Combined High-Visibility Enforcement: Determining the Effectiveness (2024)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

28 Description of Data Used for Evaluation of HVE Agencies collect a variety of metrics for high-visibility enforcement (HVE) campaigns, as illustrated in Figure 4-1. Agencies typically have two objectives for gathering data during campaigns. First, agencies are required to report productivity to NHTSA or other funding organizations to show the amount of overtime funded or how other funds were used. Second, metrics are collected to evaluate the effectiveness of HVE. Productivity metrics are usually collected during the campaign. Common productivity metrics include the following: • Agencies/officers involved, • Number and type of arrests, • Number and type of citations/warnings, • Number and type of crashes, • Number of driver contacts, • Number of media spots, • Number of stops, • Number of assists, • Officer hours/overtime hours, and • Cost of equipment used [e.g., dynamic speed feedback signs (DSFS) trailer]. As noted in Chapter 3, the project team conducted a survey to gain additional information on how agencies manage their HVE campaigns. Representatives from 29 states responded. One question focused on what metrics are normally collected during HVE campaigns. Respondents indicated that the most frequently collected metrics included arrests, citations, number of stops, and number of officers involved. The number of paid media spots was another often-collected metric, followed by the number of unpaid media spots. The second reason agencies collect data is to assess whether the campaign was effective in improving safety (e.g., fewer impaired drivers, increased seat belt use). Many of the productivity metrics (e.g., citations, crashes) can be used to assess effectiveness. However, assessing effective- ness usually requires collection of additional data before the campaign or immediately after. In some cases, similar data are also collected at a location where the campaign would not have had an impact (control location). Before/after or control data can be used as a baseline to determine whether a metric increased or decreased due to the campaign. Other metrics are also used to assess the effectiveness of a campaign. These include speed, seat belt use, cell phone use, whether drivers are stopping or yielding, etc. Some are collected through passive observational studies, which involve an observer stationed near a roadway or facility (e.g., a gas station) to record behaviors such as the number of drivers using a cell phone, number of drivers using a seat belt, number of vehicles left in a bar parking lot, etc. (NHTSA n.d.). C H A P T E R 4

Description of Data Used for Evaluation of HVE 29   A number of campaigns use driver surveys to assess driver behaviors and attitudes. These data are subjective since they measure a driver’s attitudes or self-reported behavior (e.g., seat belt use). Subjective measures of evaluation are typically those used to assess changes in attitude or acceptance of HVE activities. For instance, a survey question may ask whether the respondent remembers hearing any media messages or noticed additional enforcement (NHTSA n.d.). In many cases, the surveys are conducted at department of motor vehicles offices or other con- venient locations near campaign locations. Eleven of the studies (92%) outlined in Chapter 2 utilized a survey of drivers to assess the effectiveness of the HVE campaigns. The most common metrics collected by agencies in HVE campaigns are provided in Table 4-1. 4.1 Description of Data Utilized in Analyses Each of the following sections describes a specific type of data that may be utilized to assess the effectiveness of HVE campaigns. This includes a description of types and sources of data, as well as the types of analyses best suited to each type of data. This information is provided for agencies that may be interested in obtaining other data sources to evaluate the effectiveness of their own Enforcement Activity Data Citation, warning and complaints data (e.g., TraCS), agency man-hours, law enforcement activity logsDriver BAC Levels Crash data, citation data, checkpoints Public Awareness Public surveys, media contacts HVE Safety Analysis Data Sources Seat Belt Usage Data Seat belt surveys, crash data, citation data Crash Data State motor vehicle crash databases, FARS, GES, HSIS, MCMIS, insurance industry databases Speed Data Vehicle probe data, sensor data, field collected speed data (e.g., hand-held radar gun, pneumatic tubes, multilane radar traffic data collectors) Figure 4-1. Data needs.

30 Combined High-Visibility Enforcement: Determining the Effectiveness Metric Examples Theme Collection Method Agencies involved Enforcement, metropolitan planning organization (MPO), department of transportation (DOT) All Agency Arrests Operating while intoxicated (OWI), outstanding warrants All Officer reported Attitude toward enforcement activities Do drivers favor use of checkpoints to enforce OWI laws? All Survey Blood alcohol level sampling Number of drivers with high levels of breath alcohol concentration (BrAC) Impaired Officer reported Cell phone use Handheld, manipulation Distraction Officer reported, observational data collection, survey Citations OWI, speed, seat belt All Officer reported Crashes Usually select crash type specific to intervention All State database, local agency Delay through checkpoint Average, seconds per vehicle All Electronic data collection, observers Driver awareness of the campaign Driver recalls seeing sign, driver has noticed enforcement All Survey Driver contacts Officer interactions with drivers All Officer reported Media spots (paid) TV ad, radio ads, print All Agency reported Media spots (unearned)* Times mentioned on radio show, press present at checkpoints All Review of television, newspapers, and radio Motorist assists Officer assistance with stalled vehicle All Officer reported Officer hours Regular, overtime All Agency Perception of risk of citation Driver perception of likelihood of being stopped for OWI All Survey Seat belt use Percent of drivers wearing seat belts, percent of child restraints properly used Seat belt use Officer reported, observational data collection, survey Seizures Stolen vehicle All Officer reported Speed Spot speed Aggressive Electronic data collection, observers Stops Number of stops for impaired, OWI All Officer reported Volume Vehicles passing checkpoint, vehicles passing signing All Electronic data collection, observers Visibility Number of drivers who read, saw, or heard about campaign; number of drivers seeing signs All Survey Warnings Seat belt, registration All Officer reported Yielding Number of vehicles yielding to pedestrians in crosswalk Yielding Observers *Unearned media are spots where the campaign is mentioned outside of purchased spots (e.g., an unsolicited mention in a morning radio show). Table 4-1. Common metrics collected for HVE campaigns. campaigns. Additionally, each section describes what data the team had available to evaluate the frameworks in Chapters 6 through 9. The project team conducted four different types of analyses to assess the effectiveness of HVE, especially combined HVE. A detailed description of each is outlined in the corresponding chapters. They include the following: • Data visualization (Chapter 6): Uses graphics to display data to show relationships and data- driven insights and can show trends, patterns, outliers, or progress toward a performance metric. • Simple before-and-after analysis (Chapter 7): Collects data on metrics of interest before an HVE campaign is conducted as well as after each wave and/or at the conclusion of the

