Not so black
and white

An American-Statesman analysis of DPS troopers' stops and searches finds wide racial disparities

Between 2009 and mid-2014, a Texas Department of Public Safety trooper patrolled a three-county area between San Antonio and Houston, stopping 6,500 motorists. The officer, a 30-year veteran of the agency, was more than seven times more likely to search African-American and Hispanic drivers than white motorists — yet he was only half as likely to find contraband on the minority drivers.

The officer, whose patrol area centered around Victoria, Dewitt and Jackson counties, is an outlier. According to the American-Statesman’s first-of-its-kind analysis of the outcomes of 14 million DPS traffic stops by individual officers, most troopers searched minority and Anglo motorists at or near the same rates.

But experts say the dramatic racial disparity of a significant number of troopers’ stop and search rates raises questions about how well the state policing agency keeps an eye on, and responds to, individual officers with troubling track records. The newspaper’s analysis found that:

  • 35 percent of the 1,138 troopers included in the analysis searched black and Hispanic motorists at least twice as often as white drivers.
  • 231 of the officers who searched black and Hispanic motorists at two times or more the rate at which they searched white drivers were less likely to find contraband while searching the minority drivers.
  • 65 DPS officers searched minority drivers at least three times more often than the white motorists they stopped yet found contraband less often.
  • 16 officers searched minority motorists more than four times as often as Anglos, with lower contraband hit rates.

The Statesman studied traffic stops made by 3,700 DPS officers from 2009 to mid-2015. The paper narrowed its analysis to those troopers who made at least 100 stops and 20 searches of minority motorists, a methodology recommended by University of North Carolina professor and national racial profiling expert Frank Baumgartner to avoid statistical anomalies.

DPS officials strenuously objected to the paper’s findings. “Implying or insinuating racial profiling without concrete evidence is slanderous and unfairly tarnishes the reputation of the commissioned men and women of the Texas Department of Public Safety, who proudly risk their lives every day to protect and serve all residents and travelers in Texas,” a spokeswoman wrote in response to the newspaper’s questions.

The DPS reviewed several troopers identified by the paper as having high search rate disparities, analyzing their complaint history and video of traffic stops, and found “zero evidence of racial profiling,” said DPS spokesman Tom Vinger. He said that troopers being periodically deployed to the Rio Grande Valley as part of the agency’s border operations could result in some racial disparities.

Experts caution that proving a law enforcement officer treats citizens differently according to race is extraordinarily complicated. Even glaring variations in stop and search rates by race don’t necessarily prove bias, said Lorie Fridell, a criminology professor at the University of South Florida and author of “By the Numbers: A Guide for Analyzing Race Data from Vehicle Stops.”

“It’s easy to measure disparity,” she said. “It’s hard to prove out the causes.”

Yet others said that unusually skewed search rates are at least a sign that the DPS needs to take a more detailed look at its troopers’ track record of stops and searches by race. State Rep. Garnet Coleman, D-Houston, said the newspaper’s findings suggest that the state’s racial profiling law, which requires departments to publicly report a limited amount of traffic stop data, needs to be overhauled to make it more useful.

“Under the current definition of the law, DPS can show that they are not racial profiling,” Coleman said. “Let’s make sure it’s not just a law in spirit, but that the letter meets the reality. Some people say we’ve solved this problem; well, obviously not.”

Alex del Carmen, executive director of Tarleton State University’s School of Criminology, Criminal Justice and Strategic Studies, added that analyzing individual troopers’ stop and search patterns would be a valuable supervision and training tool. Many police departments already track other information signaling an officer could be headed for trouble, such as a high number of use-of-force incidents or a steady stream of citizen complaints.

“This should be in the forefront of what police departments are doing,” he said.

A 'sense' of profiling

The Department of Public Safety produces a report every year showing that the agency’s stops and searches of motorists are not out of line with state demographics. Steven McCraw, DPS director since 2009, also has noted that his agency has fielded just 40 citizen complaints alleging racial profiling in recent years — and that only one resulted in officer discipline.

Yet a review of that case demonstrates both the difficulty in proving bias, and the department’s reluctance to accuse its troopers of racially motivated behavior.

Agency records show that trooper Brandon Smith pulled over Patrick Strong on Feb. 2, 2013, on a Harrison County road, allegedly for swerving. When Smith discovered Strong is a licensed gun owner and had his pistol in the car, the trooper demanded it to verify its registration. Strong refused, telling the officer — correctly — that Texas has no gun registration requirement.

After moving to handcuff Strong at one point, and demanding access to Strong’s car and being refused, Smith called for a police dog to sniff around the vehicle for drugs. The dog registered a “hit,” but when the trooper searched Strong’s car, he found nothing.

Strong, an African-American attorney from Houston, accused Smith of racial profiling. (Strong didn’t return calls seeking comment for this article.) The department’s Office of Inspector General, charged with investigating allegations of DPS wrongdoing, issued its report in Dec. 2013.

The inspector general concluded that Smith “stopped minorities (African Americans, in particular) and vehicles displaying out-of-state license plates at an exceptionally high rate.” Investigators also noted that the justifications the trooper used for the traffic stops often were vague violations, such as driving too long in the passing lane or following another vehicle too closely. A review of Smith’s dashboard camera suggested that several drivers he stopped seemed not to have violated any laws.

Investigators also compared Smith’s stops against other local departments’ and the area’s demographics. “The evidence revealed Smith’s rate of contact with African Americans was exceptionally high and contrary to all other collected data,” their report concluded. (The Statesman’s trooper-by-trooper analysis found that Smith searched black drivers at nearly double the rate of white drivers.)

