Research Briefs

Robbing banks: Crime does pay–but not very much.,

Reilly, Rickman, & Witt, Significance, (2012)

While pop media and culture portrays bank robbery as exciting and lucrative, the authors conducted an examination of the actual economic profits and costs of bank robbery in the UK, both to the robber and the bank. The authors were given access to confidential information on the amount of cash lost in bank robberies as well as other data in a cost/benefit analysis of bank robbery. They report that in 2007, there were 80,000 robberies reported to police, 7500 occurred at businesses, but with over 10,500 bank locations in the UK, only 106 bank robberies were attempted, accounting for only .13 % of robberies. Compared to the 2006 data for the US over the 4440,00 robberies reported, 12,000 occurred at banks (2.7%), indicating that in even in the US, bank robbery only makes up a small proportion of robberies.

The authors examined some factors that may influence whether a bank was targeted like size of municipality (defined by distance from a police station), size of the bank based on number of employees and, how busy the bank was based on the number of customers, but none of these factors had a significant effect. Of the robbery attempts in the study, very few were repeat location attempts and the amount stolen also wasn’t related to the size of the bank. Without clear indicators as to which bank branches may be targeted, improvements in bank security must be borne by the whole branch system, the costs of which are often ultimately borne by the customers.

The profitability for the robber was also considered. Using UK data from 2005-2008, of the 364 attempted bank robberies examined, the average loss was £20,330 (five times as much as their US counterparts), however 1/3 of  the attempts were unsuccessful with no money stolen, so the average haul for successful robberies was £30,000 pounds. The authors note that with approximately 20% of bank robberies solved, and some of the haul recovered, this does reduce the financial gain for robbers.

The authors constructed an equation using economic constructs to help identify the robbers’ input into the robbery balanced against expected gains and losses from the attempt. Robbers’ input include labor costs (number in the gang), whether a firearm was displayed (capital costs) and a difficulty vector accounting for size of branch, number of customers, and branch security measures. The output (value of money stolen) is a function of all these factors, however the output may also be influenced by robber apprehension and incarceration. The difficulty vector had little significant influence on the take; average travel distance from a police station was 4.5 minutes, the average number of customers was two, and alarms were activated 85% of the time  

While the average number of robbers were 1.6 (60% were lone robbers) increasing the number in the gang had a significant effect on increasing the amount stolen. For each additional robber employed, the expected value of the haul increased by over £9,000 pounds (with all other factors being equal). Working in teams indicates more professionalism (including planning and surveillance) and task division and while additional gang members increase the total take, it also reduces the take of each individual gang member.

A firearm was displayed in 36% of the robberies. The display and inherent threat of firearm use was also a significant factor in the size of the haul; with all other factors being equal, displaying a weapon increased the average take by £10,000. However, the use of a firearm in some cases may ultimately affect output as the incarceration costs are higher for armed robbery if they are apprehended.

One factor reduced the profit output. Being more common in the UK, as compared to the US, 12% of bank branches utilized fast-rise security screens. Their use reduced the expected value of robbery haul by over £24,000, all other factors being equal. In a regression model to determine whether the screens foil robberies or just reduce the take, on average, and all other factors being equal, using fast-rise screens reduced the probability of a successful robbery attempt by 32% compared to other security systems. Two thirds of the reduced output effect was from a reduced probability of robbery success when using the screens, while the other third was attributable to the screens reducing the financial value of the robbery haul.

As discussed in the criminological Rational Choice theory as well as in economic models, cost/benefit analysis should be a factor in the commission of the act. If the costs outweigh the benefits, a rational choice would be to not engage in the act. The authors constructed a mathematical formula, utilizing the probability of success, the probability of apprehension, the average take for one robber, and a monetary calculation of the cost of imprisonment in terms of lost income, to determine the tipping point where the costs outweigh the benefits. The formula demonstrated the monetary cost should be in excess of £33,000, which is roughly equivalent to three years’ income from a full-time minimum-wage job. Thus a three year incarceration would represent that monetary cost, as well as have a deterrent effect, however it should be noted that the average sentence for a convicted bank robber in England is a little less than three years.

As a means of income, bank robbery may not be very lucrative though the actual work to income ratio is large. The average annual full-time employment income is £26,000 and at an average of £12,706 per robber per robbery, one bank robbery can support  a modest life style for six months. However, with only an 80% success rate, after three robberies (a year and half average income) the likelihood of a success is roughly 50/50 and each subsequent robbery will typically increase the odds of apprehension and incarceration. This overall smaller take may be influencing the overall decrease in UK bank robberies but also may explain an increase in their armored truck/security van robberies as a competing criminal opportunity that provides more returns. The low take in robbery also affects the implementation of security measures. While the authors noted the effectiveness of fast-rise security screens in foiling robberies or reducing their costs, suggesting widespread use could cut successful bank robberies in half, the cost of £4,500 per teller window would suggest the cost of the robbery is a smaller financial cost itself when compared to the security measure.

Reilly, B., Rickman, N., & Witt, R. (2012). Robbing banks: Crime does pay–but not very much. Significance, 9(3), 17-21.

