Saint Paul Traffic Stop Data Analysis
- Short Report -
|
Introduction The Institute on Race & Poverty was asked by the Saint Paul Police Department on January 11, 2001 to analyze the traffic stop data collected by the Department from April 15 to December 15, 2000. This analysis has multiple purposes: to attempt to determine whether the Department engages in racial profiling, and if so, what the dimensions of the problem are; and to formulate suggestions for improvements to the Department's data collection program that will allow for more comprehensive analysis of future data. The Department has adopted the definition of racial profiling found in the U.S. Department of Justice Resource Guide on Racial Profiling Data Collection Systems:
The focus of the inquiry here is on racial profiling in traffic stops, both in the decision to initiate stops, and in subsequent decisions to search drivers and/or their vehicles.
Scope and Limitations of the Data We analyzed data from 41,249 traffic stops conducted between April 15 and December 15, 2000.2 For approximately the first five months of this eight-month data collection period, the Selective Enforcement Unit of the Department's Traffic and Accident Division3 recorded data on its stops in writing. Sometime in September, Traffic and Accident Division officers began recording data through the computer-aided dispatch (CAD) system used by other officers. Of the 41,249 stops for which we have data, 9,233 are stops for which data was recorded by hand by Traffic and Accident Division officers between April 15 and September 2000. The remaining 32,016 are stops for which data was recorded using the CAD system, including Traffic and Accident Division stops from September through December 15. For each of the 41,249 stops we have the following information: the gender and race/ethnicity of the driver, whether the officer conducted a pat down search of the driver's person, whether the officer conducted a search of the vehicle, and whether the driver was issued a tag (ticket). For the 32,016 CAD-recorded stops we also have the following information: the date of the stop, and the location of the stop by street address or street intersection. We compared the stop data to United States Census data on the adult residential population of St. Paul for the year 2000. The adult residential population of the city is not the ideal benchmark to which to compare traffic stop data. There are several reasons for this: St. Paul's driving population includes drivers who live outside the city; not every adult who lives in St. Paul drives, and driving frequency varies among those who do drive; resident drivers drive outside their neighborhoods of residence; and various parts of the city, such as major thoroughfares and the downtown business district, have various levels of non-residential drivers. Although not ideal, we assume for purposes of this report that comparisons of traffic stop demographics to adult residential population demographics will provide useful information about the interactions between St. Paul police officers and the city's driving population. The only census data currently available from the 2000 census is for the general population and for the population aged eighteen and over. When more comprehensive data from the 2000 census becomes available, particularly data broken down into more specific age categories, and data for automobile ownership rates, a more precise comparison of stop data to estimated driving population data will be possible. Ultimately, however, the only way to get a perfect comparison of stop data to driving population would be to do a specific empirical study of the driving population of the city of St. Paul. The available data for the period from April 15 to December 15 2000 is limited in several ways. Elements of potentially useful data that were not recorded for these stops include the reason for the stop, the age of the driver, and the results of the search if a search was conducted. One important caveat to the analysis is that the data does not include traffic stops that resulted in arrest. When a traffic stop results in an arrest, the Department reclassifies the incident as something other than a traffic stop, and removes the record from the traffic stop records. We do not know how many of these records there were in the eight-month period from April 15 to December 15, nor do we have any information about the racial breakdown of these incidents. The analysis that follows, therefore, covers only those traffic stops in which the driver was either tagged (ticketed), or released with neither a tag nor an arrest. Summary of Report Contents
Summary of Conclusions
Comparison of Citywide Stops to 2000 City Population According to the U.S. Census for the year 2000, the population of the city of Saint Paul is 65.58% non-Hispanic white, 13.28% Asian, 12.25% black, 7.29% Hispanic, and 1.6% Native American.5 The 2000 census data available at this time contains only one age breakdown, between persons under age eighteen, and those aged eighteen and over.6 The population aged eighteen and over does not correspond perfectly with the driving age population, but it is a closer match to the driving age population than is the total population. So, for purposes of comparison to the traffic stop data, we use the census figures for the population aged eighteen and over. That population in Saint Paul is 73.18% non-Hispanic white, 10.22% black, 8.99% Asian, 6.27% Hispanic, and 1.35% Native American.
