Tools and Strategies for Eliminating Assaults Against Transit Operators, Volume 1: Research Overview (2018)

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.

112 The route-based risk calculator allows an agency to compare operator assault risks between routes. For each route, the risk of driver assault can be calculated based upon the risk factors along the route, in conjunction with region and population density risk factors. The route-based risk calculator provides transit owners and operators with a structured and viable risk manage- ment capability that can perform both “what if” and “trade off” decisionmaking. The route-based risk calculator includes pre-defined and pre-weighted operator assault risk factors in an easy-to use look-up table (Table E-2) that makes it possible to develop sound risk estimates on a route basis. In addition to these factors, there is another characteristic common to all transit systems i.e., terminals or route end-points or turnarounds. Each route has an origination and termination terminal (which may or may not be a physical terminal). There are essentially two calculations that have to be done to derive the Risk Ranks for terminals—derive the rank for each terminal on a route (i.e., Terminal A and B), then sum those ranks to derive the route terminal risk which is to be used in the route-based risk calculator. The same approach is to be taken for transfer stations. A P P E N D I X E Route-Based Risk Calculator RISK FACTOR FACTOR CHARACTERISTICS RISK FACTOR SCORE SYSTEM FACTORS REGION South, Midwest, West, Northeast POPULATION DENSITY Metropolitan area, cities, Nonmetropolitan areas ROUTE FACTORS INCIDENT HISTORY Aggravated and simple assault rates, previous driver incidents NUMBER OF BARS/CRIME PRONE SPOTS Bars, sports venues, gang territories, juvenile crime areas KNOWN THREATS OPERATION FACTORS HOURS OF OPERATION Graveyard, morning/mid-day, school dismissal times, peak traffic, evenings TERMINALS ��� TRANSFER STATIONS TOTAL ROUTE RISK SCORE Table E-1. Route-based risk calculator template. Tools and Strategies for Eliminating Assaults Against Transit Operators, Volume 1: Research Overview Copyright National Academy of Sciences. All rights reserved.

Route-Based Risk Calculator 113 System Factor Region and Population Density Risk Factor Rank South • Metropolitan areas • Cities • Nonmetropolitan areas 5 5 3 1 Midwest • Metropolitan areas • Cities • Nonmetropolitan areas 2 3 2 1 West • Metropolitan areas • Cities • Nonmetropolitan areas 3 3 6 2 Northeast • Metropolitan areas • Cities • Nonmetropolitan areas 2 3 5 3 Route Factor Incident History Factor Risk Factor Rank <1 Aggravated Driver Assault/60 Months 11 1 Aggravated Driver Assault/60 Months 2 1 Aggravated Driver Assault/48 Months 3 1 Aggravated Driver Assault/36 Months 4 1 Aggravated Driver Assault/24 Months 5 1 Aggravated Driver Assault/18 Months 6 1 Aggravated Driver Assault/12 Months 7 <1 Simple Driver Assault/60 Months 1 1 Simple Driver Assault/60 Months 2 1 Simple Driver Assault/48 Months 3 1 Simple Driver Assault/36 Months 4 1 Simple Driver Assault/24 Months 5 1 Simple Driver Assault/18 Months 6 1 Simple Driver Assault/12 Months 7 <1 Minor Driver Incident/60 Months 1 1 Minor Driver Incident/60 Months 2 1 Minor Driver Incident/48 Months 3 1 Minor Driver Incident/36 Months 4 1 Minor Driver Incident/24 Months 5 1 Minor Driver Incident/18 Months 6 1 Minor Driver Incident/12 Months 7 <1 Generalized Driver Threat/60 Months 1 1 Generalized Driver Threat/60 Months 2 1 Generalized Driver Threat/48 Months 3 1 Generalized Driver Threat/36 Months 4 1 Generalized Driver Threat/24 Months 5 1 Generalized Driver Threat/18 Months 6 1 Generalized Driver Threat/12 Months 7 Table E-2. Look-up table: risk factor and risk factor rank. (continued on next page) 1It is important to keep in mind that these ranks simply identify the likelihood of a particular type of event, not the con- sequence of that event. For example, 1 means a low likelihood of occurrence and 7 means a high likelihood of occurrence. Tools and Strategies for Eliminating Assaults Against Transit Operators, Volume 1: Research Overview Copyright National Academy of Sciences. All rights reserved.

114 Tools and Strategies for Eliminating Assaults Against Transit Operators: Research Overview 1 Simple to Aggravated Driver Assault/18 Months 7 1 Simple to Aggravated Driver Assault /12 Months 7 <1 Minor Incident to Simple Driver Assault/60 Months 1 1 Minor Incident to Simple Driver Assault/60 Months 3 1 Minor Incident to Simple Driver Assault/48 Months 4 1 Minor Incident to Simple Driver Assault/36 Months 5 1 Minor Incident to Simple Driver Assault/24 Months 6 1 Minor Incident to Simple Driver Assault/18 Months 7 1 Minor Incident to Simple Driver Assault/12 Months 7 <1 Repeat Aggravated Driver Assault/60 Months 1 1 Repeat Aggravated Driver Assault/60 Months 3 1 Repeat Aggravated Driver Assault/48 Months 4 1 Repeat Aggravated Driver Assault/36 Months 5 1 Repeat Aggravated Driver Assault/24 Months 6 1 Repeat Aggravated Driver Assault/18 Months 7 1 Repeat Aggravated Driver Assault/12 Months 7 <1 Repeat Simple Driver Assault/60 Months 1 1 Repeat Simple Driver Assault/60 Months 3 1 Repeat Simple Driver Assault/48 Months 4 1 Repeat Simple Driver Assault/36 Months 5 1 Repeat Simple Driver Assault/24 Months 6 1 Repeat Simple Driver Assault/18 Months 7 1 Repeat Simple Driver Assault/12 Months 7 <1 Repeat Minor Driver Incident/60 Months 1 1 Repeat Minor Driver Incident/60 Months 3 1 Repeat Minor Driver Incident/48 Months 4 1 Repeat Minor Driver Incident/36 Months 5 1 Repeat Minor Driver Incident/24 Months 6 1 Repeat Minor Driver Incident/18 Months 7 1 Repeat Minor Driver Incident/12 Months 7 <1 Repeat Generalized Driver Threat/60 Months 1 1 Repeat Generalized Driver Threat/60 Months 3 1 Repeat Generalized Driver Threat/48 Months 4 1 Repeat Generalized Driver Threat/36 Months 5 1 Repeat Generalized Driver Threat/24 Months 6 1 Repeat Generalized Driver Threat/18 Months 7 1 Repeat Generalized Driver Threat/12 Months 7 <1 Repeat Aggravated Assault Threat of Driver/60 Months 1 1 Repeat Aggravated Assault Threat of Driver/60 Months 3 1 Repeat Aggravated Assault Threat of Driver/48 Months 4 1 Repeat Aggravated Assault Threat of Driver/36 Months 5 1 Repeat Aggravated Assault Threat of Driver/24 Months 6 1 Repeat Aggravated Assault Threat of Driver/18 Months 7 1 Repeat Aggravated Assault Threat of Driver/12 Months 7 Route Factor Incident History Factor Risk Factor Rank <1 Simple to Aggravated Driver Assault/60 Months 1 1 Simple to Aggravated Driver Assault/60 Months 3 1 Simple to Aggravated Driver Assault/48 Months 4 1 Simple to Aggravated Driver Assault/36 Months 5 1 Simple to Aggravated Driver Assault/24 Months 6 Table E-2. (Continued). Tools and Strategies for Eliminating Assaults Against Transit Operators, Volume 1: Research Overview Copyright National Academy of Sciences. All rights reserved.

Route-Based Risk Calculator 115 Route Factor Incident History Factor Risk Factor Rank 1 Repeat Simple Assault Threat of Driver/18 Months 7 1 Repeat Simple Assault Threat of Driver/12 Months 7 <1 Repeat Minor Incident Threat of Driver/60 Months 1 1 Repeat Minor Incident Threat of Driver/60 Months 3 1 Repeat Minor Incident Threat of Driver/48 Months 4 1 Repeat Minor Incident Threat of Driver/36 Months 5 1 Repeat Minor Incident Threat of Driver/24 Months 6 1 Repeat Minor Incident Threat of Driver/18 Months 7 1 Repeat Minor Incident Threat of Driver/12 Months 7 Number of Bars/Crime Prone Spots Risk Factor Rank Bars, Nightclubs, and Entertainment: 1/block 1 Bars, Nightclubs, and Entertainment: 2/block 2 Bars, Nightclubs, and Entertainment: 3/block 3 Bars, Nightclubs, and Entertainment: 4/block 4 Bars, Nightclubs, and Entertainment: 5/block 5 Bars, Nightclubs, and Entertainment: 2 blocks with 5/block 6 Bars, Nightclubs, and Entertainment: 3 blocks with 5/block 7 Bars, Nightclubs, and Entertainment: 4 blocks with 5/block 8 High Incident Aggravated Assault Venues—Taverns, Bars, Nightclubs, and Sports Bars/Stadiums High concentrations of liquor-licensed establishments within a one- block area (>17/one-block radius) 7 High Incident Aggravated Assault Gang Areas 5 High Juvenile Crime Aggravated Assault Areas 1 Known Threats Risk Factor Rank 1 Known Threat of Aggravated Assault 1 2 Known Threats of Aggravated Assault 2 3 Known Threats of Aggravated Assault 3 4 Known Threats of Aggravated Assault 4 5 Known Threats of Aggravated Assault 5 Operation Factor Hours of Operation Risk Factor Rank Graveyard Shift—2:00 a.m. to 5:00 a.m. 1 Morning to Midday—5:00 a.m. to 2:00 p.m. 2 School Dismissal Hours—2:00 p.m. to 4:00 p.m. 3 Peak PM Traffic Period—4:00 p.m. to 7:00 p.m. 4 Evening/Late Night/Early Mornings—7:00 p.m. to 2:00 a.m. 5 1 Repeat Simple Assault Threat of Driver/24 Months 6 <1 Repeat Simple Assault Threat of Driver/60 Months 1 1 Repeat Simple Assault Threat of Driver/60 Months 3 1 Repeat Simple Assault Threat of Driver/48 Months 4 1 Repeat Simple Assault Threat of Driver/36 Months 5 Table E-2. (Continued). Tools and Strategies for Eliminating Assaults Against Transit Operators, Volume 1: Research Overview Copyright National Academy of Sciences. All rights reserved.

116 Tools and Strategies for Eliminating Assaults Against Transit Operators: Research Overview TERMINAL RISK RANK RISK FACTOR TERMINAL A ROUTE TERMINAL CHARACTERISTICS RISK FACTOR RANK REGION P OPULATION DENSITY INCIDENT HISTORY NUMBER OF BARS/CRIME PRONE SPOTS KNOWN THREAT HOURS OF OPERATION TERMINAL A RISK FACTOR TERMINAL B ROUTE TERMINAL CHARACTERISTICS RISK FACTOR RANK REGION POPULATION DENSITY INCIDENT HISTORY NUMBER OF BARS/CRIME PRONE SPOTS KNOWN THREAT HOURS OF OPERATION TERMINAL B RISK SCORE Total Route Terminal Risk Score = A + B TRANSFER STATION RISK RANK RISK FACTOR TRANSFER STATION 1 ROUTE TRANSFER STATION CHARACTERISTICS RISK FACTOR RANK REGION POPULATION DENSITY INCIDENT HISTORY NUMBER OF BARS/CRIME PRONE SPOTS KNOWN THREAT HOURS OF OPERATION TRANSFER STATION 1 RISK SCORE RISK FACTOR TRANSFER STATION N ROUTE TRANSFER STATION CHARACTERISTICS RISK FACTOR RANK REGION POPULATION DENSITY INCIDENT HISTORY NUMBER OF BARS/CRIME PRONE SPOTS KNOWN THREAT HOURS OF OPERATION TRANSFER STATION N RISK SCORE RISK SCORE Total Route Transfer Station Risk Score = 1 + 2 + … + n Table E-3. Terminal and transfer station risk rank template. Tools and Strategies for Eliminating Assaults Against Transit Operators, Volume 1: Research Overview Copyright National Academy of Sciences. All rights reserved.

