The effect of changing from secondary to primary safety belt enforcement on police harassment

Accident Analysis and Prevention 36 (2004) 819–828

The effect of changing from secondary to primary safety belt enforcement on police harassment David W. Eby∗ , Lidia P. Kostyniuk, Lisa J. Molnar, Jonathon M. Vivoda, Linda L. Miller Social and Behavioral Analysis Division, University of Michigan Transportation Research Institute, 2901 Baxter Road, Ann Arbor, MI 48109-2150, USA Received 20 March 2003; received in revised form 29 May 2003; accepted 29 May 2003

Abstract The purpose of the study was to investigate whether changing Michigan’s safety belt law from secondary to primary (standard) enforcement resulted in police harassment. The study investigated safety-belt-enforcement-related harassment by considering three measures: citizen complaints arising from enforcement of the safety belt law; citation over-representation, that is, a statistical determination of whether certain groups received more citations than would be expected based upon their presence in the driving population and their rate of violating the safety belt use law; and self-reported harassment among the population of people who receive safety belt citations. Safety-belt-related harassment complaints were very uncommon both before and after primary enforcement. Implementation of primary enforcement did not lead to an increase in citation over-representation, and, therefore, safety-belt-related harassment by sex, race, or age. The vast majority of people receiving safety belt citations reported officer behavior as professional and did not feel that they were singled out for their citation. However, a sizeable minority of Blacks and young people report perceptions of safety-belt-related harassment. Results suggest that states with secondary enforcement should continue their efforts to change to primary enforcement, but should also make a strong effort to educate both law enforcement and the public about the harassment issue. © 2003 Elsevier Ltd. All rights reserved. Keywords: Safety belt enforcement; Police harassment; Michigan’s safety belt law

1. Introduction Although safety belts were introduced into motor vehicles on a widespread basis during the 1960s, very few people used them. As recent as the early 1980s, only about 13% of automobile drivers in the US used safety belts (NHTSA, 1997), prompting states to consider legislation to make use of safety belts mandatory. In July 1984, New York passed the first mandatory safety belt use law followed soon after by New Jersey, Illinois, and Michigan (Lund et al., 1986). As these states considered this legislation, citizens began to express concern that safety belt laws might be used by police to harass motorists. In response, New Jersey legislators limited their safety belt law to secondary enforcement, that is, police officers could issue safety belt citations only after stopping a vehicle for some other violation (Moffat, 1998). Thus, in order to address concerns about possible police harassment, New Jersey created a distinction between secondary and primary (or standard) enforcement which allows an officer to pull over a motorist solely for safety belt non-use (NHTSA, ∗ Corresponding author. Tel.: +1-734-763-2466; fax: +1-734-936-1076. E-mail address: [email protected] (D.W. Eby).

0001-4575/$ – see front matter © 2003 Elsevier Ltd. All rights reserved. doi:10.1016/j.aap.2003.05.007

1999a). This distinction can be found in no other traffic law. Once the precedent for secondary enforcement was set in New Jersey, most other states, including Michigan, enacted safety belt use laws with secondary enforcement in order to curtail concerns about possible police harassment. As of 2002, more than 60% of the US (31 states) had enacted secondary enforcement safety belt laws. Research has shown, however, that safety belt use is significantly lower in states with secondary enforcement than in states with primary enforcement (e.g. Solomon et al., 2001), leading many states to try to change to primary enforcement. There is clear evidence that changing to primary enforcement is an effective way to increase use of safety belts (Campbell, 1987; NHTSA, 1999a; Ulmer et al., 1994). In 10 of the first 11 states to make this change, increases in statewide safety belt use ranged from 8 to 22 percentage points (see Eby et al., 2002b).1 In order to change to primary enforcement, each of these states had to address concerns about police harassment. 1 In the 11th state, Indiana, safety belt use decreased after primary enforcement probably because of a well-publicized law suit challenging the law that resulted in a decrease in safety belt law enforcement (NHTSA, 2001).

820

D.W. Eby et al. / Accident Analysis and Prevention 36 (2004) 819–828

During the legislative debate about this change in Michigan, concerns were raised about the potential for law enforcement officers to use primary enforcement as an opportunity to harass drivers in some way. Indeed, newspapers in Michigan and other parts of the country have reported on studies that reached a conclusion of traffic-law-related harassment, based on some group’s (e.g. a racial group) over-representation on some measure of enforcement activity. However, close analysis of these studies shows that the presence of over-representation is either not supported scientifically or is only one of several possible conclusions. One major flaw in many of these studies (see McGraw, 2000) is that they reach their conclusion of over-representation based simply on whether the proportion of traffic stops for a particular group (e.g. race, sex, or age) exceeds the proportion of that group in the general population. Without knowing, however, the proportion of that group in the population of people on the road, one cannot make any claims about traffic-stop harassment. Other studies have rightly considered a group’s proportion in the distribution of road users in their analyses (see Arrellano, 2000; Bullers, 2000), but have not considered the proportion of traffic law violations comprised by the group. Again, without knowing the group distribution of people who are violating traffic laws, one cannot make conclusions about the presence of over-representation because one group may violate certain laws more frequently than another group, as has been found in some studies (see Lang et al., 2001). One study that included both the group distribution (in this case, race) of drivers and the group distribution of speeding law violations (Lambert, 1998) used direct observation to determine the distribution of both White and Black2 drivers on the road and traveling more than 5 mph over the speed limit. These proportions were compared to the racial distribution of drivers pulled over by the police. Since 98% of the drivers in this sample exceeded the speed limit by more than 5 mph, nearly every driver on the road was a candidate for a speeding-related traffic stop. The study showed that Black drivers were pulled over by law enforcement for any reason, not just speeding, at about four times the rate that would be expected from the presence of Black drivers on the road. Lambert (1998) incorrectly suggested that this finding demonstrates over-representation and, therefore, harassment. While the study provided good support that harassment may have been present, this conclusion is only one of several possible reasons for the difference between the proportions. As discussed in a report by the US General Accounting Office (GAO, 2000), to identify the effects of enforcement of a certain traffic law on police harassment, 2 The racial descriptor “Black” rather than “African American” is used throughout this manuscript. Michigan is located adjacent to Canada and has a relatively large population who are not US citizens. Because we do not know the citizenship of people in our study, Black is a more accurate racial description.

