Effectiveness of Taxicab Security Equipment in Reducing Driver Homicide Rates

      Background

      Taxicab drivers historically have had one of the highest work-related homicide rates of any occupation. In 2010 the taxicab driver homicide rate was 7.4 per 100,000 drivers, compared to the overall rate of 0.37 per 100,000 workers.

      Purpose

      Evaluate the effectiveness of taxicab security cameras and partitions on citywide taxicab driver homicide rates.

      Methods

      Taxicab driver homicide rates were compared in 26 major cities in the U.S. licensing taxicabs with security cameras (n=8); bullet-resistant partitions (n=7); and cities where taxicabs were not equipped with either security cameras or partitions (n=11). News clippings of taxicab driver homicides and the number of licensed taxicabs by city were used to construct taxicab driver homicide rates spanning 15 years (1996–2010). Generalized estimating equations were constructed to model the Poisson-distributed homicide rates on city-specific safety equipment installation status, controlling for city homicide rate and the concurrent decline of homicide rates over time. Data were analyzed in 2012.

      Results

      Cities with cameras experienced a threefold reduction in taxicab driver homicides compared with control cities (RR=0.27; 95% CI=0.12, 0.61; p=0.002). There was no difference in homicide rates for cities with partitions compared with control cities (RR=1.15; 95% CI=0.80, 1.64; p=0.575).

      Conclusions

      Municipal ordinances and company policies mandating security cameras appear to be highly effective in reducing taxicab driver deaths due to workplace violence.

      Introduction

      Workplace violence remains a leading source of occupational fatalities and injuries

      Bureau of Labor statistics, U.S. Department of Labor. National Census of Fatal Occupational Injuries in 2011. www.bls.gov/news.release/archives/cfoi_09202012.pdf

      with taxicab drivers historically experiencing one of the highest homicide rates of any occupation.
      • Richardson S
      • Windau J.
      Fatal and nonfatal assaults in the workplace, 1996 to 2000. Violence in the Workplace Special Issue.
      Since the mid-1990s, workplace homicides have declined in the general working population.
      • Hendricks SA
      • Jenkins EL
      • Anderson KR.
      Trends in workplace homicides in the U.S., 1993-2002: a decade of decline.
      However, homicide rates among taxicab drivers continue to rank among the highest of any occupation.
      • Richardson S
      • Windau J.
      Fatal and nonfatal assaults in the workplace, 1996 to 2000. Violence in the Workplace Special Issue.
      Despite a tremendous need for effective safety advances in this occupation, there is a paucity of research focused on evaluating the effectiveness of safety equipment in taxicabs.
      Two safety publications

      National Institute for Occupational Safety and Health (NIOSH) Alert. Preventing homicides in the workplace. 1995; DHHS (NIOSH) Publication No. 93-109. www.cdc.gov/niosh/homicide.html.

      Occupational Safety and Health Administration (OSHA) fact sheet. Workplace violence: preventing violence against taxi and for-hire drivers. 2000. www.osha.gov/Publications/taxi-driver-violence-factsheet.pdf.

      that summarized risk factors for work-related homicides have guided the taxicab industry and its regulators in the use of safety equipment to prevent workplace violence. In the past 20 years, the use of safety equipment in taxicabs occurred through ordinances promulgated by municipal transportation regulators or policies issued by large companies. Bullet-resistant partitions were the dominant safety equipment in use in the early 1990s. Currently, cameras are in greater use and have become the security equipment of choice for industry regulators and taxicab fleet operators.
      Although a comprehensive evaluation of interventions designed to reduce robberies in the retail industry has been undertaken,
      • Casteel C
      • Peek-Asa C.
      Effectiveness of crime prevention through environmental design (CPTED) in reducing robberies.
      there have been to date only two reports examining the effectiveness of taxicab safety equipment in reducing workplace violence outcomes.

      Stone JR, Stevens DC. The effectiveness of taxi partitions: the Baltimore case. Report prepared for the Southeastern Transportation Center, University of Tennessee–Knoxville, Knoxville, Tennessee. 1999. www.taxi-library.org/stone_abs.htm.

      Taxicab driver personal safety in Seattle and King County, final report and recommendations. The report of the Taxicab Advisory Group Committee on Driver Safety to the Director of the Department of Executive Administration for the city of Seattle. June 18, 2004.

