Effectiveness of a Program Using a Vehicle Tracking System,Incentives,and Disincentives to Reduce the Speeding Behavior of Drivers With ADHD

Abstract

Objective: In this article, the authors investigated the effectiveness of a behavior modification program using global positioning system (GPS) vehicle tracking devices with contingency incentives and disincentives to reduce the speeding behavior of drivers with ADHD. Method: Using an AB multiple-baseline design, six participants drove a 5-mile stable driving route weekly while GPS devices recorded speeds. The dependent variable was percentage of feet speeding. Following an initial baseline period, five participants received treatment. One participant remained at baseline. Results: Visual inspection of individual participant graphs, reductions in mean percentages of speeding from baseline to treatment across participants (M = 82%), C-statistic analyses, and visual graphs with applied binomial formula supported a treatment effect. The between-participant analysis using R _n Test of Ranks was significant, R _n = 6, p < .01, and complemented a clean multiple-baseline result. Conclusion: Results indicated that this treatment program was effective in reducing speeding by drivers with ADHD and warrants replication.

Keywords

ADHD speeding drivers GPS contingency

Drivers with ADHD experience a higher incidence of automobile crashes, at-fault crashes, bodily injury with crashes, and traffic citations, particularly speeding, than the general population (Barkley, Murphy, DuPaul, & Bush, 2002), with speed directly related to the severity of crash injury or incidence of fatality (Ossiander & Cummings, 2002). Youth and male gender are major risk factors for risky driving behavior, and young males with ADHD are a high at-risk population for injury or death behind the wheel (Barkley, 2003).

At-risk Drivers With ADHD

Barkley, Murphy, and Kwasnik (1996) investigated negative driving outcomes of individuals with ADHD by means of a Self-Report Survey, review of driving records, and by testing participants’ driving skills using a computer simulation program. In all, 25 young adults from the community with ADHD and 23 without ADHD (ages 17-30) participated. More drivers with ADHD received speeding citations than did the comparison group (100% vs. 54%). Drivers with ADHD received an average 4.9 tickets each compared with the average 2.3 tickets per driver without ADHD. It was seen that 80% of the drivers with, compared with only 52% of drivers without, ADHD reported previous crash involvement, with drivers with ADHD averaging 2.7 crashes compared with 1.6 crashes for drivers without ADHD. Of even more concern, 60% of the ADHD group reported crashes that involved bodily injury as opposed to only 17% of the non-ADHD group, and 32% of drivers with ADHD reported having had their licenses suspended or revoked compared with only 4% of the comparison group.

Barkley et al. (2002) obtained nearly identical results in a study of 105 individuals with, and 64 without, ADHD (ages 17-28). The ADHD group received twice as many driving citations (11.7) on average as the non-ADHD group (4.8), with speeding tickets topping the list again (3.9 vs. 2.4). This time, instead of bodily injury, dollar amount determined crash severity. The average dollar cost of crashes involving drivers with ADHD was US$4,221, compared with US$1,665 for crashes involving drivers without ADHD. The authors found no differences in the driving difficulties of those with ADHD based on sex or subtype of ADHD (inattentive, hyperactive, or combined) or IQ (all participants > 80). Although results did find IQ to be a moderating factor in driving ability—IQ affected drivers with and without ADHD alike—there was no interaction effect.

Studies conducted by Nada-Raja et al. (1997) in New Zealand provided converging evidence for similar driving studies conducted in the United States. In this comprehensive and longitudinal study, Dunedin health department officials tracked 470 male and 446 female children born between April 1, 1972, and March 31, 1973, assessing these children at various junctures for 18 years. Of this cohort, the authors found that 11% had reported ADHD symptomology at the age of 15. They found that ADHD symptomology was highly associated with traffic crashes. In addition, by examining the driving records of the entire cohort, they found that individuals exhibiting ADHD symptomology committed significantly more driving offenses than their non-ADHD counterparts and that this finding was not a result of a comorbid diagnosis such as conduct disorder (CD) or oppositional defiant disorder (ODD), although both of those diagnoses were also more highly associated with traffic offenses compared with drivers without these disorders.

In one of the most recently published longitudinal studies in the United States, Fischer, Barkley, Smallish, and Fletcher (2007) analyzed data from a cohort of hyperactive children (n = 158) and a matched community control sample (n = 81). Investigators first evaluated both groups of children when they were 4 to 12 years old in 1979 to 1980, again at ages 12 to 20 years in 1987 to 1988, and for the published driving study when they were ages 19 to 25 years between 1992 and 1996, with 93% and 91%, respectively, of the original cohort participating. Based on self-report data, significantly more of the hyperactive group had experienced a license suspension/revocation, had received tickets for reckless driving and driving without a license, and had been involved in a hit-and-run collision than had members of the control group. Although the frequency of speeding tickets and total traffic tickets was not found to differ significantly between the two groups, speed may have been a factor in the significantly higher cost of first accidents reported by hyperactive members as speed at the time of accident was not reported. Department of Motor Vehicles (DMV) records indicated that more members of the hyperactive group, relative to the control group, had received a traffic citation. Based on measures of self-report and other-report safe driving behavior ratings, the hyperactive group obtained significantly lower ratings of safe driving behavior than the control group and was rated as making more impulsive errors than the control group on a behind-the-wheel road test. On a driving simulator test, the hyperactive group incurred significantly more scrapes and crashes and exhibited more variability of steering. The authors reanalyzed all data after removing participants from psychiatric medication (8% of the hyperactive group, 1% of the control group) with no change in result patterns. The study did not compare a hyperactive group on medication with one not on medication. The study supported previous findings that ADHD is associated with a higher risk of adverse driving outcomes. Self-report ratings after simulator performances suggested that this higher risk may be due to poorer impulse control and an inability to adapt quickly to driving situations. Fischer et al. pointed out that, overall, adverse driving outcomes were not as disparate between the two groups (i.e., hyperactive and non-hyperactive) as those found in previous studies. One explanation may be that the hyperactive cohort in this study may have differed qualitatively from the clinic-referred adults with ADHD found in some of the previous studies (e.g., younger mean ages and, thus, less time driving). Alternatively, or additionally, members of the hyperactive group may have underreported their adverse driving outcomes. There is some evidence in the study supporting this latter explanation. Members of the hyperactive group dramatically underreported their symptom severity compared with parental reports. The correlation in this study between these two measures was nonsignificant whereas the correlation between the parent- and self-report of clinically referred adults with ADHD had previously been found to be around .75 (K. Murphy & Barkley, 1996; P. Murphy & Schachar, 2000).

Treatment for ADHD-related driving problems has focused on the use of medication to control symptoms associated with poor driving skills (Barkley & Cox, 2007). Research suggests that stimulant medication significantly improves driving performance and reduces inattentive driving errors made by drivers with ADHD. This prompted the current investigators to require participants to remain on a physician-prescribed medication for ADHD during this study.

Psychotherapy Interventions for Symptoms of ADHD

In two comprehensive literature reviews, Barkley (2004a) and Hinshaw, Klein, and Abikoff (2002) reported low sustained efficacy from three common psychotherapeutic interventions for treating ADHD symptoms (i.e., social skills training, traditional, and cognitive-behavioral therapy). Efficacious interventions were contingency methods combined with medication. Barkley posited that ADHD is a “disorder of performance” involving a deficit of real-time application of knowledge rather than a deficit of knowledge itself (Barkley, 2004a). Barkley’s model offers two explanations for the failure to apply knowledge, providing clues for optimal interventions.

