- Email: [email protected]
Randomized trial of an official contact letter intended to increase proper licensure among unlicensed motorcycle owners
Transportation Research Part F 23 (2014) 165–172
Contents lists available at ScienceDirect
Transportation Research Part F journal homepage: www.elsevier.com/locate/trf
Randomized trial of an official contact letter intended to increase proper licensure among unlicensed motorcycle owners Kevin J. Limrick ⇑, Scott V. Masten 1 California Department of Motor Vehicles, Research and Development Branch, 2570 24th Street, MS H-126, Sacramento, CA 95818-2606, USA
a r t i c l e
i n f o
Article history: Received 14 June 2013 Received in revised form 27 January 2014 Accepted 1 February 2014
Keywords: Motorcycle Motorcyclists Unlicensed Licensure Randomized trial Educational intervention
a b s t r a c t Background: Approximately one-third of all fatality-injured motorcycle operators in California from 1999 through 2008 were not properly licensed to ride a motorcycle at the time of the crash. In availing themselves of a licensing system, motorcyclists are afforded the opportunity to acquire and demonstrate the minimum knowledge and skills necessary to operate motorcycles on public roadways, and licensing allows agencies to monitor their riding performance and apply appropriate sanctions, with the overall goal being enhanced safety. To that end, a randomized trial was conducted of an intervention recommended by the National Highway Traffic Safety Administration to increase proper motorcycle licensure. Methods: Official contact letters providing information about obtaining a motorcycle license, and the potential legal consequences of riding without one, were mailed to owners of currently-registered California motorcycles who were not licensed to ride motorcycles. Half of the 65,766 improperly licensed California motorcycle owners identified were randomly assigned to be mailed the contact letter while the remainder served as a no-letter control group. The groups were compared on 33-month post-treatment motorcycle licensure rates and 18-month post-treatment motorcycle crashes and traffic violations. Results: While a modestly higher percentage of motorcycle owners who were mailed the contact letter became properly licensed (14.5%) than those who were not mailed the letter (10.4%), the majority (85.5%) remained improperly licensed to operate a motorcycle 33 months later. The effect of the contact letter was influenced by owner age, whereby the odds of licensure for those to whom the letter was sent increased significantly for owners aged 20–34 (21%), 35–54 (47%), and 55 and older (96%). No 18-month post-treatment differences in motorcycle crash involvements or traffic violations were found. Conclusion: Given the traffic-safety neutral outcomes and relatively low cost of treatment, official contact letters should be considered in other jurisdictions to increase proper motorcycle licensure as recommended by the National Highway Traffic Safety Administration. Future letters may be more effective if they are tailored to the demographic characteristics (e.g., age and sex) of the owners. Ó 2014 Elsevier Ltd. All rights reserved.
⇑ Corresponding author. Tel.: +1 916 657 9791; fax: +1 916 657 8589. 1
E-mail addresses: [email protected] (K.J. Limrick), [email protected] (S.V. Masten). Tel.: +1 916 657 6638; fax: +1 916 657 8589.
http://dx.doi.org/10.1016/j.trf.2014.02.001 1369-8478/Ó 2014 Elsevier Ltd. All rights reserved.
166
K.J. Limrick, S.V. Masten / Transportation Research Part F 23 (2014) 165–172
1. Introduction Operating a two-wheeled motorcycle without a valid license is associated with an increased likelihood of being involved in a crash, a finding that has been demonstrated in Australia, New Zealand, Taiwan, Thailand, the United States, and other jurisdictions (Harrison, 1997; Hurt, Quellet, & Thom, 1981; Kraus et al., 1991; Lardelli-Claret et al., 2005; Lin & Kraus, 2009; McGwin et al., 2004; Reeder, Chalmers, & Langley, 1995; Sirathranont & Kasantikul, 2003; Watson & Steinhardt, 2006; Yeh, Chang, & Chang, 2008). Absolute estimates of the prevalence of improperly licensed motorcycle riding in the U.S. and other jurisdictions are difficult to establish; instead, crash involvements among such riders are typically used to infer the relative prevalence of unlicensed riding over time (Kraus et al., 1991; Reeder et al., 1995; Watson & Steinhardt, 2006; Yeh et al., 2008). In the United States, the number of motorcycle riders involved in fatal crashes increased 114% from 1999 through 2008 (National Highway Traffic Safety Administration [NHTSA], 2011). Consistent with this national trend, the number of motorcycle riders in California involved in fatal crashes during this time period increased 137%. Approximately 27% of motorcycle riders nationwide and 34% of those in California who were involved in fatal crashes during this decade were not properly licensed to ride motorcycles at the time of the crash (NHTSA, 2011). For comparison purposes, among passenger vehicle drivers involved in fatal crashes during this time period, only 14% nationwide and 20% in California were improperly licensed when they crashed. Overall, the number of improperly licensed motorcyclists involved in fatal crashes increased 89% nationwide and 146% in California between 1999 and 2008. In response to these trends, NHTSA entered into a cooperative agreement with the American Association of Motor Vehicle Administrators to develop and publish Guidelines for Motorcycle Operator Licensing, which contains recommended motorcycle safety interventions and strategies to increase proper licensure among motorcyclists (NHTSA, 2009). One such intervention recommended in these guidelines is for motor vehicle agencies to mail a contact letter to improperly licensed motorcycle owners encouraging them to complete the requisite training and obtain a motorcycle operator license (NHTSA, 2009). The rationale behind this recommendation is that in availing themselves of a licensing system, motorcyclists are afforded the opportunity to acquire and demonstrate the minimum knowledge and skills necessary to operate motorcycles on public roadways, and licensing allows agencies to monitor their riding performance and apply appropriate sanctions, with the overall goal being enhanced safety. This strategy also has the potential to increase proper licensure among motorcycle owners and perhaps reduce the prevalence of unlicensed riding for jurisdictions outside of the United States. Approximately 13% of motorcycle owners in California are not properly licensed to operate their vehicle on public roadways. Riding without the proper license or permit is a violation of California Vehicle Code §12500, and those cited for this violation risk impoundment of their motorcycle as well incurring a fine exceeding $1000. Similar to the procedure used in most U.S. states, individuals who wish to become properly licensed to ride a motorcycle on public roadways in California must submit a license application, pay a fee, and pass a visual acuity screening, a motorcycle operator knowledge test of rules of the road and safe driving practices, and a motorcycle skills test. The intent of this motorcycle rider licensing program is to increase rider competency by (1) motivating applications to acquire at least the minimum levels of knowledge and skill necessary to pass the required tests, and (2) screening out those who are unqualified to safely and knowledgably ride a motorcycle on public roadways as demonstrated by their failure of the required vision, knowledge, and/or skills tests. Motorcycle licenses are issued by the California Department of Motor Vehicles (DMV), are usually valid for 5 years, and typically can be renewed by mail in 5-year increments by paying a fee to DMV. The recommended use of contact letters to increase licensure among improperly licensed motorcycle owners was evaluated in a randomized trial in Maryland, a small state on the east coast of the U.S. (Braver et al., 2007). The results of the study indicated that 10.4% of those who were mailed the contact letter obtained a motorcycle license within 6 months after the mailing compared to 7.9% of those in the control group, and that the majority of owners remained improperly licensed to operate a motorcycle after 6 months. The short 6-month follow-up period used in the Braver et al. study may have underestimated the true impact of the contact letter intervention on licensing rates, particularly given that motorcycle learner permits in the state were valid for 6 months. In addition, the potential impact of the contact letter on subsequent motorcycle crashes or traffic violations was not evaluated as part of the study. The current study replicates the randomized trial of Braver et al. (2007) by assessing the effect of using contact letters to increase proper motorcycle licensure among motorcycle owners, but builds upon their work by comparing the licensure rates of the groups after a longer follow-up time (33 months vs. 6 months). It is conceivable that bringing more owners into legal riding status may lead to an increase in motorcycle crash or traffic violations incidents by increasing operator exposure to motorcycle travel, unless any skills gained through the process of licensure counterbalanced the risk from greater exposure. To evaluate this possibility, the present investigation expands upon the Braver et al. study by also comparing the motorcycle crash and traffic violation incidents of the groups in addition to licensure rates. 2. Methods 2.1. Sample identification and assignment Over 800,000 records for currently registered motorcycles were identified from the DMV registration database. Records were discarded if the motorcycles had more than one registered owner, were owned by military personnel or businesses
K.J. Limrick, S.V. Masten / Transportation Research Part F 23 (2014) 165–172
167