Professionals or amateurs? Revisiting the notion of professional crime in the context of cannabis cultivation.

Bouchard and Nguyen, World Wide Weed, 2016

The authors discuss how historical literature (like Edwin Sutherland’s The Professional Thief) recognizes the level of a professional criminal but more current research hasn’t explored the concept as to its validity or applicability to any great degree. The authors explore defining professionalism by constructing a four classification matrix ranging from amateurs to professionals and applying that to marijuana cultivators in Canada.

There are certain elements that constitute professionalism in all fields; skill and knowledge in the craft or occupation as well as a commitment to the craft or occupation including its function as full-time or significant employment and immersion in its social milieu. Skills would include, good growing skills like feeding, watering, breeding, cloning, trimming, and light management. Skills also involve grow operation construction, management, and security and these skills could be operationalized as outcome measures such as plant (or lamp) yield, attrition rate (proportion of plants harvested), and criminal earnings. The social milieu in this case would be association and connectedness to other growers and laborer/service providers within the industry and subculture, receiving and offering mentorship, and an appreciation for, and dedication to growing, marijuana The matrix classifies commitment  and skill level as either being low or high, resulting in four categories; Amateurs (low skills, low commitment) Average Career Criminals/Growers (low skills, high commitment), Pro-Ams (high skills, low commitment), and Professionals (high skills, high commitment).


Figure from authors’ publication


The authors apply this matrix to a series of interviews conducted with 13 marijuana cultivators in Montreal, Quebec, and Vancouver. The Montreal and Quebec growers were all white males, and the 4 growers from Vancouver were of Vietnamese descent (“the Vietnamese population has been identified as key players in Vancouver’s cultivation industry”) and two were female.

Of the 13 growers, their classification defined three Professionals, five Pro-Ams, three Amateurs, and two Average Career Growers. The authors didn’t feature all the interviews but examined two case studies as descriptive of the classifications of professional and average career grower. In defining the categories, the authors note that the professionals had been involved in growing for 7 years or more and owned or managed operations larger than 100 plants. However, also linked to professionals was a social opportunity structure. For their professional case study of a grower with 12 years’ experience, this entailed having early or intimate introduction to the social milieu through family and friends, early growing experience, and social and technical mentorship by more experienced growers. An early introduction and mentorship allowed the development of growing skills. This commitment and skill level led to more trust, responsibility, and contacts, which led to involvement in larger growing operations, which allowed for involvement and gaining experience in different industry areas and tasks. As this professional became involved in areas of hash production, clipping at large sites, cuttings production, helping manage a large site (2,000 plants), and finally moving on to managing a large site, legitimate employment tapered off, replaced by illegal income.

Their other case study involved a Vietnamese average career grower of 5 years. Different then the professional, this grower entered the industry mainly as a way to make money and had no initial indoctrination into the milieu. She did not receive any mentoring and began to learn the skills only as she became employed in an existing grow operation. While aspiring to reach a professional level, this grower was unsuccessful as she lacked good growing skills with a lot of trial and error experiences and had difficulty in the management of a grow operation. She eventually left the industry after not achieving her desired level of income. In the context of the cannabis cultivation industry, these average career growers are attracted to the potential to make money but never realized that cultivation would be so much work or would require a learning curve.

The authors didn’t discuss Pro-Am growers in much detail but they note “A review of the past typologies of growers found a similarly high prevalence of growers who were sometimes highly skilled but never transformed those skills into a full-time occupation”. The Pro-Am classification accommodates these types of growers as “labeling them as either professionals or amateurs would be unsatisfactory.” The authors consider that “Pro-ams are interested in the intangible rewards from cultivation; they are passionate and knowledgeable about the plant, its appearance, its taste, and its physiological effects. They are not amateurs but connoisseurs; not professionals but aficionados. Even if cultivation can be time consuming for pro-ams, it is not a full-time occupation, and never a means of living. Many of these participants are accomplished growers but opt to grow smaller amounts for personal consumption rather than profit.

The authors conclude that this classification system provides a useful discrimination tool that can be applied not just to growers. The classification system could be used to explore and research the professionalism of all types of criminal activity and that professionalism in criminals careers can be defined both by the skills of the offender and the degree that the offender is committed to a particular offense or type of offending. This commitment can be gauged by the characteristics noted by the authors; early introduction to, and involvement with, the milieu, an active mentoring process, and increased labor and involvement in larger activities or operations.

Bouchard, M., & Nguyen, H. (2016). Professionals or amateurs? Revisiting the notion of professional crime in the context of cannabis cultivation. In World Wide Weed (pp. 129-146). Routledge.

Differentiating contract killers: A narrative‐based approach.

Yaneva, Ioannou, Hammond, & Synnott, The Howard Journal of Crime and Justice, 2018.

