Journal of Safety Research, Vol. 14, pp. 65-82, 1983 0 1983 National Safety Council and Pergamon Press Ltd
0022-4375183 $3.00 + .OO Printed in the USA
Driver License Renewal By Mail in California Shara
Lynn
Kelsey
and Mary
K. Janke
Drivers under the age of 70 whose prior 4-year accident and conviction records showed no entries when examined 2 months prior to the date of expiration of their driver licenses were randomly assigned to either a group that renewed licenses in the normal manner or a group that was offered the opportunity to receive a 4-year license extension by mail. A related experiment involved comparing clean-record drivers of any age who were given a e-year extension with a group of drivers required to renew in person, Examination of driver records 18 and 28 months subsequent to extension uncovered no deleterious effects of the program, while savings of roughly 3 million dollars annually are expected. Findings are tentative, however, until 4-year data can be collected. The report concludes that the extension program has been suc-
cessful, that multiple extensionsshould be authorized and evaluated for drivers with continuing clean records, and that the present exclusion of drivers 70 and above does not appear to be defensible.
For some 40 years the renewal of a driver license in California has required that the person appear at a local field office periodically to pass a written and a vision screening test. In rare instances the driver may also be required to pass a road test; this only occurs, however, when the Licensing-RegistraShara Lynn Kelsey, Social/Behavioral Research Analyst, and Mary K. Janke, Ph.D., Operations Research Specialist, are members of the staff of the Research and Development Office, Department of Motor Vehicles, 2415 First Avenue, Sacramento, California 95818. This study was undertaken in response to legislative manate (AB 583 and 777, Calvo, 1978-1979). and major support was provided through contract #957994 with the California Office of Traffic Safetv. The research was conducted under the general administrative direction of Ronald S. Coppin, Chief, Research and Development Office. Technical support was provided by Raymond C. Peck, Research Program Specialist, and Michael Ratz, Research Manager.
Summer
1983Nolume
14Ahmber
2
tion Examiner observes some trait that he or she believes could adversely affect safe driving ability. For older drivers, the in-person renewal process often results in either a restricted or limited-term license. Restricted licenses may require the use of corrective aids such as prescription lenses while driving or may limit exposure by confining legal driving within certain geographic boundaries or to daylight (or other specified) hours. Limited-term licenses are those issued for less than the standard 4 years; they are used when the Department wishes to monitor a driver more closely, as, for example, when a medical condition that could affect driving becomes apparent. The in-person renewal process results in about four million drivers returning to a field office every year to requalify for a license. With this volume of annual renewals and the 65
associated escalating administrative costs, it became desirable to examine the traffic safety impact of eliminating the in-person renewal process. If in-person testing could be eliminated without negatively affecting traffic safety, substantial savings in departmental costs could be identified and reallocated to other governmental programs. The California Driver’s License Extension Study was required by AB 777 (1979-Calvo), Chapter 625, as an evaluation of a 4-year driver license extension program authorized by that bill. The program authorized by AB 777 provided for a single 4-year extension by mail of the driver license for any driver less than 70 years of age whose record at the time of renewal notification showed, over the preceding 4 years, no traffic accidents, no convictions of traffic law violations, and no failures to appear (FTAs) in court or to pay fines in connection with traffic law violations. The regular driver license fee of $3.25 was charged for the extension. The 4-year exte~ion-by-rn~ program was briefly preceded by a e-year extension plan that had no age restriction. This bill, AB 583 (1978-Calve), Chapter 658, was superceded by AB 777. Some 760,000 drivers were identified for the Z-year program before it was supplanted by the longer extension program. Data on these subjects will be included here primarily to evaluate the effects of extensions on drivers 70 years of age and older. The results of such an analysis have obvious implications on whether the 70-year age limit imposed by AB 777 is necessary or sufficient from a public safety standpoint. One effect of a license extension program is to postpone for extended drivers the usual in-office license renewal testing process, which consists, in California, of a vision test and a test of knowledge of traffic laws and traffic sign recognition. In rare instances, where it is deemed necessary, a road test may also be required. If eliminating the present renewal process were to lead to an increase in traffic accidents of convictions, the increased cost to society might cancel out any savings in administrative costs attributable to the program. Thus, two questions were of primary interest in the evaluation: 66
1. What is the traffic-safety impact of this license extension program? 2. What are the administrative savings to be expected of such a program? Evidence bearing on the above questions will be considered in detail, following a review of the pertinent literature. Previous studies to be reviewed here evaluated alternatives to the standard in-person renewal process. When the effects of various modified renewal programs on the performance of cleanrecord drivers are examined, some studies have shown no significant impact on accidents or convictions. This argues for the costeffectiveness of simplifying renewal processes. Other studies, however, have found a negative effect on such drivers. In response to ACR 67 (Wakefield), the California Department of Motor Vehicles (1971) evaluated the effects of waiving traffic-law tests for renewal applicants. Two related experiments were performed. In Experiment A, waiver of the test (announced by a congratulatory letter) was contingent upon a clean 37-month prior driving record. In Experiment B, the waiver was given randomly to drivers from the renewal population, without regard to their prior driving histories. Test waivers in Experiment A were thus rewards for good driving; in addition, Experiment A included a clean-record treatment group that received a congratulatory letter but no waiver. A control group consisted of drivers whose clean records made them eligible for waivers, but who were not contacted by the Department. Examination of 6-month subsequent driving records indicated that neither the letterwaiver not letter-only treatment had a statistically significant effect on either accidents or convictions, although in Experiment A the group that received only a letter had directionally more accidents than the no-contact group, with this difference approaching significance. The 6-month follow-up period of the above study, constrained by a legislative deadline, was insufficient to show moderatesized differences among the groups. A second evaluation (Kuan, Peck, & Sadler, 1982) exjournal
ofsajety
Research
tended the length of post-treatment followup to 1 year for subjects of the 1971 study. An overall test of the driving records of subjects in Experiment I (Experiment A of the 1971 study) indicated a significant treatment effect on subsequent accidents and a significant three-way interaction between treatment, age, and marital status. The treatment main effect was due to the greater number of accidents for the letter-only group as compared with the no-contact and letter-waiver groups. The interaction was such that for married drivers the difference existed for younger subjects only, while for single drivers the only significant difference occurred between subjects aged 45 or older in the letter-waiver and letter-only groups. There was no difference in accidents between the letter-waiver and no-contact groups. There was no significant treatment main effect on accidents in Experiment II (Experiment B of the 1971 study) and no significant interaction effect. Similarly, neither experiment showed significant main or interaction effects on subsequent convictions. Since subjects in both groups of Experiment II had anticipated taking a law test, similar proportions of subjects in each group could have been expected to prepare for it, thus refreshing their memory of traffic regulations and doing away with any potential difference between the groups on knowledge grounds. The results of the Kuan et al. (1982) study are consistent with those of previous studies (e.g., Carpenter, 1978), which found the predictive validity of written tests to be low. Waiving the written test had no negative effect on traffic safety. Acknowledgment of a good driving record unaccompanied by a waiver, on the contrary, had negative effects on the driving behavior of some subgroups. Harano and Hubert (1974) evaluated the effects of granting a l-year license extension to drivers under the age of 56 with clean l-year prior driving records. Examination of l-year driving records subsequent to treatment revealed no significant effect of the treatment on traffic convictions and a detrimental effect on accidents. Specifically, drivers given extensions in this “reward group” had 14 % more accident involvements reported by the California Highway Patrol Summer
1983Nolume
14hVumber
2