First, according to this model, persons with ADHD experience a deficit or delay in internally representing forms of information (internalization of behavior). Thus, an intervention should make needed information overt. Because this deficit leads ADHD individuals to have difficulty organizing their behavior across time (i.e., “temporal myopia”), they can be helped with tools that offer an external visual or auditory representation of time (e.g., clocks or timers). Second, persons with ADHD experience a deficit in internally generated and represented forms of motivation necessary to complete goal-directed behavior, which may help explain why contingency interventions that provide an externally generated and represented motivational source are effective. Barkley (2004b) admitted, however, that applying this model in the design of effective interventions for teen drivers is complicated by teen strivings for greater autonomy from parents and the fact that there is no one in the car to provide externalized forms of information or motivation to enhance driving safety. The current study combined innovative in-car tracking devices equipped with global positioning system (GPS) technology and a contingency reinforcement program to attain the treatment objectives as outlined by Barkley.

Contingency management (CM) is the systematic reinforcement and/or punishment of targeted behaviors to produce an increase in desired behaviors or a decrease in undesired behaviors, respectively. Although, according to Hackenberg (2009), CM stands “among the most successful behaviorally-based interventions in the history of psychology” (p. 257), detractors have argued that the use of extrinsic rewards decreases a person’s intrinsic motivation. However, a meta-analysis of approximately 100 studies by Cameron and Pierce (1994, 1996) did not support the negative view of CM. Their results indicated that it was the method of extrinsic reward implementation that created the troublesome motivational outcomes found in some studies. They found that, although extrinsic rewards applied to task-based behavior were shown to produce a negative effect, extrinsic rewards tied to performance—or achievement-based behavior—enhanced intrinsic motivation.

One type of CM system is the token economy. Hackenberg (2009) reviewed the use of token economies in CM systems and concluded that, when paired with multiple terminal reinforcers such as a store where they can be exchanged for a variety of items of one’s choosing, tokens acquired strong reinforcement properties. Hackenberg noted the attractiveness of token economies in that the incentives and disincentives used are of equal value, making their application cleaner and without the confounds inherent in the application of privileges or other natural reward systems. One weakness Hackenberg found in token economy systems was the low response rates that occurred early on before tokens had acquired their reinforcing properties. However, early weak responding could be attenuated by providing free tokens at the outset. Carlson, Mann, and Alexander (2000) cited numerous studies involving children with ADHD where the application of CM and one punishment procedure in particular—response cost—was demonstrated to be efficacious. The effect of response cost was suggested, in some reports cited, as being more effective than stimulant medication in enhancing performance and more effective than rewards in maintaining gains following treatment withdrawal. In the current study, each participant began with a certain amount of tokens (i.e., money) already credited to an on-paper account that they could either double (i.e., reward/incentive) or lose completely (response cost/disincentive), contingent on speeding behavior. At the end of the study, participants were given cash equal to their paper account balance, which they could spend at any store of their choice.

In adult populations, CM interventions have been successfully used to curb recalcitrant, impulsive behaviors such as substance abuse. Higgins and Petry (1999) outlined four principles central to the success of CM in this patient population: (a) regular testing is necessary to ensure ready detection, (b) tangible reinforcers should be agreed on for specified behavior, (c) designated incentives are withheld when the behavior is not modified as agreed, and (d) the patient works with the clinician to establish alternate and healthier activities to compete with the reinforcement formerly derived from the abusing lifestyle. Higgins and Petry noted that CM in substance abuse programs is usually not a stand-alone therapy but part of a comprehensive treatment plan that incorporates other psychosocial and pharmacological interventions. In the current study, we used in-car GPS devices to regularly monitor and test speeding behavior, satisfying the first principle. We used tokens in the form of money as a tangible reinforcer, and speeding behavior was clearly defined for incentive/disincentive purposes, satisfying the second and third principles. The fourth principle was outside the scope of our study. However, our treatment program was more comprehensive in nature than most CM systems. Our program incorporated weekly feedback sessions using information uploaded to the first author’s computer from the participants’ GPS devices and required participants to be on medication as prescribed and overseen by their personal physicians.

General Speeding Interventions

Diffuse interventions

Because the likelihood of injury and death in an automobile crash is so closely associated with the speed of the vehicle(s) involved, interventions to reduce speeding in general proliferate. Diffuse interventions target all drivers within their scope of operation. They include feedback in the form of mobile roadside speed radar trailers, visible speed enforcement cameras, police surveillance, and enforcement using disincentives in the form of fines. Although these localized interventions have succeeded in slowing traffic, in reducing the percentage of drivers that speed on entering their proximity, and in reducing speeding in the proximal population as a whole, the beneficial effects are often site specific or temporary and often least effective on the most problematic drivers (Casey & Lund, 1993; DeWaard & Rooijers, 1994; Hirst, Mountain, & Maher, 2005; Vaa, 1997).

Results from a meta-analysis by Masten and Peck (2004) found that distribution of educational or informational materials alone was an ineffective intervention resulting in no reduction in crashes or violations. A small but significant reduction in both crashes and violations was associated with warning letters, group meetings, individual hearings, and license suspension/revocation, with license suspension/revocation by far the most effective treatment. In a study funded by the World Health Organization (WHO), Peden et al. (2004) concluded that “public education without enforcement has negligible effect, but, combined with enforcement, increases compliance with laws” (p. 28).

Enforcement (fines) and suspension/revocation programs are predicated on deterrence theory. Deterrence theory posits that consequent punishment will deter people from performing behaviors in which they perceive a high likelihood of detection, and that being caught and punished, or witnessing others who have been caught and punished, will deter people from performing similar behavior in the future. Punishment certainty, severity, and immediacy are the three critical factors believed to account for behavior change through deterrence (Glendon & Cernecca, 2003; Ross, 1982). Detractors for the use of deterrence measures have pointed out that such measures used to reduce speeding are neither cost-effective nor permanent (Glendon & Cernecca, 2003; Hauer, Ahlin, & Bowser, 1982; Van Houten & Nau, 1981). However, results from a meta-analytic review by Blais and DuPont (2005) of police programs using deterrence measures provided support for their use. Of the 33 programs evaluated, all but 3 led to an average decrease of 23% to 31% in injury-causing accidents by controlling various deviant driving behaviors (e.g., speeding, red-light running, driving while intoxicated) through deterrence.

Some of the most effective deterrence strategies combine photo radar, speed cameras, or roadside speed radar trailers with an enforcement program. The strategy of these types of programs is to ensure that the driver perceives that the likelihood of being caught is either high or unpredictable. Keall, Povey, and Frith (2001, 2002) found that hidden speed enforcement cameras appeared to provide a more generalized effect that lasted up to 2 years in the follow-up study. The authors attributed the longer generalized effect, in whole or in part, to the accompanying fourfold increase in ticketing and fines. Chen, Meckle, and Wilson (2002) found that the use of mobile photo radar units, when unpredictably deployed at different times and locations, decreased the traffic speed and incidence of collisions at the photo radar locations. Furthermore, the effect carried over the entire length of the designated enforcement corridor. The authors concluded that because drivers could not discern when or where it was safe to speed, a more generalized effect occurred.

Casey and Lund (1993) noted that providing speeding feedback alone via roadside speed radar trailers or in combination with police enforcement resulted in about a 10% reduction in average vehicle speeds at the feedback (speed radar trailer) site. The difference between the two presentations of feedback (i.e., alone or combined with enforcement) was that speeding reductions associated with enforcement appeared to last through the entire length of deployment (3 weeks), whereas reductions associated with speed radar trailers deployment alone had largely dissipated by the end of the 1st week despite their continued presence. Fuller (1991) argued that such behavioral adaptations to detection methods occurs because the contingency between speeding—a rewarding behavior—and hazardous consequences are improbable and uncertain, leading to contingency conditioning in favor of speeding. Such an argument lends support to the current driver improvement program that provided a reward for driving within 5 mph of the posted speed limit and, through the use of the GPS monitoring device, made the probability of speeding detection certain.

