An increased need for improved driver safety has existed since the first motor vehicle fatality was reported over 96 years ago. That was September 13, 1899, in New York, NY, when Henry H. Bliss was struck by an automobile (Gunby, 1992a). In 1994 the U. S. had over 175 million licensed drivers and, according to U.S. Department of Transportation (1995), there were over 3.2 million injuries and over 40,000 fatalities attributed to driving accidents. With this growing number of drivers and the on-going problem of traffic safety, various campaigns, programs, and driving laws have been created and implemented over the years aimed at making our traffic system safer.

Slogans such as "buckle up" and "friends don't let friends drive drunk" have been created in hopes of curtailing these staggering driving statistics. Further, new driving laws have been implemented such as the mandatory seat belt use law. Another example includes Maryland's "zero tolerance" law that makes it illegal for drivers under age 21 to operate a motor vehicle at a blood alcohol level (BAC) of 0.02% or higher (Gebers, Romanowicz, & McKenzie, 1993). Other promising steps include graduated licensing programs for novice drivers, which gradually and systematically lift licensing restrictions, and has shown significant reductions in accidents in New Zealand and Australia (Gebers et al., 1993). Other accident countermeasures include raising the licensing age, implementing driving curfews, increased driver training, graded licensing, and license extension by mail for clean-record drivers.

Fortunately, the overall effect of such programs seems to be positive. Significant reductions in accidents over the past decades have been reported by the National Highway Traffic Safety Administration (Gebers et al., 1993). For example, the percent of alcohol-related fatalities declined from 57 percent in 1982 to 41 percent in 1994. The motor vehicle fatality rate per 100 million vehicle miles traveled has steadily declined from 5.5 in 1966 to 1.7 in 1994 (U.S. Department of Transportation, 1995).

The success and growth of such programs and campaigns is an example of the concern our nation maintains for traffic safety. Alternatives for improving driving safety are still needed, as the population of the United States rapidly approaches the 300 million mark and our nation's people continue to rely on the automobile for transportation. Since that first fatality in 1899, the U.S. traffic death toll has reached nearly 2.8 million people (Gunby, 1992a). In 1990, almost 6.5 million motor vehicle crashes were reported­about one every 5 seconds (Marwick, 1992). More than just campaigns and frightening statistics, alternatives for driver education, training, evaluation, and licensing are continually needed.

Standard Licensing Process
The typical licensing process begins with drivers' education and training, followed by on-road instruction and examination. Driver's education consists of classroom instruction, where one learns the rules of the road and receives education about driving safely. Such education is taught in high school as required by law (Department of Motor Vehicles, 1995) or by independent driving instructors through a driving school certified to do so in that state. While various methods have been used to teach driving education (e.g., books, films, discussion groups, quizzes), eventually, on-road training and evaluation is administered in preparation for the driver's licensing examination, using an actual vehicle. On-road driving performance evaluation involves scoring or otherwise measuring performance of the driver to judge his or her ability to drive safely. Such evaluations are made daily by driver educators and parents of new drivers, by department of motor vehicles staff and evaluators, throughout the research community for a variety of applications, and as part of established driving evaluation programs.

Road Evaluation for Testing
Using an automobile on the road continues to be the "gold standard" of instruction and evaluation for driving. The driver licensing examination is usually performed at the state's Department of Motor Vehicles (DMV). The licensing process often includes a knowledge test, some sort of visual acuity or perception examination, in addition to the on-road evaluation. Road evaluations are administered by DMV driver's examiners and are considered important for measuring driving performance. Indeed, in 1990, the California DMV initiated a program to increase the level of driving competency of drivers (Hagge, 1994). "A key element of this program involves the development and implementation of an improved drive test," states the introduction of this report.

Road Evaluation for Research
The research community also uses the on-road evaluation to measure driver performance. In fact, such research has contributed to improved road safety, highway and vehicle design, and driver evaluation techniques (Waller, 1991), as suggested by the falling per-mile loss rates (U.S. Department of Transportation, 1995). Specifically, safety experts and educators, scientists, human factors engineers, and other researchers have conducted a wide variety of driving performance investigations on the road. Examples of the extensive research literature include mapping eye-movement patterns to the visual driving scene (Mourant & Rockwell, 1970), driver behavior patterns in traffic (Forbes, Schmidt, Nolan, & Vanosdall, 1974), elderly drivers' perceptions of ability as compared to visual skills and driving performance (Fox, 1989), car-following in traffic (Brookhuis, De Waard, & Mulder, 1994), and an examination of an auditory route-guidance system (Green & George, 1995).