(CHP) than comparison drivers - a significant difference. (All fatal and injury, as well as some property-damage-only, accidents are reported by the CHP.) The negative effect of the program was greatest for subjects whose next renewal years were 1972 and 1973 (i.e., closest in time to the treatment). A similar but lesser negative effect was found when the total number of reported accidents was used as a criterion measure. Program effects were dependent not only on renewal year, but also on subjects’ prior driving records. This was shown by a separate analysis for drivers with 3-year clean prior records. For these subjects, the detrimental effect of the program on CHP-reported accidents was substantially reduced, although drivers with renewal years 1972 and 1973 still showed a nonsignificant negative effect. (Drivers with renewal years 1974 and 1975 showed a nonsignificant positive effect.) The present extension program is limited to drivers with 4-year clean prior records, so the relationship of program effect to length of prior clean record found by Harano and Hubert is extremely relevant here. Consideration of renewal year is also relevant to the present program, which offers extensions to drivers whose licenses are soon to expire. These drivers are equivalent in this respect to the 1972 renewal group in the Harano and Hubert study, one of the groups showing the most deleterious effect of the extension program on accidents. Nevertheless, it is entirely possible, as the authors suggested, that the effect of renewal year resulted from differences in subject samples rather than from some effect of proximity to renewal per se. Harano and Hubert included an incentive condition in their study. The effects of this treatment should be mentioned since, under a license extension program, the extension may be presumed to function as an incentive for drivers anticipating future license renewal, as well as being a reward for those who would otherwise undergo immediate renewal testing. In the incentive condition, drivers whose records were not clean were notified that extensions could be earned if they maintained clean records during the next year. Their control group consisted of drivers, also 67
having prior entries on their records, who were not contacted. At the end of the year, driver records were reviewed to see if incentive program participants were eligible for extensions; if so, they received them. Results of the incentive program were encouraging. There was no treatment effect on convictions during the year subsequent to the initial notification. For accidents, however, there was a significant interaction between renewal year and treatment: ?972-renewal drivers (those whose renewal was closest in time to the treatment) had a significant reduction in CHP-reported accident involvements, while drivers in other renewal years did not differ significantly from controls. After receiving their extensions, drivers in the incentive program had a highly significant reduction in CHP-reported accidents when compared with the control group. This was the case over all renewal years combined. One may hypothesize that these drivers had made some conscious effort over the preceding year to change to safe driving behaviors; these behaviors were then successful in keeping them out of accidents and were strengthened through being rewarded in a timely manner by the extension. As mentioned above, an ongoing extension program would be expected to have these incentive effects, although the present study does not rigorously evaluate such effects. Waller, Hall and Padgett (1977) evaluated the effects of North Carolina legislation exempting renewal applicants from knowledge and road testing (but not from vision testing) if they had no convictions for moving violations during the 4 years immediately preceding renewal and showed no physical or mental impairment that might affect their driving. (A road test was not given to all renewal applicants prior to this legislation, but it was required for renewal applicants aged 65 and over.) This and subsequent legislation covered both operators’ and chauffeurs’ licenses, and affected approximately 70 % of renewal applicants for operators’ licenses (equivalent to class 3 licenses in California). Although Waller et al. reported that they obtained accident and violation rates for control (renewing under the old procedure) and experimental (test-waiver) applicants, these 68
rates were not presented in their report. Instead, the authors presented as their measure of program effectiveness the number of months subsequent to program implementation in which either of the two groups had a higher mean accident or violation rate. This measure seems almost uninterpretable. In the first place, control and experimental subjects were clearly not randomly assigned, since control subjects were those who renewed their drivers’ licenses prior to implementation of the test waiver program and experimental subjects were those who renewed following program implementation. Thus, the control group of the study does not properly perform the function of a control group in a scientific assessment. Second, since the evaluation was made over the 23 months subsequent to program implementation, rather than to renewal, some members of the experimental group had not undergone the renewal process until almost the end of the period. It may be assumed that all subjects were aware of the program before undergoing renewal. Thus, there would have been some incentive effects for post-implementation clean-record applicants and some anticipated punishment effects for those with “dirty” records. (Under the program, all “dirty drivers”- those with much activity on their records-were required to take a road test in addition to a knowledge test. Prior to the program, road testing was not required unless the driver was over 64 years old.) Third, since the North Carolina program used freedom from convictions for moving violations, not accidents, as the criterion for test waiver, a major conclusion of the Waller et al. study-that there was a detrimental program effect on subsequent accidents both for young drivers and for dirty drivers - is not really relevant to the present assessment of the California license extension program. Stoke (1978) studied the effectiveness of knowledge testing in reducing traffic accidents and convictions in Virginia. Three experimental groups were studied. Group I received the standard Virginia Driver’s Manual at the same time that they received their renewal notices. A notice was enclosed, encouraging them to study the manual. Group II Journal of Safety Research
received a copy of the manual plus a written test to be completed at home after studying the manual and returned to the examining station at the time of renewal. Group III received a copy of the manual and a notice informing them that a written examination would be given covering the material at the time of renewal. Passing a vision test was necessary for relicensing in all three groups. It was not necessary to pass the written test in order to be relicensed, although subjects were not told this. In fact, it was not even necessary to take the written test; refusals were still included in the experimental group. The control subjects received the standard renewal notice and took a vision test (the usual relicensing procedure in Virginia); they were not given any study materials, written examination, or other special treatment. While Stoke concluded that knowledge testing does not significantly improve shortterm (subsequent 6-month) driving performance, his study did show trends toward accident reduction for some treatments. For example, the comparison of combined experimental treatments with the control produced a probability less than .20 that the better accident record of the treatment group was due to chance. The effect, if real, was small; nevertheless, it was probably cost-effective. In another Virginia study, McKnight and Edwards (1982) reported that drivers receiving and tested on an experimental safe driving manual exhibited significantly fewer culpable accidents compared to a nontested control during the 24-month period following renewal. A recent California study (Harrington & Ratz, 1978) evaluated the effectiveness of an at-home written traffic-law test for class 3 (noncommercial) drivers with clean, or nearly clean, prior driving records. A sample of renewal applicants with no accidents or convictions for moving violations in the preceding 3 years was sent a pamphlet self-test on traffic regulations, including a sheet giving the answers, as well as a renewal notice. When these subjects presented their renewai notices at a field office, their regular written tests were waived. Another sample of applicants was sent the pamphlet test and a sheet on which to mark their answers, as well as a Summer 1983Nolume
14hVumber
2
renewal notice. Members of this group had one accident or conviction on their records over the preceding 3 years. They were required to present the self-completed answer sheet, as well as the renewal notice, at a field office; at that point their written test was waived. Results for this latter group will not be discussed, since the program, even in its application to clean-record drivers, differed markedly from the present license extension program; that is, subjects still had to go to a field office to renew their license, and vision tests were not waived. Results for clean-record drivers showed significant interactive effects between treatment (experimental vs. control) and classification status. With respect to accidents, drivers correctly classified as having clean records (the great majority of subjects in this group) showed no effects of the treatment. “Misclassified’ drivers, who appeared to have clean records at the time of selection, but who had incurred accidents or violations leading to convictions so close in time to their selection as subjects that these incidents were not yet reflected on their records, had a (significant) 100 % increase in accident involvements over the control rate. With respect to convictions, correctly classified clean-record drivers who were male and under 30 showed a significant detrimental effect of the treatment. Among misclassified drivers (“late record update” drivers in the present study), there were no significant treatment effects on convictions within age-sex categories. The increase in accidents for drivers classified as having clean records, but who had incurred recent accidents or violations, suggested to the study’s authors that there was some adverse effect of a positive communication about one’s driving record when the memory of an accident or infraction is still fresh. This seems likely (although, with regard to the positiveness of the communication, one must assume-the report does not explicitly state - that clean-record drivers were informed in advance that their written tests might be waived). Where detrimental effects of experimental renewal programs have been found, the affective value of the departmental communication to the driver must always be consid69