Description of Data Used for Evaluation of HVE 31   campaign. Then, the impact of the campaign is evaluated by comparing the increase or decrease in metrics of interest. • Classical statistical evaluations (Chapter 8): Includes methods such as full Bayes, empirical Bayes, logistic regression, generalized linear models, etc. These methods can be used to com- pare outcome measures from before, during, and/or after an intervention. Rather than simply comparing the outcome measures from the two time periods, these methods typically use comparison groups with similar characteristics and can adjust for potential biases such as regression to the mean. • Spatial/temporal analyses (Chapter 9): Techniques that use geographic characteristics such as location and geometry to explain the relationships among data. Geospatial analysis encompasses a range of statistical techniques that are used to quantify the geographical rela- tionships between incidents and characterize the spatial and environmental factors associated with them. Temporal analyses are able to evaluate trends over time. When conducting tempo- ral analyses, a sufficient number of months (or other unit of time) should be used, since short time frames are not likely to show trends. 4.2 Basic HVE Campaign Metrics 4.2.1 Description of Metrics Agencies typically collect the previously listed productivity metrics during campaigns (i.e., officer resources, arrests/stops, citations/warnings, crashes, media spots, officer hours/ overtime hours). These data are typically collected by the participating agencies based on reporting requirements. 4.2.2 HVE Campaign Data Utilized for Framework Evaluations Two different campaign data sets were obtained and used in examples for the various frameworks. 4.2.2.1 Iowa sTEP Campaign Data Iowa conducts annual HVE campaigns termed the Special Traffic Enforcement Program, or sTEP. Campaigns are conducted on the federal fiscal year (FFY) calendar—October to September. The research team met with several members of the Iowa Department of Public Safety’s Governor’s Traffic Safety Bureau (GTSB) to determine availability of the data for project needs. The Iowa GTSB provided calendars and productivity data for its sTEP program for fiscal year (FY) 2017 to FY 2022 (Figures 4-2 and 4-3). More than 176 law enforcement agencies participated in HVE efforts during this 6-year period. Approximately five waves occur annually. The sTEP efforts typically focus on one behavior at a time, usually seat belt use or impaired driving. Data included the following, and an example of the data provided was previously shown in Figure 4-2: • Campaign dates, • Nature of the HVE efforts (e.g., general, impaired driving, seat belt use), • Agencies involved, • Officers worked, • Overtime officer hours, • Number and types of citations and warnings, • Number and type of arrests,

32 Combined High-Visibility Enforcement: Determining the Effectiveness Source: Iowa GTSB. Figure 4-2. Typical data collected and reported in Iowa sTEP campaigns.

Description of Data Used for Evaluation of HVE 33   Source: Iowa GTSB. Figure 4-3. FY 2022 sTEP calendar.