Holland and Conner (1996) found one group (i.e., young men) for whom feedback via roadside speedometers and police enforcement resulted in an increase in intention to speed that was suggestive of a reactance effect. Reactance is a motivational state thought to be triggered by perceived threats to behavioral freedoms that results in an effort to protect or restore those threatened freedoms (Brehm, 1966). Because crash and violation repeaters have been found to respond differentially to various types of interventions depending on gender and the authority source imposing restrictions, a more customized approach to driver improvement programs that took into account characteristics of the targeted group has been advocated (Donelson & Mayhew, 1987; Pennebaker & Sanders, 1976; Woller, 2000). The investigators in the current study with a young male driver population took these factors into consideration by offering both incentives and disincentives without threatening the driver’s freedom to make speed choices. In addition, in consideration of the study’s ADHD population, the contingencies provided an externally generated and an externally represented motivational source as suggested by Barkley (1994).

Individually targeted interventions

The search for a cost-effective way to automatically reduce or override drivers’ propensity to speed has been fueled in large part by the weaker site-specific or temporary gains provided by diffuse interventions (Comte, 2000). Interventions that target the individual driver most often have utilized a mechanical or electronic device that is installed in the driver’s car intended to either make the driver more aware of their speed or mechanically prohibit speeding altogether. Examples of such devices include the active accelerator pedal, electronic speed controller, (ESC), adaptive cruise control (ACC), and intelligent speed adaptor (ISA). Research on such devices has been mixed, with most devices resulting in beneficial effects, but often at the cost of inducing unwanted behavioral adaptations (Comte, 2000; Garvill, Marell, & Westin, 2003; Hjälmdahl & Várhelyi, 2004; Organisation for Economic Co-Operation and Development [OECD], 1990; Saad et al., 2004).

Várhelyi, Hjälmdahl, Hydén, and Draskcózy (2004) got more sanguine results with long-term use of an ACC device that produced the highest reduction of speed in the fastest drivers, improved compliance, and resulted in fewer negative behavioral adaptations. In their study, the authors used a GPS device to monitor the speed of individual drivers with driving data logged for a baseline period before the ACC device was activated. Várhelyi et al. argued that automatic speed adaptation devices may simply require an initial adjustment period to eliminate negative behavioral adaptations but that more research is needed before such conclusions can be drawn. Although Várhelyi et al. used a GPS device in their study to monitor speeding, the device and the data it generated were not used as an intervention tool. The current study used a GPS device to provide feedback and to tie contingencies to the feedback data to manipulate the speeding behavior of drivers. To our knowledge, this is the first study in which GPS technology was used as part of the treatment program, and not simply a means of gathering data or monitoring behavior. The multiple-baseline design of our study allowed drivers an initial adjustment period for having the device in their car. The use of contingencies, tied to feedback, was an important part of our program, as speed feedback alone, without the threat of speed enforcement, has been found to have limited, if any, effect on reducing speeding (Briziarelli & Allan, 1989; Casey & Lund, 1993; Parker, Manstead, Stradling, & Reason, 1992).

Hutton, Sibley, Harper, and Hunt (2002) successfully used feedback on targeted driving behavior when delivered by passengers in a field case study using a multiple-baseline, across-behaviors ABA design. Comparing baseline with intervention periods, both drivers improved on their two targeted behaviors (i.e., mirror checking and either following distance or speeding). Mirror checking and following distance remained improved at follow-up. However, for the male, appropriate speed behavior fell to near baseline levels by follow-up. This case study is important in that it demonstrated the usefulness of applied behavior analysis in constructing and evaluating a customized intervention program as proposed by Donelson and Mayhew (1987), while demonstrating, once again, the recalcitrant nature of speeding behavior.

Current Driver Improvement Program

Our study examined the effectiveness of a behavior modification program for reducing the speeding behavior of young male drivers with ADHD. GPS units installed in participants’ cars detected occurrences of speeding with a color-coded visual representation of driver speed. Incentives and disincentives were used to reduce the frequency of speeding on individualized, preselected driving routes incorporating both city and highway driving conditions.

Using a multiple-baseline AB design, six young male drivers with ADHD, between the ages of 19 and 31, drove an individually determined stable driving route (SDR), approximately 5 miles in length, two or more times a week for 8 weeks. After an initial baseline period (Phase A), five of the six drivers entered the treatment phase (Phase B) in a staggered fashion. One driver remained at baseline. The first author monitored the speeding behavior of all drivers while in baseline (Phase A). As each experimental driver entered his treatment phase, she informed him that the purpose of the study was to reduce speeding on the SDR and demonstrated how his car’s GPS tracking device would be used to monitor his speeding behavior. She outlined the treatment program consisting of incentives for not speeding and disincentives for speeding on the SDR that the driver would earn weekly. We analyzed Phase A and B data to determine treatment effect.

Hypotheses

The dependent variable (DV) was percentage of feet speeding (i.e., number of feet speeding on the SDR divided by number of feet in the SDR). See the “Materials” section for how the GPS vehicle tracking device (i.e., TravelEyes2) collected these data. Using a significance level of p < .05 for all analyses, we hypothesized the following:

Hypothesis 1: There would be a downward trend (decrease) in the percentage of feet that drivers sped on the SDR during the treatment phase (Phase B) compared with the percentage of feet they sped during baseline (Phase A). The C statistic was used for this trend analysis.

Hypothesis 2: A visual graphic using the split-middle technique would show a corresponding downward trend during Phase B. A binomial formula applied to Phase B data points would show a significant rate of change (slope) for each participant when compared with a projected slope (celeration line) based on Phase A behavior.

Hypothesis 3: The R _n Test of Ranks would detect a significant treatment effect across participants. Inspection of a multiple-baseline graphic would provide visual analysis of between-participant results that would complement the R _n Test of Ranks.

Method

Participants

The Institutional Review Board of Old Dominion University approved this research, and we followed established ethical guidelines (American Psychological Association, 2001). Using convenience sampling, we recruited seven adult (ages 19-31; M = 24) male participants with ADHD (six Caucasian and one Hispanic) from a southeastern metropolitan area through a university disability services office, private practices, and by word of mouth. We assigned participants by their order of enrollment, sequentially assigning the first six enrolled participants to staggered treatment positions, with the last participant assigned to a baseline-only position. One participant (P2; Caucasian) dropped out before data collection began. Of the six remaining participants, five (P1, P3, P4, P5, P6) underwent a baseline monitoring phase for a minimum of 2 weeks, followed by a treatment phase implemented in a staggered fashion. One participant (P7) remained at baseline throughout the study and opted to receive treatment after data collection ended.

All participants reported that they had received an ADHD diagnosis (five—inattentive and one—combined type) from a licensed mental health clinician as determined by a formal clinical interview. Five (P3-P7) of the six participants’ treating physicians confirmed their diagnoses by signature. The physician mailed a form attesting that the participant’s ADHD diagnosis had been arrived at by clinical interview and that medication supervision would be ongoing. One (P1) participant’s form was not returned. Five participants (P1, P3-P6) reported a diagnosis of ADHD–inattentive type; one (P7) reported combined type. All six met Diagnostic and Statistical Manual of Mental Disorders (4th ed., text rev.; DSM-IV; American Psychiatric Association, 2000) diagnostic criteria for ADHD as determined by an 18-item symptom checklist (M = 50, SD = 8.7) and by a screening version of the same checklist (M = 17, SD = 3.9; WHO, 2003; Kessler et al., 2005). Cutoff scores above 24 and 11, respectively, meant symptoms were highly consistent with ADHD. All agreed to remain on physician-prescribed ADHD medication (one—Ritalin, four—Adderall, and one—Strattera) with ongoing physician supervision for the duration of the study. Participants reported a medication compliance rate of 82%. On the Wechsler Abbreviated Scale of Intelligence (WASI; Wechsler, 1999), participants scored in the average to very superior range (M = 123, SD = 7.97). All participants were either in, or had graduated from, college, and all held part- or full-time jobs in a chosen field or profession. Average age at which participants received their driver’s license was 15 years 11 months.