The authors desired to construct a psychological-based typology framework that will assist in the analysis and understanding of contract murders. Previous attempts at typologies have utilized professional, semi-pro and amateur categories, solicitor or contractor designations, or more behavior-based categories of aggressive behaviors, inept behaviors, and criminally sophisticated behaviors. However the authors consider a psychologically based framework will also serve as a n investigative tool in profiling contract killers through the crime scene indications of offender characteristics. They utilized the Narrative Action System that suggests that for the full range of human action, four different fundamental narrative action modes are utilized: Professional’s adaptive adventure, Revenger’s conservative tragedy, Victim’s integrative irony, and Hero’s expressive quest. These typologies represent behavioral styles, attitudes, outlooks, and behavior motivations. “In psychological terms, criminal and deviant behaviors form a distinct subset of this general set of human actions, distinguished from other behavior by the absence of legal or moral code.” The themes have been utilized on other studies of different types of criminal behavior and the authors attempted to see if it can be utilized with contract killers. The characteristics of these typologies are discussed below in the context of contract killer behaviors.

Using media accounts as a source for data, 75 cases from nine different countries were subjected to content analysis. A total of 83 individual variables were derived from the analysis, 65 behavioral variables (crime scene behaviors) which included things like location and types of wounds delivered, body position, how the victim was controlled, and how the victim was isolated. For example they found the most frequently engaged in crime scene behaviors were: weapon brought to crime scene (88%), victim was ambushed (84%), victim’s body left at crime scene (83%), murder happened as planned (83%), victim was found as fell (80%), victim was shot (68%), victim had head wounds (58%) and offender used a vehicle for transportation (53%). They incorporated 56 of these variables into multidimensional scaling to explore subsets of behavior. The remaining 18 variables (victim, instigator, and offender characteristics) like gender, age, instigator motive, victim-instigator relationship, instigator-offender relationship, and offender’s previous convictions were used for descriptive purposes.  For example, instigators were 51% female, financial gain was the instigator’s primary motive (43%), and only 44% of offenders were strangers to the instigators.

The authors used Smallest Space analysis which “is a non-metric multidimensional scaling procedure in which the relationships between variables are represented as distances in space and the resulting spatial configuration is examined to determine whether meaningful regions can be identified”. The most common features to all contract killings, as noted above, form the core of behavior typical to all the offenders while the other behavioral variables in the periphery reflect the different aspects of the contract killer. This core behavior demonstrates these murders are typically very instrumental and centered on monetary gain and not the victim or the murder itself.

The authors’ analysis of the data describes the characteristics of different contract killer typologies linking their psychological and behavioral components as shown below.

The Professional’s adaptive adventure-View themselves as professionals and display methodical and criminally sophisticated behavior. They view each job as an adventure, seek to master the environment, and take pride in a job well done. Offenders tend to be masked and forensically cognizant, and use firearms at close distance.

The Revenger’s conservative tragedy-Sets out to right a wrong or address a problem, has a selfish or ambivalent nature, and rationalizes the crime as necessary or the right thing to do. Their extreme violence is not expressive but instrumental, often to send a message from the instigator. The offender and victim may be acquainted and are often abducted or lured. This typology is usually found in organized crime.

The Victim’s integrative irony-Generally inexperienced, and dispassionate, they hold a pessimistic world view and lack a moral code. There is evidence of incompetence at the crime scene and they may commit the murder for drugs/drug money or some other support. This typology is associated with mistakes in the job, and kills in the home and while the victims were sleeping, typically slashing their throats. There may be expressive behaviors at the crime scene that aren’t readily apparent.

The Hero’s expressive quest-A mix of instrumental and expressive behaviors; characterized by excessive, extreme violence that demonstrates their power and demands respect but instrumental in their preparedness. They may depersonalize the victim and use available weapons at the scene that includes vehicles, stabbing, and bludgeoning.

The characteristics for each typology as derived from the analysis are shown in the authors’ figure below.

Figure from authors’ publication

The authors conclude that this typology can be used as a basis for examining the individual type of contract killers so as to examine the connection between the psychological construct and behaviors, as well as an investigative tool to link possibly similar offenses together, and as a basis for therapeutic treatment approaches.

Yaneva, M., Ioannou, M., Hammond, L., & Synnott, J. (2018). Differentiating contract killers: A narrative‐based approach. The Howard Journal of Crime and Justice, 57(1), 107-123.

Welcome to Criminal Justice Access

Criminal Justice Access Mission Statement

Catering to practitioners, scholars and the public, Criminal Justice Access (CJA) brings historical, original, and current criminal justice research, practitioner interviews, and crime data together in an easily accessible and user-friendly format. The field of criminal justice is broad so CJA is devoted toward focusing on issues in policing, Part One Crimes, drugs, gangs, and deviance. By aggregating and summarizing data and information from literature in the criminal justice field, CJA tries to simplify the process of keeping abreast of current criminal justice research and information. I will be publishing content monthly so check back at see what’s new.