34 Combined High-Visibility Enforcement: Determining the Effectiveness • Motorist assists, • Vehicle inspections, • Crashes (total and by severity—property damage only, injury, fatality), • Before-and-after seat belt surveys, • Number of media advertisements by format (e.g., print, television), • Interdictions/canine searches, • Motor Carrier Safety Assistance Program inspections, • Warrants served, • Costs for equipment and overtime, and • Type of equipment used by agencies (e.g., radar, DSFS trailer). With the exception of the seat belt survey, these data only include those collected during the campaign. The project team primarily used violations, citations, arrests, and media data for the data visualization framework (Chapter 6). The team also included media contacts as an independent variable in the crash analyses. It should be noted that law enforcement agencies only report GTSB overtime activity during sTEP campaigns. The number of media contacts was available for each agency for each sTEP wave (included in the data from the GTSB) for FY 2017, FY 2018, FY 2019, and FY 2022. The team determined the number of media contacts by county by summing up media contacts for all agencies within each county for each time period. The team assumed media contacts would have had an impact beyond the city itself. Media contacts included TV spots, radio spots, printed material (i.e., flyers), and digital media. 4.2.2.2 Maryland HVE Campaign Data The Maryland Department of Transportation (DOT) Motor Vehicle Administration’s High- way Safety Office (MHSO) is the central point of coordination for safety programs at all levels of government and the private sector within Maryland. The agency receives funding from NHTSA for use at the statewide and local levels. The MHSO submits its plan for allocating these funds to NHTSA by way of its highway safety program (HSP), utilizing formulas and strategic planning models. Funds are allocated to jurisdictions and grant-funded projects that meet the state’s traffic safety goals, as outlined in the state’s HSP. The plan outlines the upcoming activities and priority areas for the next FY. The MHSO provides an HVE calendar (Figure 4-4) to grant-recipient law enforcement agencies, so they can plan operations and actively conduct enforcement during HVE waves throughout the FY. The MHSO law enforcement services staff work closely with local police agencies around the state to maximize the impact of traffic safety enforcement programs. They do this through data analysis utilizing websites dedicated to Vision Zero and other state highway safety information, resources, and initiatives, for instance state crash data and citation data. Citations are issued throughout the campaigns. The citation data are recorded in a grant activity tracking system, which also compiles year-end results for all law enforcement partners that participated in the federally grant-funded HVE waves. Crash report data are recorded in a similar fashion and maintained by the Maryland State Police, and traffic records citation data are maintained by the Motor Vehicle Administration, housed within the MHSO. Table 4-2 presents Maryland’s annual highway safety enforcement calendar, with different weeks each month designated for the indicated HVE effort. An example month was presented in Figure 4-4. The MHSO provided citation data consisting of output reports from the aforementioned grant tracking system. Due to changes in data extraction and grant activity tracking, year-to-year per- formance comparisons were not advised.

Description of Data Used for Evaluation of HVE 35   Data collected during the campaign waves, specifically, grant performance totals for law enforcement citations, included the following for FFY 2019, FFY 2020, FFY 2021, and FFY 2022: • Participating agency name. • Total number of citations for the following: – Belt, – Cell phone, – Child restraint, – Criminal arrests, – Driving under the influence (alcohol, drugs), – Pedestrian, – Speed, – Texting, and – Vehicle contacts. A total of 62 local law enforcement agencies were involved in FFY 2022. Results are displayed as a year-end summary report. 4.2.3 Utility of Campaign-Collected Data Most data collected by agencies during campaigns are used to measure productivity, since metrics are frequently only collected in relation to the campaign. As a result, there is often Source: MHSO 2023. Figure 4-4. FFY 2021 HVE wave calendar.

36 Combined High-Visibility Enforcement: Determining the Effectiveness Month Enforcement Initiatives January Impaired Driving February Impaired Driving Distracted Driving Seat Belt/Occupant Protection March Impaired Driving Pedestrian Safety Speed/Aggressive Driving April Distracted Driving Pedestrian Safety Seat Belt/Occupant Protection May Distracted Driving Impaired Driving Seat Belt/Occupant Protection Speed/Aggressive Driving June Distracted Driving Impaired Driving Seat Belt/Occupant Protection Speed/Aggressive Driving July Impaired Driving Speed/Aggressive Driving August Distracted Driving Impaired Driving Seat Belt/Occupant Protection Speed/Aggressive Driving September Distracted Driving Impaired Driving Pedestrian Safety Seat Belt/Occupant Protection October Distracted Driving Impaired Driving Pedestrian Safety Seat Belt/Occupant Protection November Impaired Driving Pedestrian Safety December Impaired Driving Data source: https://zerodeathsmd.gov/law-enforcement/law- enforcement-calendar. Table 4-2. Maryland yearly highway safety enforcement. not sufficient data to conduct evaluations of the effectiveness of campaigns. However, use of data visualizations, as described in Chapter 6, can show patterns in the data that may be useful for agencies. Metrics such as speed studies, seat belt surveys, or consumer surveys may also be collected in conjunction with campaigns but are covered in subsequent sections. 4.3 Crash Data 4.3.1 Description of Crash Data Crash data are a direct measure of safety. Most states and large agencies maintain electronic crash data. Several variables are usually associated with a crash database. Some variables relevant to evaluation of HVE include location, severity, type (e.g., rear-end, head-on), major contribut- ing factors (e.g., impairment, speeding), and seat belt use. 4.3.2 Crash Data Utilized for Evaluation of Combined HVE 4.3.2.1 Iowa Crash Data Since the crash analyses conducted for this project required a number of different data sets, the data collection and reduction involved a significant amount of resources. As a result, the research team only used Iowa data in crash analyses.