As noted above, we analyzed data from 41,249 traffic stops conducted between April 15 and December 15, 2000, including 32,016 CAD-reported stops and 9,233 stops for which data was recorded by hand by officers of the Traffic and Accident Division. Because the racial demographics of the 9,233 Traffic and Accident Division stops differ significantly from the other 32,016 stops, we have done separate comparisons between the residential population and three groupings of stops: the 9,233 Traffic and Accident Division stops, the 32,016 CAD-recorded stops, and all 41,249 stops. Looking at all 41,249 stops for the eight-month period, 57.67% of the drivers stopped were white, 26.25% were black, 8.87% Asian, 6.71% Hispanic, and 0.50% Native American. As TABLE 2 shows, whites, Asians and Native Americans were stopped at rates lower than their representation in the over-eighteen population, Hispanics at a rate slightly higher than their proportion of the population,7 and blacks at a rate substantially higher than their representation in the population.
*Not a significant difference. For all races other than Native American, the racial demographics of the 9,233 stops done by the Traffic and Accident Division, which consist primarily of stops resulting from radar checkpoints, differ substantially from the total numbers. In this group of stops, 74.78% of the drivers were white, 15.60% black, 5.79% Asian, 3.55% Hispanic, and .28% Native American. As TABLE 3 shows, the racial disparities in stops of blacks and whites are less pronounced in the Traffic and Accident Division stops than in the total stops. Whites were stopped by the Traffic and Accident Division at a rate roughly equivalent to their representation in the population. Blacks were stopped at a rate higher than their share of the population, but lower than their rate in the total stops. Asians and Hispanics were stopped at rates lower than their share of the population, and lower than in the total stops.
Finally, when the 9,233 Traffic and Accident Division stops are subtracted from the total, the remaining 32,016 stops have greater disparities in the stops of blacks and whites than do the total stops. In this group of stops, whites were stopped at a rate that is barely more than two thirds of their proportion of the population, and blacks were stopped at a rate almost three times their representation in the population. Hispanics and Asians were also stopped at rates slightly higher than their population rates in this group of stops.9
1990 Automobile Ownership Rates by Household As noted above, the best benchmark for comparison of traffic stop demographics would be an empirical study of the driving population throughout the city of St. Paul. Another possible way to arrive at an estimated driving population is to look at the rates of automobile ownership by race. The Census Bureau has not yet released the household automobile ownership rates for the year 2000, however, so the 1990 census figures are the most recent available. In 1990, just over 90% of the households in St. Paul owning one or more automobiles were white households. Households of all races other than white constituted smaller proportions of the automobile-owning population than of the general population. (See TABLE 15) When the automobile ownership rates from the 2000 census are released, a closer estimate of the driving population will be possible. Until then, it is worth noting that automobile ownership rates are historically higher for whites than for people of other races. Automobile ownership is strongly tied to income level. Because relative income levels among people of different races have remained fairly consistent, we can presume that disparities in rates of automobile ownership persist. This suggests that the disparities between the driving population and the people being stopped may be more pronounced than comparisons to residential populations suggest.