As this is my first month of publishing, there are no archived posts, however be sure to check the site categories. For November:

Research Briefs covering a possible new role for detectives, clearance rate differences in gun homicides vs gun assaults, reluctance in talking to the police, and differences in attitudes towards stop and search

At Issue looks at marijuana driving impairment and roadside testing

For Discussion explores recognizing a beat management philosophy called beat integrity

US Crime Data focusing on seven Part One Crimes from the UCR

Original Research is featuring past academic research by the author with this month featuring my PhD dissertation, a qualitative study of patrol officer behavior and decision making

Editorials and Opinions examines a possible deviance continuum from motorcycle enthusiast to outlaw through the mechanism of differential association

Marijuana Driving Impairment and Roadside Testing

Introduction

            This literature review will examine marijuana’s effects on driving performance, determining impairment levels, and the use and effectiveness of Drug Recognition Experts (DREs) and roadside testing in making determinations of driving under the influence of marijuana. Marijuana use and acceptance is on the rise and this has contributed to more incidence of drugged driving. 34 states have legalized medical marijuana, and while some strains are produced with relatively higher CBD to THC ratios to optimize their medical use, some strains ostensibly sold for medical purposes may have a relatively high THC level that has the potential for misuse. Legalization at the state level has been expanding. Currently 11 states and Washington D.C. have legalized recreational marijuana, while in 15 states decriminalization has been an ongoing process. When combined with a change in public perception, this can lead to increased incidence of driving under the influence of marijuana. A 2001 study examining roadside testing found that of 209 positive drug tests, 113 indicated cannabis (Steinmeyer, et al). A ten-year Swedish study found measurable THC levels between 18 and 36 percent of suspected drugged drivers (Jones, et al. 2008) and a UK lab sample of over 3,600 suspected drugged drivers showed 58% were positive for cannabis (Wolff and Johnston, 2014).

What was once an underlying marijuana subculture decades ago, has now reached a new level of acceptance in overall society as research has dispelled some of the myths about marijuana use and its negative consequences, and revealed suggested medical uses and benefits. Marijuana awareness, and its use, are also becoming more normalized with depictions in popular media in the form of news reports and specials, programming focusing both specifically and tangentially on its use, movie and TV characters using marijuana in a normalized context similar to how drinking has been portrayed, and websites that describe and rate different strains of marijuana as well as provide growing and usage tips. This normalization may contribute to some users minimizing the risks of driving while high.

Research into Driving Effects

Marijuana users have believed, and research has shown, that frequent users who are experienced drivers are able to compensate to a certain degree for marijuana’s effects while driving (Robbe, 1998; Kelly, 2004; Sewell, et al, 2009; Wolff and Johnston, 2014; Micallef et al, 2018) and some users forward the argument that driving high is safer than driving drunk (Kelly, 2004) . However, marijuana can impair driving performance, but typically in a way that differs from the effect of alcohol on driving performance. In research, impairment identification testing while under the influence of THC typically measures car following (maintaining distance with a car under varying speeds) and the standard deviation in lateral position (SDLP), or lane tracking (amount of weaving off center within the lane), as well as some other driving related tasks. Studies typically provided THC in two ways, either in 100-300 microgram/kilogram of subject body weight, or in 10-30 milligram doses of THC, irrespective of subject body weight.

Many studies when examining driving performance find significant effects on car following and SDLP. Robbe (1998) found at higher doses of THC tracking and following were significantly negatively affected compared to a control group and Ramaekers et al’s (2000) research revealed that low doses only moderately impair drivers actual driving performance in those areas Micallef, et al (2018), also found THC had a significant effect on lane tracking. However, a 2001 Netherlands study didn’t find any significant difference in visual search frequency with a 100 microgram/Kg THC dose (Lamers, et al). In 2009, Sewell and colleagues reviewed marijuana driving impairment studies and found, beside marijuana effects on driving being dose dependent, that there were differences in the type of impairment between alcohol and marijuana. Marijuana’s effects are more pronounced with tasks requiring a higher automatic driving function as opposed to alcohol’s more pronounced effects on conscious, cognitive function. SLP is affected by marijuana and not subject to compensatory behavior the way other driving functions are, and other effects like poor speedometer monitoring, increased decision time in passing, and increased time in responding to a light change or sudden sound were also present. Alcohol was implicated more in pursuit tracking, divided attention, signal detection, hazard perception, and lowered concentration, reaction time, and hand-eye coordination.

Research has shown the driving impairment effects are dose, and frequency of past use, dependent, as well as demonstrating significant effects differences from placebo or control groups at the THC levels used for testing. But how marijuana contributes to car crashes also requires examination. In 2014, Romano and Pollini, examining 12 years of crash data, found 26% of single car fatal crashes involved a drug other than alcohol, with cannabinoids present in approximately a quarter of those cases. However, previous work has not provided a clear indication of marijuana’s influence. Kelly (2004) in a review of driving studies found inconsistent associations of THC with traffic accident involvement. A 2005 multivariate study found that once accounting for things like speed, seatbelt use, and BAC, marijuana use was no longer a significant factor in car crash injury. However, it was a significant factor in those that identified as habitual marijuana users (Blows, et al). In 2014, Wolff and Johnston reviewing a meta-analysis showed that drivers were twice as likely to be involved in a car accident with a THC blood plasma level of 5 micrograms/liter. Huestis (2015) in her review of two recent meta analyses found a significantly increased crash risk (2-2.6 times more likely) with a measurable amount of THC and a six-fold increased risk at a 5 microgram/liter level. However, in 2016 Rogeberg and Elvik found that marijuana involvement in crash risk is low to moderate, similar to a BAC level of .04-.05.