ered as a possible cause. In the Harrington and Ratz study (making the assumption mentioned above), subjects with late record updates may have experienced anxiety or guilt upon receiving a reward which might presumably be snatched away as soon as the mistake was discovered. Also they may have come to the conclusion that departmental recordkeeping is so unreliable as not to constitute a threat. In the Harano and Hubert (1974) study, a rather effusive letter of praise was sent to l-year clean-record drivers. They were referred to as being members of a “select group” - a considerable overstatement, since more than 70% of California drivers qualified for the program. This exaggerated praise may have led to irritation or, alternatively, to overconfidence. These are all afterthe-fact speculations; nevertheless, in the present study an attempt was made to minimize negative emotional reactions to the departmental message. (Although the inclusion of late record update subjects was unavoidable, its effect was assessed.) When the effects of various modified renewal programs on the performance of drivers selected for their clean records are considered, most studies have shown no particular impact on accidents or convictions. This argues for the cost-effectiveness of simplifying renewal processes. Three studies have found a negative effect on such drivers, however. In the study of Harano and Hubert (1974), drivers receiving extensions (in the reward group) showed a worse subsequent accident record than their controls. This negative effect was considerable lessened, however, when results were analyzed for drivers with clean 3-year, rather than clean l-year, prior driving records. Waller et al. (1977) found that the North Carolina test-waiver program had a negative impact on the accidents of young drivers. Stoke (1978) found trends toward fewer accidents for subjects receiving, and sometimes tested on, a drivers’ manual. Harrington and Ratz (1978) found that men under 30 showed more convictions than controls after a written test waiver and also found that, for late record update drivers classified as having clean records, waiving the test had a similarly negative effect on accidents. This negative effect, it should be 70
noted, was found only for drivers under 30 and over 50. It seems possible that the value of renewal tests may be greatest for drivers at the extremes of the driving-age continuum. The present study therefore evaluates the effect of the license-extension program, not only on the entire sample of clean-record drivers, but on subgroups of this sample formed on the basis of age. Test Performance As Harrington
and Driving Record and Ratz (1978) stated:
The necessity of retesting of drivers license applicants at periodic intervals has long been a controversial topic. Many other countries issue lifetime licenses. In the United States until recently, many states permitted renewal of drivers licenses by mail, so that there was no retesting. Since passage of the Highway Safety Act of 1966, setting federal standards for drivers licensing, most states have required eye examinations and written tests of driving knowledge approximately every four years. California has long required such retesting. There is, however, no compelling evidence for the [appropriateness of the] federal standards. (p. 1) Christensen, Glad, and Pedersen (1976) attempted to determine the value of license-renewal tests. After reviewing the international literature, they concluded that it is impossible, on the basis of existing research data, to assert that any known relicensing test will have measurable safety effects if its aim is to weed out unsafe drivers. This is to some extent a result of the presence of a large random component in the occurrence of accidents. For example, in one study (Peck, McBride, & Coppin, 1971) the correlation for men between accidents in two contiguous l-year periods was only .056. As the driving record becomes longer the random component becomes relatively smaller, but it is always substantial. Law and knowledge tests. In assessing the value of knowledge tests, Christensen et al. referred to a study by Wallace and Crancer (1971), which presented correlations of trafJournal of Safety Research
fit-law test scores with driver category (ranging from 1 for three or more accidents and/or citations to 5 for no accidents or citations) and with number of violation points. The former correlation was - .123, not significant for 235 subjects; the latter was .144, significant at the .05 level but clearly rather low. In comparing means and standard deviations of scores on these knowledge tests for “good” and “bad” drivers, Wallace and Crancer found that the test scores gave them no basis for discriminating between the two groups. Creech and Grandy (1974) analyzed the relationship between driving records and renewal knowledge tests constructed for use in North Carolina. Relationships between test scores for about 12,000 renewal applicants and various driving-record measures were usually in the expected direction, but weak. For example, the correlation between one knowledge test form and accidents plus violations (4-year prior record) was - .07 for men and - .08 for women. Both coefficients were significant at the .05 level, but, like the others found, were too low for the tests to be of practical value as screening devices. Similar results were found by Carpenter (1978). In his study, a new traffic-law test stressing knowledge of safe driving principles and recent law changes was administered to a group of California renewal applicants. A control group received the standard law test based on the California Vehicle Code. Correlations between each test and driving record variables were computed for the 3 years prior and 6 months subsequent to testing. Although all coefficients were significant (p < .05) they were very low, ranging from .03 to .07 for the old test and from .02 to .07 for the new one. (Correlations were positive because the number of errors on the test was correlated with, e.g., the number of accidents.) Records of accidents and convictions over the 6 months following testing showed no differences between applicants receiving the new test and applicants receiving the old. All of this evidence leads to the conclusion that law tests are not useful as screening devices to eliminate unsafe drivers from the driving population. This does not necessarily mean that such tests have no merit, since testSummer
1983Nolume
14/Number
2
ing, in general, may serve at least two other functions. Tests may have diagnostic value, so that the results may be used to identify persons who need further training in certain areas. This is probably more a function of the driving test than of the law test and would have its primary application to previously unlicensed drivers. Tests may also have an educational value, motivating applicants, in the case of the law test, to review the rules of driving. In fact, some researchers (e.g., McKnight & Green, 1977) insist that the primary purpose of the driver license examination is to maintain or improve the knowledge level of the driving population and that studies which evaluate licensing examinations by determining whether test scores correlate significantly with accident rates are essentially irrelevant. According to this view, the requirement of passing a test containing taskrelevant information motivates preparation and the acquisition of information that could lead to reduced accident likelihood, irrespective of whether test scores correlate with accident frequency. In this way law tests may have a traffic-safety impact that does not manifest itself in test scores, but lack of which would appear if the requirement for testing were done away with. In the earlier discussion, however, it has been seen that some negative effects have been found following a test waiver (Harrington & Ratz, 1978) and, moreover, that trends toward positive effects have been found by exposing drivers to the types of materials covered in the written test (Stoke, 1978). Also pertinent is the study by McKnight and Green (1977) which found support for the effectiveness of tests and manuals in reducing accidents associated with convictions for renewal applicants and older drivers. (McKnight’s and Green’s conclusions are more sweeping than this, but methodological flaws impair their arguments - for example, the use of a nested design with an independent-measures significance test in the case of young license applicants.) All of these studies support to some extent the traffic-safety value of a written test, although at the same time they suggest that the effect is not great. Written testing may have its greatest value in indicating to applicants that officialdom is 71
concerned with the amount of their safety-related knowledge and, by extension, with their safe driving performance. This concern would be meaningless if official records were so inaccurate that safe drivers could not be distinguished from unsafe ones. It might be expected, then, that when officialdom appears not to know what it is doing, as in the case of misclassified drivers rewarded for having a clean record (Harrington & Ratz, 1978), negative effects will emerge.