Road Evaluation for Occupational Therapists
Occupational therapists (OTs), the focus of the present study, represent another group that uses on-road evaluation. Occupational therapy (OT) is based on helping challenged individuals to develop and function independently within their environment. The OT has traditionally been involved with the daily living task of mobility, including bed and wheelchair mobility, transfers, and functional ambulation (Cox, 1989), allowing an individual to live independently and move about from one location to another. OTs are also involved in assisting with automobile driving. The ability to drive allows a person to remain independent by providing the mobility and freedom for daily living (Cox 1989), such as going to work, play, or shopping. For people who may have specific conditions that inhibit them from driving safely, OTs often help make driving easier by providing skills and/or necessary equipment (Cox, 1989).

OTs are involved in driver training and evaluation of people with a variety of disabilities, including the head-injury population, individuals with diabetes, dementia, Parkinson's disease, Alzheimer's, and the elderly. These may be people who have moderate to severe physical or cognitive impairments, multiple medical conditions, neuromuscular or neurological impairments, or progressive diseases, and they often require a referral (usually from a medical doctor) to a formal OT driver-evaluation program. People who have mild impairments require referrals only if there is specific cause to question their driving safety (Lillie, 1993). From a medical point of view, Miller and Morley (1993) state that driver-training programs that utilize the services of OTs to assist physicians in the evaluation of drivers can be helpful in directing driving recommendations; the goal of such programs is to keep persons functioning in society as independently as possible, without sacrificing safety.

OTs evaluate abilities related to driving skill; they teach appropriate attitudes, road courtesy, and defensive driving, identify and teach safe compensatory techniques, and select and train in the proper use of adaptive equipment. This equipment may include special adaptive devices, including hand controls for braking and accelerating, modified steering systems, left-foot accelerators, and special mirrors (Cox, 1989). While cognitive and other driver performance measures are widely utilized in evaluating drivers, an on-road test is used in most OT evaluation programs. A typical OT road test consists of 30-90 minutes of driving, including vehicle control orientation, parking lot, residential, and business driving, a variety of maneuvers (e.g., lane changes, turns, stops), and sometimes freeway driving. Typical DMV road tests range from 10-60 minutes with an average examination lasting 20-30 minutes (Shumaker, 1994).

Off-road tests. OTs who become certified to teach driver training are involved in more than just on-road testing. Off-road tests, including assessment of visual perception and motor and cognitive skills, are also necessary (Irwin, 1989). For example, findings from a nationwide questionnaire sent to 200 OTs showed that they vary widely in their assessment of driving performance. While many rely on clinical judgment rather than standardized tests to determine driving ability, sixty-five percent of OTs reported using cognitive evaluations of their own design (Hunt, 1994). This included an assortment of simple tests to detect sensory and perceptual abilities related to driving (e.g., reaction time, cognition) such as clinical vision tests, (e.g., peripheral visual field, depth perception, color sensitivity, static and dynamic visual acuity, and figure-ground discrimination), a range of motion and strength tests, such as neck rotation and grip strength (Cox, 1989), reaction time tests, and cognitive and memory tests (e.g., trail making, sign recognition, mental status, selective attention, and judgment). Another simple test, commonly used for static visual acuity, utilizes an eye chart. While the use of static acuity as a prediction of clarity of vision for driving does not seem realistic, dynamic visual acuity, which seems better, is used by OTs (Fox, 1989). Cox (1989) adds that neck rotation, grip strength, and reaction time are used as part of an OT evaluation of drivers in examining motor skills and driving ability.

The question of which tests predict actual driving performance has been ongoing, but generally the tests are used to assist the OT in making a clinical judgment about the patient's ability to drive. If such tests indicate that driving ability is sufficient, then the evaluation and training process continues to the on-road test. While standardization of evaluation procedures is lacking, most OTs report that a road test is still necessary in order to determine driving fitness (Hunt, 1994).