(e.g., sales dealerships or rental companies), or if they did not have a California driver license number recorded for the owner. Mopeds or motorized bicycles were also excluded because such vehicles do not have annual registration renewal cycles, and hence often have outdated owner information. Finally, only one motorcycle was retained per owner to remove duplicate ownership and create a sample of unique owners. The driver license numbers for the registered owners were then matched to a file obtained from the DMV licensing database containing records for all individuals with a valid motorcycle license, endorsement, or learner permit. This matching process was used to identify owners of registered motorcycles who did not have a valid motorcycle license, endorsement, or learner permit. After removing those who were deceased, a final sample of 65,766 improperly licensed motorcycle owners remained for use in the study (Fig. 1). These individuals were randomly assigned to either be sent the official contact letter (treatment group; n = 33,068) or not (control group; n = 32,698) based on the terminal digit of their driver license numbers. 2.2. Creation and mailing of contact letters An official DMV contact letter was created based on the NHTSA (2009) recommendations. This correspondence notified owners that the DMV was aware that they were not properly licensed to ride their motorcycle on public roadways, encouraged them to become properly licensed, warned of the legal and financial consequences of being cited for riding without a motorcycle license (i.e., impoundment of their motorcycle and a fine exceeding $1000), and provided information on motorcycle training courses and the procedures for obtaining a motorcycle license or permit. The contact letters were mailed over a 4-day period starting on November 10th, 2009. Some individuals (0.2%; n = 57) responded to the letter by mail, most claiming that their actions did not warrant receiving the letter. Several of these individuals said they no longer owned the motorcycle(s) in question, only rode three-wheeled motorcycles, or owned but did not ride motorcycles, and thus were exempt from motorcycle licensing requirements. Although some letters were returned as undeliverable (1.4%; n = 476), and others were erroneously mailed to registered owners who were not required to have a motorcycle license, the integrity of the study groups, as randomly assigned, was maintained for analysis purposes. This is sometimes called ‘‘intent-to-treat’’ analysis and is frequently used to avoid introducing selection bias (Hollis & Campbell, 1999). 2.3. Outcome data sources Data on the study participant’s demographics, motorcycle licensing status, crash involvements, traffic violations, and other variables used in the analyses were obtained from DMV driver records. Traffic violations included incidents for which the driver was convicted, failed to appear in court, or had a moving violation dismissed as a result of attending a traffic violator school. Crash involvements and traffic violations were counted for the 2-year period before the letters were mailed and separately for the 18-month period afterwards. The starting letter mail date (November 10th, 2009) was used as the reference date for counting crashes and violations within these periods. Crash involvements and traffic violations were divided into those that were motorcycle-related (i.e., in which the owner was riding the motorcycle at the time of the incident) and
All registered CA motorcycles (N = 847,937)
All licensed/permitted CA motorcyclists (N = 1,722,066) Removed: Mopeds (n = 102,210) Owned by business/trust (n = 7,164) Multiple owners (n = 115,838) Missing owner driver license (n = 32,387) Registered to out-of-state address (n = 2,417) Owned by military personnel (n = 384) Duplicate owners (n = 84,647)
Registered CA motorcycles with one in-state owner (n = 502,890)
Merge based on driver license number
Improperly licensed motorcycle owners (n = 65,766)
Treatment (letter) (n = 33,068)
Control (no letter) (n = 32,698)
Fig. 1. Formation of the study sample.
168
K.J. Limrick, S.V. Masten / Transportation Research Part F 23 (2014) 165–172
those that were not motorcycle-related incidents. Total traffic violations (motorcycle-related or not) were categorized based upon whether the infraction was a major (2-point) or minor (1-point) violation. The driver record data were extracted from the DMV licensing database on August 10th, 2012, which provided 14½ months following the end of the 18-month criterion period (May 10th, 2011) for crash and violation incidents occurring during the criterion period to be reported to DMV and added to driver records, and hence be included in the analyses. This ‘‘lag period’’ was intended to provide sufficient time for crashes occurring during the 18-month post-treatment period to be reported to DMV and recorded in the database. Historically, a 3-month lag has been more than ample for this purpose, but as of October 2011, the California Highway Patrol was over 13 months behind in reporting crashes to DMV. The August 10, 2012 driver record extract date also enabled the licensing status of each owner to be determined 33 months after the mailing of the letter on November 10th, 2009. 3. Results 3.1. Descriptive statistics for letter and no letter groups Table 1 presents descriptive statistics for various demographic and pre- and post-treatment driver record variables for participants in the treatment (letter) and control (no letter) study groups. The pre-treatment driver record variables represent incidents that occurred during the 2 years preceding the mail date of the contact letter, whereas the post-treatment driver record variables cover the 33-month licensing and 18-month crash and conviction periods subsequent to the mailing date of the letter. The groups appear to be similar on the demographic and pre-treatment driver record variables, as would be expected if the random assignment was successful. Also shown is that the percentage of participants who received the official DMV contact letter and who subsequently obtained a motorcycle license or learner permit was 14.46% compared to 10.40% among those who did not receive the contact letter. 3.2. Assessment of study group equivalency at random assignment To evaluate whether the method used to assign participants into the treatment (letter) and control (no letter) study groups produced equivalent pre-treatment groups with respect to age, sex, number of non-motorcycle endorsements/certificates, and prior driver record, these variables were entered into a logistic regression analysis predicting study group assignment (treatment vs. control). The results are presented in Table 2. The fit of the model predicting study group assignment was non-significant (p = .9434), indicating that the predictors, as a set, did not reliably distinguish between drivers in the two study groups. In addition, none of the individual predictors were
Table 1 Description of the treatment (letter) and control (no letter) study groups. Variable
Study group Treatment (letter) n = 33,068
Control (no letter) n = 32,698
Demographics Age (mean) Male (%) Non-motorcycle certificates/endorsements (per 100)
45.77 72.76 14.23
45.85 73.05 14.15
2-year pre-treatment driver record (per 100) Total crash involvements Total violations Two-point violations One-point violations Motorcycle crash involvements Motorcycle violations Suspended or revoked license
10.65 49.99 1.95 34.36 0.81 2.08 8.32
10.79 49.32 1.98 33.90 0.81 2.01 8.15
18-month post-treatment driver record (per 100) Total crash involvements Total violations Two-point violations One-point violations Non-motorcycle crash involvements Motorcycle crash involvements Motorcycle violations
6.85 33.74 1.33 21.83 6.29 0.56 1.44
6.66 33.32 1.26 21.55 6.03 0.64 1.43
33-month post-treatment motorcycle licensure Motorcycle license/permit (%)
14.46
10.40
169
K.J. Limrick, S.V. Masten / Transportation Research Part F 23 (2014) 165–172
related to group assignment (ps > .05). These results indicate that the study group assignment was indeed random and resulted in equivalent pre-treatment groups with respect to demographic, licensure, and driver history variables. 3.3. Effect of the contact letter on subsequent motorcycle licensure Results of a chi-square test of independence indicated that 33-month subsequent-to-mailing motorcycle licensure differed between the study groups, with participants who received the official DMV contact letter being 39% more likely to be properly licensed to operate a motorcycle (14.46%) than those who did not receive the letter (10.40%), v2(1, N = 65,766) = 249.89, p < .0001. However, the majority (85.5%) of motorcycle owners who received the letter still remained improperly licensed to operate a motorcycle 33 months later. To further evaluate the effect of the letter on motorcycle licensure, age, sex, pre-treatment driver record variables, and study group were used as predictors in a logistic regression model predicting 33-month subsequent-to-mailing motorcycle licensure. The intent of this analysis was to determine whether the letter was still significantly associated with subsequent motorcycle licensure after adjusting for any nominal differences between the study groups on potential confounders. The logistic regression results presented in Table 3 indicate that the model significantly predicted subsequent motorcycle licensure (p < .0001). Age, sex, prior one-point traffic violations, and motorcycle-related traffic violations were all statistically significant predictors of subsequent motorcycle licensure (ps < .05). Importantly, study group (p = .0445) and the study group age (continuous) interaction (p < .0001) also significantly predicted motorcycle licensure, indicating that the effect of the contact letter on subsequent motorcycle licensure varied according to driver age. To explore this interaction, an additional logistic analysis was conducted in which the age variable was categorized into the following age groups in accordance with how licensing data by age are conventionally grouped for reporting by the DMV: 19 and younger (n = 362), 20–34 (n = 13,544), 35–54 (n = 35,767), and 55 and older (n = 16,093). Separate odds ratios reflecting the effect of the contact letter on subsequent motorcycle licensure were then calculated for each age group. The results, presented in Table 4, indicate that the letter did not significantly affect motorcycle licensure status for the youngest group (under 20 years), though it did significantly increase the odds of subsequent motorcycle licensure for owners aged 20–34 (21%), 35–54 (47%), and 55 and older (96%). 3.4. Effect of the contact letter on subsequent motorcycle crash involvements To evaluate the effect of the official DMV contact letter on subsequent crash involvements, age, sex, pre-treatment driver record variables, and study group were used in a logistic regression analysis predicting 18-month subsequent motorcycle crash involvements. Table 5 presents the results of the logistic regression predicting motorcycle-related crash involvement. The model significantly predicted motorcycle crash involvement (p < .0001), with age, sex, and pre-treatment motorcycle crashes, one-point violations, and motorcycle violations all being significant predictors (ps < .05). However, the contact letter was not significantly related to subsequent motorcycle crash involvements (p = .2609). 3.5. Effect of the contact letter on subsequent motorcycle traffic violations The effect of the official DMV contact letter on subsequent traffic violations was assessed by using age, sex, pre-treatment driver record variables, and study group (letter vs. no letter) to predict, in linear regression analysis, 18-month subsequent motorcycle-related traffic violations. Table 6 presents the linear regression model results for predicting motorcycle-related traffic violations. The fit of the overall model was statistically significant (p < .0001), with age, sex, and pre-treatment motorcycle crashes, one-point violations, and motorcycle violations all being significant individual predictors (ps < .05). The contact letter was not significantly related to subsequent motorcycle traffic violations (p = .9953).