As noted above, St. Paul's population is 65.58% white, 13.28% Asian, 12.25% African American, 7.29% Hispanic, and 1.6% Native American. The residential population of St. Paul is not evenly distributed by race throughout the city. Map 1 and Map 2 show the residential distribution of the general population and the adult population of St. Paul, respectively. As can be seen in Maps 1 and 2, St. Paul's African American population is largely concentrated in the central and northern parts of the city, in the neighborhoods of Downtown, Summit University, Thomas-Dale, Midway, North End, and Payne-Phalen. There are also significant African American populations in the southwestern tracts of the Greater Eastside neighborhood, and in the southeastern neighborhood of Highwood. The Asian population is largely concentrated in the central and northeastern sections of the city, in the neighborhoods of Summit University, Thomas-Dale, North End, Payne-Phalen, Dayton's Bluff, and Greater Eastside. The Hispanic population is concentrated mainly in the West Side neighborhood, and in the central Thomas-Dale neighborhood and the northern neighborhood of Payne-Phalen. The Native American population is also concentrated in the central and northern parts of the city, but the total population of Native Americans is so small that there is no significant concentration of Native Americans in any one neighborhood. Whites make up a majority of the population in most parts of the city, with the only exceptions being the Thomas-Dale and Summit University neighborhoods, and a few tracts in the North End, Payne-Phalen and West Side neighborhoods. Most of the western part of the city is almost exclusively white, including the neighborhoods of Como Park, St. Anthony, Merriam Park, Macalester-Groveland, Highland Park, Summit Hill, and West Seventh. Map 3 shows the distribution of traffic stops by race throughout the city.10 This map shows that the distribution of stops of Asian and Hispanic drivers roughly matches the residential distribution of those populations in the city. In other words, stops of Asian and Hispanic drivers occurred mainly in the neighborhoods in which Asians and Hispanics live, and in roughly the same proportion as their representation in those neighborhoods. In contrast, stops of African American drivers are more evenly distributed throughout the city, and are not as closely correlated with the distribution of the African American residential population. In all but two of the city's 82 census tracts, African Americans are stopped at rates greater than their proportion of the population. The two tracts in which stops of African Americans occurred at rates lower than the African American share of the population are tracts 335 and 336 in the Summit University neighborhood, the two tracts with the highest proportions of African American residents in the city. Over sixty-nine percent of the residents of tract 335 are African American, and 62.41% of the stops in that tract were of African Americans. Sixty-two percent of the residents of tract 336 are African American, and 44.19% of the stops in that tract were of African Americans. Stops of white drivers are also distributed through the city
somewhat more evenly than is the white residential population,
but the situation here is the converse of the situation with
African American stops. African American stops are more evenly
distributed than the African American population because in the
areas where their population rates are low, their stop rates
are higher than the population rates. White stops, on the other
hand, are more evenly distributed than the white population because
in areas where the white population predominates, their stop
rates are lower than the population rates. Whites were stopped
at rates lower than their proportion of the population in 74
of the city's 82 census tracts. Comparison of Stop Rates and Expected Stop Rates, by Census Tract In order to determine whether racially disproportionate stops were more common in certain areas of the city, we did two comparisons using "expected stop rates." The expected stop rate for people of a particular race in a given area is the rate at which people of that race would be stopped if the stops were absolutely proportional to the adult residential population of the area. For example, in a census tract that is 50% white, 30% black and 20% Native American, the expected stop rates would be 50% white drivers, 30% black drivers, and 20% Native American drivers. After calculating the number of expected stops for each race
in each census tract, we compared those numbers to the actual
stops in the tract. We then mapped the differences between expected
stops and actual stops in two ways. Maps 4, 5, 6, and 7
show the absolute differences between stops and expected stops
for African Americans, Hispanics, Asians and whites.11 For each race, these
maps show the number of stops above or below the expected number
of stops in each census tract. Maps 8, 9, 10, and 11 show
the differences between the expected stops and the actual stops
for each race as a percentage. Maps 7 and 11, showing the differences in absolute
numbers and percentage rates between stops and expected stops
for white drivers, are the converses of Maps 4 and 8,
dealing with stops of black drivers. Two notable exceptions to
this statement are tract 305 in the North End neighborhood, and
tract 361 in the West Side neighborhood, in which stops of both
whites and blacks were higher than expected. Tract 305 has one
of the highest rates of Asian population in the city, and Asian
drivers were stopped at a rate substantially lower than their
expected rate in that tract. Similarly, tract 361 has the highest
rate of adult Hispanic residents of any tract in St. Paul, and
Hispanic drivers were stopped at a rate substantially lower than
the expected rate in that tract. The tracts with the highest numbers of stops of black drivers over expected stops were generally tracts with above-average proportions of black residents and high numbers of stops. Examples of this are tract 315 in the Payne-Phalen neighborhood, with 935 total stops, an adult residential population that is 15.2% black, and 173.67 stops over expected stops for black drivers; and tract 325 in the Thomas-Dale neighborhood, which has an adult population that is 23.5% black, and had the highest total number of stops of any tract in St. Paul - 1488 - and the highest number of stops of black drivers over expected stops - 519.88. As noted above, the only two tracts in which black drivers were stopped at rates lower than their proportion of the residential population are tracts 335 and 336 in the Summit-University neighborhood, the two tracts with the highest proportions in the city of African American residents. In tract 335, disproportionate stopping of black drivers does not appear to be a problem. Black drivers were stopped at a rate lower than their population share in this tract. Yet, because such the total number of stops in tract 335 was so high - at 1293 the fourth highest stop total in the city - and the residential population in the tract is mostly black, a large number of black drivers were stopped in this tract. Eight hundred and seven black drivers were stopped in this tract, more than in any other tract in the city except tract 325. Although stops of blacks were disproportionately low in tract 335, the situation in this tract illustrates one of the reasons for the citywide disproportionality of stops of black and white drivers: the concentration of traffic stops in areas of the city in which the residential population is largely non-white. The ten tracts with the highest traffic stop totals account for 37.25% of the traffic stops in the city. Each of these ten tracts had more than 900 stops. Of these ten high-stop tracts, only tract 342 in the Downtown neighborhood has a white adult population close to the city average of 73.18%. The other nine tracts have white populations ranging from 14.3% to 62.2%, and six of the nine have white populations of 50% or less. Nine of the ten high-stop tracts have above average proportions of black residents. Nine of the ten have above average Asian populations, eight have above average Native American populations, and seven have above average Hispanic populations. (See TABLE 16) The next two sections of this report look at possible explanations for the high concentrations of traffic stop activity in these ten tracts.