The OT evaluation paradox. On-road driving evaluations are the main platform for OT assessment due to a variety of benefits. Benefits include actually seeing the driver in action (as opposed to paper and pencil, knowledge tests) where specific behaviors, habits, and skills can be observed. Further, when teaching or evaluating driving, immediate feedback can be given. Behavior reflecting an understanding and incorporation of such feedback can be monitored quickly, and driving skill can be assessed during the evaluation session.

However, there are some serious limitations to on-road driver evaluations including those related to weather and traffic restrictions, vehicle maintenance and expense, pollution, and perhaps most importantly, safety and liability. Not surprisingly, on-road driving evaluations can be dangerous: the combination of a potentially-impaired driver with an unfamiliar vehicle (the vehicle used is seldom owned by the person being evaluated) creates a potentially hazardous situation. Indeed, the opinion expressed by many OTs is that although the on-road evaluation seems to be the ultimate in evaluating driver performance, the risk associated with each ride is high. Consequently, most evaluations are limited to daytime driving, in relatively safe areas (e.g., residential neighborhoods, parking lots). Further, for reasons of cost, safety, and liability, few evaluations are administered in challenging, stressful, complex, or "unexpected" driving conditions (e.g., driving at night, in rain or fog, or in heavy traffic).

Thus, the paradox of on-road evaluation exists: The road evaluation is the most commonly used and presumably the most effective form of driver evaluation, but it is limited to driving during the daytime, in safe environments where driving is easy, unchallenging, stressless, and least representative of potential accident situations. Driving in such conditions maximizes the safety of the evaluation but greatly limits the value of conducting on-road evaluations. Stated another way, the potentially most informative method of driver evaluation is rendered uninformative by the fact that it does not represent the range of real driving conditions and cannot be carried out in a manner that challenges the relevant abilities of the driver being evaluated.

On-road safety. There have been few reported accidents during on-road evaluation, although little data exists on this topic (C. Beatty, personal communication, November 17, 1995). Accidents during evaluation are few because the amount of on-road evaluation time is relatively small and is spent mainly on basic maneuvers in safe, "normal" conditions (Andre, 1993). Also, there is a reduced chance of accidents due to the extensive training of the instructors (e.g., 300 hours) in the proper use of dual controls, such as instructor gas and brake pedals, and steering wheels that are added to many evaluation vehicles.

With little data available on the OT accident rates in driver evaluation sessions, the number of near misses reported necessitating evaluator intervention (e.g., controlling steering wheel, braking) and the level of stress experienced by driver assessors is high (S. Lillie, personal communication, November 8, 1995).

With the continuous growth in the number of drivers on the road and a rise in the number of people needing driver evaluations, an increasing need for evaluation is apparent. Further, with a small number of driver evaluators available and the expense involved, the need for a more efficient evaluation process is dictated. Budget cuts and the limited resources for training, combined with an increased need for qualified driving evaluators puts pressure on existing evaluation programs. For example, the 3 full-time OTs at the Santa Clara Valley Medical Center perform on-road evaluations daily, and have had a waiting list ranging from two weeks to twelve months for potential clients. They currently are experiencing hospital-wide budget cuts in the millions of dollars.

An Alternative Approach
With such pressure, evaluations can be inefficient because of the need for specialized evaluation vehicles (e.g., vans for wheel chair drivers, dual control automobiles), the extensive, time-consuming but necessary OT training, and driving limitations (e.g., daytime, weather). In summary, to increase the efficiency of such programs, new assessment tools are needed to complement or replace the on-road evaluation.

Driving Simulator. A promising alternative for conducting driver performance evaluation may be some form of a driving simulator. As an assessment tool, the modern, computer-controlled, interactive driving simulator has enormous potential benefit, especially when used to test drivers of unknown driving ability. With the increasing number of evaluations needed and the always present threat of mishap on the road, a driving simulator seems a reasonable choice for controlling costs and improving safety. Simulation technology is now being used or is being considered for driver evaluation, training, and licensing (e.g., see Transportation Research Board, 1992). Human factors researchers have used and are using simulation to measure driver performance objectively (e.g., steering and braking behavior, heart rate, blood pressure, eye movement behavior). In addition, observational methods, used by DMV evaluators and OTs for road evaluations, might be applied to evaluations done during an interactive driving simulation.