Average number of miles driven by participants at the start of the study was 181 miles per week. Total and mean (M) number of citations, crashes, and at-fault crashes reported by participants (n = 6) were as follows: 15 speeding (2.5), 9 crashes (1.5) with 3 at-fault crashes (0.5), 4 reckless driving (0.67), 1 running red light (0.16), and 1 failure to stop (0.16). Participants reported that they were free of any history of conditions that would impair their ability to drive, such as stroke, seizures, epilepsy, uncorrected vision problems, or difficulties with drugs or alcohol. All denied any history of driving under the influence (DUI) citations. Each participant produced a valid driver’s license and had daily, sole access to a car equipped with a cigarette lighter for the GPS unit (TravelEyes2).

Measures

ADHD screening

Adult ADHD Self-Report Scale (ASRS-v1.1) Symptom Checklist (WHO, 2003) is an 18-item self-report questionnaire that asks about the number, type, and intensity of ADHD symptoms a person is exhibiting. Individuals endorse symptom frequency on a 4-point Likert-type scale with 0 representing never and 4 representing very often. The ASRS Screener-v1.1 (ASRS-v1.1; Kessler et al., 2005; see reference for website where this measure can be obtained) is a shorter self-report measure with its 6 items drawn, based on a stepwise logistic regression analysis, from the larger 18-item scale. A dichotomized ASRS Screener outperformed a similarly dichotomized form of the full ASRS Symptom Checklist in sensitivity (68.7% vs. 56.3%), specificity (99.5% vs. 98.3%), and total classification accuracy (97.9% vs. 96.2%). Concordance rates between dichotomized symptom responses and clinical symptom ratings were also better for the screening instrument than the unweighted full checklist (Cohen’s k = .76 vs. .58). Based on the recommendations of Kessler et al. (2005), the investigator used the 18-item ASRS Symptom Checklist in conjunction with the ASRS Screener-v1.1 to qualify individuals for the study. Because the 6 items on the screener were extracted from the 18 items of the ASRS Symptom Checklist, the investigator administered the 18-item checklist but evaluated the scores produced on both instruments.

Demographics

Participants completed a Background Information Survey, based on the one used by Lovejoy (1997), to report age, gender, ethnic background, education, ADHD diagnosis, current stimulant and medication use, and the presence of comorbid psychological or neurological disorders. No reliability estimates were available. Items have face validity.

Driving history

The Driving History Survey, Part A & B is a screening tool used to qualify participants for the study by gathering information about driver’s licenses, access to driving equipment, driving habits, and willingness to perform the requirements of the study. Two questions on this survey (Items 9 and 10) were taken from a Driving History Survey used by Barkley et al. (1996) and asked about the number of times the participant was cited for various offenses, especially moving violations (e.g., speeding, reckless driving, at-fault crashes). No reliability estimates were available for this measure. Items have face validity.

Speeding

For incentive and disincentive (dis/incentive) purposes, speeding was defined as any instance in which the GPS vehicle tracking detected a driver exceeding the posted speed limit by at least 6 mph for 250 consecutive feet (approximately 83 yards) or more. Thus, drivers were given a “grace span” of 250 consecutive feet before earning a disincentive. Typically, police officers have used a 5-mph over-the-speed-limit criterion for speeding (Holland & Conner, 1996). We chose a 6-mph speeding criterion to accommodate the limitations of the GPS devices.

Percentage of feet speeding

The DV was percentage of feet speeding, defined as the total number of feet speeding on a SDR divided by number of feet in the participant’s SDR (i.e., usually about 26,500 ft). One data point was plotted for each day the participant drove the SDR. Participants drove 2 to 6 days per week (M = 4). If the participant drove the SDR more than one time in a day, data for that date were truncated by randomly choosing (using Excel Rand) 1 data point for analysis.

Setting

SDR

Participants drove an individually tailored SDR 1 time a day, 2 to 6 times per week (M = 4), for 8 weeks. All six SDRs were carved out portions of a larger route routinely driven by the participants. The first author worked with each participant to choose a driving route that balanced participant convenience with the requirements of the study. One participant experienced a change in job location so that his route no longer was a portion of the larger route he routinely drove, but he continued to drive the agreed-on route for the remainder of the study. SDRs were approximately 5 miles long and consisted of a combination of city and highway roads for five of the six participants. One participant drove an all-city route. Two participants had matching routes. All city portions of driving routes had a posted speed limit of 30 or 35 mph and all highway (interstate) portions a speed limit of 55 mph.

Materials

Vehicle tracking system and software

We used Travel-Eyes2 GPS System to monitor the speeds at which participants drove their SDRs (Advanced Tracking Technologies, Inc. 2001). Positional accuracy of the device is within 10 to 50 ft, or approximately 3 to 15 m, to the vehicle’s true geographic coordinates but averages a positional accuracy of approximately 20 ft, or 6 m. The tracking unit records data at 10-s intervals. The data display in a map format in which the driver’s route is color coded to indicate the driver’s speed along the route in either 5 or 10 mph increments. A corresponding 5- or 10-mph speed bar legend is provided. Thus, speeding could not be detected until a driver was at least 6 mph or more over the speed limit. TravelEyes2 provides printable map views of driving routes on fixed map scales of ~500 (i.e., 496) or ~1,000 (i.e., 996) ft/in. and provides a zoom feature. Printouts for measurement purposes were usually at the 1,000-ft level, with one-fourth inch = 250 ft, and one-eighth inch = 125 ft. At these scale levels, the DV of speeding (i.e., >250 ft,) could be instantly viewed by participants during the treatment phase for dis/incentive purposes and measured more precisely later in printout form for analysis. A limitation of this software is that routes driven on the same day are not stored separately, so that routes driven more than one time a day overlapped in the printout and were not reasonably distinguishable. Seven routes driven by one participant had to be excluded due to this limitation.

Procedure

Eligible volunteer participants were assigned in the order of their enrollment (i.e., to the first open position in the multiple baseline). After settling on an agreed-on SDR, the first author informed eligible participants that the study involved monitoring aspects of their driving behavior but delayed informing them about which specific driving behavior (i.e., speeding) would be monitored until the beginning of each participant’s treatment phase. A raffle for a free iPod was held at the end of the study as an incentive to remain in the study to its conclusion and to drive their route at least twice weekly.

The first author drove each participant’s SDR prior to data collection to record speed limits and make any final logistical adjustments to SDRs before installing the in-car GPS equipment. P1’s unit was installed 1 week before the others’ as a test run for equipment difficulties. None of the data from that week were used in this study. P3, P4, P5, P6, and P7 began their 1st week at the time their units were installed. P4’s unit experienced technical difficulty at installation that delayed his 1st week but did not impact analyses as he was able to catch up. On occasion, participant life events interfered with precise data collection dates so that a participant “week” sometimes stretched to more than 7 days, extending the intended 8-week data collection period to 9 to 11 weeks. The first author met with each participant weekly to download data from the participant’s unit to her computer. These meetings occurred at a time and location of the participant’s choosing. In each case, participants chose to meet near where their car was parked where they could easily retrieve their GPS device without having to remember to bring it with them to another location. In most cases, this meant meeting outside their residence, but sometimes they chose to meet in a college parking lot where they attended classes or worked. GPS units were uploaded to the first author’s computer while sitting in her car. The download process automatically erased previous data from the individual units, which were then reconnected to record the next week’s data.

During baseline (Phase A), participants were blind to the driving behavior under investigation (i.e., speeding) and did not see their downloaded GPS information. At the end of the baseline and beginning of the treatment phase, the first author followed a script that disclosed the behavior under study, the incentive/disincentive program, and operationally defined speeding for the purposes of that program. In subsequent feedback sessions during treatment phase (Phase B), participants viewed their downloaded GPS information where their driving speeds were indicated by color code and instances of speeding that had occurred on the SDR pointed out by the first author. She then informed the participant of the dis/incentive earned based on the speeding behavior just viewed.

The dis/incentive treatment program worked and was explained as follows: A certain amount of money had been credited to each participant’s on-paper “account.” His ending account balance would be paid to him in cash. Each week that he did not speed at any time when driving the SDR, he would earn an incentive of US$15.00 that would be added to his account. In any week that the participant did speed on the SDR, he would earn a disincentive of US$15.00 that would be deducted from his account. Account balances were updated weekly.