Vision tests. The following discussion will concentrate on the relationship between driving record and static fovea1 visual acuity under normal illumination, which is the only visual ability currently tested in California as a requirement for driver licensing or relicensing. Burg (1964, 1966, 1967, 1968, 1971) has performed an extensive series of studies of the relationship between visual abilities and good driving record. In his 1967 study of over 14,000 test subjects, it was found that all major vision tests used, including tests of visual acuity with a Snellen Chart and Ortho-Rater, correlated significantly (r = - .053 and - .076, respectively) with number of convictions, but in the wrong direction - i.e., good vision was associated with a higher number of convictions (probably an effect of driver age). Static acuity measured by the Snellen Chart showed a correlation (T = .031) in the expected direction with conviction rate per mile. This correlation barely missed significance at the .05 level, with a z of 1.922. The relationships between accidents and vision variables were even weaker than those for convictions. The correlations between binocular static acuity and total number of accidents for the 3 years prior to testing were - .014 (Ortho-Rater) and - .030 (Snellen), both insignificant. Again, the direction of relationship was opposite to that expected. When the dependent variable was accidents per mile, however, the corresponding correlations, .041 and .032, were statistically significant at the .05 level (though extremely low) and were in the expected direction. Burg also found evidence for major differences in the predictability of driving record variables for different age groups and prom72
ised to investigate these differences in future analyses. In a later paper (Burg, 1974)) he reported that the vision-accident relationships for various accident types agree with the age-accident relationships for these types, reflecting age-vision relationships. For example, leftrear, rear, and right-rear accidents increase with increasing age; vision worsens with increasing age, and worsening vision is positively associated with increasing left-rear, rear, and right-rear accidents. Nevertheless, correlations between age and accidents of various types, and between static acuity measures and accidents of these types, were uniformly low. None was as large as .06, although several were statistically significant, due to the large number of subjects involved (N= 12,211). Crancer and O’Neall (1969) studied two contrasted groups of male drivers aged 50 to 70. One group consisted of 108 men without accidents or “violation citations” on their driving records (nonproblem drivers). The 177 men in the second group had either two or more accidents plus two or more “violation citations” within the 24-month period immediately preceding their most recent record entries or two or more accidents within the most recent 12-month period (problem drivers). The age range under study was selected as one that would be most likely to provide a correlation in the “right” direction between vision test scores and driving performance, being a range in which visual deterioration was likely to be present. Static visual acuity was measured by means of an Ortho-Rater. Correlations between static acuity measures (right, left, and both eyes) and group membership (problem vs. nonproblem drivers) ranged from .158 to .185 and were significant at the .05 level. They were not in the expected direction, however, Problem drivers as a group were visually more competent than nonproblem drivers, a finding reminiscent of Burg’s (1967) for accidents unadjusted by mileage. Shinar (1977) presented a review of studies of the relationship between static visual acuity and accident involvement as part of a comprehensive investigation of drivers’ visual limitations. His conclusion, after surveying the literature, was that: Iourflal
of safety Research
In general, there appears to be a negative relationship between acuity and accidents, mostly at the high-accident and poorvision end of the range: Poor accident history may be related to poor vision, while a good driving record is independent of visual acuity. (p. 18) The major purpose of Shinar’s study was to evaluate the reliability and validity of the Mark II vision tester. An Integrated Driver Vision Test Device (Mark I) had been developed (Henderson & Burg, 1974). It served as a prototype in the development of the Mark II, which contains eight driving-related vision tests. Shinar tested 890 subjects on the Mark II. The relationship between performance on a particular vision test and accident involvement was calculated separately for four different age groups: 17-24,25-54,5564, and 65 or over. Static acuity under normal illumination (SA-N) was measured using Landolt rings (light on a dark background) as targets. Significant correlations occurred between SA-N and night accident frequencies for subjects aged 17-24 and between SA-N and dusk/dawn accident frequencies and rates per mile for subjects aged 65 or over. These correlations were in the expected direction (poor vision associated with more accidents), but were generally rather low, the largest being .27. Large correlation coefficients cannot be expected in such research, however, due to the inherent unreliability of the accident measure. In a multiple regression analysis of scores for subjects divided on the basis of age, Shinar found that SA-N emerged as a significant predictor of accident frequency for the nighttime accidents of subjects less than 25 and the daytime accidents of subjects aged 55-64. SAN was also a significant predictor of daytime accident rate per mile for subjects aged 2564. (A variable was considered significant so long as its inclusion in the regression equation reduced the variance around the function.) An analysis of conditional probabilities, using SA-N scores for 841 subjects, showed that the probability of a subject’s having had an accident in the past 5 years given poor vision (below the sixth percentile) was .21, less than the probability (.29) of having had an acciSummer
1983Nolume
14hVumber
2
dent given good vision (above the sixth percentile). However, SA-N was associated with accident rates higher than the study sample mean for subjects scoring below the sixth percentile; this was true both for subjects aged 17-24 and for all ages combined. Shinar’s work showed that the relationship between any single vision test and accident involvement was weak when all subjects and accident categories were pooled. Stronger relationships emerged when subjects were divided on the basis of age and suggested that static acuity testing under normal illumination may be more important for the young and the old than for the middle-aged segment of the driving population. (The issue of whether or not other vision tests would be more useful is beyond the scope of this paper.) Furthermore, Shinar’s analysis and those of other authors indicate that accident frequency is considerably less closely associated with driver vision than is accident rate per mile. The latter measure- the number of accidents as a function of exposure to risk - is a more rational measure than accident frequency for many purposes. Nevertheless, accident frequency better reflects the total risk to society posed by any group of drivers. This consideration is relevant in making decisions based on the societal costs and benefits stemming from any new program. Age as Related to Driving It has been noted that drivers at the extremes of the driving-age continuum may be poorer license extension risks than others. There is considerable evidence that both young and old drivers have more accidents than “middle-aged” drivers relative to the number of miles they drive. It may be hypothesized that whatever value licensing tests or observation by licensing officials have resides chiefly in their application to the young and old. Younger drivers. In California, as well as nationally, young drivers are overrepresented in total traffic convictions, accidents, and fatalities. This overrepresentation is greatest for drivers under 20, who would not ordinarily be renewal applicants, but exists to a 73
lesser extent for drivers up to about the age of 26 (California Department of Motor Vehicles, 1981). The types of traffic-law convictions a driver incurs indicate the types of potentially dangerous errors he or she is making. For young drivers (under 26)) 40 % of all convictions are for speed violations (Harrington & McBride, 1970). Other common types of violations are failure to stop for traffic signs and signals (21% ) and driving vehicles that have defective equipment (19 % ) . It may be questioned whether the requirement of coming to a field office and taking traffic-law and vision tests has any effect on the unsafe driving behaviors of youthful drivers. However, the fact that Waller et al. (1977) found that young drivers whose licenses were extended had more subsequent accidents than controls, as well as more reckless driving and alcohol-related violations, suggests that something about the licensing agency’s renewal process curbs the tendencies of young drivers to engage in unsafe behavior. (It is true that young drivers whose records were dirty, and who therefore had to undergo additional testing, also showed adverse effects, so that interpretation of the results is complex.) In addition, it has been noted above that Harrington and Ratz (1978) found that the negative effects of misclassifying drivers as having clean records (and waiving their written tests) were limited to drivers under 30 and over 50. Also, correctly classified young men (under 30) showed an increase in traffic convictions following the test-waiver treatment. Another point to be made regarding young drivers is that people who become 21 are enabled to drink alcohol legally in California. These people, if they are drivers, will usually have received their first license renewal at age 20. While many young drivers have experience with alcohol long before the age of 21, there still is a likelihood that some, during the term of their first renewal license, will be learning to drink and to drive while under the influence of alcohol. (This may be considered a skill in itself, involving, for example, compensation for slower reaction time.) Thus, the first license renewal may be a particularly inappropriate time for a license extension to be given, if the renewal process has any effect on alcohol-related unsafe driving behavior. 74