2This table can be reproduced to cover as many routes as are in a transit agency’s system – as denoted by n in the last row. ROUTE-COMPARISON SUMMARY TABLE—PART A2 ROUTES ROUTE INCIDENT FREQUENCY SELECT RISK FACTOR RANK FROM LOOK-UP TABLE ROUTE RISK SCORE — PART A AGG ASLT SIMP ASLT MIN INC THR SIMP TO AGG MIN TO SIMP REP AGG REP SIMP REP MIN REP THR THR AGG THR SIMP THR MIN 1 2 3 4 5 6 7 8 9 10 .. .. .. n REGION SELECT RISK FACTOR RANK FROM LOOK-UP TABLE (ENTER SAME VALUE FOR ALL ROUTES) POPULATION DENSITY SELECT RISK FACTOR RANK FROM LOOK-UP TABLE (ENTER SAME VALUE FOR ALL ROUTES) Table E-4. Route-comparison summary table, Part A. T ools and S trategies for E lim inating A ssaults A gainst T ransit O perators, V olum e 1: R esearch O verview C opyright N ational A cadem y of S ciences. A ll rights reserved.

3This table can be reproduced to cover as many routes as are in a transit agency’s system – as denoted by n in the last row. ROUTE-COMPARISON SUMMARY TABLE—PART B3 ROUTES ROUTE FACTOR SELECT RISK FACTOR RANK FROM LOOK-UP TABLE OPERATION FACTORS SELECT RISK FACTOR RANK FROM LOOK-UP TABLE ROUTE RISK SCORE— PART B TOTAL ROUTE RISK SCORE— A+B BARS, NIGHTCLUBS, ENTERTAINMENT HIGH INCIDENT VENUES – TAVERNS, BARS, NIGHTCLUBS, AND SPORTS BARS/STADIUMS HIGH INCIDENT GANG AREAS HIGH JUVENILE CRIME AREAS KNOWN THREATS TERMINALS AND TRANSFER STATIONS HOURS OF OPERATION 1 2 3 4 5 6 7 8 9 10 .. .. .. n Table E-5. Route-comparison summary table, part B. T ools and S trategies for E lim inating A ssaults A gainst T ransit O perators, V olum e 1: R esearch O verview C opyright N ational A cadem y of S ciences. A ll rights reserved.

119 Introduction In this Appendix, we briefly describe the risk factors for operator assault, along with the framework for calculating the overall risk for any given route within any given system. Appendices G and H provide detailed discussions of the underlying model and assumptions that drive the specific risk estimates for each factor. Because of data limitations, the following discussion of risk factors and risks of assault on drivers along any given route is constrained by having to make certain simplifying assumptions so as to address the data limitations. To make the task of transit operators’ developing risk esti- mates for the respective systems and routes easier, we have created a series of look-up tables that identify the various factors and identify the likelihood of bus driver assault in a given system on a given route. As transit agencies adopt the incident reporting procedures outlined elsewhere in this project, the ability of the transit staffs to develop their own look-up tables for their respec- tive systems will be increased such that they can adjust the risk estimates for the factors shown in the tables below—i.e., create their own system and route specific estimators. The procedures for making these risk estimate modifications are discussed below. First, we briefly discuss the look-up tables by type of factor. System Factors Two generalized system factors that impact the likelihood of violence or assaults for the gen- eral population and, more specifically, bus drivers have been identified: geographic region and population density—characterized as metropolitan areas, cities outside metropolitan areas, and nonmetropolitan counties. The latest available data (2009) for aggravated assaults in the U.S. compiled by geographic region are: South – 359,045 (44.5%); West – 181,540 (22.5%); Midwest – 152,493 (18.9%); and Northeast – 114,572 (14.2%).4 However, simple numbers of assaults do not reveal the per capita exposure to assault, that is, to have a better assessment of the risk of aggravated assault by region, we need to control for total regional population. Based upon the 2010 population data by region, and assuming each reported assault occurs to one individual (i.e., the same person A P P E N D I X F Operationalizing Risk Factors Identified in the Literature 4According the National Crime Victimization Survey, 41 percent of the crimes committed were reported to authorities in 2005. About 47 percent of all violent crimes were reported, while only 40 percent of property crimes were reported. Only 38 percent of rapes and sexual assaults were reported to the police, lowest among violent crimes. [Cited in Legislative Ana- lyst’s Office. (2007). California’s Criminal Justice System: A Primer. http://www.lao.ca.gov/2007/cj_primer/cj_primer_013107. aspx#chapter5]. Tools and Strategies for Eliminating Assaults Against Transit Operators, Volume 1: Research Overview Copyright National Academy of Sciences. All rights reserved.

120 Tools and Strategies for Eliminating Assaults Against Transit Operators: Research Overview was not assaulted more than once in 2009), the risk of aggravated assault by region is shown in Table F-1. Further, the population density in the regions is not uniform, therefore regional population data are compiled by metropolitan areas, cities, and nonmetropolitan areas and risk of assault is normalized by population density, as shown in Table F-1. Thus, for example, the overall risk of bus driver assault in a Metropolitan transit system in the South is 0.0041/capita. In short, just on a system basis for such a transit system, all on-duty drivers have an exposure of 0.0041/capita for being assaulted in any given year. However, as is self-evident, the system is not the route. So, we turn to looking at what impact specific route factors have on the likelihood of bus driver assault. Route Factors In this section, we focus on the route factors that are largely beyond the control of the transit system management or drivers. These factors are drawn from Appendix B. The order of the specific factor does not signify importance or priority. In fact, while some factors may be more important in any given instance, the importance of the factors may vary along a route and shift in terms of likely impacts and likely effective countermeasure implementation. It is important to keep in mind that data with regard to these factors are quite limited, in contrast to the system- wide effects, since those either exist or they don’t, whereas, route factors, while they may exist or not, there are not repositories of data that could inform us with regard to existence of each factor and the extent to which any given factor has been evident in the case of driver assault(s) for any given transit authority/agency. Thus, the proposed numerical ratings given below are a first cut at assigning risk ratings to given factors. In all cases, first the factor has to be identified as present or not (it either receives a 1 for present or 0 for not present). For example, there are either bars or no bars along a route. If there are no bars, then, a 0 is entered into the calculation and the factor is null for this particular route. If there are bars, then the number of bars per block becomes a determinant in the risk level along a route and those risk levels are given in Table F-3 below. Route Incident History First, we consider assault incident history. It is no surprise that the likelihood of a bus driver assault along any given route is related to the history of bus driver assaults along that route. Thus, bus routes that have histories of assault, repeat crimes, minor incidents or threats of assault are more likely to experience driver or passenger assault than those routes that have no history of these kinds of incidents. If we have no incident data, then we assume that the likelihood of bus driver/ passenger assaults is similar to the population as a whole along the route—bus driver assault rates tend to mimic general population assault rates.5 Table F-2 presents the risk of assault, by type of assault, based upon the frequency of that type of assault over a five-year period of time. That is, Regional Assault Risk Population Density Based Assault Risk Metropolitan Areas Cities Nonmetropolitan Areas South – .0032/capita 0.0041/capita 0.0024/capita 0.0012/capita West – .0025/capita 0.0025/capita 0.0054/capita 0.0022/capita Midwest – .0023/capita 0.0029/capita 0.0021/capita 0.0009/capita Northeast – 0021/capita 0.0026/capita 0.0030/capita 0.0025/capita Table F-1. Bus driver assault risk by region and population density. 5Pearlstein, A., and Wachs, M. (1982). Crime in Public Transit Systems: An Environmental Design Perspective. Transportation, 11 (3): 277-297. Tools and Strategies for Eliminating Assaults Against Transit Operators, Volume 1: Research Overview Copyright National Academy of Sciences. All rights reserved.

Operationalizing Risk Factors Identified in the Literature 121 for example, if the incident history for aggravated assault along a given route is less than 1 incident in 60 months (five years), the risk of aggravated assault is 0.000. If the frequency of aggravated assault along that route is 1 incident in 60 months, then the risk of bus driver aggravated assault is 0.017, and so on, such that if the frequency of aggravated assault is 2 or more incidents in a 12 month period, the risk of aggravated assault on that route is 0.999 in any given year. For each type of assault—Aggravated, Simple, Repeat Crime and Minor Incident—the risks are calculated based upon the five-year incident history along the route for that type of incident. To account for the progression of type of incident to the next level of incident, Simple Assault to Aggravated Assault, for example, those risk levels are similarly calculated over a five-year period and are shown in Table F-2 as Simple to Aggravated, etc. Again, this mimics the continuum in Figure 2. For a detailed explanation of how these rates are derived, please see Appendix G. Bars, Nightclubs and Entertainment Venues Blocks with bars have higher levels of reported crime than blocks with no bars. Certain types of bars, such as dance clubs, have higher levels of reported violence. Neighborhood bars and social clubs have lower levels of reported violence. Stadia, arenas, sporting grounds, and con- cert halls are conducive to aggravated assault. Based upon studies6 considering the relationship Route Incident Incident Frequency <1/60 mo 1/60 mo 1/48 mo 1/36 mo 1/24 mo 1/18 mo 1/12 mo 2+/12 mo Aggravated Assault 0.000 0.017 0.034 0.068 0.136 0.272 0.544 0.999 Simple Assault 0.000 0.017 0.034 0.068 0.136 0.272 0.544 0.999 Repeat Crime 0.000 0.017 0.034 0.068 0.136 0.272 0.544 0.999 Minor Incident 0.000 0.017 0.034 0.068 0.136 0.272 0.544 0.999 Threat 0.000 0.017 0.034 0.068 0.136 0.272 0.544 0.999 Simple to Aggravated 0.000 0.034 0.068 0.136 0.272 0.544 0.999 0.999 Minor to Simple 0.000 0.034 0.068 0.136 0.272 0.544 0.999 0.999 Repeat Aggravated 0.000 0.034 0.068 0.136 0.272 0.544 0.999 0.999 Repeat Simple 0.000 0.034 0.068 0.136 0.272 0.544 0.999 0.999 Repeat Minor 0.000 0.034 0.068 0.136 0.272 0.544 0.999 0.999 Repeat Threat 0.000 0.034 0.068 0.136 0.272 0.544 0.999 0.999 Threat Aggravated 0.000 0.034 0.068 0.136 0.272 0.544 0.999 0.999 Threat Simple 0.000 0.034 0.068 0.136 0.272 0.544 0.999 0.999 Threat Minor 0.000 0.034 0.068 0.136 0.272 0.544 0.999 0.999 Table F-2. Bus driver assault risk by incident type and frequency. 6See, for example, Scott, M. S., and Dedel, K. (2006). Assaults in and Around Bars, 2nd Edition. Center for Problem-Oriented Policing, University at Albany, New York; Sytsma, V. (2011). A Pilot Application of Risk Terrain Modeling: Aggravated Assault in Rutgers Center for Public Security Brief; and Drucker, J. (2011). Risk Factors of Aggravated Assault. RTM Insights. Rutgers Center for Public Security. Tools and Strategies for Eliminating Assaults Against Transit Operators, Volume 1: Research Overview Copyright National Academy of Sciences. All rights reserved.

