeldey 1987), but scores on these tests may be related to driving behavior across a larger proportion of the population. Chapter three discusses Way- Point’s use in selecting safe fleet drivers. Medical Status and Conditions The past decade has seen increasing interest, research, and regulatory activity relating to the issue of commercial driver health and medical conditions. This activity has reflected concerns about driver wellness and longevity, and also con- cerns about driving safety. Two previous synthesis reports (Orris et al. 2005; Krueger et al. 2007) have addressed commercial driver health issues. Commercial drivers as a group compare unfavorably to other Americans in measures of personal health (Roberts and York 2000; Krueger et al. 2007). FMCSA’s medical program acknowledges these con- cerns. The following is a characterization of U.S. commer- cial drivers, excerpted from the FMCSA Medical Examiner Handbook: The Average Driver. The [commercial] driver population exhibits characteristics similar to the general population, including an aging work force. Aging means a higher risk exists for chronic diseases, fixed deficits, gradual or sudden incapacitation, and the likelihood of comorbidity. All of these can interfere with the ability to drive safely, thus endangering the safety and health of the driver and the public (FMCSA 2010). The following is the profile of the average truck or bus driver: • Male • More than 40 years of age • Sedentary • Overweight • Smoker • Poor eating habits. The following is the medical profile: • Less healthy than the average person • More than two medical conditions • Cardiovascular disease prevalent. Although a detailed review of the safety relevance of spe- cific medical conditions is beyond the scope of this report, medical conditions can reduce driver and fleet safety in three primary ways (Knipling 2009a). The first two relate to driv- ing performance and crash risk while driving, whereas the third relates to more to the long-term stability of a carrier’s driving workforce: • Chronic performance decrements. Medical conditions could affect driver safety by causing general decreases in psychomotor skill and cognitive functions. Such chronic performance decrements might include decreases in flexibility, decreases in alertness, or

16 increases in reaction time. Psychomotor/cognitive per- formance has a weak relationship to crash risk unless a driver has significant deficits. Therefore, predicting crash risk based on medical conditions causing such deficits has been difficult. Moreover, most physicians do not have the time or the tools to assess functional impairments associated with illness. • Catastrophic performance failures. Medical condi- tions can cause episodic losses of the ability to control a vehicle, usually by loss of consciousness. Medical crises such as heart attacks, seizures, or diabetic insu- lin shock are significant proximal causes of serious large truck crashes. Sleep disorders such as obstruc- tive sleep apnea are often a root cause of asleep-at-the- wheel crashes. In the LTCCS, truck driver physical failures, primarily falling asleep and heart attacks, were the CR of 12% of truck at-fault crashes and 6% of all truck crashes (Starnes 2006). The major purpose of commercial driver medical qualifications is to prevent such crises. Medical screenings beyond the minimum qualifications can help carriers to reduce the risk of such crashes and associated losses. • Absenteeism and reduced employment longevity. This effect on fleet safety is less obvious and dramatic, but may be comparable in its long-term effects on carrier and industry safety. Chronic medical conditions are the most obvious signs of the poor health of many commer- cial drivers. Many of these individuals would be high- performing and reliable long-term employees were it not for their health problems. In the LTCCS, commer- cial drivers aged 51+ were 17% less likely than younger drivers to be at fault in multivehicle crashes, and yet these drivers are those most like to have reduced ser- vice owing to chronic medical conditions. Cardiovascular Illness Cardiovascular illness, the number-one cause of death in the United States, includes hypertension (high blood pressure), arteriosclerosis, coronary artery disease, angina (heart pain), heart attacks, and congestive heart failure. Cardiovascular illness is associated with both catastrophic performance failures while driving (principally heart attacks) and with shortened careers among middle-aged commercial drivers. In the 1990 NTSB study of 182 fatal-to-the-driver truck crashes, 17 (9%) were found to involve a heart attack or other cardiac incident as the primary cause. In the LTCCS, about 6% of large truck single-vehicle crash involvements and 3% of all involvements had a CR of heart attack or other physi- cal impairment (not including asleep-at-the-wheel). A 2007 report by the FMCSA Medical Review Board reviewed eight prior studies and estimated the relative crash risk of drivers with cardiovascular disease (all types