While the possibility of increased risk of car crashes may yet to be determined because of dose dependent effects and other factors, research has been consistent on the effects of combining a small amount of alcohol with marijuana. BACs of .04-.05 when combined with even low levels of THC significantly increased the severity of driving impairment, resulting in impairment equal to a BAC of .08-1.9 (Robbe, 1998; Ramaekers, et al, 2000; Blows et al, 2005; Sewell, 2009; Huestis, 2015).

Determining and Setting Intoxication and Impairment Standards and Limits

            As a practical application, identifying and utilizing THC levels in determining impairment for drivers in the real world is important for state legislatures and municipalities in legalization states, as well as for those adjoining states, and other states considering legalization, as the country continues to progress toward greater medical and recreational legalization. Standards are typically expressed as THC parts per volume of oral fluid or serum/plasma concentrations, either as nanograms/milliliter or as micrograms/liter (note that these are equivalent measures-1 nanogram/milliliter = 1 microgram/liter). Setting driving limits typically take two approaches; a per se approach, where any amount of THC above a preset limit indicates a violation of the law, similar to the .08 BAC limit currently used in the U.S. but could also resemble a zero-tolerance level where any detectable THC amount is a violation of law. The other approach is to determine a THC fluid concentration that indicates impairment, often based on driving performance deficits comparable to a specified BAC level.

The approaches vary around the world, and approaches and levels can change or be set by a combination of research, politics, and public opinion. In 2007, a toxicology lab survey in 24 states found cutoff and confirmation levels of THC ranged from 2 to 70 ng/ml. While recommending that a 2 ng/ml confirmation level be used in drugged driving investigations, it is a level based more on good analytic methodology and lab practice than in an indication of impairment (Farrell, et al, 2007). While European workplace testing typically has a cutoff (the drug screening point where a concentration above the cutoff indicates a positive test for the substance, and below that point is a negative result) of 10 ng/ml (Moore, 2012) European driving limits varied widely; Sweden (a zero-tolerance country) sets a cutoff at .3 micrograms/liter, Germany-.5, France-1.0, Norway-1.3(declaring it an impairment limit), Switzerland-1.5 (declaring it a prosecution limit), and Portugual-3. A Swiss study of drug impaired drivers found that a 5 microgram/liter fluid level correlated to observable driving impairment (Wolff and Johnston, 2014). However, a German study from 2007 suggested a range of 7-10 micrograms/liter as a non-zero per se limit after finding that it correlated to an impairment comparable to a .05 BAC and that concentrations below 10 micrograms/liter were not associated with increased accident risk (Grotenherman, et al). Huestis (2015), in reflecting on an appropriate measure for the U.S. suggests a two-tiered approach, similar to what is used in some European countries and Australia. that when combined with driver education, has been shown to deter drugged driving. It provides for a small fine and driver education for exceeding a per se limit and provides for a more severe penalty and criminal prosecution for demonstrated impairment.      

Testing for Marijuana Intoxicated Drivers

            Regardless of where the cutoff points are established, the concern for law enforcement is utilizing the tools at their disposal to identify drivers who are intoxicated or impaired. Drug Recognition Experts (DREs) and roadside testing can be utilized both on active patrol and at driving checkpoints. DRE use is expanding and the options for roadside testing include oral fluid (saliva) testing with point of collection testing (POCT) that can provide results in minutes. Time can be an important factor as THC concentration levels peak approximately 15 minutes after smoking but then go on a rapid decline over the first hour after smoking, and drops even further between one and three hours after smoking. However, while THC levels drop off quickly, the subjective high of the user declines at a much slower rate, meaning people who have a broad range of THC levels may exhibit the same level of being high, and thus impairment.

DRE Effectiveness

            According to Brown, (2001), the DRE program was first started as a partnership between the LAPD and National Highway Traffic Safety Administration (NHTSA) in the 80’s and the program became more popular in the 90’s, resulting in the majority of states utilizing DREs in law enforcement agencies. The International Association of Chiefs of Police (IACP) is the national certifying agency for DREs and instructors.

There are three phases of training and DREs must re-certify based on the past two years of their evaluations. Typically, DREs are called in during a DUI traffic stop when the suspected driver appears intoxicated from something other than alcohol. The DRE evaluation is a 12-step process that involves a BAC test, pulse checks, eye examinations, temperature, muscle rigidity, and ingestion exams, interviews with the suspect and arresting officer, a standardized field sobriety test (SFST), the DRE’s opinion, and a toxicology test. Some studies have demonstrated moderate to high effectiveness in determining intoxication and drug classification. A 1995 Phoenix PD study compared DRE opinions to the toxicology reports and found DREs identified at least one of the drugs in 91% of the samples confirmed as containing drugs and overall the DREs impaired decisions were correct 83.5 %. Marijuana was the drug most often missed though, and the study notes it’s unknown whether this was due to DRE accuracy, or the time course of the drug (Burns and Adler, 1995). A 2002 study found that when DREs reports were stripped of all information except DRE observations, the DREs tested were 95% accurate in determining intoxication and 81% accurate in identifying the cannabis category. Beirness, et al (2009) examined over 1,300 Canadian DRE evaluations and found the DREs were 95% accurate in determining true positives, 80% accurate in determining true negatives, and 92% accurate on drug classification, compared to lab results.