      (In 1999, transportation researchers reported a 56% decrease in assaults after 12 months of mandatory partition installation in a pilot group of taxicab drivers.

      Stone JR, Stevens DC. The effectiveness of taxi partitions: the Baltimore case. Report prepared for the Southeastern Transportation Center, University of Tennessee–Knoxville, Knoxville, Tennessee. 1999. www.taxi-library.org/stone_abs.htm.

      ) Further, taxicab drivers with partitions experienced fivefold fewer assaults than taxicab drivers without partitions.

      Stone JR, Stevens DC. The effectiveness of taxi partitions: the Baltimore case. Report prepared for the Southeastern Transportation Center, University of Tennessee–Knoxville, Knoxville, Tennessee. 1999. www.taxi-library.org/stone_abs.htm.

      Subsequently, a comprehensive report in 2004 presented case studies for two cities using cameras in taxicabs and claimed the use of cameras resulted in decreases in both robberies and assaults.

      Taxicab driver personal safety in Seattle and King County, final report and recommendations. The report of the Taxicab Advisory Group Committee on Driver Safety to the Director of the Department of Executive Administration for the city of Seattle. June 18, 2004.

      Although both reports presented data supporting the use of partitions and security cameras as effective in reducing assaults, the findings were based on a short time period and a single city’s experience. Evaluating workplace violence incidents among taxicab drivers in multiple cities over a shared, longer time span would contribute to this limited body of research and provide stronger conclusions and generalizability of the findings.
      The study objective was to evaluate the effectiveness of safety equipment in reducing taxicab driver homicide rates. Specifically, it was hypothesized that installing cameras in taxicabs resulted in a reduction in citywide taxicab driver homicide rates (1) post-installation and (2) in comparison to cities without cameras. Second, it was hypothesized that cities with partitions installed in taxicabs experienced reduced taxicab driver homicide rates compared with cities without partitions.

      Methods

      A city was selected for inclusion in the analysis based on the following criteria: (1) being the most populated city within a metropolitan statistical area (MSA) with a population >250,000 and (2) maintaining taxicab licenses or playing a formal role in taxicab regulation. A list of the most-populated metropolitan statistical areas in the U.S. was generated from the U.S. Census Bureau.

      Census 2000 PHC-T-29. Ranking tables for population of metropolitan statistical areas, micropolitan statistical areas, combined statistical areas, New England city and town areas, and combined New England city and town areas: 1990 and 2000. www.census.gov/population/www/cen2000/briefs/phc-t29/tables/tab03a.pdf.

      One major city within each MSA was identified for every MSA on the list. If there was only one city for an MSA, data for that city were collected. Cities meeting Criteria 1 and 2 that did not have a substantial taxicab presence as determined by the city regulator were excluded in the evaluation.
      Homicide data were retrieved by conducting a Lexis-Nexis Boolean search designed to locate electronically published newspaper reports during 1996 through 2010 describing taxicab driver homicides, using the following algorithm: ‘cabdriver or cab driver or cabbie or (taxi w/2 driver) or livery driver or (limo! w/2 driver)’ w/15 dead or death or die or died or dies or slay! or slain or kill! or murder! or fatal! or mortal!) and ‘and not compiled by or obit! or subject (jury trial or mistrial or testimony or sentencing or verdict or decisions rulings or settlements or decisions)’. The and not section was designed to exclude articles on ongoing litigation. Each article was reviewed for duplication by trained data extractors. Data extracted on each taxicab driver who was ascertained to be a homicide victim were recorded for subsequent aggregation. To check for completeness, the name and date of each taxicab driver homicide was compared with a comprehensive list (www.taxi-library.org) memorializing the drivers created and maintained by a taxicab driver

      Rathbone C. Taxi Driver Memoriam List. www.taxi-library.org/murdrate.htm.

      in addition to verifying each city’s homicides with the city transportation regulator.
      Taxicab driver homicide rates consisted of the number of taxicab driver homicides (news clippings) divided by the number of licensed taxicabs (provided by municipal transportation regulators). Licensed taxicabs included medallions, liveries, and paratransit but excluded shuttles. Unauthorized taxicabs or taxicabs with expired vehicle licenses were excluded to the extent possible. Transportation regulators also described the type of security equipment installed in taxicabs (cameras, partitions, or neither) and provided the year most city taxicabs, if any, were installed with the security equipment. City homicide rates per 100,000 population were obtained from the annual Uniform Crime Reports published by the U.S Federal Bureau of Investigation (FBI) and represented the background crime rates for each MSA.