combined) to be 1.43, or a 43% increase over other drivers. Obstructive Sleep Apnea Obstructive sleep apnea (OSA) is a medical condition of great concern to motor carriers and many others involved in truck and bus safety. It is a common illness among middle- aged males, the principal commercial driver demographic. OSA is a breathing disorder that disrupts sleep and causes often-severe daytime drowsiness. OSA is associated with obesity, which is prevalent, and perhaps the norm, in the U.S. commercial driver population (FMCSA 2010). The increase in crash risk associated with OSA is probably substantial. Various studies of noncommercial drivers with OSA put the increase in crash likelihood from two- to sixfold. A case control study by Young et al. (1997) placed the increase at fourfold. However, a case control study of commercial drivers found no increased crash risk among truck drivers with OSA (FMCSA 2004). This unexpected finding is questionable because the study involved mostly short-haul drivers and had unverified mileage exposure data. In the 10 safety manager interviews conducted for the report case studies, OSA was the most frequently cited driver medical concern. Carriers know that OSA is not always detected through the medical qualifications process and that it can be a cause of major crashes with high human and financial consequences. Chapter five describes several carrier medical programs addressing OSA and other driver health problems. Individual Differences in Fatigue Susceptibility OSA and other sleep disorders are major causes of individ- ual differences in susceptibility to drowsiness while driv- ing. However, these differences are also seen among drivers without known sleep disorders (Knipling 2005). For exam- ple, sleep-deprived healthy adults show wide variations in their progressions of performance deterioration and in over- all degree of performance impairment (Van Dongen et al. 2004). Moreover, these differences are consistent over time and, based on twin and family studies, have a partial genetic basis (Van Dongen et al. 2005). Many different patterns and features of human sleep, wakefulness, and sleepiness seem to vary widely among individuals. Moore-Ede (2007) has introduced the term chronotype to refer to an individual’s vulnerability to drowsiness and other sleep- and alertness- related characteristics. Although there appears to be a partial genetic basis, envi- ronmental and lifestyle differences also play a role. They include differences in the sleep setting (e.g., room quiet- ness and darkness, bed comfort) and in sleep hygiene habits. Although such differences may persist over time, they are potentially changeable.

17 The Driver Fatigue and Alertness Study (Wylie et al. 1996) is one of many to show wide variations in driver fatigue susceptibility. Eighty long-haul commercial driv- ers in the United States and Canada were monitored over a week of driving. Video segments were scored for drowsi- ness based on drivers’ eyelid droops, facial expressions, and facial muscle tones. Eleven of the 80 drivers (14%) were responsible for 54% of all observed drowsiness episodes. At the other extreme, 29 of the drivers (36%) were never judged to be drowsy. Figure 5 shows the skewed frequency distribu- tion of drowsiness episodes among the 80 drivers, plotted with five frequency bins. Notice the classic skewed shape of the frequency distribution, characteristic of differential driver risk. The two drivers in the far right bin had 78 total drowsiness episodes, which was more than the total number of drowsiness episodes exhibited by the best 51 drivers in the study. Two driver subjects among the 80 were diagnosed with OSA, but they were not the two highest-risk drivers. Behavioral History Psychologists widely regard past behavior as the best single predictor of future behavior (Ajzen 1991; Parker et al. 2001). Behavioral history, sometimes called biodata (which might also include medical data), includes both driving events and nondriving events and indices relevant to safety behavior. This section reviews both areas. Driving Behavioral History There are at least two reasons to expect drivers’ past driving behaviors and events to be predictive. The first is the “metap- rinciple” of behavioral consistency over time. The second, of less interest here, is that driving environments and mileage exposure levels tend also to be consistent. A driver’s history of crashes, violations, and other incidents is a well-documented predictor of future crash involvements, and also whether the driver will be at fault in future crashes. Using a sample of more than 200,000 driv- ers (mostly noncommercial), Chandraratna and Stamatiadis (2004) were able to predict the at-fault driver in crashes with 88% accuracy based on past crash involvements and viola- tions. Having recent at-fault crash was one factor that made drivers more likely to be at fault in another crash. Miller and Schuster (1983) followed 2,283 drivers in Cali- fornia and Iowa for 10 years or more. They found that past traffic violations