Description of Data Used for Evaluation of HVE 37   Traffic and Criminal Software (TraCS) is predominantly used in the state of Iowa to collect crash and citation data. The Iowa DOT indicated that 387 agencies are currently using TraCS in the state and that TraCS captures 99.7% of reported crashes and 92% of citations. TraCS includes a geolocation tool, which is used to geocode all reportable crashes on public roads in Iowa (approximately 115,000 miles). A reportable crash in Iowa is defined as “one in which all damages (vehicle and property) are combined and estimated to be $1,500 or more, and/or an injury or fatality has occurred” (Iowa DOT Form 433033) (Iowa DOT 2012). The Iowa DOT Traffic and Safety Bureau makes current year-to-date crash data and the prior 10 years of crash data available to the public via several different methods, such as the Iowa Crash Analysis Tool (https://icat.iowadot.gov/). The geocoded crash data received through this channel include reported crashes, as well as several derived attributes and unit- and person-level details, excluding personally identifiable information (PII). The project team obtained crash data for FY 2017, FY 2018, FY 2019, and FY 2022 but excluded FY 2020 and FY 2021 due to the COVID-19 pandemic. Then, the team determined dates for each sTEP campaign for the given years, which allowed them to designate the during period of a cam- paign and note the dates between each sTEP campaign. The team further identified the between periods as before the proximate campaign wave, which consisted of the 10 days prior to the cam- paign, or after the proximate campaign, which consisted of the 10 days following the campaign. The before period reflected a baseline, and the after period allowed the team to test for a residual effect immediately after the campaign. The project team modeled data for crash and violation analyses at the county level rather than jurisdiction level for two reasons: (1) in many cases, the county sheriff ’s office participated in the sTEP campaign, so the team expected that enforcement could have occurred anywhere in the county, and (2) due to privacy concerns, the team agreed not to provide results at the level of individual jurisdictions. Iowa has 99 counties, so this provided a large number of data points for the analyses. The sTEP data provided by the Iowa GTSB included all agencies that participated in each wave. Then, the team determined the number of participating jurisdictions for each county for each wave. The team noted each county as participating or not participating for each sTEP period for FY 2017, FY 2018, FY 2019, and FY 2022. If some agencies within a county participated in one campaign but not all, the team determined percent participation by vehicle miles traveled (VMT) as described in Section 4.8.2.1. The project team used crashes as the dependent variable in the framework analyses and extracted several different categories of crashes to be included as dependent variables (modeled separately) using the following: • Total (all crashes). • Impairment-related (crashes that are alcohol- or drug-related). • Speeding-related (crashes indicated as speeding or speeding-related). • Seat belt-related (crashes where any passenger or driver is unbelted). – Due to the complexity in determining whether passengers were unbelted, only unbelted drivers were included. • Distraction-related (crashes involving a distraction such as texting/using mobile phone while operating commercial vehicle, use of electronic communication device, etc.). The team extracted crash data for each time period for each county and included all severity levels. It included all severity levels because the number of severe crashes was small when

38 Combined High-Visibility Enforcement: Determining the Effectiveness modeled for the short time periods before, during, and after each campaign. An example of the aggregated data is shown in Table 4-3. 4.3.3 Utility of Crash Data for Evaluation of HVE Crash data are widely collected and reported by agencies during HVE campaigns. However, crashes are rare and random events. As a result, it is unlikely that the impact of HVE can be deter- mined using short periods such as a single period before and after a campaign. Most crash analyses use multiple years of data and statistical models that account for issues such as regression to the mean. Crash analyses can be conducted but are best analyzed using robust statistical techniques such as those outlined in Chapters 8 and 9. Regression to the mean describes the situation where variables that are much higher or much lower than average in one period tend to move closer to the average in the next period. In terms of crash analyses, this means a period of high crashes in one period is likely to be followed by fewer crashes in the next period (Nikolopoulou 2023). Examples of how crash data were used to compare effectiveness of an HVE campaign are provided in Sections 8.7.3, 8.7.5, and 9.6. * OT = overtime hours. County FY Period Start Date End Date Total Speeding Distraction Impaired Unbelted County X 2017 Normal 10/01/2016 10/31/2016 22 3 16 2 1 County X 2017 Normal 11/01/2016 11/19/2016 12 5 0 6 1 County X 2017 Seat Belt OT* 11/20/2016 11/27/2016 6 0 0 1 0 County X 2017 Normal 11/28/2016 11/30/2016 2 2 0 0 0 County X 2017 Normal 12/01/2016 12/31/2016 24 8 2 2 1 County X 2017 Normal 01/01/2017 01/31/2017 14 7 10 2 2 County X 2017 Normal 02/01/2017 02/28/2017 12 2 2 5 0 County X 2017 Normal 03/01/2017 03/14/2017 5 2 0 0 0 County X 2017 Impaired OT* 03/15/2017 03/18/2017 1 0 2 0 0 County X 2017 Normal 03/19/2017 03/31/2017 4 0 0 0 0 County X 2017 Normal 04/01/2017 04/30/2017 16 0 3 2 0 County X 2017 Normal 05/01/2017 05/21/2017 10 2 2 0 0 County X 2017 Seat Belt OT* 05/22/2017 05/31/2017 4 1 0 0 0 County X 2017 Seat Belt OT* 06/01/2017 06/04/2017 4 0 0 0 0 County X 2017 Normal 06/05/2017 06/30/2017 12 0 0 0 1 County X 2017 Impaired OT* 07/01/2017 07/04/2017 4 0 2 0 0 County X 2017 Normal 07/05/2017 07/31/2017 17 1 13 0 1 County X 2017 Normal 08/01/2017 08/17/2017 8 0 0 0 0 County X 2017 Impaired 08/18/2017 08/31/2017 6 1 5 0 0 Table 4-3. Example of reduced crash data.