122 Tools and Strategies for Eliminating Assaults Against Transit Operators: Research Overview between bars and nightclubs and spatial proximity, and assuming data are available regarding the number and type of bars/nightclubs that are arrayed along any given bus route, Table F-3 provides the risk level for aggravated assault per year associated with the distribution of these establishments within a single city block. Thus, the risk of bus driver aggravated assault for a route with 1 bar, nightclub, etc. in a block is 0.1 per annum, for 2 bars in a block is 0.116, etc. up to 5 bars in a block. When there are two blocks with 5 bars per block, the risk increases to 0.362, etc. up to 4 blocks on a route with 5 bars per block having a risk of 0.999 of aggravated assault per annum. In short, a route traversing through an area that is crowded with a large number of bars, etc. (for example, an entertainment district in a city or metropolitan area), it is highly likely there will be a bus driver aggravated assault occurring in a year. (See Appendix G for the detailed explanation of these calculations and factor.) Proximity to Crime Hot Spots Extensive research has shown that occurrences of social disorder, crime and law enforcement activity tend not to be randomly scattered in space, but are clustered in certain areas. So certain places (locations) and spaces (areas) may provide a high-risk setting for a disproportionate number of certain kinds of criminal incidents.7 These so-called “crime hot spots” may simi- larly pose differential risks of assault on bus drivers. Crime hot spots are often linked with other activities or facility types, such as liquor serving establishments, transit stops, sports venues, etc. Table F-4 identifies the types of Crime Hot Spots that pose risks for bus driver assaults. Each type of Hot Spot is defined and factor weightings per annum are specified. For example, High Incident Venues have high concentrations of liquor licenses within a 1-block area. The probability of bus driver assault for a route traversing such an area in a year is very high—0.999. Table F-4 contains the risk estimates for High Incident Venues, High Incident Gang Areas, High Juvenile Crime Areas, and Known Threats. For detailed discussions of each of these route factors, see Appendix B. Operation Factors Operation factors are those within the transit company’s purview to manage and/or change in a system or on a specific route. These include such things as fare collection policies, installing monitoring equipment or driver protection barriers, driver training practices, incident report- ing and management practices, etc. Unfortunately, there are little to no data on the impacts of these operation factors, thus developing factor weightings is problematic at best. Given the data and factor weightings identified in Tables F-1 through F-5, the following discussion focuses on how to utilize these tables to calculate the risks of bus driver assault incidents. Route Factor Factor Frequency for Aggravated Assault/Annum Bars, Nightclubs, and Entertainment 1/bl 2/bl 3/bl 4/bl 5/bl 2 bls w/5/bl 3 bls w/5/bl 4 bls w/5/bl 0.1 0.116 0.135 0.156 0.181 0.362 0.724 0.999 Table F-3. Bus driver assault risk for routes with bars, nightclubs, and entertainment. 7See Block, R. L., and Block, C. R. (1995), in Eck, J. E., and Weisburd (Eds.) (1995). Crime and Place. Willow Tree Press, Inc., pp. 145-184, for discussion of the extensive literature on this subject going back to the early Chicago School of Social Ecology research. Tools and Strategies for Eliminating Assaults Against Transit Operators, Volume 1: Research Overview Copyright National Academy of Sciences. All rights reserved.

Operationalizing Risk Factors Identified in the Literature 123 Framework for Calculating Overall Risk for Routes within a System As stated previously, the risk of driver assault(s) on a given route may be conceptualized as a tripartite function—that is, route risk is affected by the potential for an assault (threat), the probability that an assault would be successful (vulnerability), and the severity of an assault (consequences). Appendix F details the theoretical underpinnings of the following simplified framework for how these various factors–system, route and operation—may be combined to Route Factor – Crime Hot Spots Definitions & Factor Weighting/Annum High Incident Venues – Taverns, Bars, Nightclubs and Sports Venues High concentrations of liquor licenses within 1 block area -- >17/1 block radius – treat like area of 4 blocks with 5 or more bars, etc. per block – probability = 0.999/annum of aggravated assault on bus drivers.8 High Incident Gang Areas Areas of concentrated gang activity – 48% aggravated assault committed in such areas, 60% of victims gang members9 – treat like area w/ 5 or more bars factored for skewedness vis-à-vis gang members – probability = 0.19/annum for aggravated assault on bus drivers High Juvenile Crime Areas Areas of high juvenile crime other than gang- related are most commonly impacted by gang member activity against other gang members10 – very little likelihood of aggravated assault on bus drivers – mostly it will be property damage or threats against a customer. Probability of aggravated assault on bus driver = 0.017. Known Threats The most common “known threat” that precedes aggravated assault is a verbal threat (see Figure 1)11 – 84% of aggravated assaults have been preceded by verbal threats. While, it would be natural to assume, therefore, that we could simply use this information to adjust the likelihood of assault on bus drivers, this would be incorrect since we do not have baseline data for the history of verbal threats vis-à-vis assaults, nor do we have any temporal data regarding whether the verbal threat occurred immediately prior to the assault or at some other point in time and then was subsequently acted upon. Thus, in our following framework we simply adopt a numerical rating for threats on a scale of 1 to 5. Table F-4. Bus driver assault risk for routes with crime hot spots, gang areas and known threats. 8Scott, M. S., and Dedel, K. (2006). Assaults in and Around Bars 2nd Edition. Center for Policy-Oriented Policing, University at Albany, New York. 9Federal Bureau of Investigation. (nd). 2011 National Gang Threat Assessment—Emerging Trends. http://www.fbi.gov/ stats-services/publications/2011-national-gangthreat-assessment (cited in Bilal Sevinc and Irfan Çiftci. 2015. Theoretical Approaches To Violent Criminal Acts Of Street Gangs In USA. Akademik Bakis Dergisi, Sayi: 48, Mart–Nisan 2015. 10Legislative Analyst’s Office. (1995). Juvenile Crime—Outlook for California Part III, http://www.lao.ca.gov/1995/050195_ juv_crime/kkpart3.aspx. 11Bruyere, D., and Gillet, J. M. (2005). National Operator Assault Survey Results, referenced in Nakanishi, Y., and Fleming, W. (2011). TCRP Report 93: Practices to Protect Bus Operators from Passenger Assault. Transportation Research Board, National Academy of Sciences, Washington, D. C. Tools and Strategies for Eliminating Assaults Against Transit Operators, Volume 1: Research Overview Copyright National Academy of Sciences. All rights reserved.

124 Tools and Strategies for Eliminating Assaults Against Transit Operators: Research Overview Operation Factors Definitions & Factor Weighting/Annum Operations Hours of operation: 48% of bus driver assaults occur in evening/late night/early morning; 38% occur in the PM Peak Traffic Hour period; and 28% occur in school dismissal period.12 Again, we do not have temporal baseline data regarding these assault patterns, therefore, in the framework presented below, we again treat these as numerical ratings from 1 to 5 to account for hours of operation. State of Good Repair (SGR): Anecdotal evidence suggests that delays in bus service due to breakdowns, etc. may lead to greater risks of bus driver assaults.13 However, there are no known data that demonstrate the linkage between SGR and assaults, threats, etc. Nevertheless, it is reasonable to assume that minimizing delays resulting from breakdowns would lessen this as a risk factor. Changes or Delays in Schedules: It is suggested that assaults can be precipitated by cutbacks in bus frequency, elimination of routes, service problems and other causes of rider frustration. Further, delays in transportation; poor information following delays; the quality of environmental surroundings; and failure to meet passenger expectations, are likely to incite anger and frustration in the public, and increase the risk of aggression.14 However, again, there are no known data that demonstrate the linkage between these types of delays and assaults, threats, etc. Nevertheless, it is reasonable to assume that minimizing such delays would lessen these risk factors. Fare Structure and Disputes: Fare enforcement is cited by transit agencies as the most common cause of driver assault, followed by intoxicated passengers and drug users. In a survey of transit agencies, 67% cited fare enforcement as a contributor to driver assaults.15 In an ATU legislative presentation featuring newspaper articles on transit worker Table F-5. Operation factors and factor definitions and weightings. 12Nakanishi, Y. J., and Fleming, W. C. (2011). TCRP Synthesis 93: Practices to Protect Bus Operators from Passenger Assault. Transportation Research Board, National Academy of Sciences, Washington D.C. 13Ibid. 14Boyd, C. (2002). Customer violence and employee health and safety. Work, Employment and Society, 16 (1): 151-169; Chappell, D., and Di Martino, V. (2006). Violence at Work. Geneva: International Labor Office; Essenberg, B. (2003). Violence and Stress at Work in the Transport Sector. Geneva: ILO. 15Nakanishi, Y. J., and Fleming, W. C. (2011). Practices to Protect Bus Operators from Passenger Assault. Transportation Research Board. National Academy of Sciences, Washington, D. C. Tools and Strategies for Eliminating Assaults Against Transit Operators, Volume 1: Research Overview Copyright National Academy of Sciences. All rights reserved.

Operationalizing Risk Factors Identified in the Literature 125 Operation Factors Definitions & Factor Weighting/Annum enforcement policies. However, it has been stated that where fare payment processes occur off-board the transit vehicle, or the fare payment process is not otherwise identified with the operator, operator assaults caused by fare disputes are minimal.18 Training & Skill Level of Operators: It has been asserted that the prevention of an assault begins with the transit agency’s hiring process. Further, operator training in customer relations, conflict mitigation, diversity, stress management, and verbal techniques such as verbal judo is vital for new bus operators in facing the daily challenges of their job. Refresher training for current operators is important as well in preventing operator assaults. Self-defense training and tools provide bus operators with a protection measure that is immediately available to the operator during an attack.19 However, we have no data to clarify the extent such hiring practices and training deter/reduce bus driver assaults. Security Personnel Countermeasures: There is a general consensus that deployment of uniformed police and other security personnel is one of the most effective deterrents to crime and assaults.20 However, it is also one of the most costly countermeasures and cannot be deployed across all bus routes. Nevertheless, in the case of NYC and NYCTransit, the use of Compstat over the period 1994 to 2009 to allocate police officers led to a 64.5% decrease for Index offenses (homicide, rape, robbery, assault, burglary, assaults, roughly 1/3 of the newspaper reports cite fare enforcement as the factor inciting the assault, some of which were fatal.16 Fare enforcement policies vary significantly between transit agencies.17 There are no data regarding bus driver assaults vis-à-vis Table F-5. (Continued). 16Amalgamated Transit Union. (nd). Ripped from the Headlines: Bus Drivers Under Attack. 17Nakanishi, Y. J., and Fleming, W. C. (2011). TCRP Synthesis 93: Practices to Protect Bus Operators from Passenger Assault. Transportation Research Board, National Academy of Sciences, Washington, D. C. 18Ibid. 19Ibid. 20McElvain, J. P., Kposowa, A. J., and Gray, B. C. (2012). Testing a Criminal Control Model: Does Strategic and Direct Deploy- ment of Police Officers Lead to Lower Crime? Journal of Criminology, Vol. 2013; Walsh, W. F. (2001), quoted in Ozdemir, H. (2011). Compstat: Strategic Police Management for Effective Crime Deterrence in New York City. International Police Execu- tive Symposium. Working Paper; and Interactive Elements, Inc. (1997). Guidelines for the Effective Use of Uniformed Transit Police and Security Personnel. (continued on next page) Tools and Strategies for Eliminating Assaults Against Transit Operators, Volume 1: Research Overview Copyright National Academy of Sciences. All rights reserved.

126 Tools and Strategies for Eliminating Assaults Against Transit Operators: Research Overview Operation Factors Definitions & Factor Weighting/Annum conducting exercises and putting procedures in place for handling bomb threats and suspicious objects are not costly undertakings and can be expected to reduce the risk of bus driver assault. However, there are no data to assess how much coordination and relationship reduce the risk of bus driver assault along any given route. Random Passenger Security Inspections: Suspicion-less inspections are expected to deter assaults by discouraging attackers that seek to avoid detection by increasing the uncertainty that they will be detected by a random search.22 Vehicle Security Countermeasures Security Cameras: To discourage violent behavior and identify the perpetrator if an assault does occur – the evidence is mixed regarding deterrence of violent crime, however, identification of perpetrators is made easier. There are no clear data available on the effectiveness of security cameras as a deterrent.23 There is anecdotal evidence that security cameras are effective in the prosecution of fraud perpetrators, etc. Audio Recorders: Do not appear to deter assaults, but do provide for rapid response when assaults occur.24 There are no data on deterrence due to audio surveillance. security personnel through the use of rigorous data analysis of crime patterns and locations can dramatically lower the incidence of bus driver assaults. On an annualized, non- compounded basis, this equates to roughly a 4.3% reduction in bus driver assaults per annum if the same level of effectiveness were achieved. Relationships and Coordination with Local Law Enforcement: Improving liaison with local police and other emergency responders, establishing crisis management plans, larceny, motor vehicle theft, and arson).21 Thus, efficient allocation of police officers and Table F-5. (Continued). 21Ozdemir, H. (2011); also see McDonald, P. P. (2001). Managing Police Operations: Implementing the NYPD Crime Control Model Using COMPSTAT. Wadsworth Publishing, New York; Kelling, G., and Coles, C. (1997). Community-Based Crime Pre- vention. In Fixing Broken Windows: Restoring Order and Reducing Crime in Our Communities, Chapter 5, Simon and Schuster, New York. 22Countermeasure Assessment and Security Experts, Waite and Associates, and Nakanishi Research and Consulting. (2007). TCRP Report 86: Public Transportation Security, Volume 13: Public Transportation Passenger Security Inspections: A Guide for Policy Decision Makers. Transportation Research Board, National Academy of Sciences, Washington, D. C. 23See Rose, J. (2010). TriMet to get more security cameras in Portland area, but do they help fight crime? The Oregonian/ OregonLive; Cameras on Buses and Shuttles Can Reap Rewards (2002). Parking Today, 7 (9): 16. 24Nakanishi, Y. J., and Fleming, W. C. (2011). Tools and Strategies for Eliminating Assaults Against Transit Operators, Volume 1: Research Overview Copyright National Academy of Sciences. All rights reserved.