three group distributions must be compared: presence on the road, rate of law violation, and police action for violation of the law (stops or citations). The Lambert (1998) study did not consider the reasons for the stops (only stops for speeding should have been considered), and, therefore, did not unambiguously discover traffic-stop over-representation or harassment. No previous study has addressed the issue of harassment in relation to a change from secondary to primary enforcement. This is surprising since the main impetus for secondary enforcement was concerns about harassment. The purpose of the research reported here was to investigate whether changing Michigan’s safety belt law from secondary to primary enforcement resulted in police harassment. For the purposes of this study, we adopted the implied definition of harassment from the Lambert (1998) study and modified it to be specific for safety belt enforcement: “a driver being singled out for a safety-belt-related traffic citation or treated differently during the stop on the basis of race, sex, age, or other factors unrelated to the actual violation.” The study was designed to investigate safety-belt-enforcement-related harassment by considering three measures. The first was citizen complaints arising from enforcement of the safety belt law. While people who are harassed may not actually submit complaints, and others may complain when no actual harassment occurred, we include this measure because a significant increase in complaints after implementation of primary enforcement may be indicative of increased harassment. The second measure was citation over-representation, that is, a statistical determination of whether certain groups received more citations than would be expected based upon their presence in the driving population and their rate of violating Michigan’s safety belt use law. This measure is similar to the one used in previous research on harassment resulting from traffic-law enforcement, except violation, presence on the road, and police activity (citation) proportions were utilized. The final measure utilized in this study was self-reported harassment among the population of people who received a safety belt citation in the year following implementation of primary enforcement. While one cannot make a valid conclusion about the presence of harassment based only upon self-reports, these data are extremely useful for understanding perceptions of harassment and developing programs to address this issue.

2. Methods This study involved the collection of five distinct types of data: safety belt citation data from courts; electronic driver license images; traffic-stop-related complaints from Michigan law enforcement agencies and civil rights groups; a direct observation survey of safety belt use in Michigan by age, sex, and race; and a telephone survey of Michigan residents who had recently received a safety belt citation.

D.W. Eby et al. / Accident Analysis and Prevention 36 (2004) 819–828

821

2.1. Safety belt citation data

2.3. Citizen complaints

Safety belt citation data were obtained from Michigan’s district and municipal courts, to whom law enforcement agencies are required to report all traffic citations, including safety belt and child restraint citations. Letters were sent to all 163 court locations in Michigan requesting several data elements in electronic format for each citation: court name, court address, offense date, offense time, birth date, telephone number, driver license number, vehicle type, original charge (i.e. driver safety belt, passenger safety belt), sex, disposition, plea, vehicle make, and vehicle year. Because few courts had the computer expertise to generate electronic files, most of the data came in paper format and were entered electronically by temporary research assistants. Of the 163 courts contacted, 161 (99%) agreed to participate in the study and provided citation data. Two courts refused to participate because of time and staffing constraints, and concerns about protecting the confidentiality of violators, respectively.

Copies of all written traffic-stop-related citizen complaints on file with each department for incidents that occurred between 10 March 1999 and 9 March 2001 were requested from local, county and state law enforcement agencies. Law enforcement agencies are required to maintain records on all reported incidents of harassment, and as a result of the primary enforcement law, reported incidents of safety-belt-related harassment must be investigated. Agencies not responding to the initial request letter were contacted a minimum of three times via a combination of letter, telephone, facsimile, and/or electronic mail. A total of 551 (93%) of the 593 law enforcement agencies in Michigan participated in the study. For various reasons, 15 agencies refused to participate and 27 agencies did not respond. Departments with formal complaints on file sent us hard copies for review, which were destroyed once the review was completed. Formal requests for similar complaint information were also made to the American Civil Liberties Union, the Michigan Chapter of The National Association for the Advancement of Colored People, and the Michigan Department of Civil Rights (MDCR). Only the MDCR provided the complaint data. In order to prevent double-counting, each complaint received from the MDCR was compared to all law enforcement agency complaints with the same incident date. Complaints to both agencies on the same incident were combined. Each complaint contained all, or a subset, of the following: the statement from the complainant, a copy of the citation if one was written, complaint investigation information, and a disposition. Each complaint was reviewed for the following variables: date and time of incident; age, race, sex, and state of residence of the complainant; seating position and type of vehicle complainant was occupying; age, race, and sex of the victim if different from the complainant; relationship of complainant and victim; age, race and sex of the officer(s); stated reason for the traffic stop; types of citation(s) given, if any; the action(s) leading to the complaint; the basis of the complaint (i.e. harassment, or something else); and the agency’s disposition of the complaint. These data were entered into a spreadsheet and then converted into the proper form for analysis with the Statistical Analysis Software (SAS) package. All complaints that did not arise from a traffic stop, such as a crash or a pedestrian incident, were eliminated from the analysis. In addition, all complaints that did not meet the study definition of harassment (i.e. the complaint did not mention a person being singled out or treated differentially for any reason during the traffic stop) were eliminated.