Map 12 compares traffic stop rates and rates of calls for police service. In this map, the city is divided into grids that are smaller units than census tracts, because the Department keeps its records of calls for service by these grids. Map 12 shows a rough correspondence of stop rates and call-for-service rates. All of the grids with extremely low stop rates also have very low numbers of calls for service. However, while most of the grids with high stop rates also had large numbers of calls for service, some grids with high numbers of stops had call-for-service rates that were lower than the rates in grids with fewer stops. Grids 32, 33, 34, 35, 84, and 91 each had between 1236 and 2453 calls for service, and between 350 and 464 stops. On the other hand, grids 105, 110, 112, 131, 132, and 153 each had higher numbers of calls for service - between 2457 and 4692 - and lower numbers of stops - between 141 and 286. Increased police presence in some parts of the city due to large numbers of calls for service is a possible contributing factor to higher traffic stop rates in those areas, but it would not explain racial disparities in stops in those areas.
Comparison of Stop Rates to Traffic Accident Rates Traffic accident rates in different parts of the city provide some insight into the reasons for uneven distribution of traffic stop activity throughout the city. Traffic accident rates are an indicator of the levels of driving activity in various parts of the city. Map 13 shows levels of traffic stops and traffic accidents by grid.12 This map indicates only a moderate correlation between traffic accident rates and traffic stop rates. While accident rates are fairly high in most grids that have high numbers of traffic stops, several grids with high accident rates do not have high numbers of stops.
The overall ticketing rate for whites for all traffic stops was 70.4%, compared to 60.8% for blacks, 61.3% for Hispanics, 61.7% for Native Americans, and 64.1% for Asians, but the higher ticketing rate for whites is largely attributable to the fact that a greater proportion of stops of whites were conducted by officers of the Traffic and Accident Division, who issued tickets at much higher rates than did other officers. Virtually all of the 9,233 Traffic and Accident Division stops from April to September resulted in tickets. Every Hispanic, Asian and Native American driver stopped by the Traffic and Accident Division during that period was tagged, as were 99.9% of white drivers, and 99.8% of black drivers. (See TABLE 11) The ticketing rates for the CAD-reported stops vary slightly between races, with a slightly, but statistically significantly higher rate for white (58.4%) than black drivers (54.8%). (See TABLE 12)
Citywide Search Rates Data regarding searches of stopped drivers can be more informative than the stop data itself. In analyzing stop data, the ideal benchmark would be the driving population at the stop locations, which we can only approximate here using census data. For the search data, on the other hand, the stop data provides the most accurate benchmark. In analyzing search data, the relevant measure is the rate at which stopped drivers are searched, so the population benchmark for this analysis is the population of stopped drivers. In this case, however, it is important to bear in mind that all stops that resulted in arrests are excluded from the data. Assuming that all searches that turned up illegal drugs, illegal weapons, or other contraband resulted in arrests, the data analyzed here includes only searches in which no contraband was found. In determining search rates, we looked separately at the rates of pat down searches of drivers, and of vehicle searches. We then aggregated all incidents in which the driver, the vehicle, or both were searched. Incidents in which the vehicle was searched and the driver was not were extremely rare, representing only 1.6% of all search incidents. In 28.4% of search incidents, the driver was searched but the vehicle was not, and in 70% of search incidents, both the driver and the vehicle were searched. In the total 41,249 stops, African American, Hispanic and
Native American drivers were more likely than were whites and
Asians to be subject to pat down searches and vehicle searches
in which no contraband was found. This is also true of the 9,233
Traffic and Accident Division stops for April-September, and
of the 32,016 CAD-recorded stops, but the pat down and vehicle
search rates were substantially lower in the Traffic and Accident
Division stops. TABLE 5 compares the pat down search rates
for all three groups of stops. TABLE 6 compares the vehicle
search rates, and TABLE 7 compares the aggregated pat
down and vehicle search rates for all three groups of stops.