Description of Data Used for Evaluation of HVE 39   4.4 Statewide Citation/Violation Data 4.4.1 Description of Citation/Violation Data A traffic citation is the written notice (ticket) given by an officer when a driver has committed a traffic violation. A violation is the offense committed by the driver when a citation is issued (e.g., failure to stop at a stop sign). In some cases, the terms citation and violation are used interchangeably. The number of citations and the associated violations issued during an HVE campaign are commonly collected metrics. Most agencies use this for reporting productivity. Citations can be aggregated by violation type (e.g., speeding, distraction, seat belt, impaired) to determine how officers targeted resources. If citation data are collected before a campaign, they can also be used to compare whether citations increased during the campaign and what violations were targeted. Most agencies participating in an HVE are required to tabulate and report citations issued during the campaign. As a result, this information is usually readily available. Obtaining data for other periods may be more challenging, especially if multiple agencies participate. 4.4.2 Citation/Violation Data Utilized for Evaluation of Combined HVE 4.4.2.1 Iowa Statewide Citation/Violation Database As introduced in Section 4.3.2, Iowa TraCS is used throughout Iowa to collect citation data. Data are available for all 99 counties in Iowa. Similar to crash data, current year-to-date citation data and the prior 10 years of citation data are available for most agencies. Example attributes reported include code section violation and description, the user recording the citation, and the date and time of the violation. Some agencies, such as the Iowa State Patrol and Iowa Motor Vehicle Enforcement Agency, also geocode their citation, warning, and inspection data. Traffic-related citation data in Iowa are recorded by the Iowa Department of Human Rights Division of Criminal and Juvenile Justice Planning. The division’s Justice Data Warehouse, specifi- cally the Easy Access to Adult Criminal Data system (https://disposedcharges.iowa.gov/), provides public access to disposed charges and convictions. More detailed aggregate data (e.g., at the county level) are currently available for 2010 through 2022. Data may be refined by various categories, including crime subtype, through which operating while intoxicated (OWI) and traffic violations may be identified. However, the type of traffic charge is not available. The research team obtained disaggregated non-PII statewide violation data from the Iowa Depart- ment of Human Rights Division of Criminal and Juvenile Justice Planning. Approval was provided by the state court administrator at the Iowa Judicial Branch. The team obtained data for FY 2017, FY 2018, FY 2019, and FY 2022. The data contain all traffic-related violations and include the following: • Year, • Arresting agency pin, • Arresting agency name, • Case ID, • County name, • Initiated date, • Offense date, • Charge count, • Charge code, • Charge description, • Disposition date, and • Disposition. Disposition date and outcome were included in the violation data. Disposition is the outcome of the violation and may include vacated (charge withdrawn), acquitted (found not guilty), convicted

40 Combined High-Visibility Enforcement: Determining the Effectiveness (found guilty), dismissed (charges do not move forward), etc. As a result, some citations are given that do not result in a penalty. For instance, if a speeding citation is dismissed, the driver does not pay the fine, and the violation is not recorded on the person’s driving record. The team discussed whether to include all citations or only those that resulted in a penalty, and it decided to keep all citations, since all would have resulted in a stop and are still likely to have some impact on the driver and potentially surrounding traffic (perception of risk). Additionally, HVE campaigns report citations given rather than tracking outcome through the system. The project team used citations as a dependent variable in the analyses or as an explanatory variable in models using crashes as the dependent variable. The team used relevant categories of violations associated with citations and determined the violations that fit under each category in conjunction with a former Iowa law enforcement officer and include the following: • Total (all traffic violations), • Impairment-related (alcohol- or drug-related), • Speeding-related (drag racing, eluding—speed 25 mph over limit, etc.), • Seat belt-related (failure to maintain safety belt, failure to use seat belt—minor, failure to secure child, etc.), and • Distraction-related (texting/using mobile phone while operating commercial vehicle, use of electronic communication device, etc.). The team extracted citation data by time period and by county with the following violation categories: total, distraction, seat belt, and speeding. The team further designated time periods during the corresponding FY using the sTEP calendar provided by the Iowa GTSB. The team also noted the number of days during each period to normalize the data. For instance, if a normal period (i.e., not proximate to an HVE campaign) was 15 days, the number of violations could be reported as violations by day, which allowed data from time periods with varying days to be normalized. Additionally, the team extracted citation data by type of violation (e.g., impairment). Violation data included any traffic violation issued in the state of Iowa for FY 2017, FY 2018, FY 2019, and FY 2022. The team did not use FY 2020/2021 data due to the impact of COVID-19. Figure 4-5 shows an example of the percent of violations by category for FY 2017, FY 2018, and FY 2019. Figure 4-5. Type of traffic violations issued.