were a better predictor of future crashes than were past crashes. Past traffic violations seem to be a better predictor of future crashes (1) because they are more numerous and thus more statistically reliable than crashes, and (2) because violations clearly imply misbehavior and fault, whereas a driver may not been at fault in past crashes. Murray et al. (2005) analyzed the records of more than 500,000 U.S. commercial drivers to determine factors most predictive of future crash involvements. Principal data sources were the Motor Carrier Management Information System (MCMIS) and the Commercial Drivers License Information System (CDLIS). Three driver history risk indicators were roadside inspection violations, traffic viola- tion convictions, and crashes. Rates of involvement in these behaviors over a 3-year period were correlated with future crash involvement. Table 1 shows the percentage increase in driver crash likelihood associated with the top behavioral predictors. Note in the table that six different violation and conviction types were more predictive than were past crashes themselves. This finding likely reflects the two advantages noted earlier. It is not surprising that an egregious violation like reckless driving is predictive of future crashes, though the strength of the relationship may surprise some. TABLE 1 INCREASES IN CRASH LIKELIHOOD ASSOCIATED WITH PAST DRIVER BEHAVIORS Behavioral Predictor Increase in Crash Likelihood Reckless driving violation 325% Improper turn violation 105% Improper or erratic lane change conviction 100% Failure to yield right-of-way conviction 97% Improper turn conviction 94% Failure to maintain proper lane conviction 91% Past crash 87% Improper lane change violation 78% Failure to yield right-of-way violation 70% Driving too fast for conditions conviction 62% False or no log book violation 56% Any conviction 56% Speeding > 15 mph over speed limit 56% Source: Murray et al. (2005). Although violation history appears to be better than crash history as a predictor of future crashes, a history of one par- FIGURE 5 Frequency distribution of long-haul truck driver high-drowsiness episodes among 80 drivers. (Source: Wylie et al. 1996.)

18 ticular crash type might be considered a “red flag” for future crash risk. This crash type is single-vehicle crashes. Single- vehicle crashes generally occur as a result of a catastrophic loss of vehicle control, resulting in a road departure, rollover, or jackknife. In contrast, multivehicle crashes are usually trig- gered by a traffic interaction mistake such as “looked but did not see” or false assumption. Thus, single-vehicle crashes suggest a more profound failure of driving safety. In the LTCCS (Starnes, 2006; Knipling, 2009b), truck single-vehicle involvements were much more likely than at-fault multivehicle involvements to involve asleep-at-the-wheel, driver physical failure (e.g., a medical event), excessive speeds, aggressive driving (as associated factor), response execution errors, and vehicle maintenance failures (for which drivers are responsi- ble). Further, single-vehicle crash involvements seen in driver records almost always imply culpability, whereas multive- hicle crash involvements may not have involved any fault by the commercial driver (i.e., the other driver was at fault). The text box shows LTCCS truck/truck driver CR comparisons between single-vehicle crash involvements and multivehicle involvements. The latter includes both at-fault and non-at-fault involvements, consistent with driver records that show crash involvements but not necessarily principal fault or cause. Truck/Truck Driver CR Percentages for Single- vs. Multi-Vehicle (SV vs. MV) Involvements in LTCCS • Asleep-at-the-wheel (CR): SV: 12.8%; MV: 0.4% • Other physical failure (CR): SV: 7.5%, MV: 0.9% • Too fast for conditions or curve (CR): SV: 28.7%; MV: 5.4% • Aggressive driving: SV: 2.1%; MV: 0.2% • Response execution error (CR): SV: 8.2%; MV: 1.1% • Vehicle failure (CR): SV: 12.7%; MV: 2.9% Note: Includes all MV involvements. Nondriving Behavioral History This section focuses on nondriving biographical information that might predict driving safety, including criminal record, credit history, past bankruptcies, workers’ compensation claims, or other predictive behavioral indicators. Although these are not personal traits, they could be valid predictors of risk. The principle of behavioral consistency suggests that peo- ple will tend to behave similarly across different types of situa- tions. Comparing people’s lifestyles to their driving styles—or, more specifically, their personal problems and transgressions to their driving mishaps—may predict driving safety. Criminality and personality traits like aggressiveness and impulsivity are related to unsafe driving. In his book Traffic Safety, Leonard Evans (2004) reviews studies show- ing greatly elevated crash risks—twofold or more—associ- ated with nondriving criminality. In Australia, Brace et al. (2009) reviewed studies linking criminal history and road safety. The study