The initial monetary amount credited to each participant as he began Phase B was calculated by multiplying his number of Phase B weeks by US$15.00. For example, P1 had 6 weeks to drive in Phase B, so he began with an account balance of US$90.00. He could then increase or decrease that amount—contingent on speeding behavior—by US$15.00 per week during the 6 Phase B weeks. This would allow him to finish the program with an ending balance in the range of US$0.00 to US$180.00. Thus, each participant could double his initial award amount, lose it all, or earn some amount in between, depending on his speeding behavior. (Note: P7 remained at baseline for the duration of the study. At the conclusion of the study, he was offered and chose to receive 1 week of treatment at the same level as P6. That is, he began with US$15.00 in his account with the chance to earn a balance of US$0.00 to US$30.00.)

Results

To determine whether our contingency program, coupled with the use of in-car GPS tracking devices to monitor speed, reduced the speeding behavior of drivers with ADHD, we conducted both case study and between-participant analyses. Case studies were first analyzed using the C statistic accompanied by baseline and treatment line graphs (Tryon, 1982; Young, 1941). The study supplemented these foundational analyses with celeration line graphs drawn using the split-middle technique with the application of a binomial formula to determine significance (Barlow & Hersen, 1984). Baseline/treatment and celeration line graphs were merged here to conserve space. These two analyses assessed the first two hypotheses of the study. To assess the third hypothesis, a between-participant analysis was performed using the R _n Test of Ranks, accompanied by a multiple-baseline graphic to provide a pictorial analysis (Revusky, 1967).

Case Studies

The first hypothesis of this study was that drivers would reduce their SDR speeding behavior as indicated by a downward trend (decrease) in the percentage of feet sped during the treatment phase (Phase B) compared with baseline (Phase A). Primary analysis of the data utilized the C statistic, which is a simplified time-series statistical procedure requiring considerably fewer data points (i.e., 8 data points; Tryon, 1982; Young, 1941) than the more commonly used and more complex time-series analysis (i.e., 50-100 data points needed for the autoregressive integrated moving average (ARIMA); Jones, 2003). In the current study, participants produced a range of 8 to 23 data points per phase. Because the required minimum of 8 data points per phase was collected for both baseline and treatment phases of P1, P3, P4, and P5, there were no missing data to replace. By design, P6 obtained fewer than 8 data points during his 1 week of treatment, and P7 remained at baseline throughout the study, so no C statistic was planned for these two participants.

The ratio of C to its standard error is the z statistic. The statistical significance of C was evaluated using a one-tailed z statistic, with a critical z value of 1.64 when alpha = .05. This critical value was applied to all C-statistic trend analyses in this study. The C statistic assumes no trend in baseline. When there is no trend in baseline, but a downward trend in the aggregate data (i.e., when treatment data are appended to baseline data), the indication is that treatment had an effect. When a trend is found in baseline, Jones (2003) recommended that other forms of analyses be used to aid in the interpretation process. Because other forms of analyses, such as the split-middle technique with binomial formula, R _n Test of Ranks, and multiple-baseline graph, were planned a priori (i.e., whether or not a trend was found in baseline), these other forms of analyses were readily available as supplemental analytical aids.

Some drivers drove their routes more than once in 1 day. The C statistic was first performed with nontruncated data and then with truncated data, with no change in outcome. Because only 1 data point would be allowed per date on the multiple-baseline graph, and because the outcome did not change when data were truncated, the results presented here are for truncated data in order that data used in statistical analyses would match data used in graphs. Data were truncated by using the Excel Rand function to randomly select 1 data point to include in analysis. There were 13 such truncated data points in the analysis of the total SDRs, with 3 falling in a treatment phase (i.e., 1 for P1 and 2 for P6) and the remaining 10 falling in a baseline phase (i.e., 1 for P5, 2 for P6, and 7 for P7). P2 dropped out before data collection, so he is not included in the analyses presented here. Because of the strong similarity in graphed results between total, city, and highway data, the figures provide visual results only for participants’ total SDRs to avoid redundancy.

Participant 1

Mean percentage of total feet sped during baseline was 82% (M = 0.82, SD = 0.08) and dropped to 2% (M = 0.02, SD = 0.04) during treatment. P1’s total SDR showed a trend in baseline phase (z = 2.70, p < .01) as well as a trend when treatment data were appended (z = 4.37, p < .001). Although a trend was found in baseline, requiring other statistical means to secure statistical significance (i.e., celeration line with binomial formula and the multiple-baseline Test of Ranks presented later), the drop between baseline and treatment was so striking as to infer effective treatment (Figure 1).

Figure 1.

Participant 1’s baseline and treatment phase, projected celeration line, and binomial formula results.

Participant 3

Mean percentage of total feet sped during baseline was 60% (M = 0.60, SD = 0.20) and 0% (M = 0.00, SD = 0.00) during treatment. P3’s total SDR showed no trend in baseline (z = 1.00, n.s.) but showed a trend when treatment data were appended (z = 4.42, p < .001; Figure 2).

Figure 2.

Participant 3’s baseline and treatment phase, projected celeration line, and binomial formula results.

Participant 4

Mean percentage of total feet sped during baseline was 66% (M = 0.66, SD = 0.20) and 3% (M = 0.03, SD = 0.03) during treatment. P4’s total SDR showed no trend in baseline (z = −1.54, n.s.) but showed a trend when treatment data were appended (z = 3.91, p < .001; Figure 3).

Figure 3.

Participant 4’s baseline and treatment phase, projected celeration line, and binomial formula results.

Participant 5

Mean percentage of total feet sped during baseline was 46% (M = 0.46, SD = 0.17) and 1% (M = 0.01, SD = 0.02) during treatment. P5’s total SDR showed no trend in baseline (z = 0.77, n.s.) but showed a trend when treatment data were appended (z = 3.73, p < .001). Although there was a noticeable downward slope of P5’s trend line that occurred during his first 2 weeks of baseline—possibly due to an observation effect—his trend line eventually stabilized so that the overall baseline trend was nonsignificant (Figure 4).

Figure 4.

Participant 5’s baseline and treatment phase, projected celeration line, and binomial formula results.

Participant 6

Mean percentage of total feet sped during baseline was 41% (M = 0.41, SD = 0.22) and 1% (M = 0.01, SD = 0.01) during treatment. P6’s total SDR showed no trend in baseline (z = −1.46, n.s.). Not enough treatment data points (n = 4) were available to analyze his truncated appended data with the C statistic.

Participant 7

Mean percentage of total feet sped during baseline was 59% (M = 0.59, SD = 0.18) and 0% (M = 0.00, SD = 0.00) during the week he received treatment after the study ended. P7’s total SDR showed no trend in baseline (z = −1.00, n.s.). Treatment data were not analyzed as P7 served as the multiple-baseline control, receiving treatment only after the study ended.

Visual Analyses of Case Studies’ Results With Applied Binomial Formula

The second hypothesis of this study was that an additional visual graphic, using the split-middle technique with applied binomial formula as outlined by Barlow and Hersen (1984), would support C-statistic findings by also showing a significant downward trend in percentage of feet sped on the SDR during treatment compared with during baseline. The split-middle technique involves plotting two trend lines for each participant, one for baseline and one for treatment. The baseline trend line is then extended, or projected, into the treatment portion of the graph, called the celeration line. A binomial formula is then used to determine whether there is a significant difference between the treatment trend line and this projected celeration line. When significantly more treatment points fall below the celeration line than chance (i.e., 50%, or p = .5) would allow, treatment is considered effective.