Table 2 Summary of logistic regression predicting study group assignment. Predictor Intercept Age Sex Certificates/endorsements Prior suspension/revocation Prior non-motorcycle crashes Prior motorcycle crashes Prior two-point violations Prior one-point violations Prior motorcycle violations
Regression coefficient 0.0332 0.0003 0.0172 0.0054 0.0278 0.0153 0.0023 0.0345 0.0087 0.0187
Standard error
Wald v2
p
Oddsratio
95% confidence interval
0.0314 0.0006 0.0180 0.0145 0.0315 0.0241 0.0867 0.0584 0.0120 0.0505
1.12 0.24 0.92 0.14 0.78 0.41 0.00 0.35 0.52 0.14
.2902 .6277 .3378 .7119 .3769 .5245 .9788 .5540 .4699 .7105
– 1.00 0.98 1.00 1.03 0.99 1.00 0.97 1.01 1.02
– 1.00–1.00 0.95–1.02 0.98–1.03 0.97–1.09 0.94–1.03 0.84–1.18 0.86–1.08 0.99–1.03 0.92–1.13
Note. Overall model fit was not statistically significant, v2(9, N = 65,766) = 3.46, p = .9434. Coding: Sex 0 = women, 1 = men; Study Group 0 = control, 1 = treatment.
170
K.J. Limrick, S.V. Masten / Transportation Research Part F 23 (2014) 165–172
Table 3 Summary of logistic regression predicting 33-month subsequent motorcycle licensure. Predictor
Regression coefficient
Intercept Age Sex Prior non-motorcycle crashes Prior motorcycle crashes Prior two-point violations Prior one-point violations Prior motorcycle violations Study group Study group age
2.2162 0.0258 1.4305 0.0481 0.1767 0.0097 0.0417 0.3361 0.1730 0.0129
Standard error
Wald v2
p
Odds ratio
95% confidence interval
0.0733 0.0015 0.0392 0.0359 0.1120 0.0731 0.0169 0.0590 0.0861 0.0019
915.09 313.37 1334.82 1.80 2.49 0.02 6.11 32.42 4.04 46.28
.0001* .0001* .0001* .1803 .1148 .8945 .0135* .0001* .0445* .0001*
– – 4.18 1.05 1.19 1.01 1.04 1.40 – –
– – 3.87–4.51 0.98–1.13 0.96–1.49 0.88–1.17 1.01–1.08 1.25–1.57 – –
Note. v2(9, N = 65,766) = 2546.75, p < .0001. Coding: Sex 0 = women, 1 = men; Study Group 0 = control, 1 = treatment; Licensure Status 0 = unlicensed, 1 = license or permit. A group-by-sex interaction term was included in the original model, but was removed due to non-significance (p > .05). * p < .05.
Table 4 Odds ratios illustrating how the effect of the contact letter on 33-month subsequent motorcycle licensure differed within owner age groups.
*
Age group
Wald v2
Standard error
p
Odds ratio
95% confidence interval
19 or younger 20–34 35–54 55 or older
0.02 14.78 142.59 141.46
0.2694 0.0604 0.0475 0.1108
.8942 .0001* .0001* .0001*
0.96 1.21 1.47 1.96
0.56–1.67 1.10–1.34 1.38–1.57 1.75–2.19
p < .05.
Table 5 Summary of logistic regression predicting 18-month subsequent motorcycle-related crash involvement. Predictor Intercept Age Sex Prior non-motorcycle crashes Prior motorcycle crashes Prior two-point violations Prior one-point violations Prior motorcycle violations Study group
Regression coefficient 6.2201 0.0153 1.9305 0.1882 0.9871 0.1415 0.1722 0.7341 0.1152
Standard error
Wald v2
p
Oddsratio
95% confidence interval
0.2914 0.0041 0.2363 0.1346 0.2706 0.2505 0.0588 0.1287 0.1025
455.68 14.29 66.74 1.95 13.31 0.32 8.58 32.54 1.26
.0001* .0002* .0001* .1622 .0003* .5722 .0034* .0001* .2609
– 0.96 6.89 1.21 2.68 1.15 1.19 2.08 0.89
– 0.98–0.99 4.34–10.95 0.93–1.57 1.58–4.56 0.71–1.88 1.06–1.33 1.62–2.68 0.73–1.09
Note. v2(8, N = 65,766) = 227.30, p < .0001. Coding: Sex 0 = women, 1 = men; Study Group 0 = control, 1 = treatment. Group-by-age and group-by-sex interaction terms were included in the original model, but were removed due to non-significance (p > .05). * p < .05.
4. Discussion 4.1. General discussion of findings Thirty-three months after the contact letters were mailed, the percentage of motorcycle owners holding a motorcycle license or learner permit was significantly higher for participants who were mailed the contact letter (14.5%) than for participants who were not (10.4%). This 39% increase in subsequent motorcycle licensure associated with the contact letter is similar to the 32% increase in motorcycle licensure found by Braver et al. (2007); the higher licensure rate among contacted owners in the current study likely reflects the longer follow-up period. Nonetheless, the majority (85.5%) of motorcycle owners in the present study who received the letter remained improperly licensed to operate a motorcycle 33 months later. The increases in motorcycle licensure associated with the contact letter were moderated significantly by owner age (see Table 4), which is consistent with the results obtained by Braver et al. (2007). For those aged 19 or younger, the probability of becoming licensed was not significantly different between the letter and no-letter conditions. Significant differences emerged however when comparing the odds of motorcycle licensure for owners age 20 or older. For participants aged 20–34, the odds of becoming licensed were 21% higher for those who were mailed the contact letter compared to those who were not. Similarly, the odds of licensure for owners 35–54 were higher (47%) for the letter condition compared to the no-letter condition. This effect is further strengthened when evaluating motorcycle licensure for the oldest comparison
171
K.J. Limrick, S.V. Masten / Transportation Research Part F 23 (2014) 165–172 Table 6 Summary of linear regression model predicting 18-month subsequent motorcycle-related traffic violations. Predictor Intercept Age Sex Prior non-motorcycle crashes Prior motorcycle crashes Prior two-point violations Prior one-point violations Prior motorcycle violations Study group
Regression coefficient 0.0228 0.0005 0.0169 0.0015 0.0143 0.0052 0.0049 0.0601 0.0000
Standard error 0.0021 0.0000 0.0012 0.0016 0.0056 0.0035 0.0008 0.0033 0.0010
t
p 10.90 12.99 14.63 0.98 2.54 1.48 6.30 18.36 0.01
.0001* .0001* .0001* .3255 .0110* .1398 .0001* .0001* .9953
Note. R2 = .01; F(8, 65,758) = 115.68, p < .0001. Coding: Sex 0 = women, 1 = men; Study Group 0 = control, 1 = treatment. Group-by-age and group-by-sex interaction terms were included in the original model, but were removed due to non-significance (p > .05). * p < .05.