2.2. Images Because race information for citations is not recorded by the police, courts, or the Michigan Department of State (DOS), the race of those receiving safety belt citations was determined visually from the electronic driver license images. A similar method for race identification has been used by other researchers (see Lang et al., 2001). The DOS extracted images for each valid driver license number for which there was an image available, and provided these images in JPEG format for analysis. Images were linked to the appropriate records in the safety belt citation database using the driver license number. Each image was viewed by two project staff, working separately, who made a judgement about the race of the person in the image and recorded that judgement as White, Black, or some other race. Reliability between data entry personnel was measured throughout the race-judgement process. Overall agreement was found to be above 95%. In cases of disagreement, a third person judged the race, and the race that was agreed upon between two of the three was permanently assigned to the record. If all three people disagreed (well under 1% of cases), two additional people jointly reviewed the image and made a final decision on race. The database was also examined to ensure that there was agreement regarding race for individuals who appeared in the database more than once (due to multiple citations). When disagreements were found, the race most frequently assigned to that individual was used as the permanent race assignment. If conflicting races were equally assigned, a final determination was made by a two-person team. Five temporary employees completed the task of race identification and were monitored by project supervisors throughout the data entry process.

2.4. Direct observation survey In order to determine whether certain groups of motorists experienced safety-belt-related harassment after Michigan’s

822

D.W. Eby et al. / Accident Analysis and Prevention 36 (2004) 819–828

change to primary enforcement, information was needed about their violation rates and presence on the road. Because this information can be determined visually on the road, a direct observation survey of safety belt non-use in Michigan was designed and conducted to estimate non-use by sex, age, and race and to determine presence on the road. The goal of the sample design for the direct observation survey was to select observation sites that accurately represent front-outboard vehicle occupants in eligible commercial and non-commercial vehicles (i.e. passenger cars, vans/minivans, sport-utility vehicles, and pickup trucks) by race, age, and sex in Michigan. An ideal sample minimizes total survey error while providing sites which can be surveyed efficiently and economically. To achieve this goal, the following sampling procedure was used. All 83 Michigan counties were rank ordered by population (US Bureau of the Census, 1992) and the low population counties were eliminated, reducing the sample space to 28 counties. To ensure representativeness of the major racial groups in Michigan, the 28 counties were rank ordered by their percentage of Blacks, and counties with a percentage smaller than 6% were removed, reducing the sample space to 13 counties. Safety belt use rates by county were calculated using data from a recent statewide direct observation survey of safety belt use in Michigan (Eby et al., 2000a). Wayne county was chosen as a separate stratum because of its disproportionately high percentage of Blacks as well as high vehicle miles traveled (VMT). The remaining counties were rank ordered by their percentage of Blacks and divided into three additional strata, defined by a high, medium, and low percentage of Blacks. Observations were conducted at a total of 400 sites, the number required to achieve a statistical precision of less than 5% relative error, to ensure adequate representation of safety belt use and non-use for each day of the week and all daylight hours, and to ensure that a representative number of Black motorists were observed in the sample. The total number of sites and the distribution of sites by road type were adjusted to account for differences in VMT across strata. Site selection also took into account the percent of Black residents in city and non-city areas within each stratum. Within each stratum, observation sites were randomly assigned to a location and a direction of travel for observation was chosen randomly for each site. The day of week and time of day for site observations were quasi-randomly assigned to sites so that all days of the week and all daylight hours (7:00 a.m. to 7:00 p.m.) had essentially equal probability of selection. Sites that were located spatially adjacent to each other were treated as a cluster and within each cluster, a shortest route between all of the sites was chosen. Observations were made at all sites in each cluster during a single day. Data collected in the study included shoulder belt use, sex, race, and estimated age of drivers and front-right passengers; as well as vehicle type and vehicle purpose (commercial or non-commercial). Observations were made on vehicles as

they stopped at traffic control devices during daylight hours from 8 April to 1 May 2001. The sample design was constructed so that each observation site was weighted by the traffic volume at the site. This was accomplished by selecting sites with equal probability and by setting the observation period to a constant duration (50 min) for each site. However, since all vehicles passing an observer could not be surveyed, a vehicle count of all eligible vehicles on the traffic leg under observation was conducted for a set duration (5 min) immediately prior to and immediately following the observation period (10 min total). The vehicle count was used to estimate the traffic volume at each site. During the observation period, observers were instructed to record data on vehicles in only the lane immediately adjacent to the curb, regardless of the number of lanes present. Observations began immediately after completion of the first vehicle count and observers recorded data for as many eligible vehicles as they could observe. If traffic flow was heavy, observers recorded data for the first eligible vehicle they saw, and then looked up and recorded data for the next eligible vehicle they saw, continuing this process for the remainder of the observation period. Field observers participated in 5 days of intensive training, including both classroom review of data collection procedures and practice field observations, at several sites not included in the sample, but representative of sites and situations that would actually be encountered in the field. Observers practiced observations in rotating teams of two, until each pair had achieved consistent interobserver reliability of at least 85% for all measures on drivers and front-right passengers. During actual data collection, observers were monitored by the field supervisor through on-site visits, telephone contacts, and personal contacts in the office where forms were dropped off. Incoming data forms were examined by the field supervisor and problems (e.g. missing data, discrepancies between the site description form and site listing or schedule) were noted and discussed with field staff. The data were entered into an electronic format and verified for accuracy through double entry, review of data from randomly selected sites by a second party, and a check for inconsistent codes across all data. Errors were corrected after consultation with the original data forms. For each site, computer analysis programs determined the number of observed vehicles, belted and unbelted drivers, and belted and unbelted passengers. Separate counts were made for each independent variable in the survey (i.e. site type, time of day, day of week, weather, sex, race, age, seating position, and vehicle type). This information was combined with the site information to create a file for analysis. 2.5. Telephone survey A telephone survey of people who had received a safety belt citation during the year following primary enforcement was conducted from 1 October to 4 November 2001. The