Operationalizing Risk Factors Identified in the Literature 127 estimate the potential for an assault, the probability that an assault will be successful, and the severity of an assault. Further, any countermeasure (operation factors) may affect at least one of these risk compo- nents. There are various countermeasures that may be implemented by a transit authority/agency to reduce or eliminate risks. However, each countermeasure may, in fact, only address a single component of the risk equation, or it may impact multiple components. That is, a counter- measure may only address, for example, the Threat (T) component, or it may address, for exam- ple, the Vulnerability (V) and Consequence (C) components of the risk. Appendix C describes each countermeasure and its relationship to the T, V and C components of the risk function. Clearly, how these various factors and components are combined to develop an estimate of the risk of bus driver assault on any route could become a very complex undertaking for an agency/ authority, even if that organization implemented the self-assessment and incident protocols developed and reported elsewhere in this project. This situation becomes even more difficult and complex when there are no true baseline data for a transit system, let alone a route within a system. One of the objectives of this project is to provide transit operators with tools that can be implemented, even when there are no system or route data available, to estimate the risk of bus driver assault on any route in its system. To this end, we propose a framework that builds upon the look-up tables and data presented above (Tables F-1–F-5) and presented in greater detail in Appendices G and H. Simple Route Risk Calculation Table To illustrate and simplify, we assign a Risk Factor Rank to three of the factors identified in Tables F-1 through F-5: System Factor, Route Factor of Bars or other Crime Spots and Incident Table F-5. (Continued). Operation Factors Definitions & Factor Weighting/Annum Barriers (partitions) between Operators and Passengers: Many bus transit operators are installing barriers. To date there are no data regarding the effectiveness with regard to deterring bus driver assault.25 Other Vehicle Security Systems: Such as, Automatic Vehicle Location Systems (AVL), Emergency Communications Systems, etc. are essentially incident response systems, not deterrence systems.26 There are no known data regarding the effectiveness of these systems in deterring assault. There is anecdotal beneficial evidence regarding prosecution of perpetrators of assault, etc. Improved Interior and Bus Stop Lighting: Allows the operator to be aware of passenger behavior. There are no available data on the effectiveness of this countermeasure. Intuitively it should reduce the risk of assault to some degree, perhaps comparable to CCTV. 25Ibid. 26See Nakanishi, Y. J., and Fleming, W. C. (2011). for descriptions of such systems. Tools and Strategies for Eliminating Assaults Against Transit Operators, Volume 1: Research Overview Copyright National Academy of Sciences. All rights reserved.

128 Tools and Strategies for Eliminating Assaults Against Transit Operators: Research Overview History (these Risk Factors are selected because they are the ones for which the Project Team has the best data).27 Table F-6 illustrates these three factors. Step 1: Probability of an Assault In this table, the minimum sum of risk factor rank is 3 (1 + 1 + 1), and the maximum is 6 + 4 + 7 = 17. These two represent the scenarios with the highest/lowest likelihood of an assault. From above, we may let P(assault) = 0 for the rank sum being 3, and P(assault) = 1 for the rank sum being 17. For other rank sum scores, for example, if the rank sum score = 15, the probability is assigned as 0.88 (= 15/17). Step 2: Probability that an Assault is Successful In the worst-case scenario, we can assume the probability is 1 (all the conducted assaults are successful). If we consider countermeasures, the estimated probability can be reduced by an adjustment factor. Examples of Risk Calculations for Specific Routes Example 1: A Route Traverse in a Northeast City – the Route has 2 Bars and 1 Assault Over the Past 48 Months, What is the Route Risk? Step 1: Sum of risk factor rank is 5 (Northeast City) + 3 (two bars) + 3 (one assault over 48 months) = 11. The estimated probability of an assault is 0.65 (11/17). System Factor – Region and Population Density Risk Factor Rank South – Nonmetropolitan area 1 Midwest – Cities 2 West – Nonmetropolitan area 3 Northeast – Nonmetropolitan area 4 Northeast – Cities 5 West – Cities 6 Route Factor — Number of Bars/Crime Prone Spots Risk Factor Rank 0 1 1 2 2 3 3 or more 4 Route Factor — Incident History Factor Risk Factor Rank <1 Simple Driver Assault/60 Months 1 1 Simple Driver Assault/60 Months 2 1 Simple Driver Assault/48 Months 3 1 Simple Driver Assault/36 Months 4 1 Simple Driver Assault/24 Months 5 1 Simple Driver Assault/18 Months 6 1 Simple Driver Assault/Year 7 Table F-6. Risk factors and risk factor rank. 27For any agency, these Risk factor Ranks can be based upon local conditions and data, or can be assigned based upon regional and national data. Procedures for making these choices are discussed in the User Guide. Tools and Strategies for Eliminating Assaults Against Transit Operators, Volume 1: Research Overview Copyright National Academy of Sciences. All rights reserved.

Operationalizing Risk Factors Identified in the Literature 129 Step 2: Probability that the assault is successful, P = 1 (worst case) The risk is 0.65 * 1 = 0.65. Example 2: A Route Traverse in a Midwest City – the Route has 1 bar and 1 Assault Over the Past 60 Months, What is the Route Risk? Step 1: Sum of risk factor rank is 2 (Midwest City) + 2 (one bar) + 2 (one assault over 60 months) = 6. The estimated probability of an assault is 0.35 (6/17) Step 2: Probability that the assault is successful, P = 1 (worst case) The loss risk is 0.35 * 1 = 0.35. Calculating Total Route Risk with Countermeasures In this section we show what happens when countermeasures are added to the calculus. Table F-7 groups the Countermeasures according risk components, i.e., Threat (T), Vulner- ability (V) or Consequence (C) and weight or effectiveness. In some cases, the Countermeasure may impact some combination of risk components and these are shown as combined Counter- measure weights or effectiveness. As noted previously, these weights are drawn from the litera- ture, however, similar to the risk factors, there are no baseline data to develop and test assigned weights—in short, the weights shown in Table F-7 represent the consensus estimates of the Project Team based upon what is available in the literature. Should an agency/authority have data or alternative hypotheses as to the weightings, those can be substituted in Table F-7 and the method of calculation remains the same.28 So, for example, using the above illustrations, if, on a route with 2 bars in a Northeast City during the previous 48 months there has been 1 assault, if we implement off-board fare collec- tion policies, how much will this reduce the risk of bus driver assault on that route? Similarly, if we implement bus stop lighting on the route in the Midwest City that has 1 bar and 1 assault over the last 60 months, how much will this reduce the risk of bus driver assault on that route? Example 3: Risk Reduction on Northeast City Route after Implementing Off-Board Fare Collection: Step 1: Sum of risk factor rank is 5 (Northeast City) + 3 (two bars) + 3 (one assault over 48 months) = 11. The estimated probability of an assault is 0.65 (11/17) The implementation of off-board fare collection could affect the likelihood of an assault. Without detailed data at hand, we make the following assumptions: • The transit agency may implement a maximum of three countermeasures for V, T, C, respectively • The maximum effectiveness score for each countermeasure is 5, so the maximum total score for each risk component (V, T, C) is 15 • It is assumed that the reductions of likelihood of an assault (T), the probability that an assault is successful (V), and the consequence (C) are linearly correlated with the total effectiveness score29 28For example, an agency may have sufficient experience with assault legislation in its legislative district over several years to believe that the effectiveness weight should be 2 instead of 3, as has been found in the literature and shown in Table F-7. In that case, the weight of 2 would be substituted for the weight of 3 in Table F-7 and would be used in subsequent calculations of that particular countermeasure effectiveness. 29The research team makes the simplifying assumption that there is a linear relationship with effectiveness. While it is entirely arguable there are non-linear relationships among these variables, we have no a priori information as to what form such relationships might have and further, from a computational standpoint, introducing non-linearity introduces computational complexity significantly in an arena where there remains a great deal of uncertainty regarding these phenomena, with no further gain in insights and understanding. Tools and Strategies for Eliminating Assaults Against Transit Operators, Volume 1: Research Overview Copyright National Academy of Sciences. All rights reserved.

130 Tools and Strategies for Eliminating Assaults Against Transit Operators: Research Overview Threat (T), Vulnerability (V) and Consequence (C) Countermeasure Effectiveness Countermeasure TVC Effectiveness Weight Countermeasures Impacting Threats Policies, Plans, Protocols Fare Collection Policy and Procedures T 5 Passenger Screening T 5 Operator Assaults Zero Tolerance Workplace Violence Policy Coverage T 4 Passenger Code of Conduct T 4 Assault Legislation T 3 Barring Systems T 3 Passenger Awareness Programs T 3 Surveillance and Observation Systems Visible Surveillance Systems— Cameras in Plain Sight T 4 Driver Protection Services Public Address System and Signage T 3 Countermeasures Impacting Vulnerabilities Police or Security Staffing Staffing On Board Conveyance V 5 Driver Protection Services Physical Barriers— Compartment Barriers or Shielding, Full or Partial V 5 Driver-Side Exit Doors V 4 Defensive Weapons V 3 Training Driver Operator Security V 5 Driver Operator Self-Defense V 3 Countermeasures Impacting Consequences Policies, Plans, Protocols Communication Protocol for Violent Incidents C 5 Violent Incident Emergency Response Plan C 4 Police or Security Staffing Centralized On-Board Alarms, Panic Buttons with Immediate Force Response C 5 Data Communications and Telemetry Systems Vehicle Disabling C 4 Anti-Theft—Secure Driver Sign On C 4 Anti-Theft—Enroute C 4 Electronic Distress Signs C 3 Table F-7. Threat (T), vulnerability (V), and consequence (C) countermeasure effectiveness. Tools and Strategies for Eliminating Assaults Against Transit Operators, Volume 1: Research Overview Copyright National Academy of Sciences. All rights reserved.

Operationalizing Risk Factors Identified in the Literature 131 Threat (T), Vulnerability (V) and Consequence (C) Countermeasure Effectiveness Countermeasure TVC Effectiveness Weight Countermeasures Impacting Combined T/V Surveillance and Observation Systems Bus Stop Lighting T/V 3 Driver Protection Services On-Board Vehicle Fire Suppression Equipment T/V 5 Countermeasures Impacting Combined T/C Policies, Plans, Protocols Post Incident Action Steps T/C 4 Driver Protection Services DNA Swipe Kits T/C 2 Countermeasures Impacting Combined V/C Policies, Plans, Protocols Operator Assault Committees/Task Forces V/C 5 Police or Security Staffing Centralized Surveillance with Immediate Force Response V/C 4 Shadowing Vehicles V/C 3 Centralized Remote Sensors with Immediate Force Response V/C 3 Voice Communications Technology Two-Way Radio—3G/4G/LTE/ V/C 4 Satellite Mobile Broadband Least Cost Routing V/C 4 Cellular Telephone—Texting— Email V/C 2 Real-Time Audio V/C 2 Data Communications and Telemetry Systems Mobile Data Terminals (MDT) with DVRs V/C 4 Vehicle Locator Systems (AVLs)—Global Positioning System (GPS) V/C 4 Tracking and Monitoring— Global Positioning System (GPS) V/C 4 Surveillance and Observation Systems Video Surveillance Using On- board Computer/DVR V/C 2 Training Driver/Operator Handbook V/C 5 Driver Operator Security Awareness V/C 5 Driver Operator Security Communications V/C 4 Table F-7. (Continued). Tools and Strategies for Eliminating Assaults Against Transit Operators, Volume 1: Research Overview Copyright National Academy of Sciences. All rights reserved.