However, in 2009 Porath-Waller, Beirness, and Beasley were examining whether a simpler approach to the DRE evaluation could be utilized, focusing on the most relevant of the twelve physiological testing criteria for the three most common classifications; stimulants, cannabis, and narcotics. They found nine of the criteria were more predictive of drug category including pulse rate, condition of the eyes and eyelids, lack of convergence, hippus, reaction to light, rebound dilation, systolic blood pressure, and the presence of injection sites with an overall accuracy of 81% across the three drug classifications, however only 72% accuracy in identifying cannabis.

Other research has focused on the SFST. In 2014, Porath-Waller and Beirness examined three components of the SFST in determining THC impairment; the Horizontal Gaze Nystagmus (HGN), the One Leg Stand (OLS), and the Walk And Turn (WAT). Their results showed only the OLS was a significant predictor of THC. In a two-year study, Declues et al (2016) found that while SFTSs are sensitive to marijuana impairment that there was no correlation between SFTS performance and an average 9 ng/ml THC level, however the number of SFTS cues were greater the higher the THC level. Their study found that the WAT was a better indicator of marijuana intoxication that the OLS (contrary to DRE evaluation studies) and the Lack Of Convergence (circular tracking with eye cross) was also a strong indicator. Hartman et al (2016) arrived at somewhat different conclusions. In their study they found no difference in SFST performance between those above and below 5 ng/ml. They found the Finger To Nose test was the best predictor of cannabis intoxication, along with MRB eyelid tremors, OLS sway, pupil rebound dilation and two or more WAT cues.

Brown, in 2001, did raise an issue related to the accuracy of DRE’s and whether they can provide valid, reliable, expert testimony. He argues because DREs are labeled as experts and are construed by juries to be giving expert testimony then the basis of their evaluations should be subjected to Daubert or Frye tests of evidence admissibility. Brown considers the program would or should not pass these tests as they have not faced peer reviewed evaluations, or general acceptance in the relevant scientific community. However, some court rulings have concluded that the twelve-step evaluation is not scientifically novel enough to warrant putting it to the test of admissibility. Brown believes that with reviews of different DRE programs indicating correct hit rates of between 47 and 86%, and with only a 75% correct hit rate needed for re-certification, that DRE evaluations should not be considered reliable scientific, technical, or specialized knowledge and should continue to face challenges of admissibility in court.

ROADSIDE TESTING

THC can be found in high levels in the saliva immediately after smoking, and similar to blood serum levels, declines quickly in the first hour and steadily downward over a few hours (Cone, 1993). Roadside testing has focused on oral fluid testing as opposed to urine or blood for a number of reason including, it is less intrusive, a smaller amount of fluid is needed for a sample, it’s easier to administer, can identify parent drugs, not just metabolites, has a close correlation to serum drug concentrations, and provides quicker results (Cone, 1993; Verstraete, 2005; Bosker, 2009; Huestis, 2011). A 2001 German study of roadside testing found they were 97.6% accurate compared to lab results, with no false negatives and only 2.4% false positives. In a 2009 Australian study of random roadside testing, cannabis was the third most common drug, found in 26% of drugged drivers. Their results showed that oral fluid testing correctly identified targeted drugs 98% of the time.

But there are some caveats to oral fluid testing. Delays in testing may yield negative results if cutoffs are set too low, (Cone and Huestis, 2007) and there is a lack of conclusive evidence tying THC levels to the subjective high, (Cone, 1993) and thus, possible impairment. Verstraete (2005) considers that there should be a better correlation between drug presence and impairment, better discrimination for cannabis, better proficiency in testing, and safeguards against passive contamination before oral fluid testing will be forensically reliable. Earlier devices suffered from poor accuracy and current devices still lacked adequate reliability (Boker et al, 2009) but Huestis (2011) predicts that the move to oral fluid testing will expand as better tests become available.