D.W. Eby et al. / Accident Analysis and Prevention 36 (2004) 819–828

survey questionnaire consisted of questions on: general perceptions of safety belt use and importance of safety belt use; stated reasons for the stop; police actions during the stop; other citations issued; police attitude during the stop; number of previous stops; perceptions of reason for the stop and citation; and respondent demographics, including race, age, and sex. Only questions related to perceptions of harassment arising from a traffic stop resulting in a safety belt citation are reported here. Telephone interviews of about 10 min were conducted by a professional survey research company. The primary objective of the sampling design was to obtain a sample of respondents representative of the population who received safety belt citations in Michigan. Because the distribution of safety belt violators by race was not known at the start of the project, obtaining racial proportions in the sample similar to those in the overall safety-belt-law violator population was important. It was also important to have sufficient numbers of Blacks to allow analyses by race. The population sample consisted of Michigan residents who received safety belt citations in the state in the year following primary enforcement. Because the population of Blacks is not uniform throughout Michigan, a simple random sample from the overall list of people receiving safety belt citations would not result in the required number or proportion of Black respondents. Thus, a multi-stage clustering sample design was used that included five strata based on geographic location and the percentage of Blacks residing in the jurisdictions of the courts from whom safety belt citation data were obtained. A quota for each stratum was then set, based on the total population in that stratum and the chances that Blacks would be driving within the area. Table 1 shows the location of strata used for sampling, the percentage of Blacks in the population, the number of District/Municipal courts, and target number of completed interviews. Within each stratum, a fixed number of courts were selected randomly, proportional to the number of records in each court. Within each selected court, the same number of records was sampled across all selected courts within a stratum. A simple random sample was then used to sample the selected records within a stratum.

Table 1 Descriptive characteristics of the five strata used for telephone questionnaire sampling Stratum Location

1 2 3 4 5 Total

Black Number Target population of courts number of (%) respondents

Urban Detroit area 79.4 Southeastern Michigan 8.6 Urban southern Michigan 25.1 Rural southern Michigan 4.3 Northern Michigan 1.4

5 50 15 51 40

200 300 100 100 100

161

800

823

Telephone numbers for the sample were obtained through Telematch, a service that matches names and addresses with telephone numbers. A pretest of 30 interviews was conducted and minor changes were made to the wording of some questions. A total of 803 interviews were completed. Each telephone number was called up to six times. Based on equations from the American Association for Public Opinion Research (1998), the minimum response rate was 14.6% and maximum response rate was 29.6%. To account for the complex sample design, survey data were analyzed using IVEWare software (Raghunathan et al., 2002).

3. Results 3.1. Citizen harassment complaints This analysis was designed to answer the question: Did the frequency of safety-belt-related harassment complaints from citizens increase in the year following primary enforcement as compared to the year prior to primary enforcement? A total of 259 traffic-stop-harassment-related complaints were received for the 2-year period between 10 March 1999 and 9 March 2001. Of these, 43.6% (113) were from traffic stops in the year prior to primary enforcement of the safety belt law and 56.4% (146) were from the year after implementation of primary enforcement. Because this project was concerned with determining changes in the number of incidents of harassment resulting from the enforcement of the safety belt law, each complaint was analyzed for the presence of traffic-law enforcement activity and categorized based upon this activity. Safety-belt-related harassment complaints were separated out and analyzed separately (i.e. complaints involving a traffic stop in which a vehicle occupant was pulled over, warned, or cited for a lack of safety belt use). A total of 43 safety-belt-related harassment complaints were identified; 19 from the pre-year period and 24 from the post-year period. While this difference represents an increase in absolute numbers, it is possible that factors unrelated to police behavior toward citizens are responsible, such as an increase in the number of licensed drivers, number of safety belt citations written, or the number of traffic-stop-related harassment complaints in general. Table 2 shows the rates of safety-belt-related harassment complaints for the pre- and post-year study periods as a function of several measures. Table 2 Safety-belt-related harassment complaints rates for pre- and post-years Year