      Federal Bureau of Investigation Uniform Crime Reports. Crime in the U.S. (annual publication). www.fbi.gov/about-us/cjis/ucr/ucr.

      All data elements were recorded annually by city, spanning the years 1996 through 2010. Safety equipment was indicated by two mutually exclusive dichotomous variables—safety cameras or partitions—and was recorded annually according to installation status in the majority of taxicabs. A city was considered a “camera city” if more than 70% of the taxicabs were equipped with cameras. Similarly, a city was considered a “partition city” if more than 70% of the taxicabs were equipped with partitions. A “control city” was defined as having less than 10% of the taxicabs equipped with either a camera or a partition. These cut-points were used as they represent the distribution of safety equipment implemented as a company policy for cities without ordinances mandating safety equipment.
      A retrospective longitudinal time-series analysis was employed to evaluate the association of safety equipment type with taxicab driver homicide rates. The outcome variable was city taxicab driver homicide rate. The main effect independent variables were safety equipment type. Safety equipment status for each dichotomous variable representing safety equipment type was designated “1” beginning in the first full year safety equipment was implemented. In each city, the use of safety cameras or partitions was mutually exclusive. There were no lag periods created, as it was not expected that there would be a delayed effect of security equipment on taxicab driver homicide rates. A variable designating calendar year was included to control for the declining trend in homicide rates among taxicab drivers that began prior to 1996.
      • Hendricks SA
      • Jenkins EL
      • Anderson KR.
      Trends in workplace homicides in the U.S., 1993-2002: a decade of decline.
      All data were collected and analyzed in 1-year increments, with city being the analytic unit. Analyses were conducted in 2012 using PROC GENMOD in SAS, version 9.2. Generalized estimating equations were used to account for the serial correlation of the time series and allow for the clustering of data within cities. The natural logarithm of the number of licensed taxicabs by city each year was used as an offset variable.
      The taxicab driver homicide counts were assumed to follow a Poisson distribution; the offset variable provided the denominator used to calculate the homicide rates. The data were tested for dispersion and found to be slightly under-dispersed (scale=0.9), so that all reported CIs can be considered conservative in their range. Annual city-specific taxicab driver homicide rates were modeled on camera installation status to test the hypothesis that cities with cameras experienced a decline in taxicab driver homicide rates compared to cities with neither cameras nor partitions. The Wald test statistic determined significance.
      The same statistical model, restricted to camera cities, was used to test for the reduction in taxicab driver homicide rates post-installation compared with pre-installation. Taxicab driver homicide rates were modeled on partition installation status to test whether the hypothesis cities with partitions experienced lower taxicab driver homicide rates compared to cities with neither cameras nor partitions. The timing of the partition installations relative to the years examined precluded analysis of homicide rates post-installation versus pre-installation.