group; more specifically, owners aged 55 years or older who were mailed the contact letter were 96% more likely to become licensed compared to owners in that age group who were not. It was of concern that increasing motorcycle licensure as a result of the letter might lead to an increase in motorcycle riding (i.e., exposure) and hence an increase in crash and/or violation rates. However, the lack of a significant effect of the letter on these rates suggests that while receiving the letter may have increased the likelihood of an owner becoming licensed, it did not appear to impact the crash or violation risk of those owners compared to individuals to whom the letter was not sent. Whether or not the letter affected riding exposure or led to a change in the competency of the treated motorcyclists, or both, is unknown given that neither exposure nor competency were measured in the present study. Two strengths of this study compared to that completed prior by Braver et al. (2007) are the longer period of follow-up (33 months vs. 6 months) and the fact that crash and violation rates were also compared between the groups. That no significant difference in crash or violation rates was found between the study groups suggests that the increases in motorcycle licensure among the owners were gained without negative aggregate traffic safety consequences. This suggests at least three possibilities: (1) the contact letter resulted in increased legal licensure among the owners who were already riding, (2) the contact letter resulted in increased legal licensure among the owners who were not riding and who continued not riding, or (3) receiving the letter convinced some owners who were riding illegally to stop doing so, or perhaps to ride less often or more carefully. Whether any of these possibilities occurred is unknown, but the contact letter nonetheless appears to have modestly increased licensure among unlicensed owners without affecting their net risk of crash involvement. The findings suggest that the contact letter did not entice owners who were not riding previously to begin riding. The significant age-by-treatment interaction suggests that future motorcycle licensing interventions, such as the treatment used here, may be more effective catalysts for change if they target specific groups of riders. For instance, this study showed that while the intervention was successful in increasing licensure for owners over the age of 20, it had no impact on motorcycle licensure rates of younger owners. One way to better understand this particular result would be to obtain and compare descriptive data and pre- and post-intervention crash and violation rates for younger owners who obtained licensure to those for owners who did not. This would help to establish a profile of the target intractable population (i.e., unlicensed, young owners) and provide guidance for the development of future interventions that might better serve this particular group. It should be mentioned however that the number of unlicensed owners under the age of 20 in the present study represented a very small proportion (<1%) of the total sample compared to the other age classifications. Aside from age-related differences in licensure rates, post hoc analyses also revealed that 30% of the owners who remained unlicensed (across both study groups) by study’s end were women. This finding seems rather anomalous given that women comprise only 11% of licensed motorcyclists in California (DMV, 2008–2012), but an apparent overrepresentation of women among owners who remained unlicensed was also noted by Braver et al. (2007). The fact that the percentage of unlicensed women owners is almost three times the percentage of licensed women motorcyclists in the state suggests that the likelihood of not legally riding a motorcycle one owns is more likely among women than among men. If true, one possible explanation is that men may sometimes register their motorcycle(s) in the names of women confederates (e.g., wives, mothers, or girlfriends) to lower their insurance rates (particularly if the men are under age 25, and hence subject to higher insurance rates), or for some other reasons that are advantageous to them. 4.2. Cost–benefit of contact letter Creating and sending the letter to the 33,068 improperly licensed motorcycle owners in the treatment group cost a total of $31,705. The excess proportion of these owners who became properly licensed relative to the comparison group who did not receive the letter was 0.0406 (0.1446 0.1040 = 0.0406), which amounted to 1384 additional motorcycle licensees in the treated group. Therefore, the cost of the contact letter per additional properly-licensed owner was $23.89. The department collects a $24 fee for each motorcycle endorsement application, but this fee is not sufficient to cover the $25.92 cost to the department to test the knowledge and skills of motorcycle endorsement applicants, resulting in a net unit cost of $1.92 to the
172
K.J. Limrick, S.V. Masten / Transportation Research Part F 23 (2014) 165–172
department for each completed endorsement. Adding this net cost of the endorsement to the cost of the contact letter per additional licensed owner results in a total net cost of $25.81 per additional properly-licensed owner. Given that there was no net change in crash or traffic violation rates, any non-monetary benefit of the letter appears to be solely in bringing additional previously unlicensed owners into legal riding status. 4.3. Overall conclusion and recommendation The contact letter modestly increased the number of previously unlicensed owners who became legal motorcycle operators without increasing their crash or traffic violation rates, but at a total net cost of $25.81 per additional owner who became properly licensed as a result of sending the letters. While the letter treatment significantly increased the motorcycle licensure rate, the overwhelming majority of treated owners in the present study (85.5%) remained improperly licensed to operate two-wheeled vehicles on public roadways, which is consistent with the results from a similar study conducted in Maryland (Braver et al., 2007). Given the traffic-safety neutral outcomes and relatively low cost of treatment, official contact letters should be considered in other jurisdictions to increase proper motorcycle licensure as recommended by NHTSA (2009). Future letters may be more effective if they are tailored to the demographic characteristics (e.g., age and sex) of the unlicensed owners. Acknowledgments This study was made possible through the support of the California Office of Traffic Safety and the National Highway Traffic Safety Administration. The findings, opinions, and conclusions are those of the authors and not necessarily those of the California Office of Traffic Safety, the National Highway Traffic Safety Administration, or the State of California. References Braver, E., Kufera, J., Volpini, K., Lawpoolsri, S., Joyce, J., Alexander, J., et al (2007). Persuasion and licensure: A randomized controlled intervention trail to increase licensure rates among Maryland motorcycle owners. Traffic Injury Prevention, 8, 39–46. Harrison, W. A. (1997). An exploratory investigation of the crash involvement of disqualified drivers and motorcyclists. Journal of Safety Research, 28, 213–219. Hollis, S., & Campbell, F. (1999). What is meant by intention to treat analysis? Survey of published randomised controlled trials. BMJ, 319(7211), 670–674. Hurt, H. H., Quellet, J. V., & Thom, D. R. (1981). Motorcycle accident cause factors and identification countermeasures. Volume 1: Technical report. Washington, DC: US Department of Transportation, National Highway Traffic Safety Administration. Kraus, J. F., Anderson, C., Zador, P., Williams, A., Arzemanian, S., Li, W. C., et al (1991). Motorcycle licensure, ownership, and injury crash involvement. American Journal of Public Health, 81, 172–176. Lardelli-Claret, P., Jimenez-Moleon, J. J., de Dios Luna-del Castillo, J., Garcia-Martin, M., Bueno-Cavanillas, A., & Galvez-Vargas, R. (2005). Driver dependent factors and the risk of causing a collision for two wheeled motor vehicles. Injury Prevention, 11, 225–231. Lin, M. R., & Kraus, J. F. (2009). A review of risk factors and patterns of motorcycle injuries. Accident Analysis and Prevention, 41(4), 710–722. McGwin, G., Whatley, J., Metzger, J., Valent, F., Barbone, F., & Rue, L. W. (2004). The effect of state motorcycle licensing laws on motorcycle driver mortality rates. Journal of Trauma, 56, 415–419. National Highway Traffic Safety Administration, United States Department of Transportation (2009). Guidelines for motorcycle operator licensing (DOT HS 811 141). Washington, DC: National Highway Traffic Safety Administration. National Highway Traffic Safety Administration, United States Department of Transportation (2011). Fatality analysis reporting system. Washington, DC: National Highway Traffic Safety Administration. Accessed 20.08.11. Reeder, A. I., Chalmers, D. J., & Langley, J. D. (1995). Young on-road motorcyclists in New Zealand: Age of licensure, unlicensed riding, and motorcycle borrowing. Injury Prevention, 1(2), 103–108. Sirathranont, J., & Kasantikul, V. (2003). Mortality and injury from motorcycle collisions in Phetchaburi Province. Journal of the Medical Association of Thailand, 86(2), 97–102. Watson, B. C., & Steinhardt, D. A. (2006). A comparison of the crash involvement of unlicensed motorcycle riders and unlicensed drivers in Queensland. In 2006 Australasian road safety research, policing, education conference, 25th–27th October, 2006, Gold Coast, Queensland. Yeh, T. H., Chang, H. L., & Chang, H. W. (2008). Initial age of unlicensed motorcycling experience for a cohort of high school students. Accident Analysis and Prevention, 40(2), 511–517.