What is a simulator? The word "simulate" is defined as fabricate, feign, pretend, copy, mimic, or imitate (Transportation Research Board, 1995) or as referenced by Sanders (1991), simulation is defined as a "technique of substituting a synthetic environment for a real one, so that it is possible to work under laboratory conditions of control." Simulation was created in order to provide a safe, alternative representation believed to be a valid substitute for various aspects of a particular task (Allen et al., 1995; Andre, 1993; Bookout, 1993; Repa & Bertollini, 1990; Wachtel, 1995). In such an experimentally controlled environment, performance measures can be defined, collected, and repeated in a cost-effective manner.

Types of simulators. Simulators have been categorized as interactive and non-interactive (McKnight & Stewart, 1990). An advantage of an interactive driving simulator is that the simulator responds to the driver just as the driver responds to the simulator. This is also known as a closed-loop driving simulator (Cox, Gressard, Quillian, Westerman, & Gonder-Frederick, 1992). In a non-interactive system, the driver responds to the simulator, but the simulator does only what it was programmed to do, regardless of the driver's actions. Only limited evaluation information can be obtained in this system type, often referred to as an open-looped system (Cox et al., 1992). For this study, an interactive (closed-loop) simulator was used.

Interactive driving simulation represents actual roadways through computer generated imagery (CGI), auditory feedback, and realistic vehicle instruments and controls such as brake pedal, steering wheel, turn signal indicator, speedometer, and mirrors. The roots of driving simulation first appeared at least 85 years ago (Wachtel, 1995) with Munsterberg's response tests. Since then, simulation methods have continued to be used successfully. According to Marowitz (1991), during this same period Stern had participants press a key when certain letters appeared on a moving conveyor belt, and Sachs used patterns punched in the belt to represent pedestrians or other vehicles. Driving ability was measured in correct responses under varying conditions. Such conveyor belt simulation was modified and used through the 1970's to investigate major causes of actions leading to accidents. However, much of the driving simulation research was temporally halted, due to the need for researchers and designers to assist with the re-design of our nation's airplanes. This emphasis on aviation has led to extensive development and usage of flight simulation for training, certification, etc. (Bookout, 1993).

Aviation simulation: Lessons learned. While flight simulators are expensive and on the high end of the cost spectrum of simulators, a wide range of simulators exists. Caro (1988) gives a variety of examples, from the early Link trainer, which was a 1929 training device that simulated the instrument environment of an aircraft, to a multi-million dollar Phase II flight simulator as used by pilots of major airlines such as United Airlines. Imagine training, assessing, and certifying pilots on a regular basis in not-so-friendly skies in airliners alone! Pilots are now regularly trained by the utilization of simulation technology. Such technology, which has been advanced largely due to the funding by the military for advanced aircraft training systems (Bookout, 1993), has made flying safer, pilots and their crews more effective, and airplane (cockpit) designs that are significantly enhanced (e.g., see Caro, 1988). Further, training simulators have been used widely and accepted in the aviation community for over forty years because of their proven ability to enhance the crew's performance in a safe and cost-effective manner (Andre, 1993). For example, Caro (1988) notes that Federal Aviation Administration (FAA) rules now permit experienced pilots to complete all of their initial training and checking for particular aircraft in simulators. FAA inspectors use simulators for pilot performance evaluations for pilot certificates. This includes witnessing piloting skills that are either identical to the skills the applicant would evidence with the actual aircraft or approximations that would transfer without further practice or training (Caro, 1988).

Uses of Driving Simulators
In a similar fashion, simulation technology as used for driving evaluation, and eventually driving licensing, can include various realistic traffic situations such as one would encounter in every-day and, perhaps emergency driving situations. Although, ideally, one could perform driver evaluation in a realistic manner on-road, such assessment amid public traffic is dangerous, uncontrollable, and for ethical reasons should not be conducted unless risk of exposure to threats or dangers of the road are minimized (Alm & Nilsson, 1994). A driving simulator offers a safe, standardized, and controllable alternative for conducting such evaluations.