Older drivers. Accident frequency uncorrected by mileage remains relatively stable after the age of 26 (California Department of Motor Vehicles, 1981). Beginning at about age 55, however, the average number of accidents per driver per 100,000 miles begins to rise above the level for drivers in midlife (California Department of Motor Vehicles, 1981). The number of accidents relative to mileage for drivers over 75 is almost as great as the number for drivers under 20, the group with the highest accident risk. This finding is not unexpected, for certain functions related to driving ability have been shown to deteriorate with age. Decline of visual functions and hearing in the elderly is well known (e.g., Marsh, 1960; Shinar, 1977). Older people have poorer night vision than younger people and need a longer time to recover dark adaptation after exposure to bright light (McFarland, Tune, & Welford, 1964). There is also much evidence for a general slowing of sensorimotor response in the elderly, which is not limited to any one sense modality nor to any particular motor response (e.g., Birren, 1965). The pattern of traffic convictions for the aging driver differs from that for young drivers. According to Harrington and McBride (1970), drivers over 65 commit relatively few (11% ) speed violations compared to young drivers; 40% of all the convictions of older drivers are for failure to stop at traffic signs and signals. Right-of-way violations are much more prominent for older drivers (14 % of convictions) than for the young (5%). Making illegal turns or failing to signal turns is a relatively common type of violation (18 %) for older drivers. These violation types may reflect sensorimotor deficiencies, such as those referred to above. They may in addition reflect a relative inability of the elderly to process information rapidly (Crossman & Szafran, 1956), a central rather than a peripheral limitation. The evidence lightly touched on here suggests that, because of age-related sensorimotor and cognitive decline, license renewal testing may be particularly important for older drivers. (Although, again, it must be stressed that older drivers, as a group, are not exceptionally hazardous in terms of the total number of accidents they incur.) Moreover, Journal of Safety Research
aside from formal testing, it is sometimes the case that, on the basis of observed apparent feebleness, slowness, or confusion in a renewal applicant, a driving test is required in addition to the usual tests. The driving privilege is suspended for drivers who cannot pass this test after three attempts. Suspension under these conditions has been shown (Janke, Peck, & Dreyer, 1978) to have traffic safety value, reducing the accident rate for such drivers by more than 80 % . Consideration of these points underlines the necessity for determining the effect of any license-extension program not only on the driving population as a whole, but also on drivers in different age categories within that population. METHOD
Selection Late in 1979, 407,000 drivers under the age of 70 whose 4-year licenses would expire in January, February, or March of 1980 were randomly assigned to a group offered a 4-year extension by mail (treatment) or to standard renewal processing (control). These drivers, at the time of selection, had no accidents or convictions on their driving record for the previous 4 years. Once identified, drivers remained in the group to which they were assigned, regardless of how they accomplished the renewal process or even whether or not they completed it. This was done in order to avert a possible compromise in the validity of the random assignment scheme. In order to assess the true effects of any program, one must examine the entire population which that program affects and not just those persons who respond in the expected or desired fashion. Of the drivers who qualified, 90% were offered license extensions by mail upon payment of a fee; the other 10% were required to go through the standard in-person renewal process, including appearance at a field office, passing vision and law tests, and payment of the fee. In a parallel evaluation, some 760, 000 drivers were selected and randomly assigned to treatment or control for the e-year extension evaluation. Their licenses were due for renewal during the period August-December Summer 1983Nolume
14/l\rumber 2
1979. There were no age limitations on this sample, and drivers up to the age of 99 were included. Treatment The notice sent to extension group drivers was a package consisting of (1) a statement offering a 4-year license extension and containing a health question like the one on the regular renewal application form; (2) a sheet of instructions telling subjects where to send a check or money order for $3.25 to obtain their extensions; and (3) an unstamped return envelope. In particular, subjects were instructed not to come to field offices to pay for extensions. The notice sent to comparison group drivers was the standard departmental license renewal form. The fee for license renewal was the same as the fee charged to extension subjects. After receiving their notices, drivers in the extension group could either send in the fee for an extension, go to a field office to renew in the usual manner, or do nothing, in which case their licenses would become invalid upon expiration. Drivers in the comparison group had only the latter two options. Drivers in the extension group who sent in their payment for extensions were mailed extension certificates to be kept with their licenses; included was information on vehicle code changes over the preceding 4 years. The sending of this information was mandated by AB 777. A driver who was offered an extension, but renewed in the standard manner, was not allowed to then complete the extension statement and extend on top of the renewed license term. Moderate Variables Two variables that might possibly moderate the effect of treatment were age and record update classification status-late record update (LRU) versus others. These variables were therefore included in the analysis. Record update status refers to the completeness of the driver record at the time of selection. There is inevitably a lag between occurrence of an accident or traffic violation conviction and its entry into the driver record by DMV; this lag may be several months in 75
length. The delay is sometimes due to the necessary adjudication period between citation and conviction, and in some cases is exacerbated by court reporting delays. Some lag time is due to present departmental procedures - for example, accidents reported under the financial responsibility laws for legal reasons do not go into the driving record until the 65th day after the accident. Because of such problems, some drivers selected as having 4-year clean records actually did not; their records were updated later, after selection had taken place. These are called LRU drivers in this study, and it was one aim of this evaluation to determine if the effect of an extension offered to LRU drivers differed from the effect of an extension offered to those who actually had clean records. It will be recalled that Harrington and Ratz (1978) found a significant detrimental effect on subsequent driver records resulting from waiving the law test for LRU drivers. Variables
Showing Program Effectiveness
The primary criteria for determining program effectiveness were the l&month driving records of drivers subsequent to selection. Follow-up statistics were gathered at 28 months. Included were total accidents and traffic convictions (including failures to appear in response to citations). Prior to analyzing the impact on total accidents, the programs were also compared in terms of the distribution of accident types: property damage only (PDO), injury, and fatality. Since there was no evidence of differential effects on accident severity, no additional analyses were made of program impact by accident subtype. The following additional variables related to the licensing process were examined: 1. Proportion of subjects with valid licenses; 2. Vision restrictions; 3. Restrictions other than vision; 4. Proportion of subjects receiving licenses with a term less than 4 years (limited term); 5. Proportion of subjects receiving any kind of hearing (any of the Department’s scheduled interviews or reexaminations on the grounds of physical or mental impairment or negligent operator status; and 6. Proportion of subjects receiving a new 76
copy of their license because of changes in the information thereon or because their old license was lost or damaged (duplicate litenses) . All comparisons reflect the individual’s status 18 months subsequent to renewal notification. Data analysis was performed separately for drivers in the 2- and 4-year extension samples. When the 2-year extension study data confirm the results obtained in the 4-year extension analysis, they are not reported in detail here. Data from the 2-year evaluation are presented on the effect of extensions for drivers over the age of 69, because no such individuals were granted an extension in the 4-year program. The Department’s Management and Operations Analysis Section calculated the reducible cost for the in-person renewal and the license extension programs. RESULTS
AND DISCUSSION
The samples were examined to detect any possible bias in random assignment to either treatment or control conditions. The results are given in Table 1. The difference in age between the two groups did prove to be statistically significant, but this difference (roughly a month and a half) may be safely disregarded. When exceptionally large sample sizes and computation of several significance tests are emTABLE 1 OF PRIOR RECORD VARIABLES BY GROUPS (4.YEAR PROGRAM, N=406,915)
COMPARISON
BY COMPARISON
FACI‘OR
Mean age Percent late record updatesa Percent males Prior 4-year collision mean per 100 drivers Prior 4-year conviction mean per 100 drivers
MAIL
Fmvl3vAL
(N = 366,373)
(N = 40,542)
41.65’
41.52’
3.11 41.08
3.14 41.01
0.92
0.87
2.44
2.51
aThe proportion, per 100 drivers, who had an accident or conviction just prior to renewal notification. ‘p<.lO.