Contents lists available at ScienceDirect
Transportation Research Part F journal homepage: www.elsevier.com/locate/trf
Randomized trial of an official contact letter intended to increase proper licensure among unlicensed motorcycle owners Kevin J. Limrick ⇑, Scott V. Masten 1 California Department of Motor Vehicles, Research and Development Branch, 2570 24th Street, MS H-126, Sacramento, CA 95818-2606, USA
a r t i c l e
i n f o
Article history: Received 14 June 2013 Received in revised form 27 January 2014 Accepted 1 February 2014
Keywords: Motorcycle Motorcyclists Unlicensed Licensure Randomized trial Educational intervention
a b s t r a c t Background: Approximately one-third of all fatality-injured motorcycle operators in California from 1999 through 2008 were not properly licensed to ride a motorcycle at the time of the crash. In availing themselves of a licensing system, motorcyclists are afforded the opportunity to acquire and demonstrate the minimum knowledge and skills necessary to operate motorcycles on public roadways, and licensing allows agencies to monitor their riding performance and apply appropriate sanctions, with the overall goal being enhanced safety. To that end, a randomized trial was conducted of an intervention recommended by the National Highway Traffic Safety Administration to increase proper motorcycle licensure. Methods: Official contact letters providing information about obtaining a motorcycle license, and the potential legal consequences of riding without one, were mailed to owners of currently-registered California motorcycles who were not licensed to ride motorcycles. Half of the 65,766 improperly licensed California motorcycle owners identified were randomly assigned to be mailed the contact letter while the remainder served as a no-letter control group. The groups were compared on 33-month post-treatment motorcycle licensure rates and 18-month post-treatment motorcycle crashes and traffic violations. Results: While a modestly higher percentage of motorcycle owners who were mailed the contact letter became properly licensed (14.5%) than those who were not mailed the letter (10.4%), the majority (85.5%) remained improperly licensed to operate a motorcycle 33 months later. The effect of the contact letter was influenced by owner age, whereby the odds of licensure for those to whom the letter was sent increased significantly for owners aged 20–34 (21%), 35–54 (47%), and 55 and older (96%). No 18-month post-treatment differences in motorcycle crash involvements or traffic violations were found. Conclusion: Given the traffic-safety neutral outcomes and relatively low cost of treatment, official contact letters should be considered in other jurisdictions to increase proper motorcycle licensure as recommended by the National Highway Traffic Safety Administration. Future letters may be more effective if they are tailored to the demographic characteristics (e.g., age and sex) of the owners. Ó 2014 Elsevier Ltd. All rights reserved.
⇑ Corresponding author. Tel.: +1 916 657 9791; fax: +1 916 657 8589. 1
E-mail addresses: [email protected] (K.J. Limrick), [email protected] (S.V. Masten). Tel.: +1 916 657 6638; fax: +1 916 657 8589.
http://dx.doi.org/10.1016/j.trf.2014.02.001 1369-8478/Ó 2014 Elsevier Ltd. All rights reserved.
166
K.J. Limrick, S.V. Masten / Transportation Research Part F 23 (2014) 165–172
1. Introduction Operating a two-wheeled motorcycle without a valid license is associated with an increased likelihood of being involved in a crash, a finding that has been demonstrated in Australia, New Zealand, Taiwan, Thailand, the United States, and other jurisdictions (Harrison, 1997; Hurt, Quellet, & Thom, 1981; Kraus et al., 1991; Lardelli-Claret et al., 2005; Lin & Kraus, 2009; McGwin et al., 2004; Reeder, Chalmers, & Langley, 1995; Sirathranont & Kasantikul, 2003; Watson & Steinhardt, 2006; Yeh, Chang, & Chang, 2008). Absolute estimates of the prevalence of improperly licensed motorcycle riding in the U.S. and other jurisdictions are difficult to establish; instead, crash involvements among such riders are typically used to infer the relative prevalence of unlicensed riding over time (Kraus et al., 1991; Reeder et al., 1995; Watson & Steinhardt, 2006; Yeh et al., 2008). In the United States, the number of motorcycle riders involved in fatal crashes increased 114% from 1999 through 2008 (National Highway Traffic Safety Administration [NHTSA], 2011). Consistent with this national trend, the number of motorcycle riders in California involved in fatal crashes during this time period increased 137%. Approximately 27% of motorcycle riders nationwide and 34% of those in California who were involved in fatal crashes during this decade were not properly licensed to ride motorcycles at the time of the crash (NHTSA, 2011). For comparison purposes, among passenger vehicle drivers involved in fatal crashes during this time period, only 14% nationwide and 20% in California were improperly licensed when they crashed. Overall, the number of improperly licensed motorcyclists involved in fatal crashes increased 89% nationwide and 146% in California between 1999 and 2008. In response to these trends, NHTSA entered into a cooperative agreement with the American Association of Motor Vehicle Administrators to develop and publish Guidelines for Motorcycle Operator Licensing, which contains recommended motorcycle safety interventions and strategies to increase proper licensure among motorcyclists (NHTSA, 2009). One such intervention recommended in these guidelines is for motor vehicle agencies to mail a contact letter to improperly licensed motorcycle owners encouraging them to complete the requisite training and obtain a motorcycle operator license (NHTSA, 2009). The rationale behind this recommendation is that in availing themselves of a licensing system, motorcyclists are afforded the opportunity to acquire and demonstrate the minimum knowledge and skills necessary to operate motorcycles on public roadways, and licensing allows agencies to monitor their riding performance and apply appropriate sanctions, with the overall goal being enhanced safety. This strategy also has the potential to increase proper licensure among motorcycle owners and perhaps reduce the prevalence of unlicensed riding for jurisdictions outside of the United States. Approximately 13% of motorcycle owners in California are not properly licensed to operate their vehicle on public roadways. Riding without the proper license or permit is a violation of California Vehicle Code §12500, and those cited for this violation risk impoundment of their motorcycle as well incurring a fine exceeding $1000. Similar to the procedure used in most U.S. states, individuals who wish to become properly licensed to ride a motorcycle on public roadways in California must submit a license application, pay a fee, and pass a visual acuity screening, a motorcycle operator knowledge test of rules of the road and safe driving practices, and a motorcycle skills test. The intent of this motorcycle rider licensing program is to increase rider competency by (1) motivating applications to acquire at least the minimum levels of knowledge and skill necessary to pass the required tests, and (2) screening out those who are unqualified to safely and knowledgably ride a motorcycle on public roadways as demonstrated by their failure of the required vision, knowledge, and/or skills tests. Motorcycle licenses are issued by the California Department of Motor Vehicles (DMV), are usually valid for 5 years, and typically can be renewed by mail in 5-year increments by paying a fee to DMV. The recommended use of contact letters to increase licensure among improperly licensed motorcycle owners was evaluated in a randomized trial in Maryland, a small state on the east coast of the U.S. (Braver et al., 2007). The results of the study indicated that 10.4% of those who were mailed the contact letter obtained a motorcycle license within 6 months after the mailing compared to 7.9% of those in the control group, and that the majority of owners remained improperly licensed to operate a motorcycle after 6 months. The short 6-month follow-up period used in the Braver et al. study may have underestimated the true impact of the contact letter intervention on licensing rates, particularly given that motorcycle learner permits in the state were valid for 6 months. In addition, the potential impact of the contact letter on subsequent motorcycle crashes or traffic violations was not evaluated as part of the study. The current study replicates the randomized trial of Braver et al. (2007) by assessing the effect of using contact letters to increase proper motorcycle licensure among motorcycle owners, but builds upon their work by comparing the licensure rates of the groups after a longer follow-up time (33 months vs. 6 months). It is conceivable that bringing more owners into legal riding status may lead to an increase in motorcycle crash or traffic violations incidents by increasing operator exposure to motorcycle travel, unless any skills gained through the process of licensure counterbalanced the risk from greater exposure. To evaluate this possibility, the present investigation expands upon the Braver et al. study by also comparing the motorcycle crash and traffic violation incidents of the groups in addition to licensure rates. 2. Methods 2.1. Sample identification and assignment Over 800,000 records for currently registered motorcycles were identified from the DMV registration database. Records were discarded if the motorcycles had more than one registered owner, were owned by military personnel or businesses
K.J. Limrick, S.V. Masten / Transportation Research Part F 23 (2014) 165–172
167