Per 10,000 safety belt/child safety seat citations

Per 1 million licensed drivers

Percent of total traffic-stop-related complaints

Pre Post

0.89 1.03

2.7 3.4

16.8 16.4

824

D.W. Eby et al. / Accident Analysis and Prevention 36 (2004) 819–828

As shown in Table 2, there were: (1) about 1 safety-beltrelated harassment complaint for every 10,000 safety belt and child safety seat citations issued by law enforcement in Michigan in both years; (2) about 3 safety-belt-related harassment complaints for every 1 million licensed drivers in Michigan each year; and (3) about 16.5% of all traffic-stop-related harassment complaints each year that were safety-belt-related. 3.2. Citation over-representation This analysis was designed to answer the question: Are certain groups receiving more citations than would be expected based upon their violation rate and presence in the driving population? This question was addressed with respect to race (White, Black, all other races), sex (male, female), and age group (16–22, 23–29, 30–64, ≥65 years). The statistical over/under a representation of safety belt citations for each group was analyzed by comparing the proportions of each group in the population of safety-belt-law violators to the proportions of these same groups in the population of people actually receiving safety belt citations. If no statistical difference between the proportions for a group was found, then we concluded that no safety-belt-citation over-representation had occurred for that group. On the other hand, if a safety-belt-violation proportion was statistically lower than the citation proportion, then we concluded that the group under consideration was receiving more citations than would be expected (over-representation) and was, therefore, experiencing differential treatment. This type of treatment, according to the study’s definition, is harassment. As described in Appendix A, Z-tests were used to compare proportions for each group. In order to reduce the chance of concluding erroneously that over-representation was present for a group, prior to analysis we set the significance level of our statistical test at P < 0.0001. Because the direct observation survey of safety belt use is conducted only during daylight hours, the violation rates in this analysis can only be applied to daylight hours. Therefore, all analyses used the

number of citations written during daylight hours only (see Eby et al., 2002a for more information on these analyses). Table 3 shows proportions of each group in the safety-belt-law violator group (PropV ), proportions of each group in the daylight-safety-belt-citation group (PropC ), the standard error of the difference between the proportions (SED ), and whether or not over-representation was present for the year before (pre) and after (post) primary enforcement. To conclude that over-representation was present, the difference between the proportions had to be statistically significant, and the proportion of citations had to be greater than the proportion of safety belt violators. Data from a previous direct observation study were used to determine violation proportions by sex and age in the pre-year (Eby et al., 2000a). Since we did not know the violation proportions for the race groups in the year prior to primary enforcement, we used the same proportions as found in the year following primary enforcement for these analyses. These are conservative estimates of the violation rates, since several studies (Eby et al., 2002b; Davis et al., 2002; Solomon et al., 2000; NHTSA, 2000a,b) have suggested that the implementation of primary enforcement may lead to a larger increase in safety belt use for Blacks than for Whites. Some significant differences between the proportions were found; for both years, males received citations at a rate significantly higher than would be expected based upon their violation rate, while women received citations at a rate significantly lower. In the pre-year, there were significant differences between proportions for both Whites and Blacks, with Blacks receiving more citations than would be expected based upon their violation rate (over-representation). In the post-year, however, there were no significant differences between proportions for either race. Note that too few people of “other race” were found in the direct observation study of violation rates to conduct analyses for that race category. The two older age groups in the pre-year are slightly different than those used in the post-year, because the only violation data available in the pre-year utilized slightly different age categories. The analysis by age showed significant

Table 3 Statewide proportions of safety-belt-law violators and citations by group before (pre) and after (post) implementation of standard enforcement Pre-year

Post-year

PropV (%)

PropC (%)

SED

Harassment

PropV (%)

PropC (%)

SED

Harassment

Sex Male Female

64.2 35.8

71.4 28.6

0.017∗ 0.017∗

Yes No

66.3 33.7

74.0 26.0

0.010∗ 0.010∗

Yes No

Race White Black

80.1 17.1

73.7 22.9

0.016∗ 0.016∗

No Yes

80.1 17.1

74.5 21.3

0.016 0.016

No No

Age (years) 16–29 30–59 (64) ≥60 (65)

33.6 57.2 9.2

51.5 44.6 3.9

0.022∗ 0.019∗ 0.011∗

Yes No No

32.8 59.4 7.9

50.0 47.5 2.5

0.009∗ 0.013∗ 0.008∗

Yes No No

∗

Significant at P < 0.0001.

D.W. Eby et al. / Accident Analysis and Prevention 36 (2004) 819–828

differences between proportions for all age groups in both years, with the youngest age group receiving more citations than expected (over-representation) for both years. 3.3. Perceived safety-belt-related harassment The final way we assessed harassment resulting from safety-belt-law enforcement was to examine the perceptions of people who have received citations for non-use of a safety belt. As described previously, we assessed perceptions of safety-belt harassment by conducting a telephone survey of drivers cited for a violation of Michigan’s safety belt law. Results were weighted to account for unequal probability of selection, response rate, and the demographics of those cited for a safety belt violation in Michigan. Results are presented for each question by all respondents, and by sex, race, and age. Note that because of University of Michigan Institutional Review Board requirements, only respondents 18 years of age or older were selected to participate in the study. The respondents were mostly male (73.2%). About three-fourths were White (76.4%), 19.4% were Black, and the rest were of some other race. About 25% were aged 18–22 years, 25% were aged 23–29 years, and about 46% were aged 30–64 years. The rest were 65 years of age or more. Slightly more than one-half had a high-school diploma or less and about 42% had completed some college or had a college degree. About 17% had a household income of less than US$ 25,000, 38% had a household income between US$ 25,000 and 49,999, and about 26% had a household income in the range of US$ 50,000–74,999. Respondents were asked about how professional the police officer behaved during the traffic stop that resulted in their last safety belt citation. As shown in Table 4, nearly 80% of respondents indicated that the officer acted in a “very professional” or “somewhat professional” manner during the stop. Blacks, people under 30 years, and people over 64 years of age appeared more likely than Whites and middle-aged