      Results

      Taxicab Driver Homicide Distribution

      News clippings data on the annual number of taxicab driver homicides, the annual number of licensed taxicabs, and city homicide rate were obtained for 26 cities. Taxicabs in eight cities were equipped with security cameras, taxicabs in seven cities had partitions installed, and 11 cities served as controls as they had neither partitions nor cameras installed. The camera and partition cities included in the analysis represent all of the cities eligible for the study. Table 1 presents the cities included in the analysis, their primary safety equipment designation, and, if applicable, the year and circumstance of widespread camera or partition installation. Cameras were installed in four of the camera cities due to a company policy, and four cities passed an ordinance mandating camera installation. Only one of the partition cities installed partitions during the time period evaluated in the study.
      Table 1Distribution of safety equipment by study cities: U.S., 1996–2010
      Camera cities (year installed)Partition cities
      All partition cities have citywide installation.
      (year installed if after 1996)
      Control cities
      Austin TX (2005)Baltimore MD (1999)
      Baltimore is the only partition city that did not have partitions installed before 1996.
      Atlanta GA
      Dallas TX (1999)Boston MACincinnati OH
      Houston TX (1999)Chicago ILColumbus OH
      Las Vegas NV (2005)
      Citywide camera installation per ordinance requirement
      Detroit MIDenver CO
      Orlando FL (2009)Los Angeles CAHonolulu HI
      Portland OR (2004)
      Citywide camera installation per ordinance requirement
      New York City NYMiami FL
      San Francisco CA (2003)
      Citywide camera installation per ordinance requirement
      Philadelphia PANew Orleans LA
      Seattle WA (2006)
      Citywide camera installation per ordinance requirement
      Reno NV
      Sacramento CA
      San Diego CA
      Tampa FL
      a Citywide camera installation per ordinance requirement
      b All partition cities have citywide installation.
      c Baltimore is the only partition city that did not have partitions installed before 1996.
      During the 15-year study period, news clippings identified 216 taxicab driver homicides in the 26 cities included in the analysis. The average number of taxicab driver homicides was 14 per year, with the minimum being three homicides (2007) and the maximum 24 (1997, 1998). Table 2 delineates the number of taxicab driver homicides per city and year, classified according to safety equipment status. Also included are the average number of licensed taxicabs per city, the average taxicab driver homicide rate per city, and the average homicide rate per city.
      Table 2Distribution of taxicab driver homicides by city and per year, U.S., 1996–2010
      Number of taxicab driver homicidesAverage number of licensed taxicabsAverage taxicab driver homicide rate
      Per 1000 taxicabs
      Average city homicide rate
      Per 100,000 city population
      199619971998199920002001200220032004200520062007200820092010
      Cameras
      City 1
      Cities with ordinance mandating taxicab cameras
      11
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      3820.352.7
      City 21
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      4490.165.5
      City 3
      Cities with ordinance mandating taxicab cameras
      1111
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      1,4220.2510.2
      City 411
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      1d
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      1
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      2d2,0920.198.7
      City 5121
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      1d
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      1d1,8610.218.0
      City 61
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      6690.103.7
      City 7
      Cities with ordinance mandating taxicab cameras
      1111
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      1,3530.225.6
      City 8
      Cities with ordinance mandating taxicab cameras
      111
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      6480.313.3
      Partitions
      City 93211d2d2d
      Designate year of camera or partition implementation
      1d1d1d2d1d
      Designate year of camera or partition implementation
      2d1d1,4000.9514.4
      City 10
      Designate year of camera or partition implementation
      2d
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      1d
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      1,6850.123.8
      City 111d5d4d
      Designate year of camera or partition implementation
      3d2d
      Designate year of camera or partition implementation
      2d1d3d1d
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      6,6460.227.1
      City 12
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      1d
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      1d1d
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      7500.2716.7
      City 13
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      3d1d3d1d1d
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      1d
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      2,1690.3110.2
      City 1414d5d4d4d11d4d5d2d3d4d1d
      Designate year of camera or partition implementation
      3d3d2d12,5170.356.6
      City 15
      Designate year of camera or partition implementation
      2d
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      1d1d1d3d
      Designate year of camera or partition implementation
      Designate year of camera or partition implementation
      1d
      Designate year of camera or partition implementation
      1d
      Designate year of camera or partition implementation
      1d1,6500.4410.4
      Control
      City 161111121111,5480.438.8
      City 1711124620.765.8
      City 18115000.276.5
      City 191118210.284.7
      City 202112,0000.132.3
      City 2121111131,9450.3410.0
      City 221511112111,6000.5825.2
      City 23
      These cities did not experience any taxicab driver homicides during the time span studied.
      2500.274.7
      City 24113320.365.5
      City 25219150.233.7
      City 26
      These cities did not experience any taxicab driver homicides during the time span studied.
      5770.005.1
      a Per 1000 taxicabs
      b Per 100,000 city population
      c Cities with ordinance mandating taxicab cameras
      d Designate year of camera or partition implementation
      e These cities did not experience any taxicab driver homicides during the time span studied.
      Figure 1 depicts the annual rate of taxicab driver homicides according to safety equipment type (neither cameras nor partitions is indicated as “control”). The taxicab driver homicide rates for partition cities and control cities were very similar for almost every year examined, peaking in 1998 and 2010. In general, the camera cities pre-installation experienced lower homicide rates than the partition cities, although for 2002 through 2004, homicide rates in camera cities were the highest of any of the groups. Finally, homicide rates in camera cities post-installation appeared to be lower for the majority of years than rates in camera cities pre-installation for comparable years.
      Figure thumbnail gr1
      Figure 1Distribution of taxicab driver homicide rates by safety equipment for cities included in analysis, U.S., 1996–2010
      Examining only the camera cities allows for a pre- and post-installation comparison of the number of taxicab driver homicide rates for each city (Figure 2). For every city, the taxicab driver homicide rate decreased post-installation. All of the cities with ordinances mandating cameras had no taxicab driver homicides after installation of cameras.
      Figure thumbnail gr2
      Figure 2Taxicab driver homicide rates for camera cities pre- and post-installation, U.S., 1996–2010
      Note: There were no fatalities post-installation for Cities 1, 2, 3, 6, 7, and 8. Actual number of homicides for each period are indicated above each column. aCities with an ordinance mandating cameras