We applied the binomial formula of p = [(n / x) (.5)] ⁿ , with n as the total number of treatment data points, and x as the number of treatment data points falling below the projected celeration line (Barlow & Hersen, 1984). For example, P3 ended with a total of 14 treatment data points (n = 14). All of P3’s 14 data points fell below the projected celeration line (x = 14). By plugging these values into the binomial formula, the formula now became p = [(14 / 14) (.5)]¹⁴, or p < .001. Thus, P3’s reduction in speeding behavior during treatment compared with his baseline behavior reached significance. Using this method, results indicated that P1, P3, P4, and P5 reduced their speeding behavior during treatment compared with their respective baseline speeding behavior. Split-middle celeration line graphs and binomial formula results for these participants are provided in Figures 1 to 4. Although P6 and P7 had all of their treatment phase data falling below their celeration lines, the binomial formula could not be applied due to insufficient number of data points (n = 4) during Phase B, with P7’s occurring after the study ended.

Between-Participant Study

The third main hypothesis was that the R _n Test of Ranks as proposed by Revusky (1967) would detect a significant treatment effect across participants. With six participants, the critical value at the .05 alpha level was R_n = 8 (i.e., R _n < 8 to be significant). To satisfy the assumption of random assignment of order, the six participants were randomly ordered (by enrollment) to receive intervention. The R _n test also assumes that baseline scores are of generally the same magnitude. As our participant mean baseline scores contained a moderate degree of variance between participants, we considered the necessity of data transformation to correct for the variance as recommended by Barlow and Hersen (1984). We determined that no such correction was necessary because no overlap occurred between the lowest mean baseline score and the highest mean treatment score, creating no ambiguity for interpretation.

After calculating each participant’s mean percentage of feet speeding (DV) for each week, we entered all baseline means and the 1st week of treatment means onto a ranking chart. We followed the ranking procedure outlined in Barlow and Hersen (1984) with results summarized in Table 1. Theoretically, if treatment is effective, participants entering treatment should register the greatest drop in percentage of feet speeding compared with their cohorts and receive a ranking of “1” for that week. Their ranking is then entered at the bottom of the chart in the “Ranks =” row. After their 1st week of treatment, participants then drop out of the ranking procedure. For example, P1 sped an average of 92% (or M = 0.92) of his SDR during Week 1 of baseline, 76% during Week 2 of baseline, and 7% during Week 3, which was his 1st week of treatment. Looking vertically down the chart, one can see that during Week 3, P1 sped, on average, the least number of feet compared with the other participants (P3, P4, P5, P6, and P7 sped 76%, 70%, 54%, 41%, and 55%, respectively), so P1 received the highest ranking of “1.” This ranking was entered in the “Ranks =” row at the bottom of Week 3’s column. P1 then dropped out of ranking consideration, and his percentage of feet speeding no longer recorded on the chart. Rankings for the remaining weeks were assigned in this manner for each participant with the exception of P7. Although P7’s treatment phase occurred after the study ended, his end ranking of “1” was assumed, as argued by Revusky (1967), as a result of being the only remaining participant. (Note: Because P2 dropped out of the study, no participant entered treatment during Week 4, so there is no ranking for that week.) Rankings across participants were then summed (Σ of ranks, or R _n = 6, p < .01). As a group, treatment was effective in reducing speeding on the SDR.

Table 1.

R_n Test of Ranks for Mean Percentage of SDR Feet Sped Each Week.

	Week
Participant	1	2	3	4	5	6	7	8
1	.92^a	.76	.07b
3	.46	.45	.76a	.70a	.00b
4	.62	.83	.70a	.53a	.74a	.04b
5	.50	.20	.54a	.50a	.46a	.55a	.00b
6	.22	.66	.41a	.49a	.50a	.33a	.43a	.01b
7	.34	.58	.55a	.56a	.68a	—	.62a	.53a
	Ranks =		1	1	1	1	1	1
Σ Ranks (R _n ) = 6, p < .01

Note: Weeks 1 and 2 served as baseline (a) for all participants and are unmarked. a = baseline, b = participant’s 1st week of treatment. Because P2 dropped out of the study, there is no b for Week 4, which would have been P2’s 1st week of treatment. Table modeled after one by Barlow and Hersen (1984, p. 310).

Table values represent percentage of feet sped on the stable driving route.

Inspection of the multiple-baseline graphic (Figure 5) provides visual analysis of between-participant results that complement the R _n Test of Ranks.

Figure 5.

Multiple-baseline results.

Discussion

The focus of this study was the efficacy of a driver improvement program using a GPS vehicle tracking device with incentives and disincentives to reduce speeding in young male drivers with ADHD. Analyses of case studies and between-participant comparisons support our three hypotheses and indicate individual and collective program efficacy in this participant pool.

We first used the C statistic to analyze individual data of participants (P1, P3, P4, and P5) with the requisite number of data points in both their A and B phases and found that each participant reduced speeding behavior during treatment (Phase B) as hypothesized. One participant (P1) had a significant trend during his baseline phase, making statistical results at this stage inconclusive as the C statistic assumes no trend in baseline. It may be that P1’s baseline phase was too brief to allow for stabilization to counteract a potential observer effect (Barlow & Hersen, 1984). Although our participants were not directly observed, they were aware that they were being “watched” by GPS satellite technology. A scan of several other participants’ baselines (e.g., P3, P5, and P7; Figure 5) suggest that a longer baseline period resulted in baseline stabilization before treatment. Nevertheless, the visual figure of P1’s results is highly suggestive of a treatment effect, and results of later analyses converged to support the visual representation that treatment was responsible for P1’s downward trend in speeding during treatment. As interesting as the changes were in participants’ pattern of performance from baseline to treatment detected by the C statistic, there was a striking change in the level of performance observed in all participants that was visible to detection by the untrained eye through inspection of the figures alone. Reduction in number of feet sped on the SDR ranged from 44% to 80% (M = 59%).

The second main hypothesis of this study was that a significant downward trend in speeding would be supported by visual graphics using the split-middle technique and applied binomial formula. Although both the C statistic and split-middle analyses work on the premise of detecting a significant change in slope between baseline and treatment, the C-statistic analysis treats appended treatment and baseline data as one entity when looking for a significant trend. The split-middle analysis treats the baseline and treatment phase data separately, comparing one against the other. This made the split-middle technique a complementary, not redundant, analysis. This supplementary analysis also found a downward trend in the percentage of feet sped during treatment by the same four driving participants (P1, P3, P4, and P5) with all treatment phase data points falling significantly below their projected celeration lines. Note that P1’s treatment phase trend line was found to differ significantly from his baseline trend line at the p < .001 level, leaving no inferential ambiguity. Due to an insufficient number of data points, the binomial formula could not be applied to the data produced by P6 and P7. However, celeration line graphs for both P6 and P7 looked promising. In addition, the plotted baseline celeration line for P7 provided another clear visual picture of a flat trend line throughout his 8 weeks, making P7 a near-ideal control against which the other participants were compared in the between-participant analyses. That P7’s 8-week baseline showed no trend lends further support to our conclusion that participants reduced their speeding behavior as a result of our GPS feedback and contingency program rather than as the result of some external event.

Our third hypothesis—that the R _n Test of Ranks would show a significant treatment effect across participants—was supported. Our program was successful in reducing the speeding behavior of our participants as a group. Results were unambiguous, with no overlap between mean percentage of feet sped on any baseline week (range = 20%-92%) and mean percentage of feet sped on any treatment week (range = 0%-7%; Table 1). Finally, the multiple-baseline graphic provided a clear visual picture of an immediate and sustained reduction in speeding as each participant entered and remained in treatment (Figure 5).