825

people to report unprofessional behavior. t-Tests, however, revealed that there were no significant differences on perceived professionalism by sex, race, or age. Several questions focused on whether respondents felt that they were singled out for the stop and safety belt citation on the basis of several different factors. Table 5 shows results for perceptions of being singled out on the basis of age, sex, race, and other factors. Note that each factor was asked about independently and that the respondent may have felt that he or she was singled out for more than one reason. Overall, about 16% of respondents thought they had been singled out on the basis of age. However, nearly 40% of the youngest age group reported being singled out for the traffic stop and citation because of their age. This proportion was significantly different than the proportions for the two middle-age groups. About 7% of respondents reported that they thought they were singled out because of their sex, with no significant difference among groups within any of the demographic categories. About 10% of respondents reported being singled out on the basis of their race. This judgment was significantly more frequent for Black and other race respondents than for Whites. We also asked respondents about several other factors that they believed might have led to their being singled out and cited for a safety belt violation (Table 5). Overall, about 7% of respondents felt that they were singled out because of the condition or type of their vehicle. This perception was more common among males, Whites, and other races but not significantly so. Very few respondents reported being singled out for personal reasons (such as a vendetta) or because they committed another traffic violation.

4. Discussion The study was designed to investigate whether changing Michigan’s safety belt law from secondary to primary enforcement resulted in police harassment. The study analyzed

Table 4 Percent and unweighted number of respondents perceptions of the professionalism of the officer who cited them for safety belt non-use Very professional

Somewhat professional

Somewhat unprofessional

Overall

45.9 (394)

32.9 (242)

14.0 (96)

7.1 (62)

Sex Male Female

43.1 (272) 53.6 (122)

33.9 (175) 30.4 (67)

16.5 (79) 7.2 (17)

6.5 (45) 8.8 (17)

Race White Black Other

45.6 (299) 48.6 (76) 39.5 (17)

35.6 (195) 21.1 (38) 34.8 (8)

11.0 (67) 25.6 (20) 19.4 (7)

7.8 (40) 4.7 (17) 6.4 (5)

Age (years) 18–22 23–29 30–64 ≥65

41.9 33.8 54.2 57.6

30.7 42.3 30.6 8.8

20.3 12.4 10.2 28.4

(66) (43) (254) (28)

(54) (40) (138) (9)

(16) (17) (56) (6)

Very unprofessional

7.2 11.6 5.0 5.2

(14) (13) (33) (2)

826

D.W. Eby et al. / Accident Analysis and Prevention 36 (2004) 819–828

Table 5 Percent and unweighted number of respondents who reported being singled out for a safety belt violation by reason for being singled out Demographic

Age

Overall

15.5 (99)

6.7 (67)

10.1 (69)

Sex Male Female

17.4 (79) 10.4 (20)

6.3 (49) 8.0 (18)

Race White Black Other

15.6 (69) 11.3 (20) 39.7 (10)

Age (years) 18–22 23–29 30–64 ≥65

39.8 9.2 5.0 14.7

∗ #

Sex

(56) (14)∗ (25)∗ (4)

Race

Personal

Other violation

7.2 (64)

0.8 (10)

1.9 (15)

12.1 (61) 4.6 (8)

8.8 (54) 2.8 (10)

1.1 (9) 0.0 (1)

2.6 (15) 0.0 (0)

4.8 (37) 10.4 (21) 34.4 (9)

1.6 (20) 42.4 (36)# 40.0 (13)#

7.0 (47) 8.1 (7) 4.0 (4)

1.0 (8) 0.0 (1) 0.1 (1)

2.2 (12) 0.1 (1) 0.7 (1)

12.6 4.7 4.4 8.0

14.7 7.0 8.7 8.0

0.8 0.5 1.0 2.0

3.3 0.0 2.3 0.1

(20) (12) (31) (4)

Vehicle

(16) (13) (37) (3)

7.0 11.6 5.4 0.0

(13) (17) (34) (0)

(2) (1) (6) (1)

(2) (0) (12) (1)

Significantly different from the 18–22 years age group at P < 0.01. Significantly different from Whites at P < 0.01.

statewide written citizen complaints arising from traffic stops to determine if the number of safety-belt-related harassment complaints changed with the introduction of primary enforcement. Safety-belt-related harassment complaints were very uncommon both before and after primary enforcement, with about 1 per year resulting from every 10,000 citations written, or 3 per year for every 1 million licensed drivers. Thus, there did not appear to be a change in the rate of safety-belt-related harassment complaints after implementation of primary enforcement. We also considered harassment from the perspective of safety-belt-citation over-representation, as was done in previous studies of traffic-law-enforcement harassment (e.g. Lambert, 1998; McGraw, 2000). These analyses showed that males received more citations than would be expected based on their violation rates, both before and after primary enforcement. Statewide analysis by race showed that Blacks were over-represented in citation activity prior to primary enforcement, but not after. Analysis by age showed that vehicle occupants under 30 years of age experienced citation over-representation before and after primary enforcement. However, no safety-belt-citation over-representation occurred in the year following primary enforcement when it was not present in the year prior to primary enforcement. Thus, we conclude that the implementation of primary enforcement did not lead to an increase in citation over-representation, and, therefore, safety-belt-related harassment by sex, race, or age. Indeed, for Blacks, primary enforcement may have lead to a decrease in safety-belt-related harassment. These results show that regardless of the enforcement provision of Michigan’s safety belt law (secondary or primary), both young people and males are over-represented for safety belt citations. By our study definition, this constitutes harassment. However, as discussed by Carter and Katz-Bannister (2000) such over-representation may, instead, constitute legitimate policing practice. These researchers point out that