      Effect of Cameras on Taxicab Driver Homicide Rates

      Model 1 in Table 3 tests Hypothesis 1a that taxicab driver homicide rates post-installation of cameras were lower than those pre-installation. The unadjusted effect of camera installation in reducing taxicab driver homicide rates was significant (RR=0.18, 95% CI=0.08, 0.43). After controlling for an annual change in taxicab driver homicide rates (“year”) and city homicide rate, the effect of camera installation remained significant (RR=0.14, 95% CI=0.07, 0.29). Model 2 describes the effect of camera installation compared to control cities (Hypothesis 1b). Both the unadjusted and adjusted effects of camera installation compared to control cities were significantly associated with reduced rate ratios: 4.8 times lower rates when unadjusted and 3.7 times lower rates after adjusting for annual taxicab driver homicide rate changes and city homicide rate (Table 3).
      Table 3Statistical models describing intervention effects on citywide taxicab driver homicide rates, U.S., 1996–2010
      Model 1
      Testing Hypothesis 1a, difference in taxicab driver homicide rates post-installation versus pre-installation of cameras
      Model 2
      Testing Hypothesis 1b, difference in taxicab driver homicide rates in camera cities compared to control cities
      Model 3
      Testing Hypothesis 2, difference in taxicab driver homicide rates in partition cities compared to control cities
      VariablesUnadjustedAdjustedUnadjustedAdjustedUnadjustedAdjusted
      Cameras installed0.18 (0.08, 0.43)0.14 (0.06, 0.29)0.21 (0.09,0.52)0.27 (0.12, 0.61)
      Partitions installed1.01 (0.64, 1.59)1.15 (0.80, 1.64)
      Year
      The rate ratio represents an associated increase in taxicab driver homicide rate for every increase of 1 year.
      1.04 (0.97, 1.11)0.96 (0.90, 1.07)0.92 (0.89, 0.95)
      City homicide rate
      The rate ratio represents an associated increase in taxicab driver homicide rate for every 1-unit increase in city homicide rate.
      1.05 (0.96, 1.16)1.04 (1.02, 1.06)1.05 (1.03, 1.08)
      Note: Values are rate ratio (95% CI).
      a Testing Hypothesis 1a, difference in taxicab driver homicide rates post-installation versus pre-installation of cameras
      b Testing Hypothesis 1b, difference in taxicab driver homicide rates in camera cities compared to control cities
      c Testing Hypothesis 2, difference in taxicab driver homicide rates in partition cities compared to control cities
      d The rate ratio represents an associated increase in taxicab driver homicide rate for every increase of 1 year.
      e The rate ratio represents an associated increase in taxicab driver homicide rate for every 1-unit increase in city homicide rate.

      Effect of Partitions on Taxicab Driver Homicide Rates

      The effect of partition installation compared to control cities (Hypothesis 2) found no significant association between citywide partition installation and taxicab driver homicide rates either before or after adjusting for annual taxicab driver homicide rate changes and city homicide rates (RRunadj=1.01, 95% CI=0.64, 1.59; RRadj=1.15, 95% CI=0.80, 1.64).