Journal of Safety Research
ployed, extremely small and inconsequential differences are sometimes found to be statistically significant. Considering all of the variables in Table 1, the makeup of the treatment and control samples is virtually the same; the small distinctions observed would not have a meaningful effect on subsequent accident or conviction rates even if they represented real differences. Table 2 presents data on variables related to the licensing process. Each comparison considered is significant, except the proportion of drivers with a restriction other than vision. The data indicate that a shift to extensions by mail has resulted in: (1) more drivers being licensed; (2) fewer drivers having a vision restriction; (3) fewer drivers receiving a license with a term less than the standard 4 years; (4) more drivers being required to attend a hearing; and (5) more drivers obtaining a duplicate license. To the extent that any of the above differences resulted in changes that involved meaningful cost implications, they were considered in the program cost analysis presented later. The only significant result in Table 2 that may appear counterintuitive was the increase in hearings for drivers offered an extension. While there are a number of departmental actions defined as hearings, most are for drivers with poor records. For drivers with good records, such as those dealt with in the present study, the great majority of hearings would be related to physical or mental conditions. In most cases these would result from “yes” answers to the questions on the extension or back of the renewal notice. It is quite possible that a number of potential hearings were prevented for drivers in the in-person renewal group when examiners at the field offices determined that a hearing would not be required. None of the drivers marking a “yes” answer on the mail-in extension certificate could be immediately cleared in this way and would therefore be scheduled for hearings. In the case of the reduction in vision restrictions for extension group drivers, it should be noted that the drivers themselves may have noticed a deterioration in visual acuity and obtained corrective lenses without the Department requiring them to do so or Summer 1983Nolume
14/Number
2
TABLE 2 LICENSING PROCESS DISPOSITIONS BY GROUP: PERCENTAGE 18 MONTHS SUBSEQUENT TO TREATMENT (4-YEAR PROGRAM, N=406,915) EXENDFXI BY
COMPARISON
FACTOR
Hearings Valid license Vision restrictiona Other restrictiona Limited term Duplicate licenses
IN-PERSON RENEWAL
MAIL
(N = 366.3731 iN = 40,542) 0.26’ 90.26” 26.14” 0.22 0.01” 4.50’
l
0.21’ 88.41” 28.73” 0.20 0.10** 2.97;.
aVision and other restrictions may have been established either prior or subsequent to the time of treatment, and therefore only partral y reflect the effect of the extension program. ‘p<.O5. **p<.oo1.
knowing that it was done. Consequently, the data presented do not necessarily prove that a greater number of extension group drivers are driving with substandard vision. The increase in valid licenses for extension group drivers would seem to suggest that some drivers who might not have renewed because of the difficulties inherent in the standard process (going to the field office and taking the vision and written tests) did renew when the only requirement was mailing in a fee. This would presumably be the case most often with senior citizens who no longer found driving, and the associated expense of owning a vehicle, to be necessary and desirable. In addition, the senior drivers are also more likely to have anxieties about passing the renewal process and to voluntarily let their driver licenses expire rather than apply for renewal at a DMV field office. The data do not necessarily demonstrate, however, that the extension process has increased the number of drivers on the road; the cost of the vehicle, gasoline, and insurance, as well as self-recognition of any decline in safe driving ability, may be far more important considerations than the obstacle the licensing process imposes. Because of these factors, it is possible that many of the additional number of older drivers who extended their license did not actually drive, or drove very little. Instead, their primary motivation may have been to retain their driver license rather than 77
obtaining a DMV ID card for identification purposes. It is important to note that the subsequent accident and conviction records of all drivers who were assigned to the group offered an extension, whether they actually received one or not (including those falling into the special disposition groups being discussed here), are reflected in the accident and conviction data reported below. Consequently, if an increase in the number of licensed drivers, a decrease in vision restrictions, etc., had resulted in an increase in the number of accidents or convictions, that increase would have been reflected in the following analyses. Subsequent Driver Performance
Evaluation
Statistical analyses of total accidents and convictions plus FTAs showed no significant difference between the extended by mail and in-person renewal groups. Results are given in Table 3. Not unexpectedly, what did make a difference in the accident and conviction rates were the biographical and prior record variables- age and record update status. For both groups, younger drivers and late record update drivers had significantly worse subsequent accident and conviction records than the groups as a whole. Examination of 1% month data on the 760,000 2-year extension group drivers produced similar results. The only significant interaction found was for age by treatment on convictions; that is, the extension treatment had a different effect on drivers depending on how old they were. (This interaction will be discussed later in connection with Table 6.) Two-year extension group data, however, did not show a similar interaction, and it had disappeared from the 2%month follow-up data for the TABLE 3 SUMMARY OF DRIVER RECORD VARIABLES 18 MONTHS SUBSEQUENT TO RENEWAL NOTICE FOR ALL DRIVERS (4-YEAR PROGRAM, N=406,915)
CBlTERlON
(MEAN
MEASURE
PER 100
DRIVERS)
Total accidents Convictions & FTAs
78
BY MAIL
RENEWAL
(N = 366,373)
(N = 40,542)
4.32 12.22
4.34 12.10