132 Tools and Strategies for Eliminating Assaults Against Transit Operators: Research Overview • Under the maximum effectiveness score (i.e., 15), each risk component would be reduced by 80 percent30 Based on these assumptions, off-board fare collection has an estimated effectiveness score 5, so the projected reduction in assault likelihood is 0.8 × (5/15) = 0.27 So the updated estimated assault probability is 0.65 × (1-0.27) = 0.475 Step 2: Probability that the assault is successful, P = 1 (worst case) The loss risk is 0.475 * 1 = 0.475. Example 4: Risk Reduction on Midwest City Route after Implementing Bus Stop Lighting. Step 1: Sum of risk factor rank is 2 (Midwest City) + 2 (one bar) + 2 (one assault over 48 months) = 6. The estimated probability of an assault is 0.35 (6/17) Bus stop lighting could reduce both the likelihood of an assault and the probability that an assault is successful. Its effectiveness score is 3. Based on the assumptions above, the estimated reduction of assault likelihood is 0.8 × (3/15) = 0.16; Similarly, the reduction of the success probability of an assault is also 0.16. So the updated estimated assault probability is 0.35 × (1-0.16) = 0.294 Step 2: Probability that the assault is successful, P = 1 (worst case) The updated estimated probability of success is 1 × (1-0.16) = 0.84 The risk is 0.294 * 0.84 = 0.247. Summary and Conclusions The focus of this appendix is the development of a framework for estimating the risk of driver assault occurring on any route in a transit agency’s system. This appendix builds on prior work completed in this project and, in particular, on the Transit Agency Operator Assault Route Factor Rating Sheet and Weighting Methodology included in Interim Report 1, Novem- ber 2014, pp 64-104 and subsequent refinements of the risk factors and countermeasures con- tained therein. In that report, 20 factors were identified that may impact the risk of bus driver assault(s) within a given bus transit system and/or along any given bus route within that sys- tem. In this appendix, these 20 factors are categorized as “system factors,” “route factors,” and “operation factors.” More specifically, system factors include whether the bus transit operation is within metropolitan areas, cities, or nonmetropolitan areas, and what geographic section of the country the system resides in. Route factors are tied to a specific route and potentially impact the risk of bus driver assault on a specific route. Route factors are those factors, such as the presence of bars, sports venues, gang territories, along any given route within a system. Operation factors are within the purview of the transit company to manage and/or change in a system or on a specific route. These include such things as fare collection policies, installing monitoring equipment or driver protection barriers, driver training practices, and incident reporting and management practices. 30Since we have no detailed data on the effectiveness of given countermeasures, and since no countermeasure can ever be com- pletely successful in mitigating all V, T and C components, we assume the maximum reduction for any given risk component will be 80 percent (0.8). If an agency has extensive experiential data regarding countermeasure effectiveness in its system, it can replace the 80 percent impact assumed in these risk estimates. Tools and Strategies for Eliminating Assaults Against Transit Operators, Volume 1: Research Overview Copyright National Academy of Sciences. All rights reserved.

Operationalizing Risk Factors Identified in the Literature 133 Using national data on aggravated and simple assault rates, by region and population density, as well as data from a variety of sources covering different cities and states, the estimated risk of bus driver assaults and other behaviors are identified for various categories of risk factors (e.g., system factors, route factors, and operation factors). The risks are presented in five tables, along with the qualifying assumptions required to make it possible to develop statistically sound risk estimates on a system and route basis. We first present the theoretical underpinnings of the proposed risk assessment framework. We then simplify the framework to allow the handing of the limited aggravated assault data related to bus drivers. To demonstrate the applicability of the proposed framework, three risk factors and risk factor ranks are presented in Table F-6 and two hypothetical routes in two regions and urban areas are characterized with different risk factors being associated with the two bus routes – one in the Northeast and one in the Midwest. For each route, the risk of bus driver assault is calculated based upon the risk factors along that route, in conjunction with the regional and population density risk factors. After the risk of assault is determined, the reduction of those risk estimates is determined by introducing certain countermeasures to yield monetary loss risk reductions resulting from the countermeasures. As seen in the two examples, it is possible to determine the reduction in loss risk associated with the implementation of different countermeasures on any given route. From an agency perspective, it is therefore possible to assess the return on investment for any countermeasure vis-à-vis alternative investments of the same amount of money on some other important agency matter. Tools and Strategies for Eliminating Assaults Against Transit Operators, Volume 1: Research Overview Copyright National Academy of Sciences. All rights reserved.

134 Introduction This Appendix operationalizes the 20 factors identified that may impact the risk of driver assault(s) within a given transit system and/or along any given route within that system. These 20 factors are categorized as “system factors,” “route factors,” and “opera- tion factors.” System factors are discussed first since conceptually, these are superordinate factors that potentially impact the risk of driver assault(s) on any route within a given sys- tem. For example, system factors include whether the bus transit operation is within an urban area, suburban area or rural area, what geographic section of the country the system resides, etc. Following the discussion of system factors, route factors are described. These represent fac- tors that are tied to a specific route and that potentially impact the risk of bus driver assault(s) on a specific route. Route factors are those factors, such as the presence of bars, sports venues, gang territories, etc. along any given route within a system. Third, we discuss operation factors that potentially impact the risk of bus driver assault(s) on a specific route. Operation factors are those that are within the purview of the transit company to manage and/or change in a system or on a specific route. These include such things as fare collection policies, installing monitor- ing equipment or drive protection barriers, driver training practices, incident reporting and management practices, etc. The risk of driver assault(s) on a given route may be conceptualized as a tripartite function – that is, route risk is affected by the potential for an assault (threat), the probability that an assault would be successful (vulnerability), and the severity of an assault (consequences). Any countermeasure may affect at least one of these risk components. In this Appendix, we show how these various factors – system, route and operation – may be combined to estimate the potential for an assault, the probability that an assault would be successful, and the severity of an assault. System Factors Two generalized system factors have been identified in the literature that impact on the like- lihood of violence or assaults more generally, and more specifically with regard to assaults on bus drivers: geographic region and population density – characterized as metropolitan area, cities outside the metropolitan areas, and nonmetropolitan areas. Data for 2009 for aggravated assault (an unlawful attack by one person upon another for the purpose of inflicting severe or aggravated bodily injury with a weapon) broken out by geographic region and population A P P E N D I X G Transit Driver Assault Risk Factors Tools and Strategies for Eliminating Assaults Against Transit Operators, Volume 1: Research Overview Copyright National Academy of Sciences. All rights reserved.

Transit Driver Assault Risk Factors 135 density show distinctly different patterns as discussed below. In 2009, there were 806,843 total reported aggravated assaults committed in the United States.31 • Region within the country where the transit system is situated – in the four regions of the US, in descending order, the number of aggravated assaults was greatest in the South and least in the Northeast: – South – 359,045 (44.5%); – West – 181,540 (22.5%); – Midwest – 152,493 (18.9%); and – Northeast – 114,572 (14.2%). Thus, based upon these data, it appears the likelihood of aggravated assault in the South is almost twice as great as in the West, slightly less than three times as great as in the Midwest and more than three times as great as in the Northeast. However, simple numbers of assaults do not reveal the per capita exposure to assault, that is, to have a better assessment of the risk of aggravated assault by region, we need to control for total regional population. Population data by region for 2010, one year later than the reported assault data show, as is well known, that the South has a much greater population base than the other regions, and in fact, the West, Midwest and Northeast are much more similar in total population than the South. – South – 114,555,744; – West – 71,945,553; – Midwest – 66,927,001; and – Northeast – 55,317,240.32 Based upon the population data by region, and assuming each reported assault occurs to one individual (i.e., the same person was not assaulted more than once in 2009), the risk of aggravated assault by region becomes: – South – .0032/capita; – West – .0025/capita; – Midwest – .0023/capita; and – Northeast – .0021/capita. All other things being equal, a transit system situated in the South will still experience more driver assaults than transit systems in any other region. However, the likelihood of driver assault in the other three regions is virtually the same, that is, a driver in the West is about as likely to be assaulted as a driver in the Midwest or Northeast. Put another way, regardless of the region, the likelihood of assault is less than 1 in 100 for any individual. Assuming that drivers are no more likely than any other individual to be assaulted, then using the rankings of: Very Low/Unlikely (0.1); Low (0.25); Medium (0.5); High (.75); and Very High/Very Likely (0.95)33, as identified in Transit Agency Operator Assault Route Factor Rating Sheet and Weighting Methodology in the Interim Report #1, it is possible to assign a risk rating to a transit system situated in any particular region. For example, we would assign a likeli- hood rating of 0.0032 to transit systems operating in the South, 0.0025 to transit systems operating in the West, 0.0023 to transit systems operating in the Midwest and 0.0021 to transit systems operating in the Northeast. In short, in the three non-Southern regions, the 31According to the National Crime Victimization Survey, 41 percent of the crimes committed were reported to authorities in 2005. About 47 percent of all violent crimes were reported, while only 40 percent of property crimes were reported. Only 38 percent of rapes and sexual assaults were reported to the police, lowest among violent crimes. (Cited in Legislative Analyst’s Office (2007). California’s Criminal Justice System: A Primer. http://www.lao.ca.gov/2007/cj_primer/cj_primer_ 013107.aspx#chapter5). 32U.S. Population by Region: 1990-2010. http://www.infoplease.com/ipa/A0764220.html. 33See Transit Agency Operator Assault Route Factor Rating Sheet and Weighting Methodology in Interim Report 1 for these ratings. (Very High/Very Likely has slightly been modified to concord with standard likelihood estimates.) Tools and Strategies for Eliminating Assaults Against Transit Operators, Volume 1: Research Overview Copyright National Academy of Sciences. All rights reserved.