Description of Data Used for Evaluation of HVE 41   An example of the data aggregated by period for one county is shown in Table 4-4. For instance, from October 1–31, 2016, 200 total violations were reported (average of 6.45 per day). During this period, no HVE campaigns occurred (normal). The first HVE campaign (Click It or Ticket) occurred for the period of November 20–27, 2016, with 106 violations issued and an average of 13.25 violations issued per day. 4.4.3 Utility of Citation/Violation Data for Evaluation of HVE Citation/violation data are often recorded during HVE campaigns. In this context, they are used to assess productivity and for reporting. Citation/violation data collected during a campaign can be evaluated to determine whether officers targeted violations according to the theme (e.g., whether seat belt violations/citations increased during a wave focused on seat belt use). Citation/violation data can be obtained for a longer period of time for a particular jurisdiction, and then trends can be evaluated (for instance, if the number of impaired citations/violations issued is increasing or decreasing over time compared to other types of citations/violations). Citations/ violations can also be used to determine effectiveness of an HVE, but one should consider whether an increase or decrease in citations is the desired impact. Officer hours usually increase during HVE campaigns (overtime), so agencies may wish to determine which violations were more likely to be given during the campaign compared to a period before the campaign or how many more citations were given during a campaign versus normal periods. However, increased enforcement does not necessarily result in increased citations. Ideally, highly visible enforcement results in improved driver behaviors. Improved behaviors would lead to fewer citations. As a result, an analysis could show that seat belt violations decreased after several HVEs targeted seat belt use, suggesting drivers were more likely to wear seat belts. In this case, a decrease in citations suggests behaviors had improved, leading to a safer condition. However, citations can decrease or increase for a number of reasons unrelated to HVE, such as decreased agency resources, a need to focus on non-traffic safety issues, etc. In addition to analyses of citation data, the number of citations/violations can be included in crash analyses as an independent variable to show the relationship between increased citations and decreased crashes. Examples of how citation/violation data were used to compare effectiveness of an HVE campaign are provided in Section 6.9.5 and 7.8.4. County Code FY Period Start Date End Date Type of Violations All Distraction Impaired Seat Belt Speeding County X 2017 Normal 10/01/2016 10/31/2016 200 – 2 9 149 County X 2017 Normal 11/01/2016 11/19/2016 124 – 4 5 75 County X 2017 Seat Belt OT* 11/20/2016 11/27/2016 106 – 4 4 72 County X 2017 Normal 11/28/2016 11/30/2016 15 – 1 – 13 County X 2017 Normal 12/01/2016 12/31/2016 173 – 5 5 112 County X 2017 Normal 01/01/2017 01/31/2017 228 2 6 7 148 County X 2017 Normal 02/01/2017 02/28/2017 275 1 9 9 173 County X 2017 Normal 03/01/2017 03/14/2017 124 – 4 1 88 County X 2017 Impaired OT* 03/15/2017 03/18/2017 19 – – 1 15 * OT = officer overtime hours. Table 4-4. Example violation data for county code County X.

42 Combined High-Visibility Enforcement: Determining the Effectiveness 4.5 Speed Data 4.5.1 Description of Speed Data High-visibility speed enforcement campaigns are common, with the majority of states providing some type of funding for speed equipment, overtime enforcement targeted to speeding, or public information messages targeted to speeding as shown in Figure 4-6 (Venkatraman et al. 2021). Speed data can be collected through a variety of methods. Most states have automatic traffic recording (ATR) stations, which are set up in permanent locations to continuously collect data. ATR stations primarily collect traffic volume data, but many are also configured to collect speed data. Traffic data collection devices, which collect speed and volume data, are also commercially avail- able and include road tubes or radar/light detection and ranging (lidar)/camera-based systems. While enforcement agencies are not likely to have these types of devices, data may be obtained from other agencies (e.g., state DOTs). Many enforcement agencies use trailer-mounted DSFS, which are positioned at a particular location to remind drivers of their speed (Figure 4-7). Most trailer-mounted DSFS have the capability to record speed data. DSFS can be placed in stealth (blank) mode to collect data before an HVE campaign or at a location not impacted by the campaign. Many officers are already familiar with use of radar or lidar speed guns, which can be used to unobtrusively collect speed before a campaign, during a campaign, or at a location not impacted by the campaign. 4.5.2 Speed Data for Evaluation of Combined HVE 4.5.2.1 MHSO Campaign Speed Data The MHSO (2022) conducted a rural speed management pilot program along MD 367 in Bishopville, Maryland, in 2021. The corridor is a rural undivided two-lane road with 6,260 average daily traffic (ADT), original speed limits of 35 to 50 mph, and 85th percentile speeds ranging from 8 to 15 mph over the posted speed limit. A control location was selected along MD 68, also an Source: NHTSA 2023a. Figure 4-6. Speed messaging.