profiling or “targeting” is a common and accepted practice used by law enforcement and is problematic only when it is not based upon objective crime trends. In relation to safety belt enforcement, Selective Traffic Enforcement Programs (STEPs) are supported by NHTSA as an effective way to increase safety belt use (see NHTSA, 1999b). STEPs are designed to target safety belt enforcement toward those groups that appear disproportionately in the unbelted crash statistics and that are most likely to violate the safety belt law; that is, certain populations are targeted for selective enforcement based upon objective crash and violation data. When non-objective data, such as stereotypes, are the basis for the targeted enforcement, then this targeting constitutes harassment. There is clear evidence that both males and young people violate Michigan safety belt law significantly more frequently than others (see Eby et al., 2000b, 2002b). As such, the over-representation found for males and young people may reflect, in whole or in part, the influence of STEPs in Michigan. On the other hand, until recently (Eby et al., 2002a), there were no objective statewide data showing significantly lower safety belt use for Blacks. Thus, the presence of over-representation in the year prior to primary enforcement for Blacks may indicate harassment. Alternatively, this finding may result from the use of conservative violation proportions by race in the year prior to primary enforcement. The study also assessed perception of harassment after implementation of primary enforcement. In general, respondents thought that officers acted professionally, with a vast majority of respondents reporting the officer’s behavior as somewhat or very professional. However, about 21% overall thought the officer’s behavior was unprofessional and about 30% of Black respondents thought that the officer’s behavior was unprofessional. Thus, while perceptions of officer behavior were generally positive among the population of people that received safety belt citations in Michigan, there is still a substantial proportion of Blacks who feel that they

D.W. Eby et al. / Accident Analysis and Prevention 36 (2004) 819–828

are not being treated professionally during enforcement of Michigan’s safety belt law. In order to further assess perceived safety-belt-related harassment, we asked respondents whether they felt they were singled out for the traffic stop because of their age, sex, race, or several other factors. About 16% of respondents indicated that they thought they were singled out because of their age. Those under 23 years of age quite frequently felt that they were singled out for this reason. About 7% of respondents thought they were singled out because of their sex, however, men and women did not significantly differ in this perception. About 10% thought they were singled out because of their race. About 42% of Blacks and 40% of other races reported this perception, whereas only about 2% of Whites felt that way. These results show that among the population of people receiving safety belt citations in Michigan, there is a sizeable minority of non-Whites who perceive safety-belt-related harassment. The study also showed that about 7% of respondents thought they were singled out for the traffic stop because of the appearance of their vehicle (condition, make, etc.). In conclusion, this study provides the first evidence that changing a state’s safety-belt-law enforcement provision from secondary to primary does not lead to increased police harassment. However, certain groups (specifically young people and Blacks) still perceived a moderate presence of harassment. These results suggest that states with secondary enforcement should continue their efforts to change to primary enforcement, but should attempt to educate both law enforcement and the public about the harassment issue. In addition, law enforcement agencies should ensure that their STEPs are based on objective data, such as crashes or safety belt non-use rates, for targeting population groups or areas.

Acknowledgements This work was sponsored by the Michigan Department of State through contract number 071B1001220. We thank Jean T. Shope, Fredrick M. Streff, and William Kennedy for their insights on this project. Hans Joksch assisted in the survey designs and analyses. The opinions, findings, and conclusions expressed in the report are those of the authors and not necessarily those of the Michigan Department of State.

Appendix A. Analysis methods for determining citation over-representation The direct observation survey of safety belt use was conducted during daylight hours only. It is unknown if safety belt use at night differs from daytime use. Therefore, the daytime safety belt use rates (which are used to obtain the proportions of violators) and citations written during daylight hours only are used in these analyses. To identify the court records of citations that were written during daylight

827

hours, the daily times of sunrise and sunset for southern Michigan were obtained for the pre- and post-years of the study from http://www.TimeandDate.com. The latest time of sunrise and the earliest time of sunset for a month were used to define the daylight time period for that month in this analysis, adjusted for daylight savings time. From the direct observation survey of safety belt use we obtained the following estimates of the proportion of safety belt violators of category a and the associated variance: Pa = proportion of safety belt violators of category a σa2 = var(Pa ) From court data, we have a sample of n citations with m going to people of category a. This gives us the following proportion (and variance) of people of category a who received citations: m P= n var(P) =

P(1 − P) n

The variance of the difference between the two proportions is σa2 + P(1 − P)/n and the ratio is:


Pa − P σa2

+ (P(1 − P)/n)