136 Tools and Strategies for Eliminating Assaults Against Transit Operators: Research Overview risk of driver assault is still lower than in the South, but the likelihood of driver assault is still categorized as being Very Low/Unlikely based upon regional location. Again, while the abso- lute number of assaults that occur in the South is large compared to the other three regions, when considered on a per capita basis, the risk of driver assault in the South is greater, but not dramatically greater. Again, considering the 2009 total reported aggravated assaults (806,843), we see a dispro- portionate share occurring in the metropolitan areas as shown below. • Population Density – urban (generally surrounding a city), metropolitan (a region consisting of a densely populated urban core area and its less-populated surrounding territories – suburban areas), cities outside metropolitan areas, and nonmetropolitan counties, the 2009 data for aggravated assaults were: – 701,454 (86.9%) occurred in metropolitan areas; – 57,750 (7.2%) occurred in cities outside metropolitan areas; and – 47,639 (5.9%) occurred in nonmetropolitan counties.34 As is well documented, the population has been urbanizing for many years. By 2010, the trend continued unabated with more and more of the population in urban areas (Urbanized Areas and Urban Clusters) – 85 percent of the nation’s population was found in those two categories. By 2010, of the nation’s four census regions, the West continued to be the most urban, with 89.8 percent of its population residing within urban areas, followed by the Northeast, at 85.0 percent. The Midwest and South continue to have lower percentages of urban population than the nation as a whole, with rates of 75.9 percent and 75.8 percent, respectively.35 – South-Urban Population – 86,833,250 – West-Urban Population – 64,607,106 – Midwest-Urban Population – 50,797,593 – Northeast-Urban Population – 47,019,654 In 2010, nationwide the Urbanized Areas (Metropolitan Areas—486) accounted for 71.2 percent of the population.36 Accounting for population density and using the Urbanized Areas and Urban Clusters nomenclature to mean Metropolitan Areas and Cities outside of Metropolitan Areas, respectively, and further assuming that 2009 regional population distri- bution and ratios of Urban Clusters to Rural look essentially like the 2010 distribution (and ratios), we find the Metropolitan population by region to be: – South-Metropolitan Population – 76,156,513 – West-Metropolitan Population – 62,185,231 – Midwest-Metropolitan Population – 45,474,888 – Northeast-Metropolitan Population – 38,722,068 Per capita exposure by population for these regions to derive risk factors associated with Metropolitan areas are adjusted to obtain: – South-Metropolitan – .0041/capita – West-Metropolitan – .0025/capita – Midwest-Metropolitan – .0029/capita – Northeast-Metropolitan – .0026/capita Assuming the 2010 definition of Urban Clusters and Rural equate to the 2009 definition of Cities outside of Metropolitan areas and Nonmetropolitan, and that the population dis- tribution and ratios in 2009 were essentially equivalent to the 2010 distribution, we find that 34Urban areas were not specifically identified in the 2009 study. See Caplan, J. M., and Kennedy, L. W. (Eds). (2011). Risk Terrain Modeling Compendium. Rutgers Center on Public Security. 35U.S. Census Bureau. (2012). Growth in Urban Population Outpaces Rest of Nation, Census Bureau Reports. www.census.gov/ newsroom/releases/archives/2010_census/cb12-50.html. 36U.S. Census Bureau. (2010). 2010 Census Urban and Rural Classification and Urban Area Criteria. www.census.gov/geo/ reference/ua/urban-rural-2010.html. Tools and Strategies for Eliminating Assaults Against Transit Operators, Volume 1: Research Overview Copyright National Academy of Sciences. All rights reserved.

Transit Driver Assault Risk Factors 137 on the average, about 9.5 percent lived in Cities outside of Metropolitan areas and 19.3 per- cent lived in Nonmetropolitan areas, although, as noted above, there are regional differences between different urban clusters and rural areas. We then find Cities and Nonmetropolitan area population by region to be: – South-Cities – 10,676,737 – South-Nonmetropolitan – 17,503,975 – West-Cities – 2,421,875 – West-Nonmetropolitan – 4,843,375 – Midwest-Cities – 5,322,705 – Midwest-Nonmetropolitan – 10,645,409 – Northeast-Cities – 2,738,203 – Northeast-Nonmetropolitan – 5,476,40737 We now adjust for per capita exposure by population for these regions to derive risk factors associated with Cities and Nonmetropolitan Areas by region. We obtain: – South-Cities – .0024/capita – South-Nonmetropolitan – .0012/capita – West-Cities – .0054/capita – West-Nonmetropolitan – .0022/capita – Midwest-Cities – .0021/capita – Midwest-Nonmetropolitan – .0009/capita – Northeast-Cities – .0030/capita – Northeast-Nonmetropolitan – .0025/capita In general, all other things being equal, a transit system situated in a southern Metropolitan area will experience more driver assaults than transit systems in any other regional Metropolitan area. However, the likelihood of bus driver assault in the other three regions’ Metropolitan areas is virtually the same, that is, a bus driver in a western Metropolitan area is about as likely to be assaulted as a driver in Midwestern or northeastern Metropolitan areas. Put another way, regard- less of the region, the likelihood of assault is less than 1 in 100 for any individual. Assuming that drivers are no more likely than any other individual to be assaulted, then using the rankings of: Very Low/Unlikely (0.1); Low (0.25); Medium (0.5); High (.75); and Very High/Very Likely (0.95)38, it is possible to assign a risk rating to a transit system situated in any particular regional Metropolitan area. (Again, we would note these ratings represent a single risk factor in the con- ditional probability estimation for any given route, so as other factors are introduced, the impact of this particular factor becomes less.) Generally, the data for Cities and Nonmetropolitan have similar patterns as those for the Metropolitan areas, although there is an anomaly regarding western Cities. Inspection of these risk factors shows that western Cities have a markedly higher level of risk bus driver assault than is the case for other regional Cities, and in fact, higher than for any other City size. This result may be the artifact from the fact that within the West, the percentage of urbanization is so high (almost 90 percent) that distinguishing between reported assaults in the Metropolitan areas and Cities is confounded by the shifting boundary definitions.39 Aside from these systems factors (which clearly are system-wide), it may be argued that some factors we have identified as “operation” may also apply system-wide, for example, driver training 37The numbers do not sum to reach the gross population figures for each category because of rounding errors, but the mar- ginal difference between the calculated values and the gross Census values is not significant. 38See Transit Agency Operator Assault Route Factor Rating Sheet and Weighting Methodology in Interim Report 1. (For these ratings, Very High/Very Likely have slightly been modified to concord with standard likelihood estimates.) 39See U.S. Census Bureau. (2010). 2010 Census Urban and Rural Classification and Urban Area Criteria. www.census.gov/geo/ reference/ua/urban-rural-2010.html for a discussion of shifting nomenclature and requirements. Tools and Strategies for Eliminating Assaults Against Transit Operators, Volume 1: Research Overview Copyright National Academy of Sciences. All rights reserved.

138 Tools and Strategies for Eliminating Assaults Against Transit Operators: Research Overview programs or incident reporting procedures. However, from our perspective, the identified opera- tion factors are clearly at the discretion of management in terms of implementation, whereas the system factors identified above are exogenous to the system and are not subject to management discretion. Route Factors Most of the risk factors identified in the literature are either route factors that are characteris- tics of the specific route or are operation factors that are under the auspices and control of transit management. In this section, the focus is on the route factors that are largely beyond the control of the transit system management or drivers. There are 20 factors identified in Transit Agency Operator Assault Route Factor Rating Sheet and Weighting Methodology in Interim Report 1. For each factor there are from one to sixteen questions (sub-factors) that have been identified as characterizing specific conditions that potentially impact the likelihood of bus driver assaults, the likelihood of assault success and the consequence of a successful assault. A transit authority/ agency assessing the risk of bus driver assault along any given route can address each question with a “Yes” or “No,” i.e., 1 or 0. The score associated with each factor then represents a compi- lation of the sub-factor scores, such that the higher the score, the greater the likelihood of that factor being an important risk factor along a given route. In the following discussion we focus on those questions that are route specific. The order of the specific factor does not signify impor- tance or priority. In fact, while some factors may be more important in any given instance, the importance of the factors may vary along a route and shift in terms of likely impacts and likely effective countermeasure implementation. One other point to keep in mind, data with regard to these factors are quite limited, in con- trast to the system-wide effects, since those either exist or they don’t, whereas, route factors, while they may exist or not, there are not repositories of data that could inform us with regard to existence of each factor and the extent to which any given factor has been evident in the case of driver assault(s) for any given transit authority/agency. Thus, the proposed numerical ratings given below are a first cut at assigning risk ratings to given factors. In all cases, first the factor has to be identified as present or not (it either receives a 1 for present or 0 for not present). For example, there are either bars or no bars along a route. If there are no bars, then, a 0 is entered into the calculation and the factor is null for this particular route. Route Factors • Route Incident History Bus routes that have histories of assault, repeat crimes, minor incidents or threats of assault are more likely to experience driver or passenger assault than those routes that have no his- tory of these kinds of incidents. To account for this factor, we first simply assume a doubling of risk as the frequency of the type of incident occurs. If we have no incident data, then we assume that the likelihood of bus driver/passenger assaults is similar to the population as a whole along the route.40 First, assuming incident data are available, we: Assume the likelihood of Aggravated Driver Assault doubles as the reported frequency of aggravated assault (on either a driver or a passenger) on the route increases: – <1 Aggravated Driver Assault/60 months 0.000 – 1 Aggravated Driver Assault/60 months 0.017 – 1 Aggravated Driver Assault/48 months 0.034 40Pearlstein, A., and Wachs, M. (1982). Crime in Public Transit Systems: An Environmental Design Perspective. Transportation, 11 (3): 277-297. Tools and Strategies for Eliminating Assaults Against Transit Operators, Volume 1: Research Overview Copyright National Academy of Sciences. All rights reserved.

Transit Driver Assault Risk Factors 139 – 1 Aggravated Driver Assault/36 months 0.068 – 1 Aggravated Driver Assault/24 months 0.136 – 1 Aggravated Driver Assault/18 months 0.272 – 1 Aggravated Driver Assault/year 0.544 – 2+ Aggravated Driver Assaults/year 0.999 Assume the likelihood of Simple Driver Assault doubles as the reported frequency of simple assault (on either a driver or a passenger) on the route increases: – <1 Simple Driver Assault/60 months 0.000 – 1 Simple Driver Assault/60 months 0.017 – 1 Simple Driver Assault/48 months 0.034 – 1 Simple Driver Assault/36 months 0.068 – 1 Simple Driver Assault/24 months 0.136 – 1 Simple Driver Assault/18 months 0.272 – 1 Simple Driver Assault/year 0.544 – 2+ Simple Driver Assaults/year 0.999 Assume the likelihood of repeat crimes doubles as the reported frequency of repeat crimes on the route increases: – <1 Repeat Crime/60 months 0.000 – 1 Repeat Crime/60 months 0.017 – 1 Repeat Crime/48 months 0.034 – 1 Repeat Crime/36 months 0.068 – 1 Repeat Crime/24 months 0.136 – 1 Repeat Crime/18 months 0.272 – 1 Repeat Crime/year 0.544 – 2+ Repeat Crime/year 0.999 Assume the likelihood of minor incidents doubles as the reported frequency of minor incidents on the route increases: – <1 minor incident/60 months 0.000 – 1 minor incident/60 months 0.017 – 1 minor incident/48 months 0.034 – 1 minor incident/36 months 0.068 – 1 minor incident/24 months 0.136 – 1 minor incident/18 months 0.272 – 1 minor incident/year 0.544 – 2+ minor incidents/year 0.999 Assume the likelihood of threats doubles as the reported frequency of threat on the route increases: – <1 threat/60 months 0.000 – 1 threat/60 months 0.017 – 1 threat/48 months 0.034 – 1 threat/36 months 0.068 – 1 threat/24 months 0.136 – 1 threat/18 months 0.272 – 1 threat/year 0.544 – 2+ threats/year 0.999 To address the relationships between threats, minor incidents, repeat crimes, simple assaults and aggravated assaults, we make the following assumptions: – As the frequency of simple assaults increases, the likelihood of an aggravated assault doubles: � <1 Simple Driver Assault/60 months = 0.000 likelihood of Aggravated Driver Assault/ 60 months Tools and Strategies for Eliminating Assaults Against Transit Operators, Volume 1: Research Overview Copyright National Academy of Sciences. All rights reserved.