4-year group. Therefore, it seems that this 4-year extension finding represented chance significance. There were five chances in a hundred that this could occur. Since prior record status had a significant effect on subsequent accident and conviction data, Table 4 was developed to look at that subset of the population known as “late record update” drivers. Although their mean accident and conviction rates were two to three times higher than those of the total sample, clearly indicating the higher risk nature of subjects having recent additions to their records, there was no evidence that the extension program increased or decreased this risk. None of the differences between the extended and control groups in Table 4 approached statistical significance. A similar analysis of 2-year extension data also failed to demonstrate evidence of a differential effect on late record update drivers. Additional analyses were performed to ensure that there was no significant shift in the proportion of accidents by type, that is, no compensating reductions in property-damage-only accidents and increases in fatal and injury accidents. No such shifts were found. Age Considerations As has been indicated, there was some concern that extensions might have negative impacts on drivers at either end of the age continuum. Since the 4-year extension program specifically excluded drivers aged 70 and over from the program, 18-month subsequent records of drivers in the &-year extension program were examined to determine if there was evidence of a significant detrimental effect of extensions among older drivers. Given the results in Table 5, there appears to be no reason to exclude this portion of the population from future extension considerations; in fact, the accident means are often directionally better for older drivers who were offered extensions. The accident variation between the extended and control drivers under and over 70 was statistically significant, indicating that extended drivers over 70 had significantly fewer accidents than their control counterparts. Also included in the table are percentages of valid licenses. As expected, the extended Journal of Safety Research
TABLE 4 SUMMARY OF DRIVER RECORD VARIABLES 18 MONTHS SUBSEQUENT TO RENEWAL NOTICE FOR LATE RECORD UPDATE DRIVERS (4-YEAR PROGRAM, N= 12,674) EXTENDED CBlTERlON
(MFAN
IN-PERSON
BY MAIL
MEAS”BE
(N=
PER 100 DRIVERS)
Total accidents Convictions & FTAs
RENEWAL
11,400)
(N= 1,274) 8.87 33.20
8.32 32.54
group had an overall higher licensure rate, and the difference becomes larger with increasing age. Normally, one would expect that increased licensure would result in increased accidents through increasing exposure. The fact that this obviously has not occurred strongly suggests that many of the additional licensed older drivers were not actually driving. Mean subsequent accident and conviction rates by age group for drivers in the d-year extension program are given in Table 6. In the 18-month data, drivers under 30 who
were extended by mail were found to have significantly more convictions than drivers under 30 who renewed in the conventional manner; this difference disappeared in a later (28-month) follow-up. As this finding was also not replicated in the e-year extension sample, it seems that the initial difference was due to sampling error. With respect to accidents, there was no evidence of a differential program impact by age. (Recall that the 2-year extension program did show a significant interaction associated with drivers 70 and over, as indicated in Table 5.) Program Costs for Fiscal Year 198142 The estimated reducible unit cost (the amount of money that would be saved if the program were eliminated) was $2.77 for the standard driver license renewal program. The reducible unit cost was estimated to be $0.71 for individuals who complete the driver license extension program. For drivers to whom extension offers are mailed, but who either do not renew or who renew in person
TABLE 5 SUMMARY OF DRIVER RECORD VARIABLES 18 MONTHS SUBSEQUENT TO RENEWAL NOTICE BY AGE GROUP EXTENDED BY MAIL
N
Valid licenses
Mean per 100 drivers
N
% Valid licenses
Mean per 100 drivers
117,366 195,157 144,794 21,846 12,861 6,066 1,768 252 500,110
86.40 89.97 93.46 89.33 83.67 74.91 63.57 50.40 89.65
5.82 4.15 3.37 3.10 3.18 3.43 1.98* 1.59 4.15
62,222 100,722 75,278 11,211 6,718 3,116 837 133 260,237
85.13 89.11 92.22 84.07 75.83 63.38 45.76 27.07 88.02
5.76 4.02 3.33 3.26 3.63 3.31 3.94 1.50 4.11
117,366 195,157 144,794 21,846 12,861 6,066 1,768 252 500,110
86.40 89.97 93.46 89.33 83.67 74.91 63.57 50.40 89.65
62,222 100,722 75,278 11,211 6,718 3,116 837 133 260,237
85.13 89.11 92.22 84.07 75.83 63.38 45.76 27.07 88.02
18.68 10.84 6.11 3.92 3.20 2.76 2.75 0.75 10.39
% CIUTEIUON MEASURE BY ACE
IN-PERSON RENEWAL
Total accidents 17-29 30-50 51-69 70-74 75-79 80-84 85-89 90-98 All ages Convictions & FTAs 17-29 30-50 51-69 70-74 75-79 80-84 85-89 90-98 All ages
18.61 10.74 5.94 3.61 3.47 2.79 2.38 1.98 10.26
*p<.o2.
Summer 1983Nolume
14LVumber 2
79
TABLE 6 SUMMARY OF DRIVER RECORD VARIABLES BY AGE AND GROUP 18 AND 28 MONTHS SUBSEQUENT TO RECORD NOTICE (4-YEAR PROGRAM, N=406,915) EXTENDED
CBITEXUON MEASURE
Total
N
BY AGE
RENEWAL
18-month
28-month mean per 100 drivers
es-month mean per 100 drivers
N
8.32 6.24 4.92 6.34
10,510 17,047 12,985 40,542
5.92 4.14 3.33 4.34
8.76 6.03 4.73 6.32
28.52 17.40 9.51 17.70
10,510 17,047 12,985 40,542
18.79 12.09 6.70 12.10
27.82 17.88 9.69 17.84
l%%%s
accidents
Under 30 30-50 Over 50 Total Convictions
Note. -The *p< .05.
93,563 154,677 118,133 366,373
5.75 4.19 3.35 4.32
93,563 154,677 118,133 366,373
19.83’ 11.96 6.52 12.22
& FTAs
Under 30 30-50 Over 50 Total 28-month
data
contain
298 fewer
subjects
at a DMV field office, the reducible unit cost was determined to be equivalent to the cost of the standard renewal program ($2.77). Data obtained during the project period indicated that approximately 68 % of the drivers selected as eligible for an extension by mail completed the required processing and received an extension. The projected volume of extension eligibles for Fy 81-82 was 1,974,895; the total reducible cost to administer the standard renewal program for drivers eligible for extension would therefore be: 1,974,895
x $2.77 = $5,470,459
The total reducible cost to administer the extension program for the same drivers would be: (1,974,895x 68% x $0.71) + (1,974,895x 32% x $2.77) = $2,704,026 yielding a total estimated (reducible) savings for FY 81-82 of: $5,470,459
- $2,704,026
due to retrieval
problems.
12814.5) did not allow for two consecutive extensions, beginning in 1984 some of the drivers whose prior record would otherwise qualify them for an extension would not have been eligible because of a prior extension. At that point, the number of extensions, and the consequent budgetary savings, would have dropped sharply. (Later legislation, SB 483, changed this by allowing two consecutive extensions, as described below .) A benefit-cost analysis comparing the departmental savings with the financial gain or loss to the public resulting from the program’s effect on accidents was not performed at this time because such an analysis should consider the total (4-year) effects of the program. Nevertheless, the data on accident effects presented in Table 3 indicate that drivers in the extension group had directionally (though not significantly) fewer accidents. Consequently, there is no indication at this point that the savings to the public in reduced program administration costs is offset by an increased cost due to accidents.