But the DPS rejected the racial profiling finding. The sample of stops examined by Office of Inspector General was too small, the department said, and comparing Smith’s numbers to other police departments’ wasn’t valid because local law enforcement has “a different responsibility than DPS.” Regional demographics — the same comparison used by the DPS in its annual reports — were a useless guide because they didn’t show the racial breakdown of the people traveling on highways, and “there is no such data available.”

A superior reviewing the investigation recommended the charges against Smith should “remove the reference to racial profiling.” In a September 2014 report, DPS executives stopped just short, concluding “the apparent targeting of African-American drivers for extensive questioning … and the high incidence of searches involving African-American drivers certainly provide the sense that Trooper Smith was profiling drivers on the basis of their race.”

They recommended he be fired for the “disturbing patterns,” as well as stopping and searching motorists, including Strong, without probable cause — but didn’t discipline him for racial profiling.

A month later, McCraw ordered Smith suspended for a month without pay and to attend specialized training.

Legitimate reasons for disparities

Several recent analyses of the agency’s stops and searches have identified other notable racial disparities. Last summer, after the suicide of Sandra Bland in the Waller County Jail after a contentious arrest by DPS trooper Brian Encinia, the Statesman found that black drivers stopped by DPS troopers in the county were more likely to be given a citation than Anglos, who more often received only warnings. (Encinia was fired earlier this year for violating department policies, including failing to be courteous and detaining Bland too long without cause.)

Two months later, a team of academics concluded that the DPS had “consistently misinterpreted” traffic stop statistics in its annual reports. The study found that troopers statewide searched black drivers at higher rates than white drivers.

A December Statesman analysis found Hispanic motorists stopped by DPS troopers were more likely to be searched than white drivers — yet those inspections of Hispanics’ vehicles were also less likely to result in the discovery of drugs, weapons or illegal currency than the searches performed on Anglos’ vehicles. It also determined that black men pulled over by the troopers were more than twice as likely to be searched as white men, even though those searches turned up only slightly more contraband. (Earlier studies found similar disparities in other Texas police departments.)

The DPS vowed to hire an outside expert to analyze its traffic stop data for signs of racial profiling. DPS spokesman Tom Vinger said the department is still preparing to solicit bids for that analysis.

Experts said there are legitimate reasons why a patrol officer might have steeply unbalanced search rates by race, including demographic differences in patrol areas, disparities in local crime rates and income inequality by race that would tend to result in less affluent drivers being stopped more frequently for vehicle violations due to poor maintenance.

Still, Baumgartner said officers who search minorities at twice the rate of whites are statistically significant “high disparity officers.” A search rate that is four or more times higher, he added, is “pretty astronomical. … It’s a very egregious ratio.”

Yet there is no statistical standard for demonstrating bias. In 2011, a federal judge agreed to certify a class-action lawsuit that claimed police around the East Texas town of Tenaha had targeted minority drivers to seize their assets. While no established line had been crossed, he concluded the numbers were impossible to overlook.

“The increase in the number of minorities stopped under the interdiction program was so remarkable,” wrote Judge T. John Ward, “that it is statistically impossible that it was the result of anything other than a decision to target racial and ethnic minorities.”

David Guillory, an attorney for Lone Star Legal Aid who represented the drivers in the suit, describes the Tenaha case as a rare “perfect storm” of evidence demonstrating police bias. In addition to the racial stop and search imbalance, fewer than 2 percent of minority motorists who had property seized were ever charged with a crime. Local officials still denied any racial profiling.

Stops, searches not part of warning systems

Baumgartner said keeping tabs on each individual officer’s stop and search rates could be an important tool for police departments. Many agencies already use so-called early warning systems that track officers’ patterns in an effort to identify problem behaviors early. Last year, President Barack Obama unveiled an initiative to urge more departments to use such information.

DPS spokeswoman Summer Blackwell said the agency’s early intervention system is programmed to spot unusual numbers of vehicle pursuits, accidents, use of force and lack-of-courtesy complaints. Supervisors also regularly conduct spot checks of troopers that include reviewing dashboard camera video of traffic stops, and they can analyze individual search rates for troopers “if warranted,” she said.

Austin Police Department’s “guidance advisory program” tracks an officer’s use of sick time, number of internal affairs investigations and use-of-force reports, said Sgt. Brandon Hanly, who heads the agency’s risk management division. The department is considering updating the program to include an analysis of how officers charge defendants — flagging, say, a high number of “Interfering with public duties” charges, which might indicate an officer is reacting emotionally to confrontation.

Hanly said the new version likely won’t include a racial breakdown of stops or searches, however, because such tracking would be too labor intensive. But it’s not impossible: Guillory noted that the settlement agreement in the Shelby County asset-seizure lawsuit required officials to review each officer’s stop and search records every quarter for signs of disparities.

Del Carmen, who owns a consulting company that trains Texas police chiefs how to measure and mitigate racial profiling, agreed that highway patrol agencies, in particular, could benefit from routinely identifying individual officers whose search, stop and hit numbers are especially skewed.

While an agency-wide analysis is interesting, “Bias is more frequent at the individual level than at the department level,” he said. “Ultimately, racism comes from one or two individuals, more often than institutionally.”

How we got the story

For this report, the Statesman analyzed 14 million records of traffic stops between 2009 and mid-2015.

To compare search rates by race, we looked at individual troopers who had at least 100 stops and 20 searches of either black or Hispanic motorists, and then compared the number of searches to stops to get a search rate for that race. We did the same for their stops of white motorists, and then divided the rates to get a ratio of minority stops. For contraband hit rates, we compared the number of positive contraband results to the total number of searches by race for each trooper. DPS misidentified many Hispanic motorists as Anglo during the period studied, meaning the paper’s conclusions are most likely conservative.