looked at a variety of criminal behaviors (e.g., assault, theft, drug offenses, and fraud) and different driving safety outcomes. It explored psychological theories explaining this relationship, including Kohlberg’s Stages of Moral Development (Kohlberg 1969). In this theory, moral behavior is not just related to knowledge of laws and con- sequences of violating them, but also related to individu- als’ internalization of social responsibilities and universal moral principles. Among the many studies cited was one by Chenery et al. (1999) in Britain where the vehicle status and driver histories of vehicles parked illegally in handicapped spaces were compared with those of nearby legally parked vehicles. The study found that 20% of illegally parked vehi- cles “would warrant immediate police attention,” compared with just 2% of legally parked vehicles. The driver compari- sons were similar: 33% of the owners/drivers of illegally parked vehicles had criminal records, versus 2% of controls. A recent Society for Human Resources Management survey (cited in Perry 2010) found that 60% of employers check credit reports for at least some of their prospective employees, up from 42% in 2006. However, only 13% check all potential employees. Federal law requires that employers obtain written permission from applicants before running a credit check on them. Cited in the same article was a 2008 survey by the Association of Certified Fraud Examiners which found that employee workplace fraud was often asso- ciated with personal debt and credit difficulties. Chapter three shows that employers must be careful not to overreach in the use of assessments like credit checks in their hiring. Selection tests must be validated in relation to job performance criteria. Credit history may be more directly relevant to driving jobs involving financial responsibilities (e.g., owner-operators who must make payments on their vehicles) than to those without such responsibilities. Cognitive (Mental) Abilities Intelligence is the ability to engage in complex thought. General level of intelligence is a persistent characteristic of individuals that shows up in many different kinds of judg- ments, choices, and behaviors. Level of general intelligence (“IQ score”) is as effective as other major individual char-

19 acteristics, such as socioeconomic status and personality, in predicting major life outcomes, such as mortality and occu- pational attainment (Roberts et al. 2007). Driving is one of the areas in which intelligence can be associated with safe performance (Knipling 2009a). There are a number of reasons to believe that intelligence might affect driving outcomes. Intelligence may be associ- ated with greater patience, greater consistency in choice, and a more accurate assessment of risk, all of which may contrib- ute to safer driving (Burks et al. 2009). Higher intelligence may also be associated with quicker and more accurate eval- uations of hazardous situations, and quicker and more effec- tive responses to them. Burks et al. (2009) studied 1,065 truckload (TL) driver- trainees and found that their general cognitive ability level was correlated with their patience and the accuracy of their evaluation of risks in small-stakes monetary games used to assess risk-taking. This study also found correlations between general cognitive ability level and consistency in choice tasks, and with social awareness as measured by the willingness to help someone else at a monetary cost. In addition, basic cognitive ability was the strongest single pre- dictor of staying for a full year of service after training in a setting in which early exit carried a significant financial penalty. Because inexperienced drivers have higher accident risk (Staplin et al. 2002; Knipling 2009a), this is one way in which general cognitive ability indirectly affects safety. Nonetheless, there is little evidence of a direct link between general cognitive skills and safer driving. Kim and Bishu (2004) suggest that this is because the real relationships may involve specific cognitive abilities rather than broad traits such as IQ. The role of these narrower traits has not been studied carefully in hazardous settings, as opposed to nor- mal driving situations. In summary, as Knipling (2009a) points out, the clear- est relationship is that between very low general intelli- gence and higher accident risk. Criminality is also higher for individuals with very low IQ, which may explain much of the relationship (Evans 2004). Associations with safety may be weak for those above a minimum level of general intelligence, for whom persistent traits such as personality characteristics matter more. MAJOR RETENTION-RELATED PERSONAL TRAITS Retention has a clear relationship to safety performance. New-to-the-industry drivers are likely to have higher acci- dent rates until they acquire experience. Retained drivers are more knowledgeable on safety goals of the company, more stable in their career path, and more likely to follow com- pany safety rules. They have learned from the