= $2,766,433
It should be remembered that the budgetary savings shown here can only be expected when all drivers with 4-year prior clean records are offered extensions. As the current extension program (California Vehicle Code 80
IN-PERSON
BY MAIL
18-month mean per 100 drivers
CONCLUSIONS
AND RECOMMENDATIONS
The preceding analysis represents 18 months of post-extension data, with a 28-month follow-up. Because drivers are extended or renewed for 4-year periods, a comJournal of Safety Research
plete analysis must necessarily reflect a 4-year post-extension interval. As such, the present report is essentially an interim analysis. Although there is some risk that the results could change by the end of the 4-year period, it is unlikely that longer term data will differ substantively from the present trends. With the preceding qualification in mind, the following conclusions and policy recommendations were offered for consideration: 1. Extending clean-record drivers for 2and 4-year periods under the specifications contained in AB 583 and AB 777 does not lead to increased accidents. 2. The extension program will save a substantial amount of public money each year. Since there is no evidence of even a slight increase in accidents, extending clean-record drivers is more cost-beneficial than requiring in-person renewal. 3. There is no justification on traffic safety grounds for the 70 year age limit imposed by AB 777. Legislation should therefore be enacted to allow the Department to include drivers over 70 in the extension program. 4. Legislation should be enacted eliminating the one-extension limit and allowing drivers to receive further extensions if their record remains clean. (Implemented.) The effect of further extensions should be evaluated. 5. Consideration should be given to expanding the extension program to include, on an experimental basis, additional groups that are currently excluded by the eligibility criteria. (Implemented.) 6. Some of the cost savings resulting from the extension programs should be retained to fund or upgrade departmental traffic safety programs having higher payoff potential. There remain one inherent qualification that must be reemphasized with respect to the current research design’s ability to measure the traffic safety impact of extending drivers. Any such analysis is necessarily limited to the specific conditions and time periods covered by the study. Demonstrating that a one-time, 2- or 4-year extension has no detrimental safety impact over a 2- or 4-year period does not prove that a policy of repeatedly extending clean drivers would not eventually have detrimental effects. It is therefore Summer
1983Nolume
14/Number
2
important that any substantial expansion of the existing extension program be evaluated in a manner similar to the programs authorized by AB 583 and AB 777. Legislation which was implemented January 1, 1983 (SB 483-Speraw) has provided that persons under the age of 70 who have a clean prior record of 2 years or more, with no DUI, reckless driving, or property damage hit-and-run convictions nor fatal accidents with the 4 previous years, may receive up to two consecutive 4-year license extensions by mail. This program will be evaluated by the California Department of Motor Vehicles (reports due in June 1986 and March 1989). It should be stressed that the findings of the present study may not generalize to all states and operational settings. The effectiveness of any renewal-process waiver program necessarily reflects the relative effectiveness of the renewal process used as a comparison program. Thus, the findings here could possibly be interpreted as showing that the California driver license renewal process, in its present form, is not very effective. This does not preclude the possibility that a different renewal process may be more effective than renewal by mail.
REFERENCES Birren, J. E. Age changes in speed of behavior: Its central nature and physiological correlates. In Wilford, A. T. & Birren, J. E. (Eds.), Behaoior, aging, and the nervous system. Springfield: Charles C Thomas, 1965. Burg, A. An investigation of some relationships between dynamic visual acuity, static visual acuity, and driving record (Report No. 64-18). Los Angeles: University of California, Dept. of Engineering, April 1964. Burg, A. Visual acuity as measured by dynamic and static tests: A comparative evaluation. Journal of Applied Psychology, 1966, 50(6), 460-466. Burg, A. The relationship between vision test sco~e.s and driving record: GeneralPndings (Report No. 67-24). Los Angeles: University of California, Dept. of Engineering, June 1967. Burg, A. Vision test scores and driving record: Additionaljindings (Final Report No. 68-27). Los Angeles: Institute of Transportation and Traffic Engineering, UCLA with California Department of Motor Vehicles, December 1968. Burg, A. Vision and driving: A report on research. Human Factors, 1971, 13(l), 79-87. Burg, A. Visual degradation in relation to specific acci-
81
dent types (UCLA-Eng 7419). Los Angeles: Institute of Transportation and Traffic Engineering, UCLA, March 1974. California Department of Motor Vehicles. Renewal written test waiver study (Assembly Concurrent Resolution 67 - Wakefield). Sacramento: Author, 1971. California Department of Motor Vehicles. The California driver fact book (Report No. 29, 5th ed.). Sacramento: Author, 1981. Carpenter, D. W. The effects of administering written tests stressing knowledge of safe driving principles to renewal driver’s license applicants: An evaluation of the renewal applicantb written test component of California’s selective testing program (Report No. 61). Sacramento: California Department of Motor Vehicles, June 1978. Christensen, P., Glad, A., & Pederson, T. The safety value of driver license Tenewals: An analysis of research results (Report ISBN-82-7133-149-3). Oslo, Norway: Institute of Transportation Economics, 1976. Crancer, A., Jr., & O’Neall, P. A. Comprehensive vision tests and driving record (Report No. 028). Olympia: Washington Department of Motor Vehicles, December 1969. Creech, F. R., & Grandy, J. An analysis of the relationships between the rules tests and the accident and violation histoy of drivers in Xorth Carolina. Princeton, NT: Educational Testing Service. March 1974. Crossman, E. R. F. W., &%zafran, i. Chang!es with age in the speed of information-intake and discrimination. Experientia (Supplement). 1956. 4. 128-135. Harano, R. M.,‘&Hubert, e: M. Ai evaluation of California’s “good driver”incentive program. Sacramento: California Department of Motor Vehicles, 1974. Harrington, D. M., & McBride, R. S. Traffic violations by type, age, sex, and marital status. Accident Analysis and Prevention, 1970, Z(l), 67-79. Harrington, D. M., & Ratz, M. The effectiveness of an at-home drivers licensing law component of California’s selective testing program. Sacramento: California Department of Motor Vehicles, 1978.
82
Henderson, R. L., & Burg, A. Vision and audition in driving (Report No. TM(L)-5297/000/00). Santa Monica, CA: Systems Development Corporation, April 1974. Janke, M. K., Peck, R. C., & Dreyer, D. R. MedicaEZy impaired drivers: An evaluation of California policy (Senate Bill 2033-Garcia). Sacramento: California Department of Motor Vehicles, 1978. Kuan, J., Peck, R. C., & Sadler, D. An analysis of two strategies for renewing driver licenses. Paper presented at Western Psychological Association Meeting, Sacramento, April 9, 1982. Marsh, B. W. Aging and driving. Traffic Engineering, 1960, 31, 11-29. McFarland, R. A., Tune, G. S., & Welford, A. T. On the driving of automobiles by older people. Journal of Gerontology, 1964, 19, 190-197. McKnight, A. J., & Green, M. A. Safe driving knowledge dissemination and testing techniques, Volume 1: General findings (Final Report, DOT-HS-4-OOE17). Washington, DC: DOT, May 1977. McKnight, A. J., & Edwards, R. An experimental evaluation of driver license manuals and written tests. Accident Analysis and Prevention, 1982,14(3), 187-192. Peck, R. C., McBride, R. S., & Coppin, R. S. The distribution and prediction of driver accident frequencies. Accident Analysis and Prevention, 1971, 2(4), 243299. Shinar, D. Driver visual limitations: Diagnosis and treatment. Bloomington: Indiana University School of Public and Environmental Affairs, 1977. Stoke, C. B. The short-term effectiveness of written driver knowledge tests. Highway Safety Division of Virginia, 1978. Wallace, J. E., & Crancer, A. Licensing examinations and their relation to subsequent driving record. Behavior Research in Highway safety, i971, 53-65. Waller. P. F.. Hall. R. G.. & Padgett. S. S. The North Carolina t&t wailer law: An evaluation of its impact. Chapel Hill: University of North Carolina Highway Safety Research Center, 1977.
Journal
of Safety
Research