140 Tools and Strategies for Eliminating Assaults Against Transit Operators: Research Overview � 1 Simple Driver Assault/60 months = 0.034 probability of Aggravated Driver Assault/ 60 months � 1 Simple Driver Assault/48 months = 0.068 probability of Aggravated Driver Assault/ 48 months � 1 Simple Driver Assault/36 months = 0.136 probability of Aggravated Driver Assault/ 36 months � 1 Simple Driver Assault/24 months = 0.272 probability of Aggravated Driver Assault/ 24 months � 1 Simple Driver Assault/18 months = 0.544 probability of Aggravated Driver Assault/ 18 months � 1 Simple Driver Assault/year = 0.99 probability of Aggravated Driver Assault/year – As the frequency of minor incidents increases, the likelihood of simple assaults doubles: � <1 minor incident/60 months = 0.000 likelihood of Simple Driver Assault/60 months � 1 minor incident/60 months = 0.034 probability of Simple Driver Assault/60 months � 1 minor incident/48 months = 0.068 probability of Simple Driver Assault/48 months � 1 minor incident/36 months = 0.136 probability of Simple Driver Assault/36 months � 1 minor incident/24 months = 0.272 probability of Simple Driver Assault/24 months � 1 minor incident/18 months = 0.544 probability of Simple Driver Assault/18 months � 1 minor incident/year = 0.99 probability of Simple Driver Assault/year – In the case of repeat crimes, the frequency is related to the type of crime committed such that it follows the likelihood estimates stated previously, e.g., a repeated Aggravated Assault adopts the pattern for Aggravated Assault, i.e., each incident doubles the likelihood. – In the case of threats, the frequency is related to the type of threat made such that it follows the likelihood estimates stated previously, e.g., a threat of Aggravated Assault adopts the pattern for Aggravated Assault, i.e., each threat of Aggravated Assault doubles the likeli- hood in that period. If there are no data available regarding Route Incident History, then, we adopt the 2014 Aggra- vated Assault Prevalence Rating for the U.S. based upon the rate of assault per 1,000 people: = ×Prevalence rate Number of victims in a specified population Number of persons in the specified population 100T T T In 2014, this prevalence rate determined that 0.26% of the population 12 years of age and over experienced Aggravated Assault, whereas 0.69% of the population 12 years of age and over experienced Simple Assault. Further, there was no statistically significant change in this rate from 2013.41 To operationalize these metrics, assume that the specified population lives within three blocks, either side of the route, i.e., a six-block corridor.42 For any given bus transit route, these estimates may be greater or lesser than would be the case if actual route incident histories were available. However, absent such route incident data, this estimator provides a mechanism for determining first cut likelihood estimates for Aggravated Assault along any given route. • Population Density Along the Route This factor is addressed by the System Factors described previously that address geographic variations by region and population density. • Bars, Nightclubs, and Entertainment Venues Blocks with bars have higher levels of reported crime than blocks with no bars. Certain types of bars, such as dance clubs, have higher levels of reported violence. Neighborhood bars 41Truman, J. L., and Langton, L. (2015). Criminal Victimization, 2014., NCJ 248973. August 2015. 42For a discussion of these criteria, see Sytsma, V. (2011). A Pilot Application of Risk Terrain Modeling: Aggravated Assault in Rutgers Center for Public Security Brief. May 2011; Caplan, J. M., and Kennedy, L. W., (Eds) (2011). Risk Terrain Modeling Compendium. Rutgers Center for Public Security. Tools and Strategies for Eliminating Assaults Against Transit Operators, Volume 1: Research Overview Copyright National Academy of Sciences. All rights reserved.

Transit Driver Assault Risk Factors 141 and social clubs have lower levels of reported violence. Stadia, arenas, sporting grounds, and concert halls are conducive to aggravated assault. Based upon studies43 considering the rela- tionship between bars and nightclubs and spatial proximity, and assuming data are available regarding the number and type of bars/nightclubs that are arrayed along any given bus route, we estimate the likelihood of bus driver/passenger assault as follows: – Assume blocks with one dance club, sports bar, social club, etc. have the likelihood of 0.1 of aggravated assault/annum.44 – Blocks with two dance clubs, sports bars, social clubs, stadia, arenas, etc. will be 116% more likely to have aggravated assaults occurring on bus routes than on blocks with no dance clubs, sports bars, etc. The risks for aggravated assault increase to 0.116/annum.45 – Blocks with three dance clubs, sports bars, social clubs, stadia, arenas, etc. will be 116% more likely to have aggravated assaults occurring on bus routes than on blocks with three dance clubs, sports bars, etc. The risks for aggravate assault increase to 0.135/annum. – Blocks with four dance clubs, etc. will be 116% more likely to have aggravated assaults occurring on bus routes than on blocks with three dance clubs, sports bars, etc. The risks for aggravate assault increase to 0.156/annum. – Blocks with five dance clubs, etc. will be 116% more likely to have aggravated assaults occurring on bus routes than on blocks with three dance clubs, sports bars, etc. The risks for aggravate assault increase to 0.181/annum. – When there are two blocks with five or more such clubs, etc., the risks double to 0.362/annum. – Three blocks with five or more clubs, etc., the risks double to 0.724/annum – Four or more blocks with five or more clubs, etc., the risk of aggravated assault increases to 0.99/annum. • Proximity to Crime Hot Spots The relationship between crime and place is neither uniform nor static. Extensive research has shown that occurrences of social disorder, crime and law enforcement activity tend not to be randomly scattered in space, but are clustered in certain areas. Further, these spatial pat- terns may evolve or change over time. Finally, various kinds of disorder or criminal activity may follow completely different spatial patterns. Just as offenders may specialize in a particu- lar crime or complex of crimes and certain potential victims may be particularly vulnerable to particular kinds of repeated victimization, so certain places (locations) and spaces (areas) may provide a high-risk setting for a disproportionate number of certain kinds of criminal incidents.46 These so-called “crime hot spots” may similarly pose differential risks of assault on bus drivers. As will be discussed below, crime hot spots are often linked with other activi- ties or facility types, such as liquor serving establishments, transit stops, sports venues, etc. Thus, these interdependencies need to be addressed in the development of bus driver assault risk estimates. – High Incident Taverns/Liquor Stores/Convenience Stores � A 1993 study of Chicago crime hot spot taverns, liquor stores and convenience stores found concentrations of these establishments associated with main streets, diagonal off- grid streets and major grid streets. These are generally regional shopping and enter- tainment areas. Hot spot tavern crime areas are frequently found at these intersections. 43See, for example, Scott, M. S., and Dedel, K. (2006). Assaults in and Around Bars 2nd Edition. Center for Problem-Oriented Policing, University at Albany, State University of New York; Sytsma, V. (2011). A Pilot Application of Risk Terrain Modeling: Aggravated Assault in Rutgers Center for Public Security Brief. May 2011; and Drucker, J. (2011). Risk Factors of Aggravated Assault. RTM Insights. Rutgers Center for Public Security. March 2011. 44See Appendix E. 45Sytsma, V. (2011). A Pilot Application of Risk Terrain Modeling: Aggravated Assault in Newark, NJ. Rutgers Center for Public Security Brief. May 2011. 46See Block, R. L., and Block, C. R. (1995). In Eck, J. E. and Weisburd, D. (Eds.) (1995). Crime and Place. Willow Tree Press, Inc., pp. 145-184, for discussion of the extensive literature on this subject going back to the early Chicago School of Social Ecology research. Tools and Strategies for Eliminating Assaults Against Transit Operators, Volume 1: Research Overview Copyright National Academy of Sciences. All rights reserved.

142 Tools and Strategies for Eliminating Assaults Against Transit Operators: Research Overview Further, taverns or liquor stores in a Hot Spot Area of tavern crime are likely to be in relatively affluent singles neighborhoods, close to many other places holding a liquor license as well as to a transit stop.47 – High Incident Bars and Nightclubs � Similarly, crime hot spot bars and nightclubs tend to be trendy nightlife areas and often located in areas with a concentration of liquor licenses (in this study, averaging over 17 licenses in a one-block radius—1/8 of a mile).48 – High Incident Sports Venues � Sports venues with high incidents of aggravated assault tend to be associated with areas of concentrations of liquor licenses or taverns.49 – High Incident Gang Areas � The National Gang Intelligence Center (NGIC) estimates 48 percent of violent crime in most jurisdictions and 90 percent of other crimes are committed by gangs.50 Block and Block (1993) who examined Chicago gang homicide data classified gang related street crimes and neighborhoods as: turf hot spots where gangs fight over territory control; drug hot spots where gangs intensively deal with drug; and drug and turf hot spots where gangs commits both of the crimes. Gang involvement in violent crimes and homicide was more often turf-related rather than drug-related.51 Hutson, Anglin & Pratts (1994) found that in Los Angeles, gang members are 60 times more likely to die because of homicide than members of the general population.52 – High Incident Juvenile Crime other than Gangs � If the experience of California is a guide to juvenile crime, it is clear that a relatively small number of juveniles commit crime. Furthermore, of those juveniles who do commit crimes, the majority of them will only commit one or two offenses. For these individuals, the experience of the juvenile justice system—being arrested by a law enforcement officer, facing their parents, having to spend a night in juvenile hall, interacting with a probation officer or a judge—is enough to keep them from offending again. Nevertheless, a small number of individuals who are chronic recidivists are responsible for a large proportion of juvenile crime. Much research has shown that these juveniles commit their first offense at an early age (usually age 11), and even at this early age, these juveniles display a variety of serious problems indicative of an “at-risk” juvenile. One of the risk factors associated with an “at-risk” juvenile is gang membership. Gang membership and gang-related crime is primarily a juvenile problem. Gang membership, especially at an early age, is strongly associated with future criminal activity. Juvenile gun possession is a factor that “magni- fies” juvenile crime by making offenses more likely to result in injury or death.53 � Thus, for non-gang member juvenile crime, the apparent likelihood of aggravated assault along a specific route that passes through such a hot spot area is low. Such hot spots, as described below, are highly likely to be associated with proximity to a school and similarly temporally driven with regard to when juveniles are in school. 47Ibid. 48Ibid. 49Ibid. 50Federal Bureau of Investigation. (n.d). 2011. National Gang Threat Assessment—Emerging Trends. http://www.fbi.gov/ stats-services/publications/2011-national-gangthreat-assessment (cited in Bilal Sevinc and Irfan Çiftci. 2015. Theoretical Approaches To Violent Criminal Acts Of Street Gangs In USA. Akademik Bakis Dergisi, Sayi: 48, Mart–Nisan 2015. 51Block, R., and Block, C. R. (1993). Street Gang Crime in Chicago. Research in Brief. U. S. Department of Justice, Office of Justice Programs, National Institute of Justice. (Cited in Bilal Sevinc and Irfan Çiftci. 2015.) 52Hutson, H. R., Anglin, D., and Pratts, M. J. (1994). Adolescents and children injured or killed in drive-by shootings in Los Angeles. The New England Journal of Medicine, 330, 5; 324–327. (Cited in Bilal Sevinc and Irfan Çiftci. 2015.) 53Legislative Analyst’s Office. (1995). Juvenile Crime—Outlook for California Part III, http://www.lao.ca.gov/1995/050195_juv_ crime/kkpart3.aspx. Tools and Strategies for Eliminating Assaults Against Transit Operators, Volume 1: Research Overview Copyright National Academy of Sciences. All rights reserved.