Blacks, Hispanics and Native Americans were searched at higher rates than white or Asian drivers. Higher vehicle search rates for black, Hispanic and Native American drivers may raise the question of whether drivers of color are more likely than white drivers to be subject to stops that are ostensibly made to enforce laws relating to traffic safety and vehicle equipment requirements, but actually intended to provide an opportunity to pursue non-traffic-related purposes. Stops of this type are closely correlated with complaints of racial profiling.13 As can be seen in TABLE 8, male drivers of every race are considerably more likely to be subject to vehicle and/or pat down searches than are female drivers. Male drivers of every race but Native American are more than twice as likely to be searched as are female drivers of the same race. Although the search rates for female drivers are consistently lower than for male drivers, the racial disparities are similar for drivers of both genders, with black, Hispanic and Native American drivers more likely than white and Asian drivers to be searched.
The racial disparities in the rates of searches in which no contraband was found, and particularly the high rates at which black and Hispanic men and Native American men and women were subject to such searches, raise serious questions. Data on the stops in which searches were productive of contraband would help to answer some of these questions. Studies done in other jurisdictions have found that "hit rates," the rates at which searches produce contraband, are generally about the same for blacks and whites.14 With data from the stops in which searches produced contraband and led to arrests, we could determine the overall search rates for drivers of each race, and the hit rates for drivers of each race. Search Rates by Location Search rates are not uniform throughout the city. For purposes
of mapping search rates, we aggregated all incidents in which
the driver, the vehicle, or both were searched. As can be seen
in Maps 14-17, search rates for all races tend to be higher
than average in two parts of the city: the area encompassing
the Thomas-Dale neighborhood and the northern part of the Summit-University
neighborhood; and the area encompassing the Dayton's Bluff neighborhood,
the southern half of the Payne-Phalen neighborhood, the southwestern
quarter of the Greater Eastside neighborhood, and the northwestern
corner of the Battle Creek neighborhood. These two areas roughly
correspond to two of the three zones in which the Department
implemented the HEAT program, an intensive enforcement program
described in a later section of this report. For Hispanic and
African American drivers, the areas of high search rates fan
out from the two core areas noted above, to encompass a larger
portion of the city. Comparison of Traffic and Accident Division Stops for April-September to Other Stops for April-September. Because the Traffic and Accident Division stops were recorded separately from other stops only for the five-month period from April 15 to mid-September, we have isolated the other stops for that five-month period from the total, in order to provide the most accurate basis for comparison of Traffic and Accident Division stops to other stops. Comparing these two groups of stops from the same time period, we see that the Traffic and Accident Division stops differ significantly from the other traffic stops. As noted above, the racial breakdown of the Traffic and Accident Division stops differs from that of the other stops. The other two significant differences between these groups of stops are in the ticketing rates and the search rates. Across all races, virtually everyone stopped by the Traffic and Accident Division received a ticket. (See TABLE 11) In contrast to this, ticketing rates in the other stops for this time period were 53.7% for whites, 53.8% for blacks, 54.8% for Hispanics, 56.5% for Asians, and 57.8% for Native Americans. (See TABLE 13) Conversely, people stopped by the Traffic and Accident Division were rarely searched. Only .3% of whites, 1.9% of Asians, 2.2% of blacks, 3% of Hispanics, and 7.7% of Native Americans 15 stopped by the Traffic and Accident Division were subject to either a pat down search, a vehicle search, or both. For people of all races, these rates are significantly lower than the search rates for the other stops.