Evaluations of more recent devices provided varying results. A 2007 study of the RapiScan tester demonstrated a 96% accuracy for positive samples and 100% accuracy for negative samples but was more sensitive to stimulants than cannabis (Dewey et al, 2007). Couch and associates’ 2008 study of ten different tests including RapiScan, DrugWipe, and OraLab showed nine of the ten devices were accurate in identifying THC but all had a much higher cutoff than the 4 ng/ml suggested by SAMSHA. A European Union 2011 study evaluated eight testers on sensitivity (ability to detect a positive result), specificity (distinguishing between negative and positive samples) and accuracy (the rate of true results compared to false results). Cutoff levels for the devices varied between 5 to 100 ng/ml. Specificity ranged between 90 and 100% but sensitivity was poor, ranging from 20-60%, resulting in accuracy between 78-88% for the five best performing devices. None of these multidrug testers had over 80% in each of the three measurement categories. A 2012 Italian study of four different devices found the sensitivity ranged from 38-92% accurate compared to lab results. In 2019, a Brazilian study analyzed four devices for ease of use, operational success, and acceptable analysis and collection time, using feedback from patrol officers. Three of the four devices were rated in the 80th percentile in these categories. To complicate things, some drugged drivers may be trying to confound the test by using different fluids or mouthwashes to dilute a possible sample. de Castro (2014) found that a commercial product designed to dilute for the test was ineffective, but a plain water mouthwash at 3 hours after use was effective in dropping THC to below 25 ng/ml. However, developments continue to be made in oral fluid testing. Plouffe et al (2017) describe a new roadside testing technique they were developing, utilizing existing fluorescent scanners, that’s highly reliable, can utilize a cutoff of 1 ng/ml or lower, and from sample insertion to readout, take less than 10 minutes to complete.

IMPLICATIONS

            As improvements are made in oral fluid roadside testing, the role of a DRE may be scaled back to assisting in drug investigations or in cases involving an indeterminate drug, as may be the case as synthetic designer drugs continue to be developed and find their way into the recreational market. But in order to make oral fluid testing a viable means of deterring, detecting and stopping drugged driving, yet distinguishing from residual THC in the system that has no effect on driving performance, reasonable, science-based decisions will need to be made regarding per se or impairment levels, and the proper testing device for the intended purpose, such as the suitability of using the device for a DUID checkpoint. Jurisdictions should also be prepared for legal challenges to the admissibility of testimony from DREs as well as challenges to the reliability of some testing devices. Despite some drawbacks still present in oral fluid testing devices and the validity in determining impairment by THC level, it is the direction that roadside testing is moving. As such, jurisdictions and legislatures will need to stay current on the best testing equipment, and best practices and research, in this continually emerging field.

REFERENCES

Beirness, D. J., Beasley, E., & Lecavalier, J. (2009). The accuracy of evaluations by Drug Recognition Experts in Canada. Canadian Society of Forensic Science Journal, 42(1), 75-79.

Blencowe, T., Pehrsson, A., Lillsunde, P., Vimpari, K., Houwing, S., Smink, B., … & Verstraete, A. (2011). An analytical evaluation of eight on-site oral fluid drug screening devices using laboratory confirmation results from oral fluid. Forensic Science International, 208(1-3), 173-179.

Blows, S., Ivers, R. Q., Connor, J., Ameratunga, S., Woodward, M., & Norton, R. (2005). Marijuana use and car crash injury. Addiction, 100(5), 605-611.

Boorman, M., & Owens, K. (2009). The Victorian legislative framework for the random testing drivers at the roadside for the presence of illicit drugs: an evaluation of the characteristics of drivers detected from 2004 to 2006. Traffic injury prevention, 10(1), 16-22.

Bosker, W. M., & Huestis, M. A. (2009). Oral fluid testing for drugs of abuse. Clinical chemistry, 55(11), 1910-1931.

Brown, S. (2001). The d.r.e.: Drug recognition expert or experiment. UMKC Law Review, 69(3), 557-586.

Burns, M., & Adler, E. V. (1995). Study of a drug recognition expert (DRE) program. In Proceedings of the 13th International Conference on Alcohol, Drugs and Traffic Safety (Vol. 1, pp. 437-441).

Cone, E. J. (1993). Saliva testing for drugs of abuse. Annals of the New York Academy of Sciences, 694(1), 91-127.

Cone, E. J., & Huestis, M. A. (2007). Interpretation of oral fluid tests for drugs of abuse. Annals of the New York Academy of Sciences, 1098, 51.

Crouch, D. J., Walsh, J. M., Cangianelli, L., & Quintela, O. (2008, April). Laboratory evaluation and field application of roadside oral fluid collectors and drug testing devices. In Therapeutic drug monitoring (Vol. 30, No. 2, pp. 188-195). LWW.

Davey, J., Leal, N., Freeman, J. (2007). Screening for drugs in oral fluid: illicit drug use and drug driving in a sample of Queensland motorists. Drug and alcohol review, 26(3), 301-307.

de Castro, A., Lendoiro, E., Fernández‐Vega, H., López‐Rivadulla, M., Steinmeyer, S., & Cruz, A. (2014). Assessment of different mouthwashes on cannabis oral fluid concentrations. Drug testing and analysis, 6(10), 1011-1019.

Declues, K., Perez, S., & Figueroa, A. (2016). A 2‐Year Study of Δ 9‐tetrahydrocannabinol Concentrations in Drivers: Examining Driving and Field Sobriety Test Performance. Journal of forensic sciences, 61(6), 1664-1670.

Farrell, L. J., Kerrigan, S., & Logan, B. K. (2007). Recommendations for toxicological investigation of drug impaired driving. Journal of forensic sciences, 52(5), 1214-1218.