Various driving simulators exist (see Diewald, 1995; Federal Highway Administration, 1995; McKnight & Stewart, 1990) that range in their cost and fidelity. It seems likely that realistic, interactive driving simulators could be used to train drivers before and during actual driving takes place, and even for driver licensing. Already, a more realistic, flight-quality driving simulator is being developed. Called the National Advanced Driving Simulator (NADS), this $32 million simulator will be used to study how motorists cope with potentially dangerous traffic situations without endangering the test subjects and to examine intelligent vehicle highway systems (IVHS) [recently renamed intelligent transportation systems (ITS) (Douglas, 1995)], and other highway and automotive designs (Gunby, 1992b). Expected to be completed by 1998 (Diewald, 1995), NADS shows the degree to which the federal government is willing to invest in driving simulation. However, the validity of this and other driving simulators remains to be demonstrated.

Low-Cost Driving Simulators
While both advanced and low-cost simulation allow safe, reproducible maneuvers that can be easily controlled and replicated in an experimental setting (Wachtel, 1995), low-cost microcomputer and display system technology has recently been developing at a rapid pace. The costs of many of the key components have fallen drastically, while the capabilities have increased significantly to support a range of simulator applications (Allen et al., 1995; Andre, 1993; Bookout, 1993). Further, low-cost driving simulators have been found to be useful for research purposes. They are continually being refined, have become more accessible, and include features that make them easily programmable (Allen et al., 1995). These refinements allow an even wider variety of investigators and other users to assess driver performance more easily.

Uses of low-cost simulation for research. Low-cost simulators have been used successfully for a variety of research applications. Examples include studies of the effects of alcohol intoxication on driver performance and choice to drive (Cox et al., 1992), evaluation of driving performance in patients with juvenile macular dystrophies (Szlyk, Fishman, Severing, Alexander, & Viana, 1993), investigation of performance, velocity, cognition (memory), and gender (Sexton, 1994), relative effects of age and compromised vision on driving-related skills and on-road accidents (Szlyk, Seiple, & Viana, 1995), and screening truck drivers for the effects of fatigue (Allen, Rosenthal, & Parseghian, 1995). They also have been used in monitoring driver drowsiness and in measurement of the interference effect of in-vehicle IVHS (ITS) tasks on driving performance (Allen et al., 1995), as well as in a driving simulation based on virtual environment technology utilizing a head-mounted display (HMD) worn by the user (Levine & Mourant, 1995).

Uses of low-cost simulation for training and evaluation. Low cost simulators are now being used for practical applications such as procedural training of police officers and for the training and evaluation of older drivers (Mauger, 1994; Nowicki, 1994; Time Warner Interactive Simulation Products, 1993, 1994).

In the future, due in large part to limitations of the current DMV road test, simulation can be used for assessing the population at large, by evaluating, training, and eventually granting licenses (McKnight & Stewart 1992; Peck & Wachtel, 1993; Transportation Research Board, 1992). For example, with the guidance and input from various police training agencies around the country, a low-cost interactive simulator has been used for training exercises emphasizing the coordination of tactics, reinforcing compliance with organizational policy, and dispatcher instructions (Olsen, 1995).

Simulator sickness: A potential downfall. One disadvantage of any simulator is that as the number of images in the virtual environment increases, "there is a concomitant increase in the transport delay" according to Frank, Casili, and Wierwille (1988). This delay is a result of the limitations in computing power of the microprocessors used for the simulator. As the number of objects (e.g., vehicles, building, signs, pedestrians) increases, the greater the calculation time and thus an increase in the delay that an image appears on the screen after a driver response input.
Such delays can cause degradation in performance and may increase the likelihood of discomfort, uneasiness, or simulator sickness experienced by the user. Simulator sickness is a form of "visually induced motion sickness," that is, "simulator sickness or discomfort is similar to motion sickness, but it can occur without active motion of the participant" (Kolasinski, 1995). While it is believed that the causes of simulator sickness are visual, it is quite likely both "polysymptomatic" and "polygenic" (Casili & Wierwille, 1986; Kennedy & Frank, 1986). For this study, a brief review of simulator sickness theories and factors of simulator devices that may influence participant discomfort is justified.