Transit Driver Assault Risk Factors 143 – High Incident Prostitution and Vice Areas � While areas of high or frequent prostitution activity are also frequently areas of high incidents of aggravated assault, data from studies in the late 1990s indicate that most of those assaults occur with the victims being the prostitutes themselves.54 Further, there is an association with gang activity in that increasingly, gangs have become involved in pimping or running prostitution or commercial vice activities.55 Thus, it appears that the likelihood of assault on bus drivers in high incident areas of prostitution and criminal vice is more related to the overall level gang related assault in the area than the fact of prostitution and vice. – High Incident Drug Trade Areas � Street gangs, outlaw motorcycle gangs (OMGs), and prison gangs are the primary dis- tributors of illegal drugs on the streets of the United States. Gangs also smuggle drugs into the United States and produce and transport drugs within the country. Gangs pri- marily transport and distribute powdered cocaine, crack cocaine, heroin, marijuana, methamphetamine, MDMA, and PCP in the United States. Large street gangs readily employ violence to control and expand drug distribution activities, targeting rival gangs and dealers who neglect or refuse to pay extortion fees.56 Thus, there tends to be a high correlation between areas of gang activity and areas of drug trade. Although, the relation- ships between gang activity, drug trade, and high levels of violence are less clear. Youth gang drug trafficking coexists with other gang crimes, mainly inter-gang turf conflicts and interpersonal violence that are unrelated or only tangential to drug trafficking. Vio- lence in adult criminal drug-trafficking organizations, cartels, and syndicates appears to be connected much more directly to the drug-trafficking enterprise. Thus, it appears that reducing drug trafficking in youth gangs is not likely to have a significant impact on violent youth gang crime (except in the case of particular drug gangs), whereas suc- cessful reduction of drug trafficking in adult criminal organizations is likely to produce a significant reduction in violent crime.57 Therefore, the likelihood of bus driver assault on routes through high incident drug trade areas is more closely related to the level of violent gang crime more generally in the area. • Known Threats Known threats represent an interesting risk factor category in that the typical use of this term is found in dealing with terrorist threats or similar kinds of events.58 However, the most common “known threat” that precedes aggravated assault on a bus driver is a verbal 54Farley, M., and Barkan, H. (1998). Prostitution, Violence Against Women, and Posttraumatic Stress Disorder. Women and Health 27 (3): 37-49. (Cited in Prostitution Statistics and Rape—Physical Abuse of Prostitutes Common. http://womensissues. about.com/od/rapesexualassault/a/Wuornos.htm). 55National Gang Intelligence Center. (Cited in Bilal Sevinc and Irfan Çiftci. 2015 Theoretical Approaches To Violent Criminal Acts Of Street Gangs In USA. Akademik Bakis Dergisi, Sayi: 48, Mart–Nisan 2015). 56National Drug Intelligence Center. (2005). Drugs and Gangs Fast Facts, http://www.justice.gov/archive/ndic/pubs11/13157/. 57James, C. H., and Scott, H. D. (1999). The Youth Gangs, Drugs, and Violence Connection, Juvenile Justice Bulletin. Office of Juvenile Justice and Delinquency Prevention. 58Caplan, J. M., Kennedy, L. W. (2010). Risk Terrain Modeling Manual. Rutgers Center on Public Security; Department of Homeland Security, NTAS Public Guide, pulled on 7/2/14 from http://www.dhs.gov/ntas-public-guide; Uniform Crime Reporting Statistics, pulled on 7/2/14 from http://www.ucrdatatool.gov/; FTA Transit Threat Level Response Recommen- dation, pulled on 7/2/14 from https://www.hsdl.org/?view&did=440811; Meloy, J. R., Hoffmann, J., Guldimann, A., and James, D. (2011). The Role of Warning Behaviors in Threat Assessment: An Exploration and Suggested Typology, Behav- ioral Sciences and the Law. Published online in Wiley Online Library; Gray, Snowden, and MacCulloch, 2004; McNiel, Gregory, and Lam, 2003; Nicholls, Brink, Desmarais, Webster, and Martin, 2006; Skeem and Mulvey, 2001. See Role of Warning Behaviors in Threat Assessment for full citations. Bruyere, D., and Gillet, J. M. (2005). National Owner Operator Assault Survey Results, referenced in Nakanishi, Y, Fleming, W. (2011). TCRP Synthesis 93: Practices to Protect Bus Operators from Passenger Assault. Transportation Research Board, National Academy of Sciences, Washington, D.C. Tools and Strategies for Eliminating Assaults Against Transit Operators, Volume 1: Research Overview Copyright National Academy of Sciences. All rights reserved.

144 Tools and Strategies for Eliminating Assaults Against Transit Operators: Research Overview threat – 84% of aggravated assaults have been preceded by verbal threats.59 While it would be natural to assume, therefore, that we could simply use this information to adjust the likeli- hood of assault on bus drivers, this would be incorrect since we do not have baseline data for the history of verbal threats vis-à-vis assaults, nor do we have any temporal data regarding whether the verbal threat occurred immediately prior to the assault or at some other point in time and then was subsequently acted upon. Thus, in our risk calculation framework we adopt a numerical rating for threats on a scale of 1 to 5. 59Bruyere, D., and Gillet, J. M. (2005). National Operator Assault Survey Results, referenced in Nakanishi, Y. J., and Fleming, W. C. (2011). TCRP Synthesis 93: Practices to Protect Bus Operators from Passenger Assault. Transportation Research Board, National Academy of Sciences, Washington, D.C. Tools and Strategies for Eliminating Assaults Against Transit Operators, Volume 1: Research Overview Copyright National Academy of Sciences. All rights reserved.

145 Executive Summary A risk analysis model is developed to assess the risk of driver assault(s) on a specific route. The model accounts for the probability of the conduct of an assault, the success probability and the consequences of an assault. Methodology Without loss of generality, suppose there is a route between the Origin (O) and destination (D). This route consists of N segments. Let Si represent the ith segment. In risk analysis, one of the most widely used definitions is that risk is the probability and consequence of an event (Kaplan and Garrick 1981). In previous practical risk analyses (Liu et al. 2013a, 2013b, 2014; Liu 2016), risk is expressed as the product of the probability of an incident multiplied with its potential consequences. Put another way, on each segment, the bus driver assault risk (denoted as Ri) is calculated as follows: = ×R P C (1)i i i Where: Ri = bus driver assault risk on the ith segment Pi = probability of a successful assault on the ith segment Ci = consequences of a successful assault on the ith segment Furthermore, the probability of a successful assault is the product of the probability of a con- duct of assault and the conditional probability that that assault is successful. Mathematically, this can be described as: ( ) ( )= ×P P A P S A (2)i i i Where: Pi (A) = the probability of a conduct of an assault Pi(S|A) = conditional probability that an assault is successful Based on Equations (1) & (2), segment-specific risk would be: ( ) ( )= × ×R P A P S A C (3)i i i i Route risk is the summation of segment risk: ∑ ( ) ( )= × × = (4) 1 R P A P S A CT i i i i N A P P E N D I X H Framework for Route Risk Analysis of Driver Assaults Tools and Strategies for Eliminating Assaults Against Transit Operators, Volume 1: Research Overview Copyright National Academy of Sciences. All rights reserved.

146 Tools and Strategies for Eliminating Assaults Against Transit Operators: Research Overview Equation (4) can be used to assess the risk of bus driver assault(s) on a specific route. The route risk is affected by the potential for an assault (threat), the probability that an assault would be successful (vulnerability), and the severity of an assault (consequences). Any countermeasure may affect at least one of these risk components. There are three basic scenarios that the route risk may be zero: • No conduct of an assault • The conduct of assault is not successful • The assault yields no consequence (although this is not very likely) For example, if a countermeasure can 1) reduce the possibility of an assault; and/or 2) reduce the chance that an assault is successful; and/or 3) reduce the consequences of a successful assault, the route risk could be mitigated. Model Implementation In practice, the probability of an assault, the probability that the assault is successful and the assault consequences are difficult to estimate. There might be a relationship between these risk components and influencing risk factors. For example, the probability of an assault may be a function of a number of risk factors: ( )= , , . . . , (5)P F Risk Factor 1 Risk Factor 2 Risk Factor M If this complex relationship is unclear, a weight-sum approach may be used. Mathematically, it can be expressed as: ∑= α λ = (6) 1 P wj j j M Where: P = probability term (or the consequence term) α = scale factor to ensure that the probability is between 0 and 1 lj = numerical rating of the jth risk factor wj = weight of the jth risk factor M = number of risk factors considered For example, suppose that the probability of an assault (P(A)) is affected by Risk Factor (RF)1 & 2. The rating of RF1 is 0.7 (l1 = 0.7), its weight is 5 (wj = 5); and the rating of RF 2 is 0.4 (l2 = 0.4), its weight is 10 (wj = 10). If the scale factor is 0.1 (α = 0.1), the estimated prob- ability would be 0.75 using Equation (6). Similar calculations can be conducted for other risk components. The next step will be to assign a numerical rating to each risk factor and understand which factors affect the probability (and consequence) of an assault. Risk Reduction by Implementation of Countermeasures The countermeasures can reduce the likelihood of an assault (threat), the likelihood that the conduct of an assault is successful (vulnerability), or the consequence of a successful assault, individual or in combination. Without loss of generality, ∑ [ ] [ ] [ ]( ) ( ) ( ) ( ) ( )= − × − × − = 1 , . . . 1 , . . . 1 , . . . (7)1 2 3 4 1 5 6R P A T CT CT P S A V CT CT C D CT CTT i i i i i N i i T o o l s a n d S t r a t e g i e s f o r E l i m i n a t i n g A s s a u l t s A g a i n s t T r a n s i t O p e r a t o r s , V o l u m e 1 : R e s e a r c h O v e r v i e w C o p y r i g h t N a t i o n a l A c a d e m y o f S c i e n c e s . A l l r i g h t s r e s e r v e d .

Framework for Route Risk Analysis of Driver Assaults 147 Where: Ti(CT1, CT2, . . .) = percentage reduction in the likelihood of an assault Vi (CT3, CT4, . . .) = percentage reduction in the likelihood that an assault is successful Di(CT5, CT6, . . .) = percentage reduction in the consequence of a successful assault The next step will be to assign a numerical rating to each risk factor and understand which factors affect the probability (and consequence) of an assault. References Kaplan, S., Garrick, B. J., 1981. On The Quantitative Definition of Risk. Risk Analysis, 1(1), 11–27. Liu, X. Crude by rail route risk analysis and decision support system. Transportation Research Record (in review). Liu, X., Saat, M. R., Barkan, C. P. L. 2014. Probability analysis of multiple-tank-car release incidents in railway hazardous materials transportation. Journal of Hazardous Materials 276, 442–451. Liu, X., Saat, M. R., Barkan, C. P. L. 2013a. Integrated risk reduction framework to improve railway hazardous materials transportation safety. Journal of Hazardous Materials 260, 131–140. Liu, X., Saat, M. R., Barkan, C. P. L. 2013b. Safety effectiveness of integrated risk reduction strategies for rail transport of hazardous materials. Transportation Research Board 2374, 102–110. Tools and Strategies for Eliminating Assaults Against Transit Operators, Volume 1: Research Overview Copyright National Academy of Sciences. All rights reserved.

Abbreviations and acronyms used without definitions in TRB publications: A4A Airlines for America AAAE American Association of Airport Executives AASHO American Association of State Highway Officials AASHTO American Association of State Highway and Transportation Officials ACI–NA Airports Council International–North America ACRP Airport Cooperative Research Program ADA Americans with Disabilities Act APTA American Public Transportation Association ASCE American Society of Civil Engineers ASME American Society of Mechanical Engineers ASTM American Society for Testing and Materials ATA American Trucking Associations CTAA Community Transportation Association of America CTBSSP Commercial Truck and Bus Safety Synthesis Program DHS Department of Homeland Security DOE Department of Energy EPA Environmental Protection Agency FAA Federal Aviation Administration FAST Fixing America’s Surface Transportation Act (2015) FHWA Federal Highway Administration FMCSA Federal Motor Carrier Safety Administration FRA Federal Railroad Administration FTA Federal Transit Administration HMCRP Hazardous Materials Cooperative Research Program IEEE Institute of Electrical and Electronics Engineers ISTEA Intermodal Surface Transportation Efficiency Act of 1991 ITE Institute of Transportation Engineers MAP-21 Moving Ahead for Progress in the 21st Century Act (2012) NASA National Aeronautics and Space Administration NASAO National Association of State Aviation Officials NCFRP National Cooperative Freight Research Program NCHRP National Cooperative Highway Research Program NHTSA National Highway Traffic Safety Administration NTSB National Transportation Safety Board PHMSA Pipeline and Hazardous Materials Safety Administration RITA Research and Innovative Technology Administration SAE Society of Automotive Engineers SAFETEA-LU Safe, Accountable, Flexible, Efficient Transportation Equity Act: A Legacy for Users (2005) TCRP Transit Cooperative Research Program TDC Transit Development Corporation TEA-21 Transportation Equity Act for the 21st Century (1998) TRB Transportation Research Board TSA Transportation Security Administration U.S.DOT United States Department of Transportation Tools and Strategies for Eliminating Assaults Against Transit Operators, Volume 1: Research Overview Copyright National Academy of Sciences. All rights reserved.

Tools and Strategies for Elim inating A ssaults A gainst Transit O perators, V olum e 1: Research O verview TCRP Research Report 193 TRB TRA N SPO RTATIO N RESEA RCH BO A RD 500 Fifth Street, N W W ashington, D C 20001 A D D RESS SERV ICE REQ U ESTED ISBN 978-0-309-39024-8 9 7 8 0 3 0 9 3 9 0 2 4 8 9 0 0 0 0 Tools and Strategies for Eliminating Assaults Against Transit Operators, Volume 1: Research Overview Copyright National Academy of Sciences. All rights reserved.