Grotenhermen, F., Leson, G., Berghaus, G., Drummer, O. H., Krüger, H. P., Longo, M., … & Tunbridge, R. (2007). Developing limits for driving under cannabis. Addiction, 102(12), 1910-1917.

Hartman, R. L., Richman, J. E., Hayes, C. E., & Huestis, M. A. (2016). Drug Recognition Expert (DRE) examination characteristics of cannabis impairment. Accident Analysis & Prevention, 92, 219-229.

Huestis, M. A., Verstraete, A., Kwong, T. C., Morland, J., Vincent, M. J., & de la Torre, R. (2011). Oral fluid testing: promises and pitfalls. Clinical chemistry, 57(6), 805-810.

Huestis, M. A. (2015). Deterring driving under the influence of cannabis. Addiction, 110(11), 1697-1698.

Jones, A. W., Holmgren, A., & Kugelberg, F. C. (2008). Driving under the influence of cannabis: a 10‐year study of age and gender differences in the concentrations of tetrahydrocannabinol in blood. Addiction, 103(3), 452-461.

Kelly, E., Darke, S., & Ross, J. (2004). A review of drug use and driving: epidemiology, impairment, risk factors and risk perceptions. Drug and alcohol review, 23(3), 319-344.

Lamers, C. T., & Ramaekers, J. G. (2001). Visual search and urban driving under the influence of marijuana and alcohol. Human Psychopharmacology: Clinical and Experimental, 16(5), 393-401.

Moore, C. (2012). Oral fluid for workplace drug testing: laboratory implementation. Drug testing and analysis, 4(2), 89-93.

Micallef, J., Dupouey, J., Jouve, E., Truillet, R., Lacarelle, B., Taillard, J., … & Philip, P. (2018). Cannabis smoking impairs driving performance on the simulator and real driving: a randomized, double‐blind, placebo‐controlled, crossover trial. Fundamental & clinical pharmacology, 32(5), 558-570.

Pechansky, F., Scherer, J. N., Schuch, J. B., Roglio, V., Telles, Y. M., Silvestrin, R., … & Sousa, T. (2019). User experience and operational feasibility of four point-of-collection oral fluid drug-testing devices according to Brazilian traffic agents. Traffic injury prevention, 20(1), 30-36.

Plouffe, B. D., & Murthy, S. K. (2017). Fluorescence‐based lateral flow assays for rapid oral fluid roadside detection of cannabis use. Electrophoresis, 38(3-4), 501-506.

Porath-Waller, A. J., Beirness, D. J., & Beasley, E. E. (2009). Toward a more parsimonious approach to drug recognition expert evaluations. Traffic injury prevention, 10(6), 513-518.

Porath-Waller, A. J., & Beirness, D. J. (2014). An examination of the validity of the standardized field sobriety test in detecting drug impairment using data from the drug evaluation and classification program. Traffic injury prevention, 15(2), 125-131.

Ramaekers, J. G., Robbe, H. W. J., & O’Hanlon, J. F. (2000). Marijuana, alcohol and actual driving performance. Human Psychopharmacology: Clinical and Experimental, 15(7), 551-558.

Robbe, H. (1998). Marijuana’s impairing effects on driving are moderate when taken alone but severe when combined with alcohol. Human psychopharmacology: clinical and experimental, 13(S2), S70-S78.

Rogeberg, O., & Elvik, R. (2016). The effects of cannabis intoxication on motor vehicle collision revisited and revised. Addiction, 111(8), 1348-1359.

Romano, E., & Pollini, R. A. (2013). Patterns of drug use in fatal crashes. Addiction, 108(8), 1428-1438

Sewell, R. A., Poling, J., & Sofuoglu, M. (2009). The effect of cannabis compared with alcohol on driving. American Journal on Addictions, 18(3), 185-193.

Smith, J. A., Hayes, C. E., Yolton, R. L., Rutledge, D. A., & Citek, K. (2002). Drug recognition expert evaluations made using limited data. Forensic Science International, 130(2-3), 167-173.

Steinmeyer, S., Ohr, H., Maurer, H. J., & Moeller, M. R. (2001). Practical aspects of roadside tests for administrative traffic offences in Germany. Forensic science international, 121(1-2), 33-36.

Strano-Rossi, S., Castrignanò, E., Anzillotti, L., Serpelloni, G., Mollica, R., Tagliaro, F., … & Chiarotti, M. (2012). Evaluation of four oral fluid devices (DDS®, Drugtest 5000®, Drugwipe 5+® and RapidSTAT®) for on-site monitoring drugged driving in comparison with UHPLC–MS/MS analysis. Forensic science international, 221(1-3), 70-76.

Verstraete, A. G. (2005). Oral fluid testing for driving under the influence of drugs: history, recent progress and remaining challenges. Forensic Science International, 150(2-3), 143-150.

Wolff, K., & Johnston, A. (2014). Cannabis use: a perspective in relation to the proposed UK drug‐driving legislation. Drug testing and analysis, 6(1-2), 143-154.