(e.g., sales dealerships or rental companies), or if they did not have a California driver license number recorded for the owner. Mopeds or motorized bicycles were also excluded because such vehicles do not have annual registration renewal cycles, and hence often have outdated owner information. Finally, only one motorcycle was retained per owner to remove duplicate ownership and create a sample of unique owners. The driver license numbers for the registered owners were then matched to a file obtained from the DMV licensing database containing records for all individuals with a valid motorcycle license, endorsement, or learner permit. This matching process was used to identify owners of registered motorcycles who did not have a valid motorcycle license, endorsement, or learner permit. After removing those who were deceased, a final sample of 65,766 improperly licensed motorcycle owners remained for use in the study (Fig. 1). These individuals were randomly assigned to either be sent the official contact letter (treatment group; n = 33,068) or not (control group; n = 32,698) based on the terminal digit of their driver license numbers. 2.2. Creation and mailing of contact letters An official DMV contact letter was created based on the NHTSA (2009) recommendations. This correspondence notified owners that the DMV was aware that they were not properly licensed to ride their motorcycle on public roadways, encouraged them to become properly licensed, warned of the legal and financial consequences of being cited for riding without a motorcycle license (i.e., impoundment of their motorcycle and a fine exceeding $1000), and provided information on motorcycle training courses and the procedures for obtaining a motorcycle license or permit. The contact letters were mailed over a 4-day period starting on November 10th, 2009. Some individuals (0.2%; n = 57) responded to the letter by mail, most claiming that their actions did not warrant receiving the letter. Several of these individuals said they no longer owned the motorcycle(s) in question, only rode three-wheeled motorcycles, or owned but did not ride motorcycles, and thus were exempt from motorcycle licensing requirements. Although some letters were returned as undeliverable (1.4%; n = 476), and others were erroneously mailed to registered owners who were not required to have a motorcycle license, the integrity of the study groups, as randomly assigned, was maintained for analysis purposes. This is sometimes called ‘‘intent-to-treat’’ analysis and is frequently used to avoid introducing selection bias (Hollis & Campbell, 1999). 2.3. Outcome data sources Data on the study participant’s demographics, motorcycle licensing status, crash involvements, traffic violations, and other variables used in the analyses were obtained from DMV driver records. Traffic violations included incidents for which the driver was convicted, failed to appear in court, or had a moving violation dismissed as a result of attending a traffic violator school. Crash involvements and traffic violations were counted for the 2-year period before the letters were mailed and separately for the 18-month period afterwards. The starting letter mail date (November 10th, 2009) was used as the reference date for counting crashes and violations within these periods. Crash involvements and traffic violations were divided into those that were motorcycle-related (i.e., in which the owner was riding the motorcycle at the time of the incident) and
All registered CA motorcycles (N = 847,937)
All licensed/permitted CA motorcyclists (N = 1,722,066) Removed: Mopeds (n = 102,210) Owned by business/trust (n = 7,164) Multiple owners (n = 115,838) Missing owner driver license (n = 32,387) Registered to out-of-state address (n = 2,417) Owned by military personnel (n = 384) Duplicate owners (n = 84,647)
Registered CA motorcycles with one in-state owner (n = 502,890)
Merge based on driver license number
Improperly licensed motorcycle owners (n = 65,766)
Treatment (letter) (n = 33,068)
Control (no letter) (n = 32,698)
Fig. 1. Formation of the study sample.
168
K.J. Limrick, S.V. Masten / Transportation Research Part F 23 (2014) 165–172
those that were not motorcycle-related incidents. Total traffic violations (motorcycle-related or not) were categorized based upon whether the infraction was a major (2-point) or minor (1-point) violation. The driver record data were extracted from the DMV licensing database on August 10th, 2012, which provided 14½ months following the end of the 18-month criterion period (May 10th, 2011) for crash and violation incidents occurring during the criterion period to be reported to DMV and added to driver records, and hence be included in the analyses. This ‘‘lag period’’ was intended to provide sufficient time for crashes occurring during the 18-month post-treatment period to be reported to DMV and recorded in the database. Historically, a 3-month lag has been more than ample for this purpose, but as of October 2011, the California Highway Patrol was over 13 months behind in reporting crashes to DMV. The August 10, 2012 driver record extract date also enabled the licensing status of each owner to be determined 33 months after the mailing of the letter on November 10th, 2009. 3. Results 3.1. Descriptive statistics for letter and no letter groups Table 1 presents descriptive statistics for various demographic and pre- and post-treatment driver record variables for participants in the treatment (letter) and control (no letter) study groups. The pre-treatment driver record variables represent incidents that occurred during the 2 years preceding the mail date of the contact letter, whereas the post-treatment driver record variables cover the 33-month licensing and 18-month crash and conviction periods subsequent to the mailing date of the letter. The groups appear to be similar on the demographic and pre-treatment driver record variables, as would be expected if the random assignment was successful. Also shown is that the percentage of participants who received the official DMV contact letter and who subsequently obtained a motorcycle license or learner permit was 14.46% compared to 10.40% among those who did not receive the contact letter. 3.2. Assessment of study group equivalency at random assignment To evaluate whether the method used to assign participants into the treatment (letter) and control (no letter) study groups produced equivalent pre-treatment groups with respect to age, sex, number of non-motorcycle endorsements/certificates, and prior driver record, these variables were entered into a logistic regression analysis predicting study group assignment (treatment vs. control). The results are presented in Table 2. The fit of the model predicting study group assignment was non-significant (p = .9434), indicating that the predictors, as a set, did not reliably distinguish between drivers in the two study groups. In addition, none of the individual predictors were
Table 1 Description of the treatment (letter) and control (no letter) study groups. Variable
Study group Treatment (letter) n = 33,068
Control (no letter) n = 32,698
Demographics Age (mean) Male (%) Non-motorcycle certificates/endorsements (per 100)
45.77 72.76 14.23
45.85 73.05 14.15
2-year pre-treatment driver record (per 100) Total crash involvements Total violations Two-point violations One-point violations Motorcycle crash involvements Motorcycle violations Suspended or revoked license
10.65 49.99 1.95 34.36 0.81 2.08 8.32
10.79 49.32 1.98 33.90 0.81 2.01 8.15
18-month post-treatment driver record (per 100) Total crash involvements Total violations Two-point violations One-point violations Non-motorcycle crash involvements Motorcycle crash involvements Motorcycle violations
6.85 33.74 1.33 21.83 6.29 0.56 1.44
6.66 33.32 1.26 21.55 6.03 0.64 1.43
33-month post-treatment motorcycle licensure Motorcycle license/permit (%)
14.46
10.40
169
K.J. Limrick, S.V. Masten / Transportation Research Part F 23 (2014) 165–172
related to group assignment (ps > .05). These results indicate that the study group assignment was indeed random and resulted in equivalent pre-treatment groups with respect to demographic, licensure, and driver history variables. 3.3. Effect of the contact letter on subsequent motorcycle licensure Results of a chi-square test of independence indicated that 33-month subsequent-to-mailing motorcycle licensure differed between the study groups, with participants who received the official DMV contact letter being 39% more likely to be properly licensed to operate a motorcycle (14.46%) than those who did not receive the letter (10.40%), v2(1, N = 65,766) = 249.89, p < .0001. However, the majority (85.5%) of motorcycle owners who received the letter still remained improperly licensed to operate a motorcycle 33 months later. To further evaluate the effect of the letter on motorcycle licensure, age, sex, pre-treatment driver record variables, and study group were used as predictors in a logistic regression model predicting 33-month subsequent-to-mailing motorcycle licensure. The intent of this analysis was to determine whether the letter was still significantly associated with subsequent motorcycle licensure after adjusting for any nominal differences between the study groups on potential confounders. The logistic regression results presented in Table 3 indicate that the model significantly predicted subsequent motorcycle licensure (p < .0001). Age, sex, prior one-point traffic violations, and motorcycle-related traffic violations were all statistically significant predictors of subsequent motorcycle licensure (ps < .05). Importantly, study group (p = .0445) and the study group age (continuous) interaction (p < .0001) also significantly predicted motorcycle licensure, indicating that the effect of the contact letter on subsequent motorcycle licensure varied according to driver age. To explore this interaction, an additional logistic analysis was conducted in which the age variable was categorized into the following age groups in accordance with how licensing data by age are conventionally grouped for reporting by the DMV: 19 and younger (n = 362), 20–34 (n = 13,544), 35–54 (n = 35,767), and 55 and older (n = 16,093). Separate odds ratios reflecting the effect of the contact letter on subsequent motorcycle licensure were then calculated for each age group. The results, presented in Table 4, indicate that the letter did not significantly affect motorcycle licensure status for the youngest group (under 20 years), though it did significantly increase the odds of subsequent motorcycle licensure for owners aged 20–34 (21%), 35–54 (47%), and 55 and older (96%). 3.4. Effect of the contact letter on subsequent motorcycle crash involvements To evaluate the effect of the official DMV contact letter on subsequent crash involvements, age, sex, pre-treatment driver record variables, and study group were used in a logistic regression analysis predicting 18-month subsequent motorcycle crash involvements. Table 5 presents the results of the logistic regression predicting motorcycle-related crash involvement. The model significantly predicted motorcycle crash involvement (p < .0001), with age, sex, and pre-treatment motorcycle crashes, one-point violations, and motorcycle violations all being significant predictors (ps < .05). However, the contact letter was not significantly related to subsequent motorcycle crash involvements (p = .2609). 3.5. Effect of the contact letter on subsequent motorcycle traffic violations The effect of the official DMV contact letter on subsequent traffic violations was assessed by using age, sex, pre-treatment driver record variables, and study group (letter vs. no letter) to predict, in linear regression analysis, 18-month subsequent motorcycle-related traffic violations. Table 6 presents the linear regression model results for predicting motorcycle-related traffic violations. The fit of the overall model was statistically significant (p < .0001), with age, sex, and pre-treatment motorcycle crashes, one-point violations, and motorcycle violations all being significant individual predictors (ps < .05). The contact letter was not significantly related to subsequent motorcycle traffic violations (p = .9953).
