How We Did It
by Alison Walker-Baird
News-Post Staff

FREDERICK — The Frederick News-Post reviewed nearly 115,000 records contained in databases of 2002-2005 traffic stops compiled by Maryland State Police, the Frederick County Sheriff’s Office and the Frederick Police Department.

The analysis included data from about 10,400 state police stops, 51,000 city police stops and 53,400 sheriff’s office stops.

The databases contained each stop’s time, reason for the initial stop, duration of stop, and the driver’s race or ethnicity, county of residence, sex and age.

Records also indicated the date of a search, the justification for a search, whether an arrest was made and whether police seized illegal contraband or property.

Population estimates

One element of the analysis compared an estimated racial breakdown of drivers in the county and city against the racial breakdown of traffic stops.

Because blacks comprise about 6.4 percent of Frederick County drivers, one would expect blacks would also make up about that same percent of traffic stops by state and county police if race were not a factor, Hood College sociology professor Laura Moore said.

The News-Post used the current population of licensed county drivers, provided by the Motor Vehicle Administration; traffic stop experts agree data on licensed drivers provide a more accurate estimate of who’s actually on the road than U.S. Census data.

However, for city police whose traffic stops are mainly within city limits, The News-Post used city Census data.

The MVA does not break down its database of licensed drivers by city. To find city-level population data, The News-Post used the county’s planning division data, which is based on the 2000 census.

The data shows in the city, non-Hispanic whites make up about 81 percent of the city population, blacks 12.2 percent, Hispanics 3.7 percent and Asians 3 percent.

Limitations

In “By the Numbers: A guide for analyzing race data from vehicle stops,” sociologist and former criminology professor Lorie Fridell emphasizes limitations in data analysis that preclude a determination of racial bias.

In the 2004 book, funded by the Department of Justice’s Office of Community Oriented Policing Services, she outlines the difficulties and even impossibility of evaluating traffic stop data to find evidence of racial profiling.

Fridell is research director for an association of local and national police agency executives, the Washington-based Police Executive Research Forum.

Individuals of a given race may be stopped at a rate that exceeds their representation in the community, but that doesn’t necessarily indicate racial bias in policing, Fridell said in an interview this month.

She cites three reasons why a given demographic group could be stopped more often: that group drives more often, drives more in areas where police activity is highest or drives more poorly.

Fridell cautioned the third possibility could mean either whites or non-whites commit traffic violations at higher rates, a factor nearly impossible to determine. Knowing the likelihood of a given race to violate traffic codes would change whether stop rates should raise red flags, she said.

Searches also pose a unique problem in identifying bias, Fridell said.

Analysts shouldn’t assume every person stopped is at an equal risk of being searched, she said — the only real benchmark to link bias to searches is whether drivers were actually exhibiting behavior that should have led to a search, statistics impossible to determine.

The News-Post did not control for age when analyzing the police agency data, which could limit interpretation of bias, Fridell said: An area with a higher proportion of young people of any race would be more likely to be stopped by police than an area with high rates of elderly persons, who may drive less often.

Categorizing race and ethnicity

Though “Hispanic” denotes an ethnicity and not a race, The News-Post considered Hispanics a racial group for purposes of analyzing the data for discrepancies in the outcomes of traffic stops for Hispanics compared to whites, blacks and Asians.

The News-Post adjusted city police data to list Hispanics stopped in a separate racial/ethnic category.
The department’s original data spreadsheet included one column for race (most entries were A for Asian, W for white, etc.), and a separate category used an indicator for Hispanic.

The News-Post created a separate column in the spreadsheet so that all individuals with the Hispanic indicator were identified as “H” in the race/ethnicity column.

A small number of individuals stopped (a fraction of 1 percent) were listed as “black” or “Asian” in the race column but also had the Hispanic indicator; to simplify our analysis into five racial/ethnic groups The News-Post excluded these individuals.

Sheriff’s office stops

Sheriff’s office data included records in which multiple citations resulting from a single stop were listed separately, meaning that if an individual received five tickets and was searched, he appeared to have been searched five separate times.

The News-Post deleted these duplicate records, about 10 percent of the original 59,400 records, reducing the number of records to about 53,400.