=Z

If we assume that Z is approximately normally distributed with a mean of 0 and a variance of 1, we can test the null hypothesis that Pa = P; that is, the proportion of people in category a not wearing safety belts is equal to the proportion of people being cited for safety belt violations by using a Z-test. References American Association for Public Opinion Research (AAPOR), 1998. Primary Definitions: Final Dispositions for Case Codes and Outcome Rates for RDD Telephone Surveys and In-Person Households Surveys. AAPOR, Ann Arbor, MI. Arrellano, A., 2000. When Race Adds up in Traffic. Detroit Free Press, 1 June. Bullers, F., 2000. ACLU Studies Citations for Racial Profiling. Kansas City Star, 27 January. Campbell, B.J., 1987. The Relationship of Seat Belt Law Enforcement to Level of Belt Use. University of North Carolina Highway Safety Research Center, Chapel Hill, NC. Carter, D.L., Katz-Bannister, A.J., 2000. Racial Profiling: Issues and Responses for the Lansing, Michigan Police Department. Lansing Police Department, Lansing, MI. http://www.lansingpolice.com/site/ profile/policy%20paper.pdf. Davis, J.W., Bennink, L., Kaups, K.L., Parks, S.N., 2002. Motor vehicle restraints: primary versus secondary enforcement and ethnicity. J. Trauma 52, 225–228. Eby, D.W., Fordyce, T.A., Vivoda, J.M., 2000a. Michigan Safety Belt Use Immediately Following Implementation of Standard Enforcement. Report no. UMTRI-2000-25. University of Michigan Transportation Research Institute, Ann Arbor, MI.

828

D.W. Eby et al. / Accident Analysis and Prevention 36 (2004) 819–828

Eby, D.W., Molnar, L.J., Olk, M.L., 2000b. Trends in driver and front-right passenger safety belt use in Michigan: 1984–1998. Accid. Anal. Prev. 32, 837–843. Eby, D.W., Kostyniuk, L.P., Molnar, L.J., Joksch, H., Vivoda, J.M., Miller, L.L., 2002a. The Effects of Standard Safety Belt Enforcement on Police Harassment: Year 2 Annual Report. Report no. UMTRI-2002-13. University of Michigan Transportation Research Institute, Ann Arbor, MI. Eby, D.W., Vivoda, J.M., Fordyce, T.A., 2002b. The effects of standard enforcement on Michigan safety belt use. Accid. Anal. Prev. 34 (6), 101–109. General Accounting Office (GAO), 2000. Racial Profiling: Limited Data Available on Motorist Stops. GAO, Washington, DC. Lambert, J., 1998. Driving While Black: A Statistician Proves that Prejudice Still Rules the Road. Washington Post, 16 August. Lang, J.E., Blackman, K.O., Johnson, M.B., 2001. Speed Violation Survey of the New Jersey Turnpike: Final Report. Office of the Attorney General, Trenton, NJ. Lund, A.K., Pollner, J., Williams, A.F., 1986. Preliminary Estimates of the Effects of Mandatory Seat belt Laws: Science Versus Politics. Insurance Institute for Highway Safety, Washington, DC. McGraw, S., 2000. Blacks More Likely to be Searched, Cops Find. APBNews.Com, 21 July. Moffat, J.M., 1998. Secondary enforcement: an inferior and deadly law. Traffic Saf. 98 (6), 14–15. National Highway Traffic Safety Administration (NHTSA), 1997. Presidential Initiative for Increasing Seat Belt Use Nationwide: Recommendations from the Secretary of Transportation. US Department of Transportation, Washington, DC. National Highway Traffic Safety Administration (NHTSA), 1999a. Standard Enforcement Saves Lives: The Case for Strong Seat Belt Laws. Report no. DOT-HS-808-846. US Department of Transportation, Washington, DC.

National Highway Traffic Safety Administration (NHTSA), 1999b. Occupant Protection Special Traffic Enforcement Programs Evaluated in 20 States. Traffic Tech: Traffic Technology Series no. 189. US Department of Transportation, Washington, DC. National Highway Traffic Safety Administration (NHTSA), 2000a. Seat Belts and African Americans. US Department of Transportation, Washington, DC. National Highway Traffic Safety Administration (NHTSA), 2000b. Standard Enforcement Saves Lives: The Case for Strong Seat Belt Laws. Retrieved 17 May 2000, from http://www.nhtsa.gov/people/ injury/airbags/seatbelt/Challenges.htm. National Highway Traffic Safety Administration (NHTSA), 2001. Standard Enforcement Saves Lives: The Case for Strong Seat Belt Laws. Retrieved 13 June 2001, from http://www.nhtsa.dot.gov/people/ injury/airbags/seatbelt/default.htm. Raghunathan, T.E., Solenberger, P.W., Van Hoewyk, J., 2002. IVEWare: Imputation and Variance Estimation Software Users Guide. Survey Research Center, Institute for Social Research, Ann Arbor, MI. Solomon, M.G., Cosgrove, L.A., Preusser, D.F., 2000. A summary of results from studies measuring the change from secondary enforcement of safety belt laws to primary enforcement emphasizing the effects on race. In: Proceedings of the Presentation at the National Highway Traffic Safety Administration’s Restraint Use Strategy Workshop, Crystal City, VA. Solomon, M.G., Preusser, D.F., Nissen, W.J., 2001. Evaluation of Maryland, Oklahoma, and the District of Columbia’s Seat Belt Law Change to Primary Enforcement. Report no. DOT-HS-809-213. US Department of Transportation, Washington, DC. Ulmer, R.G., Preusser, C.W., Preusser, D.F., 1994. Evaluation of California’s Safety Belt Law Change to Primary Enforcement. Report no. DOT-HS-808-205. US Department of Transportation, Washington, DC. US Bureau of the Census, 1992. 1990 Census of Population and Housing. University of Michigan UM-Library Gopher-Computer Database.