Table 2 Summary of logistic regression predicting study group assignment. Predictor Intercept Age Sex Certificates/endorsements Prior suspension/revocation Prior non-motorcycle crashes Prior motorcycle crashes Prior two-point violations Prior one-point violations Prior motorcycle violations
Regression coefficient 0.0332 0.0003 0.0172 0.0054 0.0278 0.0153 0.0023 0.0345 0.0087 0.0187
Standard error
Wald v2
p
Oddsratio
95% confidence interval
0.0314 0.0006 0.0180 0.0145 0.0315 0.0241 0.0867 0.0584 0.0120 0.0505
1.12 0.24 0.92 0.14 0.78 0.41 0.00 0.35 0.52 0.14
.2902 .6277 .3378 .7119 .3769 .5245 .9788 .5540 .4699 .7105
– 1.00 0.98 1.00 1.03 0.99 1.00 0.97 1.01 1.02
– 1.00–1.00 0.95–1.02 0.98–1.03 0.97–1.09 0.94–1.03 0.84–1.18 0.86–1.08 0.99–1.03 0.92–1.13
Note. Overall model fit was not statistically significant, v2(9, N = 65,766) = 3.46, p = .9434. Coding: Sex 0 = women, 1 = men; Study Group 0 = control, 1 = treatment.
170
K.J. Limrick, S.V. Masten / Transportation Research Part F 23 (2014) 165–172
Table 3 Summary of logistic regression predicting 33-month subsequent motorcycle licensure. Predictor
Regression coefficient
Intercept Age Sex Prior non-motorcycle crashes Prior motorcycle crashes Prior two-point violations Prior one-point violations Prior motorcycle violations Study group Study group age
2.2162 0.0258 1.4305 0.0481 0.1767 0.0097 0.0417 0.3361 0.1730 0.0129
Standard error
Wald v2
p
Odds ratio
95% confidence interval
0.0733 0.0015 0.0392 0.0359 0.1120 0.0731 0.0169 0.0590 0.0861 0.0019
915.09 313.37 1334.82 1.80 2.49 0.02 6.11 32.42 4.04 46.28
.0001* .0001* .0001* .1803 .1148 .8945 .0135* .0001* .0445* .0001*
– – 4.18 1.05 1.19 1.01 1.04 1.40 – –
– – 3.87–4.51 0.98–1.13 0.96–1.49 0.88–1.17 1.01–1.08 1.25–1.57 – –
Note. v2(9, N = 65,766) = 2546.75, p < .0001. Coding: Sex 0 = women, 1 = men; Study Group 0 = control, 1 = treatment; Licensure Status 0 = unlicensed, 1 = license or permit. A group-by-sex interaction term was included in the original model, but was removed due to non-significance (p > .05). * p < .05.
Table 4 Odds ratios illustrating how the effect of the contact letter on 33-month subsequent motorcycle licensure differed within owner age groups.
*
Age group
Wald v2
Standard error
p
Odds ratio
95% confidence interval
19 or younger 20–34 35–54 55 or older
0.02 14.78 142.59 141.46
0.2694 0.0604 0.0475 0.1108
.8942 .0001* .0001* .0001*
0.96 1.21 1.47 1.96
0.56–1.67 1.10–1.34 1.38–1.57 1.75–2.19
p < .05.
Table 5 Summary of logistic regression predicting 18-month subsequent motorcycle-related crash involvement. Predictor Intercept Age Sex Prior non-motorcycle crashes Prior motorcycle crashes Prior two-point violations Prior one-point violations Prior motorcycle violations Study group
Regression coefficient 6.2201 0.0153 1.9305 0.1882 0.9871 0.1415 0.1722 0.7341 0.1152
Standard error
Wald v2
p
Oddsratio
95% confidence interval
0.2914 0.0041 0.2363 0.1346 0.2706 0.2505 0.0588 0.1287 0.1025
455.68 14.29 66.74 1.95 13.31 0.32 8.58 32.54 1.26
.0001* .0002* .0001* .1622 .0003* .5722 .0034* .0001* .2609
– 0.96 6.89 1.21 2.68 1.15 1.19 2.08 0.89
– 0.98–0.99 4.34–10.95 0.93–1.57 1.58–4.56 0.71–1.88 1.06–1.33 1.62–2.68 0.73–1.09
Note. v2(8, N = 65,766) = 227.30, p < .0001. Coding: Sex 0 = women, 1 = men; Study Group 0 = control, 1 = treatment. Group-by-age and group-by-sex interaction terms were included in the original model, but were removed due to non-significance (p > .05). * p < .05.
4. Discussion 4.1. General discussion of findings Thirty-three months after the contact letters were mailed, the percentage of motorcycle owners holding a motorcycle license or learner permit was significantly higher for participants who were mailed the contact letter (14.5%) than for participants who were not (10.4%). This 39% increase in subsequent motorcycle licensure associated with the contact letter is similar to the 32% increase in motorcycle licensure found by Braver et al. (2007); the higher licensure rate among contacted owners in the current study likely reflects the longer follow-up period. Nonetheless, the majority (85.5%) of motorcycle owners in the present study who received the letter remained improperly licensed to operate a motorcycle 33 months later. The increases in motorcycle licensure associated with the contact letter were moderated significantly by owner age (see Table 4), which is consistent with the results obtained by Braver et al. (2007). For those aged 19 or younger, the probability of becoming licensed was not significantly different between the letter and no-letter conditions. Significant differences emerged however when comparing the odds of motorcycle licensure for owners age 20 or older. For participants aged 20–34, the odds of becoming licensed were 21% higher for those who were mailed the contact letter compared to those who were not. Similarly, the odds of licensure for owners 35–54 were higher (47%) for the letter condition compared to the no-letter condition. This effect is further strengthened when evaluating motorcycle licensure for the oldest comparison
171
K.J. Limrick, S.V. Masten / Transportation Research Part F 23 (2014) 165–172 Table 6 Summary of linear regression model predicting 18-month subsequent motorcycle-related traffic violations. Predictor Intercept Age Sex Prior non-motorcycle crashes Prior motorcycle crashes Prior two-point violations Prior one-point violations Prior motorcycle violations Study group
Regression coefficient 0.0228 0.0005 0.0169 0.0015 0.0143 0.0052 0.0049 0.0601 0.0000
Standard error 0.0021 0.0000 0.0012 0.0016 0.0056 0.0035 0.0008 0.0033 0.0010
t
p 10.90 12.99 14.63 0.98 2.54 1.48 6.30 18.36 0.01
.0001* .0001* .0001* .3255 .0110* .1398 .0001* .0001* .9953
Note. R2 = .01; F(8, 65,758) = 115.68, p < .0001. Coding: Sex 0 = women, 1 = men; Study Group 0 = control, 1 = treatment. Group-by-age and group-by-sex interaction terms were included in the original model, but were removed due to non-significance (p > .05). * p < .05.