There are disparities worth noting in the Traffic and Accident Division stops. Even in these stops, blacks are disproportionately represented, although the disproportionality is not as pronounced as it is in the other stops. Also, although the search rates for people of all races are low in these stops, people of all races other than white are still much more likely than whites to be searched. The disparities in the search rates in the Traffic and Accident Division stops are actually greater than in the other stops. Generally, however, it does not appear that the stops by the Traffic and Accident Division are the type of stops that typically give rise to claims of racial profiling. Stops like these in which virtually everyone is ticketed and hardly anyone is searched appear to be simple traffic stops. They do not appear to be the sort of stops generally associated with racial profiling.
Analysis of HEAT Zones During the eight-month period for which we have data, the Department implemented a program called Heavy Enforcement Activities for Thirty Days (HEAT), in three areas of the city. The first HEAT zone, in which the program was implemented from August 2 to August 21, is in the Thomas-Dale and Summit-University neighborhoods. HEAT was then concentrated in Downtown St. Paul from September 8 to September 27. The third zone, in which HEAT was implemented from September 29 to October 18, encompasses the Dayton's Bluff neighborhood and the southeastern part of the Payne-Phalen neighborhood. All three HEAT zones are in areas with high numbers of traffic stops. Map 18 shows the traffic stop rates for the city, with the three HEAT zones highlighted. Because increased police activity in particular zones is a key component of the HEAT program, we looked at the extent to which stop rates during HEAT periods might be elevating the stop rates in those zones for the eight-month period. We found that the overall stop rates in the Downtown and Dayton's Bluff HEAT zones (HEAT Zones 2 and 3) were not significantly affected by HEAT activity. Only in HEAT Zone 1 in Summit-University, was the stop rate during the HEAT period substantially higher than the overall rate. In each zone, the HEAT period was twenty days, or 8.16% of the 245 total days of the eight-month data collection period. In Zones 2 and 3, traffic stops during the HEAT periods actually comprised only 6.42% and 7.77%, respectively, of the total stops in those zones for the eight-month period. Stop rates during the HEAT periods were therefore slightly lower than the total rates in Zones 2 and 3. In Zone 1, 15.57% of the total stops occurred during the HEAT period, slightly less than twice the 8.16% share of stops that would be expected if stop rates were consistent throughout the eight-month period. (See TABLE 17) The implementation of HEAT in Zone 1 appears to have contributed to the high stop rate in that zone. Because the stop rate for HEAT Zone 1 was higher than normal during the HEAT period, we looked at the racial breakdown of the HEAT period stops, to see if the demographics of those stops affected the overall demographics of the stops in that zone. The stop rate for white drivers in zone 1 during the HEAT period was slightly lower than the total rate for that zone, and the stop rates of drivers of other races were slightly higher than average during the HEAT period. (See TABLE 17) However, the differences in stop rates by race during the HEAT period are not large enough to have a substantial effect on the total stop rates. Search rates vary widely among the three HEAT Zones, and the effect of HEAT on the search rates in the zones is inconsistent. Zone 1 had search rates for drivers of all races that were substantially higher than the city average rates. This is true both in general and during the HEAT period, but during the HEAT period, the search rates were somewhat lower for whites and higher for Asians than they were generally. Zone 2 generally had lower than average search rates for drivers of all races, but during the HEAT period the search rate for whites was considerably higher than it was generally. Zone 3 generally had a substantially higher than average search rate for Native Americans, and slightly higher than average search rates for all other races. During the HEAT period no Native American drivers were stopped in Zone 3, and the search rates increased for Asian drivers and declined slightly for drivers of all other races. (See TABLE 17)
The main issue raised by the comparison of the stop data to the city population data is the disparity between the African American population rate and the stop rate for African American drivers. African Americans constitute a significantly greater proportion of the drivers stopped in St. Paul than they do of the city's adult residential population. Hispanic drivers are also stopped in numbers higher than their proportion of the adult population, but the disparity is not as pronounced as it is for African American drivers. The search rates for African American, Hispanic and Native
American drivers are consistently higher than for white drivers.