Description of Data Used for Evaluation of HVE 43   undivided two-lane road with speed limits of 30 to 50 mph that has similar length and geography to MD 367 and an ADT of 3,610. Speeds along the control section varied from 12 to 18 mph over the posted speed limit. The control site was sufficiently far away from the test corridor that the agency did not expect any spillover from enforcement or education tactics. Speed limits were adjusted along the test corridor of MD 367; additionally, the test site included lane narrowing and the placement of DSFS as well as HVE. Speed studies were collected along the MD 367 corridor as well as the control location. Speed data were collected before the cam- paign period, during the campaign, and then after the campaign concluded. A number of speed metrics and a sample size were provided (Hu and Cicchino 2023). The project team utilized these data in several analyses as described in Chapters 6 and 7. 4.5.3 Utility of Speed Data for Evaluation of HVE Speed data are well suited for evaluation of the impact of HVE, particularly speed-focused cam- paigns. Speed data can be used to compare changes in speed before and after a campaign or to compare speeds at enforcement locations with locations not included in the HVE. Analysis of speed data is fairly straightforward, since simple before-and-after tests (i.e., t-tests) can be used to compare differences. Examples of how speed data were used to compare effectiveness of an HVE campaign are provided in Sections 6.9.4, 6.9.5, and 7.8.2. 4.6 Seat Belt Use Data 4.6.1 Description of Seat Belt Use Data Increase in seat belt use is a commonly collected metric for HVE campaigns. Seat belt data are frequently collected through observational studies, where observers unobtrusively collect seat belt use. In most cases, data are collected before and after HVE campaigns. 4.6.2 Seat Belt Use Data for Evaluation of Combined HVE 4.6.2.1 Iowa Seat Belt Data Seat belt use data have historically been collected before and after each Iowa sTEP campaign wave. The project team aggregated data campaign-wide and used data for FY 2017, FY 2018, and Figure 4-7. Trailer-mounted DSFS.

44 Combined High-Visibility Enforcement: Determining the Effectiveness FY 2019. Seat belt use data were not collected for FY 2022. However, the only data provided were the before and after values. No sample size or other information was available. 4.6.2.2 Ohio Seat Belt Data Many states conduct some type of observational seat belt study annually independent of HVE campaigns. However, many of the seat belt studies reviewed, including the Iowa data, do not pro- vide background statistics such as sample size. The Ohio Department of Public Safety conducted an evaluation of its Click It or Ticket campaign in 2022 and published its annual observations study, which included metrics such as sample size (Schneider and Ackerman 2022). Seat belt observations were conducted in early May 2022 for 14 days, which served as the baseline observation, and then data were collected in June immediately after (14 days). Retired state patrol officers were used as observers. Seat belt compliance was observed in the lane closest to the observer, and data were col- lected for 1-hour periods at a time. Both drivers and passengers were included in the observations. The pre-intervention sample was 20,061, which included 16,932 drivers and 3,129 passengers. The post-intervention sample was 23,162, which included 19,460 drivers and 3,702 passengers. Data were collected at 344 sites across all 88 counties in Ohio. 4.6.3 Utility of Seat Belt Use Data for Evaluation of HVE Seat belt use is a direct measure of safety. It can easily be measured before, during, and after campaigns. Seat belt use data are well suited for evaluation of the impact of HVE, particularly campaigns that focus on seat belts. Seat belt use data can be used to compare changes in use before and after a campaign or to compare seat belt use at locations included and not included in the HVE. Analysis of seat belt data can be conducted using simple statistics. Examples of how seat belt data were used to compare effectiveness of an HVE campaign are provided in Sections 6.9.3 and 7.8.1. 4.7 Survey Data Utilized for Evaluation of Combined HVE 4.7.1 Description of Survey Data Surveys are often utilized to assess driver attitudes before, after, and/or during campaigns. They can be designed to ask specific questions that assess how HVE campaigns impacted driver behaviors or attitudes. 4.7.2 Survey Data Utilized for Evaluation of Combined HVE 4.7.2.1 Oklahoma and Tennessee Combined HVE Survey Nichols et al. (2016) evaluated a combined HVE campaign in Oklahoma and Tennessee that addressed multiple safety issues, including impaired driving, seat belt use, and speeding. Obser- vational studies were also conducted in control areas. The campaign included surveys to assess driver behaviors, which included questions to assess the following: • Driver recognition of various campaign slogans: – More Cops, More Stops. – Click It or Ticket. – Drive Sober or Get Pulled Over. • Awareness of enforcement activities: – General traffic safety. – Speeding.