group; more specifically, owners aged 55 years or older who were mailed the contact letter were 96% more likely to become licensed compared to owners in that age group who were not. It was of concern that increasing motorcycle licensure as a result of the letter might lead to an increase in motorcycle riding (i.e., exposure) and hence an increase in crash and/or violation rates. However, the lack of a significant effect of the letter on these rates suggests that while receiving the letter may have increased the likelihood of an owner becoming licensed, it did not appear to impact the crash or violation risk of those owners compared to individuals to whom the letter was not sent. Whether or not the letter affected riding exposure or led to a change in the competency of the treated motorcyclists, or both, is unknown given that neither exposure nor competency were measured in the present study. Two strengths of this study compared to that completed prior by Braver et al. (2007) are the longer period of follow-up (33 months vs. 6 months) and the fact that crash and violation rates were also compared between the groups. That no significant difference in crash or violation rates was found between the study groups suggests that the increases in motorcycle licensure among the owners were gained without negative aggregate traffic safety consequences. This suggests at least three possibilities: (1) the contact letter resulted in increased legal licensure among the owners who were already riding, (2) the contact letter resulted in increased legal licensure among the owners who were not riding and who continued not riding, or (3) receiving the letter convinced some owners who were riding illegally to stop doing so, or perhaps to ride less often or more carefully. Whether any of these possibilities occurred is unknown, but the contact letter nonetheless appears to have modestly increased licensure among unlicensed owners without affecting their net risk of crash involvement. The findings suggest that the contact letter did not entice owners who were not riding previously to begin riding. The significant age-by-treatment interaction suggests that future motorcycle licensing interventions, such as the treatment used here, may be more effective catalysts for change if they target specific groups of riders. For instance, this study showed that while the intervention was successful in increasing licensure for owners over the age of 20, it had no impact on motorcycle licensure rates of younger owners. One way to better understand this particular result would be to obtain and compare descriptive data and pre- and post-intervention crash and violation rates for younger owners who obtained licensure to those for owners who did not. This would help to establish a profile of the target intractable population (i.e., unlicensed, young owners) and provide guidance for the development of future interventions that might better serve this particular group. It should be mentioned however that the number of unlicensed owners under the age of 20 in the present study represented a very small proportion (<1%) of the total sample compared to the other age classifications. Aside from age-related differences in licensure rates, post hoc analyses also revealed that 30% of the owners who remained unlicensed (across both study groups) by study’s end were women. This finding seems rather anomalous given that women comprise only 11% of licensed motorcyclists in California (DMV, 2008–2012), but an apparent overrepresentation of women among owners who remained unlicensed was also noted by Braver et al. (2007). The fact that the percentage of unlicensed women owners is almost three times the percentage of licensed women motorcyclists in the state suggests that the likelihood of not legally riding a motorcycle one owns is more likely among women than among men. If true, one possible explanation is that men may sometimes register their motorcycle(s) in the names of women confederates (e.g., wives, mothers, or girlfriends) to lower their insurance rates (particularly if the men are under age 25, and hence subject to higher insurance rates), or for some other reasons that are advantageous to them. 4.2. Cost–benefit of contact letter Creating and sending the letter to the 33,068 improperly licensed motorcycle owners in the treatment group cost a total of $31,705. The excess proportion of these owners who became properly licensed relative to the comparison group who did not receive the letter was 0.0406 (0.1446 0.1040 = 0.0406), which amounted to 1384 additional motorcycle licensees in the treated group. Therefore, the cost of the contact letter per additional properly-licensed owner was $23.89. The department collects a $24 fee for each motorcycle endorsement application, but this fee is not sufficient to cover the $25.92 cost to the department to test the knowledge and skills of motorcycle endorsement applicants, resulting in a net unit cost of $1.92 to the
172
K.J. Limrick, S.V. Masten / Transportation Research Part F 23 (2014) 165–172
department for each completed endorsement. Adding this net cost of the endorsement to the cost of the contact letter per additional licensed owner results in a total net cost of $25.81 per additional properly-licensed owner. Given that there was no net change in crash or traffic violation rates, any non-monetary benefit of the letter appears to be solely in bringing additional previously unlicensed owners into legal riding status. 4.3. Overall conclusion and recommendation The contact letter modestly increased the number of previously unlicensed owners who became legal motorcycle operators without increasing their crash or traffic violation rates, but at a total net cost of $25.81 per additional owner who became properly licensed as a result of sending the letters. While the letter treatment significantly increased the motorcycle licensure rate, the overwhelming majority of treated owners in the present study (85.5%) remained improperly licensed to operate two-wheeled vehicles on public roadways, which is consistent with the results from a similar study conducted in Maryland (Braver et al., 2007). Given the traffic-safety neutral outcomes and relatively low cost of treatment, official contact letters should be considered in other jurisdictions to increase proper motorcycle licensure as recommended by NHTSA (2009). Future letters may be more effective if they are tailored to the demographic characteristics (e.g., age and sex) of the unlicensed owners. Acknowledgments This study was made possible through the support of the California Office of Traffic Safety and the National Highway Traffic Safety Administration. The findings, opinions, and conclusions are those of the authors and not necessarily those of the California Office of Traffic Safety, the National Highway Traffic Safety Administration, or the State of California. References Braver, E., Kufera, J., Volpini, K., Lawpoolsri, S., Joyce, J., Alexander, J., et al (2007). Persuasion and licensure: A randomized controlled intervention trail to increase licensure rates among Maryland motorcycle owners. Traffic Injury Prevention, 8, 39–46. Harrison, W. A. (1997). An exploratory investigation of the crash involvement of disqualified drivers and motorcyclists. Journal of Safety Research, 28, 213–219. Hollis, S., & Campbell, F. (1999). What is meant by intention to treat analysis? Survey of published randomised controlled trials. BMJ, 319(7211), 670–674. Hurt, H. H., Quellet, J. V., & Thom, D. R. (1981). Motorcycle accident cause factors and identification countermeasures. Volume 1: Technical report. Washington, DC: US Department of Transportation, National Highway Traffic Safety Administration. Kraus, J. F., Anderson, C., Zador, P., Williams, A., Arzemanian, S., Li, W. C., et al (1991). Motorcycle licensure, ownership, and injury crash involvement. American Journal of Public Health, 81, 172–176. Lardelli-Claret, P., Jimenez-Moleon, J. J., de Dios Luna-del Castillo, J., Garcia-Martin, M., Bueno-Cavanillas, A., & Galvez-Vargas, R. (2005). Driver dependent factors and the risk of causing a collision for two wheeled motor vehicles. Injury Prevention, 11, 225–231. Lin, M. R., & Kraus, J. F. (2009). A review of risk factors and patterns of motorcycle injuries. Accident Analysis and Prevention, 41(4), 710–722. McGwin, G., Whatley, J., Metzger, J., Valent, F., Barbone, F., & Rue, L. W. (2004). The effect of state motorcycle licensing laws on motorcycle driver mortality rates. Journal of Trauma, 56, 415–419. National Highway Traffic Safety Administration, United States Department of Transportation (2009). Guidelines for motorcycle operator licensing (DOT HS 811 141). Washington, DC: National Highway Traffic Safety Administration. National Highway Traffic Safety Administration, United States Department of Transportation (2011). Fatality analysis reporting system. Washington, DC: National Highway Traffic Safety Administration.