This is true of the stops by officers of the Traffic and Accident
Division - stops in which the search rates are low - and of the
stops by other officers, which have higher search rates. Recommendations for Improving the Data Collection System Based on our analysis of the data recorded by the St. Paul Police Department, we have the following suggestions for changes to the data collection program that would allow for a more comprehensive and effective analysis of the data:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Footnotes 1. DEBORAH RAMIREZ, JACK MCDEVITT & AMY FARRELL, A RESOURCE GUIDE ON RACIAL PROFILING DATA COLLECTION SYSTEMS: PROMISING PRACTICES AND LESSONS LEARNED, published by the U.S. Department of Justice, November 2000. 2. In addition to the 41,249 stops for which we have usable data, another 1,241 stops were made during this period, but the records for those stops contain omissions and/or mistakes that render the data from those stops unusable. In calculating stop rates by race for comparison with census information, we excluded the 1,241 erroneous records. 3. The Selective Enforcement Unit of the Traffic and Accident Division deploys traffic cars in areas experiencing a high frequency of serious accidents, and focuses its enforcement efforts on hazardous traffic violations that contribute to such accidents. 4. The comparison of stop data to population data for Hispanics is complicated by the fact that the Department used Hispanic as a racial category, while the U.S. Census designates Hispanic as an ethnicity that crosses racial boundaries. In order to compare the stop data to census data, we assumed that a police officer would record Hispanics whose race was African American, Asian or Native Americans using those racial categories. Therefore, the census data we used to determine the Hispanic residential population of St. Paul was the data for people who designated their ethnicity as Hispanic and their race as either white or other. 5. In order to facilitate comparison of the census data
to the Saint Paul stop data, we are using the five race/ethnicity
categories used by the St. Paul Police Department in recording
the race/ethnicity of drivers stopped: white, black, Hispanic,
Asian, and Native American. We therefore include the separate
census category of Native Hawaiian/Pacific Islander in the Asian
category. The census designates Hispanic as an ethnicity that
crosses racial lines. For the purpose of comparing census data
to the stop data, we count those who identified themselves as
black and Hispanic, Asian and Hispanic, and Native American and
Hispanic in the black, Asian and Native American categories,
respectively. We include in the Hispanic category those who identified
themselves as Hispanic and designated their race as either white
or "other." 6. Census data was downloaded from the U.S. Census web site, using American Fact Finder. 7. We used the two-tailed binomial probabilities test to determine that the difference between the 6.27% Hispanic population rate and the 6.71% Hispanic stop rate is statistically significant. Without controlling for other factors, there is only a .03% probability that the difference between the Hispanic population rate and the Hispanic stop rate occurred by chance. 8. Tested Binomial Approximation of the Normal distribution to determine significance of differences between stops of all types and populations. 9. We used the two-tailed binomial probabilities test to determine that the difference between the 6.27% Hispanic population rate and the 7.62% Hispanic stop rate, and the difference between the 8.99% Asian population rate and the 9.76% Asian stop rate, are statistically significant. There is a less than .02% probability that these differences occurred by chance. 10. We do not have information about the location of the 9,233 stops for which data was recorded by hand by officers of the Traffic and Accident Division. All analysis of traffic stops by location concerns only the 32,016 CAD-reported stops. 11. We did not map this comparison for Native American drivers because the total citywide Native American population and the total number of stops of Native Americans were too small to allow a meaningful comparison of stops to expected stops. The small Native American population would lead to extremely low expected stop numbers in each census tract, so that a stop of even one Native American driver in most tracts would result in a very large percentage rate disparity between stops and expected stops. 12. For comparison of stop rates to traffic accident rates we are using a map generated by the Department, because we do not have the raw data used to generate the map. 13. See DAVID HARRIS, DRIVING WHILE BLACK: RACIAL PROFILING ON OUR NATION'S HIGHWAYS, published by the American Civil Liberties Union in 1999. 14. See DEBORAH RAMIREZ, JACK MCDEVITT & AMY FARRELL, A RESOURCE GUIDE ON RACIAL PROFILING DATA COLLECTION SYSTEMS: PROMISING PRACTICES AND LESSONS LEARNED, published by the U.S. Department of Justice, November 2000. 15. The number of Native Americans stopped by the Traffic and Accident Division was so small that the disparity between this rate and the others is not significant. Only 26 Native Americans were stopped by the Traffic and Accident Division, and two of them were searched. |