Limitations

Although these results provided clear evidence that our program reduced speeding among our participants on their respective SDRs, there were limitations to the study. The program was instituted as a package that included incentives, disincentives, and feedback via the GPS device, with all components inseparable when analyzing effects. The program likely included intangibles such as rapport between researcher and participant, an increase in driving awareness due to being in a study, or the presence of an altruistic trait inherent in volunteers and not present in the general population. We acknowledge that these factors were not partitioned. In addition, this study used a small sample size of drivers with ADHD with no non-ADHD controls for comparison. The small sample size was largely offset by the use of multiple measures and both individual and between-driver analyses. The convergence of significant results found in all measures within and between participants speaks to the strength of the effect, despite the small number of drivers with ADHD. The study’s multiple-baseline design allowed participants to serve as their own and each other’s control, and it provided one baseline participant throughout as a control for history. The composition of the sample population was atypical of the general population. All participants were male, between the ages of 18 and 35, mostly Caucasian, carried an ADHD diagnosis, were 82% medication compliant, and had a mean IQ of 123. All were college students, which suggests a higher socioeconomic status as well as a higher overall adaptive functioning than the general population. Although such a homogeneous sample limited generalizability, the study benefited from the limitation of confounding variables. Future studies should explore whether these results would hold in a larger, more diverse, pool of drivers with ADHD.

Another study limitation was the introduction of possible confounding variables related to a nonstandard driving route driven at varying times of the day. Although each participant drove a stable personal route, all but two drove routes that differed in some portion to that of all other participant routes. (Two participants’ routes matched exactly.) Route variation was an ethical and purposeful study consideration. We sought to reduce risk—especially for drivers with ADHD—associated with imposing unfamiliar routes. That said, we attempted to guide route choice to those similarly composed of both city and interstate roads. Any further variance that occurred in driving routes, traffic encountered, or varying driving times of day only served to boost generalizability.

Certain other limitations that apply to this study may impact its generalizability. This study did not examine whether the improvement in speeding behavior made by these drivers while driving the SDR generalized to other portions of their routes or to other routes. Although it was not in the scope of this study to determine generalizability of treatment outside the SDR, one participant (P7) reported that he had changed his overall speeding behavior after treatment after determining that speeding had shaved an insignificant, and even imperceptible, amount of time off of his destination arrival time. By his report, the potential cost of speeding outweighed the perceived benefit. When he reported this at the end of his treatment week, the investigator reviewed with him an interstate portion of his SDR where a side-by-side comparison of SDR and non-SDR speeding habits was possible. By visual inspection, it was clear that the participant had also reduced his percentage of feet speeding to 0% on the non-SDR portion by the end of treatment, whereas he appeared to have sped a large percentage of that same interstate portion pretreatment.

With these limitations in mind, these results present a compelling picture of a successful speed-reduction program with an ADHD population. Replication of these results is needed using a larger and more diverse ADHD population that includes females, teenagers, and persons from various socioeconomic, educational, and ethnic backgrounds, and using non-ADHD comparison controls. This program has applicability for use by traffic court judges, ADHD coaches, and parents of ADHD teenagers. An added advantage of our program was its use of the TravelEyes2 units that require no monthly subscription fee.

In conclusion, these findings suggest that drivers with ADHD—even ones who habitually speed—can reduce their level of speeding with treatment. These drivers were all taking medication to control the symptoms of ADHD with an 82% compliance rate. However, simply being on medication did not eliminate speeding in these individuals, as their baseline behaviors attest. Near elimination (i.e., at floor levels) of speeding occurred only after treatment was introduced in all six participants. Medication may allow an individual with ADHD the ability to control his or her driving habits, as the literature seems to indicate. However, affecting a change in the speeding habits of persons with ADHD, like their non-ADHD counterparts, may require an aggressive program that frequently and specifically targets the behavior by applying incentives and disincentives and includes an external feedback system.

Footnotes

Acknowledgements

The authors wish to thank the team at Advanced Tracking Technologies, Inc., for providing critical technical support for the TravelEyes2 units and software, and to the Virginia Department of Motor Vehicles for providing funding to purchase the units.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The Virginia Department of Motor Vehicles awarded a mini-grant in the amount of $1500 to go toward the funding of the GPS TravelEyes2 units (Porter & Markham, 2005). The cost per unit was $200 with no monthly fees.The first author funded all uncovered cost of the units, incentives money, and any other project expenses.

Author Biographies

Paula T. Markham, PsyD, is a clinical psychologist with the Hampton-Newport News Community Services Board where she is the Regional Deaf Services Counselor.

Bryan E. Porter, PhD, is associate professor and PhD programs director for the Department of Psychology at Old Dominion University. His area of emphasis is behavioral community psychology, focusing on community problems and psychology’s role in solving those problems.

J. D. Ball, PhD, is professor and vice chair for the Department of Psychiatry and Behavioral Sciences at Eastern Virginia Medical School where he codirects the Neuropsychology Center.

References

Advanced Tracking Technologies, Inc. (2001). TravelEyes2 Quick Start User Guide. Available from Advanced Tracking Technologies, Inc. P.O. Box 168. Sugar Land, TX.

American Psychiatric Association. (2000). Diagnostic and statistical manual of mental disorders (4th ed., text rev.) Washington, DC: Author.

American Psychological Association. (2001). Publication manual of the American Psychological Association (5th ed.). Washington, DC: Author.

Barkley

R. A.

(1994). Impaired delayed responding: A unified theory of attention-deficit/hyperactivity disorder. In Routh

D. K.

(Ed.), Disruptive behavior disorder: Essays in honor of Herbert Quay (pp. 11-57). New York, NY: Plenum.

Barkley

R. A.

(2003). Attention-deficit/hyperactivity disorder. In Mash

E. J.

Barkley

R. A.

(Eds.), Child psychopathology (2nd ed., pp. 75-143). New York, NY: Guilford.

Barkley

R. A.

(2004a). Adolescents with attention-deficit/hyperactivity disorder: An overview of empirically based treatments. Journal of Psychiatric Practice, 10, 39-56.

Barkley

R. A.

(2004b). Driving impairments in teens and adults with attention-deficit/hyperactivity disorder. Psychiatric Clinics of North America, 27, 233-260.

Barkley

R. A.

Cox

(2007). A review of driving risks and impairments associated with attention-deficit/hyperactivity disorder and the effects of stimulant medication on driving performance. Journal of Safety Research, 38, 113-128.

Barkley

R. A.

Murphy

K. R.

DuPaul

G. J.

Bush

(2002). Driving in young adults with attention deficit hyperactivity disorder: Knowledge, performance, adverse outcomes, and the role of executive functioning. Journal of the International Neuropsychological Society, 8, 655-672.

10.

Barkley

R. A.

Murphy

K. R.

Kwasnik

(1996). Motor vehicle driving competencies and risks in teens and young adults with attention deficit hyperactivity disorder. Pediatrics, 98, 1089-1095.

11.

Barlow

Hersen

(1984). Single case experimental designs. New York, NY: Pergamon.

12.

Blais

DuPont

(2005). Assessing the capability of intensive police programmes to prevent severe road accidents: A systematic review. British Journal of Criminology, 45, 914-937.

13.

Brehm

J. W.

(1966). A theory of psychological reactance. New York, NY: Academic Press.

14.

Briziarelli

Allan

(1989). The effect of a head-up speedometer on speeding behavior. Perceptual and Motor Skills, 69, 1171-1176.

15.

Cameron

Pierce

(1994). Reinforcement, reward, and intrinsic motivation: A meta-analysis. Review of Educational Research, 64, 363-423.

16.

Cameron

Pierce

(1996). The debate about rewards and intrinsic motivation: Protests and accusations do not alter the results. Review of Educational Research, 66, 39-51.

17.

Carlson

Mann

Alexander

(2000). Effects of reward and response cost on the performance and motivation of children with ADHD. Cognitive Therapy and Research, 24, 87-98.

18.

Casey

Lund

(1993). The effects of mobile roadside speedometers on traffic speeds. Accident Analysis & Prevention, 25, 627-634.

19.

Chen

Meckle

Wilson

(2002). Speed and safety effect of photo radar enforcement on a highway corridor in British Columbia. Accident Analysis & Prevention, 34, 129-138.

20.

Comte

(2000). New systems, new behaviour? Transportation Research Part F, 3, 95-111.

21.