      Discussion

      These data support the hypothesis that installing cameras in taxicabs results in a reduction in citywide taxicab driver homicide rates post-installation (seven times lower homicide rate) and compared to cities with neither cameras nor partitions (three times lower homicide rate). The data do not support the hypothesis that cities with partitions installed in taxicabs experience lower taxicab driver homicide rates than cities with neither cameras nor partitions. This is the first study to methodically collect data from a nationally representative sample of the largest taxicab cities over a 15-year time span that allows for comparison of rates pre- and post-installation of cameras.
      Cameras are effective to the extent that they are used to their optimal performance and publicized. The ordinance requirements in some cities mandate that a decal be posted on the passenger windows to make passengers aware that they are under surveillance. Consistent with Crime Prevention Through Environmental Design Theory, would-be perpetrators are less likely to commit a crime while under surveillance.
      • Jeffery C.
      Crime prevention through environmental design.
      Another crucial component to ensure optimal performance of security cameras is maintaining cameras according to manufacturer’s instructions and not allowing security cameras to be intentionally disabled. Security cameras mandated by ordinance are checked for functioning at yearly inspections organized by city regulators, if not more frequently, when taxicab vehicles are checked for safety. Two of the three taxicab driver homicides that occurred in camera cities in 2010 had disabled cameras in their taxicabs (International Association of Transportation Regulators, personal communication, 2011). Specifically, the only city that experienced an increase in number of homicides after camera installation (City 4) was the city with the two murdered taxicab drivers whose cameras were disabled.
      All six taxicab driver homicides post-installation occurred in camera cities where cameras are required by company policy instead of by city ordinance. Although company policies for security camera installation may be effective, municipal ordinances requiring that all taxicabs be equipped with operating security cameras may be more effective. Such ordinances would ensure that individual owner–operated taxicabs and smaller businesses would use cameras, as do the nationally recognized taxicab companies that make up a large share of the market. Because deterrence through identification is one effect of camera installation, it is important that news reports mention the presence of a camera in the taxicab of a murdered driver, and post photos of potential suspects, so that potential perpetrators are aware of the possibility of being identified by surveillance cameras.
      The lack of an observed reduction in taxicab driver homicides in partition cities was unexpected. Partitions were implemented citywide because of ordinance requirements before 1996 in six of the seven cities examined. These were typically the cities that were experiencing the highest number of taxicab driver homicides, and also the highest crime rates in the sample. The benefit of the bullet-resistant partition, consistent with Situational Crime Prevention Theory, is that it is designed to give more power to the driver than to the passenger in regard to control of physical space. Additionally, it separates the target (cash held by driver) from the perpetrator.
      • Jeffery C.
      Crime prevention through environmental design.
      The news reports provided only partial information on location of the shooting relative to the taxicab—on average, 30% of these data were missing. For those news reports that provided the information, on average, 75% of reported locations were inside the taxicab. Details about whether they occurred because of an open partition or through the back of the driver’s seat are difficult to obtain, yet important for understanding the limitations of partitions.
      One suggestion for improving the effect of partitions may be to incorporate complementary safety features, such as signage indicating that minimum amounts of cash are carried by the driver, accompanied by installation of a cashless system, and GPSs for driver location.

      National Institute for Occupational Safety and Health (NIOSH) Alert. Preventing homicides in the workplace. 1995; DHHS (NIOSH) Publication No. 93-109. www.cdc.gov/niosh/homicide.html.

      Occupational Safety and Health Administration (OSHA) fact sheet. Workplace violence: preventing violence against taxi and for-hire drivers. 2000. www.osha.gov/Publications/taxi-driver-violence-factsheet.pdf.

      Although there was not an observed or significant reduction in taxicab driver homicides because of partitions alone, partitions could confer a protective effect in combination with additional safety measures. At this time, it is only speculation to decide which additional safety measures are needed, and further research evaluating additional safety measures for taxicab drivers is warranted.

      Limitations

      This study is limited by its ability to confer risk to individual taxicab drivers. Thus, because of the ecologic study design, the change in taxicab driver homicide rates (or lack thereof) in response to various types of safety equipment can be attributed only to citywide homicide rates. It is not possible to speak to the individual risks of taxicab drivers who have cameras versus partitions versus neither. However, this is a well-designed ecologic analysis that included all the major taxicab cities in the U.S. over a 15-year time span to allow for observed sustainable effects; it also incorporated a pre–post intervention with comparison group study design. The present study was conducted in response to a request by the International Association of Transportation Regulators, and the observations and findings of this research have implications for taxicab driver homicides in other countries where such homicide rates are considerably higher.
      Another potential limitation is under-reporting when using news reports in constructing an outcome measure. News reports on work-related homicides where police officers, convenience store clerks, and taxicab drivers are the victims tend to be well documented in electronic media. The search strategy used in the current study is methodologically rigorous and was conducted within a comprehensive electronic database. The results were validated by municipal taxicab regulators and were compared with data received from police departments for a separate phase of the overall study (covering 20 of the cities).
      An additional limitation is the use of secondary safety equipment, such as GPS devices that geographically track the taxicab, and alert devices were not included as potential covariates. These data were very difficult to obtain in order to record them annually, as most regulators do not document when they install secondary safety equipment. However, most of the cities have taxicabs equipped with GPS devices and alerts, and the use of secondary safety equipment is not predominantly associated with camera or partition cities.