Several simulator sickness theories exist (see Kennedy & Frank, 1986). Perceptual conflict or cue conflict theory is the most widely accepted theory of simulator sickness (see Kennedy, Kettinger, & Lilienthal, 1988 for a minor distinction between the two). Perceptual conflict occurs in a fixed-base simulator where there is disparity between the visual system and the vestibular system. While driving the simulator, the eyes perceive motion while the body senses no motion, thus a conflict occurs. However, according to Frank, Casali, and Wierwille (1988), the precise cause of simulator sickness is not known. Other factors such as wide field of view (FOV) may elicit simulator sickness. A system offering a wide FOV provides the opportunity for more stimulation in the periphery. Such perceived movement has been known to induce discomfort and may have been an influence in wide FOV devices such as some Air Force, Navy, and driving simulators (Casili & Wierwille, 1988). A variety of other factors, such as update rate, temperature, ethnicity, gender, experience, and cab enclosure may affect likelihood of simulator sickness (Casili & Wierwille, 1986; Kennedy & Frank, 1986; Kolasinski, 1995).

While the exact percentage of people reporting simulator sickness seems to be unknown and possibly under-reported, Kennedy, Fowlkes, Berbaum, and Lilienthal (1992) report that the incidence among Navy pilots may vary from 10% to 60%, and depends on pilot and simulator characteristics. Participants of a recent workshop discussing driving simulation reported that a range of 1% to as high as 20 or 25% of subjects were affected by such sensory conflicts (Transportation Research Board, 1995).

The benefits of driving simulators. Much driver research indicates that the utilization of simulation technology may have the potential to save billions of dollars in life, property, and insurance costs (Transportation Research Board, 1992) in the next 30 years. Already, simulators are being used because of safety and cost concerns. For instance, the wide-spread popularity of low-cost simulation for police training has stemmed from an alarming increase in pursuit-related costs, accidents, and fatalities. Costs associated with the entire nation's police pursuit costs are measured in billions of dollars (Auten, 1994). A large portion of this could be saved by improving police driving training alone. With costs and potential savings such as these, it can be easily seen that the benefits of simulation for driver safety are important to explore.

The Present Study
The purpose of this study is primarily to determine if a driving simulator can be used to accurately and efficiently measure driving performance as compared to an on-road evaluation, and additionally, to identify differences between road and simulated driving. In order to properly evaluate drivers using a driving simulator in place of currently used driving evaluations, the following question must be answered: Can a simulator serve the purpose of an on-road evaluation? Using a driving simulator for evaluation purposes may assist OTs to be more efficient in their client evaluation work, allow more people to continue to drive in a safe manner, and maximize the effectiveness of the transportation system of our society. By assessing the skills and abilities of drivers with a simulated driver evaluation as compared to an on-road evaluation, it can be determined if a simulator could be used effectively by OTs in place of an on-road evaluation.
For this study, a low-cost interactive simulation was used. This driving simulator is completely interactive in that actions of the subject or failures to act in certain situations influence the visual display that closely reflects experiences of actual driving (Janke, 1994). Various versions of this driving simulator have been used for a variety of experimental and pragmatic applications. The current study hopes to expand upon previous and on-going work in yet another application of such technology. While flight, driving, and other simulators have been and are being used extensively, questions remain as to the appropriateness and validity of their application. Can a simulator be used for longer, whole-task driver evaluation or assist part-task assessment? Most applications thus far have been limited to either short scenarios or jumps (e.g., 3-5 minutes) or for longer sessions (10-30 minutes) with very limited maneuvers, such as with straight road or highway driving (e.g., McGehee, Dingus, Papelis, & Bartelme, 1995; Sexton, 1994). This is not to say that useful findings have not been obtained, but this study attempts to assess driving in a realistic, whole task setting such as one would encounter when driving on the road, utilizing a low-cost, interactive driving simulator.

One problem with driving simulation studies has been the inability to compare on-road driving performance to simulator performance. Such comparisons are important in determining if a driving simulator can be used to measure driving performance. However, the few studies that have directly compared road performance to simulator performance have done so mostly on very expensive, elaborate, motion-base simulators. Now, with low-cost, interactive simulators available, such comparisons are more feasible. While the problems of simulators, and specifically driving simulators, continue to exist, it is to be determined whether the potential benefits will outweigh the limitations of using simulation.