Description of Data Used for Evaluation of HVE 45   – Seat belt use. – Alcohol-impaired driving. • Perceived risk of being stopped and charged by type of violation: – Alcohol-impaired driving. – Non-use of seat belts. – Speeding. Response rates and sample size were provided for various outcomes, and the project team utilized these for data visualization and simple statistical methods. 4.7.3 Utility of Survey Data for Evaluation of HVE Surveys are highly effective in evaluating the impact of HVE, since they can tailor questions to specific areas of interest. They can be conducted at various time periods (before, during, after) to assess changes in behavior due to the campaign. Examples of how survey data were used to compare effectiveness of an HVE campaign are provided in Sections 6.9.5 and 7.8.3. 4.8 Exposure Data 4.8.1 Description of Exposure Data Many crash analyses utilize a metric to represent exposure. For instance, a roadway with higher volume would be expected to have more crashes than a lower volume roadway that had similar characteristics. Common exposure metrics include annual ADT, VMT, or population. All states and many jurisdictions have a regular traffic count program, so VMT or annual ADT is usually available for at least major roadways. Census population estimates are also widely available. 4.8.2 Exposure Data Utilized for Evaluation of Combined HVE The project team used several different types of exposure data in the various analyses as described in the following sections. 4.8.2.1 Iowa VMT Data The Iowa DOT maintains annual VMT estimates by individual roadway segment. Addition- ally, the Iowa DOT has annual VMT for each incorporated city in the state available through its website. Data are provided for all municipal intersections, municipal primary roads, and municipal roadways for each city. VMT for the entire city is also provided. VMT data are publicly available for 2008 through 2021. Between 2016 and 2019, statewide VMT on average varied by approximately 1%. Between 2019 and 2020, VMT decreased by 11.54%, and in 2021, VMT increased by 11.2% and was near 2019 levels. This information confirms that 2020 and 2021 were unusual years, and thus the team chose to exclude them from analyses. It also sug- gests VMT between 2017 and 2019 was reasonably static, so one year was selected and used for the VMT estimates. VMT is consistent for 2015 through 2019, so the team chose 2018 as the base year. The VMT estimates for cities and counties are provided as an annual estimate. However, some of the analyses utilized in subsequent chapters used periods less than 1 year. As a result, the annual estimates were adjusted to account for time periods of less than a year. In addition to the VMT data mentioned previously, the Iowa DOT also has statewide VMT estimates by month. Figure 4-8 shows VMT by month for FY 2018.

46 Combined High-Visibility Enforcement: Determining the Effectiveness Monthly estimates vary from around 7% in the winter months (December through February) to around 9% for the summer months (May through August). The project team modeled data at the county level since violation data can only be provided at that level. VMT estimates by city and county were available as noted previously. Countywide estimates for VMT include VMT for cities within the county as well as all roadways outside city jurisdiction. The measure of exposure used for the analyses was the amount of VMT that was expected to have been impacted by a sTEP campaign. VMT for each wave was calculated for each county by summing the VMT according to the following logic. First, if the county sheriff ’s office participated, it was assumed county law enforce- ment focused on small cities and rural areas. Non-city VMT for a county was determined by sub- tracting the sum of the city VMT for cities located within that county. For instance, the 2018 VMT for Story County, Iowa, was listed as 856,092. Fourteen cities are located within Story County, and the VMT for those cities is 348,552. Story County VMT (non-city) was calculated as 507,540 (856,092 – 348,552). Next, VMT for a particular wave was summed for all jurisdictions within each county that participated. For instance, if two cities in Story County participated (combined VMT = 39,729) and the Story County Sheriff ’s Office participated (non-county VMT = 507,540), the total VMT used was 507,540 + 39,729 = 547,269. The project team also applied adjustments based on monthly estimates for the month the campaign took place as noted previously in Figure 4-8. For instance, statewide VMT in June is approximately 9% of total VMT. 4.8.2.2 Iowa Population by County VMT by city may be difficult for an agency to obtain. As a result, the team also used population as an exposure measure since U.S. Census Bureau and other population data are readily available. The team obtained population data for each county and city in Iowa and tested population as a measure of exposure in the various analyses. The team calculated population estimates for each wave using similar logic as for VMT. 4.9 Other Data Utilized 4.9.1 Location of Alcohol Establishments in Iowa The team obtained a list of bars and businesses that allow on-premises alcohol consump- tion from the Iowa Alcoholic Beverages Division’s website (https://elicensing.iowaabd.com Figure 4-8. Statewide Iowa VMT by month.

Description of Data Used for Evaluation of HVE 47   /OnDemandReport.aspx) and aggregated by county. The seven license types included in the search were as follows: • Class A Liquor License, • Class B Liquor License, • Class C Liquor License, • Special Class C Liquor License, • Class C Native Wine, • Class B Beer Permit, and • Class D Liquor License. The count of these businesses varied a bit annually between 2017 and 2019. However, the details for 2022 were not fully available when queried on June 6, 2023. As a result, the team used only the FY 2017 data for analysis. The license statuses included in the study encompassed active, renewed, or approved by local authority. Other statuses like application withdrawn, canceled, and denied by the Alcoholic Beverage Division, among others, were not taken into consideration. 4.10 Resources Additional resources related to speed studies include the following: • The Federal Highway Administration has a guide for speed study data collection that describes different data collection tools, tools to calculate needed sample size, and basic methods to calculate statistics such as 85th percentile speeds (Forbes et al. 2012). • The Institute for Transportation at Iowa State University developed a guidebook for local agencies to conduct traffic studies (Smith 2002). It includes a chapter on speed studies that describes the basics for conducting a spot speed study, data collection tools including a simple stopwatch method, and simple equations to calculate speed metrics such as 85th percentile or mean speed. • The National Association of City Transportation Officials has a guide for collecting speed data and conducting before and after studies (NACTO n.d.). Additional resources related to seat belt studies include the following: • NHTSA provides an annual seat belt use study (Boyle 2023). This includes statistics for how metrics such as confidence interval and p-value were collected.