      Conclusion

      The data suggest that citywide installation of security cameras in taxicabs may result in a sustainable reduction of the homicide rate among taxicab drivers. The current results are likely generalizable to countries with similar issues of taxicab safety and similar taxicab driver robbery and assault risk factors. Current research is planned to evaluate the effect of cameras in reducing robbery and assault rates by interviewing individual drivers.

      Acknowledgments

      The authors thank the International Association of Transportation Regulators (IATR) for their partnership in this considerable endeavor. In particular, Craig Leisy, with the Consumer Affairs Unit in Seattle WA provided several invaluable critical reviews of the study during the proposal and writing phases. Malachi Hull and Matt Daus, Presidents of IATR, and Karen Cameron were crucial for building partnerships with municipal agencies. The authors also express gratitude to Charles Rathbone, taxicab driver and assistant fleet manager who maintains the Taxi Library website (www.taxi-library.org), who was the inspiration for this study.
      The findings and conclusions in this report are those of the authors and do not necessarily represent the views of the National Institute for Occupational Safety and Health.
      No financial disclosures were reported by the authors of this paper.

      References

      1. Bureau of Labor statistics, U.S. Department of Labor. National Census of Fatal Occupational Injuries in 2011. www.bls.gov/news.release/archives/cfoi_09202012.pdf

        • Richardson S
        • Windau J.
        Fatal and nonfatal assaults in the workplace, 1996 to 2000. Violence in the Workplace Special Issue.
        Clin Occup Environ Med. 2003; 3: 673-690
        • Hendricks SA
        • Jenkins EL
        • Anderson KR.
        Trends in workplace homicides in the U.S., 1993-2002: a decade of decline.
        Am J Industrial Med. 2007; 50: 316-325
      2. National Institute for Occupational Safety and Health (NIOSH) Alert. Preventing homicides in the workplace. 1995; DHHS (NIOSH) Publication No. 93-109. www.cdc.gov/niosh/homicide.html.

      3. Occupational Safety and Health Administration (OSHA) fact sheet. Workplace violence: preventing violence against taxi and for-hire drivers. 2000. www.osha.gov/Publications/taxi-driver-violence-factsheet.pdf.

        • Casteel C
        • Peek-Asa C.
        Effectiveness of crime prevention through environmental design (CPTED) in reducing robberies.
        Am J Prev Med. 2000; 18: 99-115
      4. Stone JR, Stevens DC. The effectiveness of taxi partitions: the Baltimore case. Report prepared for the Southeastern Transportation Center, University of Tennessee–Knoxville, Knoxville, Tennessee. 1999. www.taxi-library.org/stone_abs.htm.

      5. Taxicab driver personal safety in Seattle and King County, final report and recommendations. The report of the Taxicab Advisory Group Committee on Driver Safety to the Director of the Department of Executive Administration for the city of Seattle. June 18, 2004.

      6. Census 2000 PHC-T-29. Ranking tables for population of metropolitan statistical areas, micropolitan statistical areas, combined statistical areas, New England city and town areas, and combined New England city and town areas: 1990 and 2000. www.census.gov/population/www/cen2000/briefs/phc-t29/tables/tab03a.pdf.

      7. Rathbone C. Taxi Driver Memoriam List. www.taxi-library.org/murdrate.htm.

      8. Federal Bureau of Investigation Uniform Crime Reports. Crime in the U.S. (annual publication). www.fbi.gov/about-us/cjis/ucr/ucr.

        • Jeffery C.
        Crime prevention through environmental design.
        Sage Publications, Beverly Hills CA1971
      9. Clarke RV Situational crime prevention: successful case studies. Harrow and Heston Publishers, New York1992