DeWaard

Rooijers

(1994). An experimental study to evaluate the effects of different methods and intensities of law enforcement on driving speed on motorways. Accident Analysis & Prevention, 26, 751-765.

22.

Donelson

Mayhew

(1987). Driver improvement as post-licensing control: The state of knowledge. Toronto, Ontario, Canada: Ontario Ministry of Transportation and Communications.

23.

Fischer

Barkley

R. A.

Smallish

Fletcher

(2007). Hyperactive children as young adults: Driving abilities, safe driving behavior, and adverse driving outcomes. Accident Analysis & Prevention, 39, 94-105.

24.

Fuller

(1991). Behavior analysis and unsafe driving: Warning—Learning trap ahead. Journal of Applied Behavior Analysis, 24, 73-75.

25.

Garvill

Marell

Westin

(2003). Factors influencing drivers’ decision to install an electronic speed checker in the car. Transportation Research Part F, 6, 37-43.

26.

Glendon

A. I.

Cernecca

(2003). Young drivers’ responses to anti-speeding and anti-drink-driving messages. Transportation Research Part F, 6, 197-216.

27.

Hackenberg

(2009). Token reinforcement: A review and analysis. Journal of the Experimental Analysis of Behavior, 91, 256-286.

28.

Hauer

Ahlin

Bowser

(1982). Speed enforcement and speed choice. Accident Analysis & Prevention, 14, 267-278.

29.

Higgins

S. T.

Petry

N. M.

(1999). Contingency management: Incentives for sobriety. Alcohol Research & Health, 23, 122-127.

30.

Hinshaw

Klein

Abikoff

(2002). Childhood attention-deficit hyperactivity disorder: Nonpharmacological treatments and their combination with medication. In Nathan

Gorman

(Eds.), A guide to treatments that work (2nd ed., pp. 3-23). London, England: Oxford University Press.

31.

Hirst

Mountain

Maher

(2005). Are speed enforcement cameras more effective than other speed management measures? An evaluation of the relationship between speed and accident reductions. Accident Analysis & Prevention, 37, 731-741.

32.

Hjälmdahl

Várhelyi

(2004). Speed regulation by in-car active accelerator pedal: Effects on driver behaviour. Transportation Research Part F, 7, 77-94.

33.

Holland

Conner

(1996). Exceeding the speed limit: An evaluation of the effectiveness of a police intervention. Accident Analysis & Prevention, 28, 587-597.

34.

Hutton

Sibley

Harper

Hunt

(2002). Modifying driver behaviour with passenger feedback. Transportation Research Part F, 4, 257-269.

35.

Jones

W. P.

(2003). Single-case time series with Bayesian analysis: A practitioner’s guide. Measurement and Evaluation in Counseling and Development, 36, 28-39.

36.

Keall

Povey

Frith

(2001). The relative effectiveness of a hidden versus a visible speed camera programme. Accident Analysis & Prevention, 33, 277-284.

37.

Keall

Povey

Frith

(2002). Further results from a trial comparing a hidden speed camera programme with visible camera operation. Accident Analysis & Prevention, 34, 773-777.

38.

Kessler

R. C.

Adler

Ames

Demler

Faraone

Hiripi

. . .Walters

E. E.

(2005). The World Health Organization Adult ADHD Self-Report Scale (ASRS). Psychological Medicine, 35(2), 245-256. Retrieved from http://webdoc.nyumc.org/nyumc/files/psych/attachments/psych_adhd_screener.pdf

39.

Kessler

Adler

Barkley

R. A.

Biederman

Conners

Faraone

. . .Zaslavsky

A. M.

(2005). Patterns and predictors of attention-deficit/hyperactivity disorder persistence into adulthood: Results from the National Comorbidity Survey Replication. Biological Psychiatry, 57, 1442-1451.

40.

Lovejoy

(1997). Neuropsychological assessment of frontal lobe (executive) functioning in adults with attention deficit hyperactivity disorder (Doctoral dissertation, Virginia Consortium for Professional Psychology). Dissertation Abstracts International, 58(09B), 5128.

41.

Masten

Peck

(2004). Problem driver remediation: A meta-analysis of the driver improvement literature. Journal of Safety Research, 35, 403-425.

42.

Murphy

Barkley

R. A.

(1996). Adults with ADHD: Comorbidity and adaptive functioning. Comprehensive Psychiatry, 37, 393-401.

43.

Murphy

Schachar

(2000). Use of self-ratings in the assessment of symptoms of attention deficit hyperactivity disorder in adults. American Journal of Psychiatry, 157, 1156-1159.

44.

Nada-Raja

Langley

J. D.

McGee

Williams

Begg

Reeder

(1997). Inattentive and hyperactive behaviors and driving offenses in adolescence. Journal of the American Academy of Child & Adolescent Psychiatry, 36, 515-522.

45.

Organisation for Economic Co-Operation and Development. (1990). Behavioural adaptations to changes in the road transport system. Paris, France: Author.

46.

Ossiander

Cummings

(2002). Freeway speed limits and traffic fatalities in Washington state. Accident Analysis & Prevention, 34, 13-18.

47.

Parker

Manstead

Stradling

Reason

(1992). Determinants of intention to commit driving violations. Accident Analysis & Prevention, 24, 117-131.

48.

Peden

Scurfield

Sleet

Mohan

Hyder

A. A.

Jarawan

Mathers

(2004). World report on road traffic injury preventions: Summary. Geneva, Switzerland: World Health Organization. Retrieved from http://www.who.int/violence_injury_prevention/publications/road_traffic/world_report/summary_en_rev.pdf

49.

Pennebaker

Sanders

(1976). American graffiti: Effects of authority and reactance arousal. Personality and Social Psychology Bulletin, 2, 264-267.

50.

Porter

B. E.

Markham

P. T.

(2005). Testing a Driver Monitoring System to Reduce Speeding Behavior of ADHD Drivers. The Virginia Department of Motor Vehicles. Award of $1500. [Proposal # 06-141].

51.

Revusky

S. H.

(1967). Some statistical treatments compatible with individual organism methodology. Journal of the Experimental Analysis of Behavior, 10, 319-330.

52.

Ross

H. L.

(1982). Deterring the drinking driver. Lexington, MA: Heath & Co.

53.

Saad

Hjälmdahl

Canas

Alsonso

Garayo

Macchi

. . .Bekiaris

(2004). Literature review of behavioural effects: Behavioural effects of driver assistance systems. Information society technologies (ISP) programme: Adaptive integrated driver-vehicle interface. Retrieved from http://www.aide-eu.org/pdf/sp1_deliv_new/aide_d1_2_1.pdf

54.

Tryon

(1982). A simplified time-series analysis for evaluating treatment interventions. Journal of Applied Behavior Analysis, 15, 423-429.

55.

Vaa

(1997). Increased police enforcement: Effects on speed. Accident Analysis & Prevention, 29, 373-385.

56.

Van Houten

Nau

P. A.

(1981). A comparison of the effects of posted feedback and increased police surveillance on highway speeding. Journal of Applied Behavior Analysis, 14, 261-271.

57.

Várhelyi

Hjälmdahl

Hydén

Draskcózy

(2004). Effects of an active accelerator pedal on driver behaviour and traffic safety after long-term use in urban areas. Accident Analysis & Prevention, 36, 729-737.

58.

Wechsler

(1999). Wechsler Abbreviated Scale of Intelligence. San Antonio, TX: Psychological Corporation.

59.

Woller

K. M. P.

(2000). The combination of two scales of psychological reactance. Dissertation Abstracts International, 61, 2-B, 1131.

60.

World Health Organization. (2003). Adult ADHD Self-Report Scale (ASRS-v1.1) Symptom Checklist. Retrieved from http://webdoc.nyumc.org/nyumc/files/psych/attachments/psych_adhd_checklist.pdf

61.

Young

L. C.

(1941). On randomness in ordered sequences. Annals of Mathematical Statistics, 12, 153-162.