training they have received and can put that training in use over a longer period. They also tend to be older drivers, an added associa- tion with safety (Staplin et al. 2002; Knipling 2009a). Turnover is a management problem many firms face, but it is especially difficult in some segments of the trucking indus- try. TL carriers that provide medium- and long-haul service have a more significant turnover problem than do carriers in parcel or less-than-truckload (LTL) operations. Until the deep economic recession that began in 2008, the annualized turn- over rate at large TL firms (more than $30 million in revenue per year) had never dropped below 100% per year. Smaller TL firms did slightly better. The annualized turnover rate at large TL firms hit an all-time low of 39% in the first quarter of 2010, and began to rise again from that point (Watson 2010). TL drivers are paid by the mile, and the rate is mod- est because the highly competitive nature of the segment prevents raising prices in order to raise wages. TL drivers often have irregular work schedules, work long hours per week, and have uncertain and limited time at home. From the driver’s point of view, the number of miles a TL driver can complete depends on many factors besides the driver’s own effort. This can be frustrating for new drivers. An experienced driver can earn substantially more than the U.S. median household income (approximately $50,000 per year in recent years). New drivers normally make significantly less than this, and many do not stay long enough to become experienced. Jobs at parcel and LTL carriers tend to be bet- ter on all of these dimensions, because of the organization of the work around fixed company terminals at which freight is handled. Firms in these segments historically have lower turnover rates than do TL carriers (Burks et al. 2008). Management can control some aspects of the job in ways that lead to better retention. Some of these aspects include a clear career path, performance-based promotions, and perceived driver equality. These factors can foster attitudes toward management, dispatchers, and other companies that increase job attachment. These are all components of job sat- isfaction that have been shown to be effective predictors of turnover (Tett and Meyer 1993; Griffeth et al. 2000). How- ever, for any given level of pay and set of working conditions, some drivers are more likely to leave than others. The focus here is on the persistent personality characteristics of indi- viduals that affect their likelihood of quitting. The “Big Five” Personality Traits The “Big Five” personality traits are extraversion, open- ness to experience, conscientiousness, agreeableness, and emotional stability, and cognitive skills. Zimmerman (2008) recently conducted a major meta-analysis of the relationship of these personality factors to turnover, examining studies by 86 authors at a large number of different firms in differ- ent industries. Burks et al. (2009) studied the relationship between cognitive skills and retention among TL drivers.

20 Agreeableness had the strongest correlation with turnover: −0.25. Conscientiousness had a −0.20 correlation, and emo- tional stability a −0.18 correlation. Openness (the trait most directly connected to cognitive skills) was positively related, with a correlation of 0.10. However, to the extent that open- ness is associated with cognitive skill, this relationship is likely to be reversed for truck drivers. Extraversion showed a small but inconsistent relationship to turnover. Predicted self-reported intentions to quit versus predicted actual quits varied with the personality trait levels. Low emotional stability was most closely connected to employ- ees’ intentions to quit, whereas low conscientiousness and agreeableness best predicted actual turnover decisions. Zim- merman (2008) developed a path model that showed impor- tant direct effects from personality to intentions to quit and turnover behaviors that were not captured through job satis- faction or job performance. Employees with low emotional stability may intend to quit for reasons other than dissatisfac- tion with their jobs or poor job performance. Employees who are low on agreeableness or high on openness may engage in unplanned quitting. The data also showed that personal- ity traits had stronger relationships with outcomes than did other measures of job complexity and job characteristics. These five personality factors are not completely inde- pendent of each other, so it makes sense that they may have a systematic relationship. Several authors (Digman 1997; DeYoung 2006) have identified a higher order structure: Emotional stability, conscientiousness, and agreeableness have a common predictive power, and the “meta-factor” this identifies has been labeled stability. This is sensible when thinking about turnover, because Zimmerman (2008) found all three factors in this meta-factor to be moderately strongly negatively correlated with quitting. In addition, openness and extraversion share a common predictive power, and the meta-factor of these two factors has been labeled plasticity. This appears to be less important for the behavior of quitting, according to these results, and may be reversed in truckers to the extent that plasticity is associated with cognitive skill, according to the results of Burks et al. (2009). Cognitive Ability Burks et al. (2009) studied 1,065 new-to-the-industry TL driver trainees, measuring their personality traits, demo- graphic characteristics, past job experience and job attach- ment history, and risk aversion and time preferences. These drivers received training at no upfront cost, but they all signed a credit contract that made them liable for the commercial cost of the training if they did not stay for 1 year of service. A major finding was that out of all the characteristics measured at study intake, the level of basic cognitive (mental) skills was the strongest single predictor of staying on the job for 1 year. Those in the top quarter of cognitive skill were almost twice as likely to complete a year as those in the bottom quarter. Conscientiousness involves self-discipline, planning, and dutifulness. Zimmerman (2008) hypothesized that those high in conscientiousness are less likely to quit because they are more likely to perceive a contractual or moral duty to stay. Another connection is that those high in conscientiousness are more likely achieve success at the job and therefore have higher job satisfaction. A third linkage is to the traits of impulsivity and risk-taking owing to their higher accident risk. Impulsivity and risk-taking are likely to be low in those with high conscientious- ness, who are better able to control short-term impulses to leave. Extraversion is the trait of seeking social relationships. Extraverts may experience more positive emotions and perceive their surroundings positively. Zimmerman (2008) hypothesized that high extraversion would lead to lower voluntary turnover because it would be associated with a greater level of job satisfaction and more social ties within the firm. This is not as clearly relevant to drivers, however, as drivers often do most of their work alone. Many of their interactions are with customers who may change from day to day, not with a stable group of coworkers. Openness to experience is the trait of seeking variety, new experiences, and being curious and imaginative. Zim- merman (2008) hypothesized that those high in this trait are more likely to quit in order to try out new job opportunities. This is also the trait most closely connected with cognitive skill and intelligence, which can be thought of as the capac- ity to analyze and make use of new experiences. However, cognitive skill is strongly associated with job success and job attachment among truckers. Agreeableness is the trait of being compassionate and caring toward others, as well as optimistic about human nature. Zimmerman (2008) hypothesized that those high in agreeableness are less likely to leave because they will be more understanding of the negative aspects of a job, have more successful relationships with coworkers, and be more likely to see a contractual obligation to stay. In addition, people high in agreeableness are less likely to be impulsive and therefore less likely to quit on an impulse. Emotional stability (known in its negative form as neu- roticism) is the trait of having positive emotions and being calm. Those low in this trait frequently experience anger, anxiety, or depression. Zimmerman (2008) hypothesized that those low in emotional stability (that is, high in neu- roticism) are more likely to quit because they have more negative views of their job, and have higher doubts and more stress about being able to do it. They would be more likely to avoid stressful situations, including stressful jobs. The trait of aggressiveness/anger/hostility is also likely to be associ- ated with low emotional stability (i.e., high neuroticism). Zimmerman (2008) found the following relationships between the Big Five personality factors and voluntary exits.

21 The authors argued that the main reason for this find- ing was the need for on-the-job self-management by TL drivers. The runs that drivers are assigned typically vary over time, with many details of scheduling, routing, and deliveries. The ability to schedule oneself to meet the needs of shippers and consignees, while taking account of HOS rules and changing traffic and weather conditions, requires cognitive skill. From the driver’s point of view, the number of miles a TL driver can complete depends on many factors besides the driver’s own effort. This gen- erates stress and frustration, especially for new drivers. When drivers are paid by the mile but cannot make enough miles, they are likely to quit. Burks summed up this find- ing by saying that “doing well financially requires a driver who is not only willing to work hard but also is able to work ‘smart’ in a competitive environment” (2009). Thus, higher mental skills are associated with stable employ- ment among truckers.