SCPD Traffic Stop Data Reforms

All consideration and discussion of SCPD raw traffic stop data recording and analysis must take into account the fact that SCPD deliberately and surreptitiously altered the 2016 and 2019 raw traffic stop data files that it posted to its website by adding more than 112,000 duplicate traffic stop records to them.   Those deliberately altered traffic stop records remained posted to the SCPD website for years, until spring 2021 – after UJPLI requested an explanation for their inclusion in the data sets.   

• The duplicate records were generated in bulk, in iterations of 826, 1,652, and 2,478 between 2016 and 2019.

• They were identical in all respects (date, time, location, unit, license plate, gender, race, age, etc.) with one exception, the purported precinct of occurrence. The precinct designation was consistently allocated among the various precincts according to the same prescribed percentage scheme.

• The scale of the duplicate records is noteworthy:

  • More than 13% of total raw traffic stop records generated between 2016 – 2019.

  • Nearly 35% of total raw traffic stop records generated in calendar year 2017

    • 22% of Q1 2017 traffic stop records

    • 30% of Q2 2017 traffic stop records

    • 39% of Q3 2017 traffic stop records

    • 43% of Q4 2017 traffic stop records

  • 13% of total raw traffic stop records generated in calendar year 2018

    • 39% of Q1 2018 traffic stop records

• Given the magnitude of the deception, Commissioner Hart’s claim that the effort was ‘a misguided attempt to make the data more digestible’ is implausible on its face.

• UJPLI analysis revealed the actual, and likely intended, impact of the duplicate traffic stop records:

  • The duplicates overwhelmingly purported to represent discretionary enforcement interactions with Whites, thereby, inflating the actual number of discretionary enforcement interactions with Whites and distorting the actual rates of enforcement interaction experienced by and between the various racial/ethnic cohorts.

  • The principal impact of the duplicate records on digital analysis of the data is that they materially diminished the statistical disparity in discretionary enforcement interactions between Whites and minorities.

• The refusal of the office of the Suffolk County Executive, the Chairman of the Legislature’s Public Safety Committee and the Police Commissioner to explain how, when, why and by whose authority the duplicates were generated goes to the heart of their individual and collective integrity. It amounts to a shocking and inexcusable abuse and betrayal of public trust.

• The question of agency must be answered if public trust is to be restored.

  • Who authorized and initiated the practice?

  • If the practice was undertaken at the direction of or with the authorization of the office of the Police Commissioner for legitimate purpose, and not in violation of NYS PL § 175.20 Tampering With Public Records in the Second Degree, why was the addition of the duplicate records not disclosed to the public that relied on the accuracy of the data reported?

  • If it was undertaken without the express authorization of the office of the Police Commissioner, who had the delegated authority to make such a decision?

    • Who is responsible?

    • Why were the duplicate records removed from the raw traffic stop data records that are posted to the SCPD website?

    • Who has been held accountable?

    • Given the review and audit requirements of Chapter 13 Section 9, and the provisions of the Settlement Agreement that require annual review of website data for accuracy, how could unauthorized duplicate records have been posted to and remained on the website for years undetected?

    • What internal controls, if any, have been implemented to prevent a recurrence?

In addition, familiarization with SCPD’s Traffic Stop Data Analysis Timeline is essential for comprehensive understanding of SCPD’s reform effort.

What did the September 13, 2011 DOJ Technical Assistance Letter to Suffolk say about traffic stops?

SCPD Should Revise the Use of Roadblocks in Latino Communities 

We understand that SCPD has used a vigorous program of roadblocks and police sobriety checkpoints. As law enforcement tools, these are established methods of addressing drunk driving and other potential criminal activity. We have received reports, however, that SCPD’s checkpoints have been used primarily to request documentation of citizenship. While we continue to investigate these claims, if true, this is not an acceptable practice. We recommend that SCPD ensure that officers at checkpoints inspect only for sobriety or other specific illegal conduct and do not conduct identity checks or otherwise ask for documentation without a basis for believing that a crime or violation has been committed. 

SCPD may wish to review research conducted by the Rand Corporation as part of a collaborative agreement with the City of Cincinnati, Ohio. Such research can assist the department in learning the cause, nature, and extent of community disengagement. Another consideration would be to transform the new “research section” of SCPD as part of a dedicated audit function and inspection section. Specifically, all data that are currently entered into all databases (all enforcement actions and procedures, investigation, complaints, incident reports, etc.) should have dedicated software to articulate trend analysis by each individual variable. This form of predictive policing enables departments to anticipate trends and better serve the community.

See Ridgeway, et al., “Police-Community Relations in Cincinnati,” prepared by the Rand Corporation, available at http://www.rand.org/pubs/monographs/MG853 (last visited July 12, 2011).

SCPD may wish to work with local universities to examine police empirical data and outside independent analysis. 

What does Rand Corporation say about analyzing traffic stop data for the influence of racial bias?

RAND’s analysis of traffic stop data for potential race bias employs an iterative process that examines the data at the department level, at the individual officer level and in post-stop outcomes to assess the potential influence of race bias in police officer decisions to stop, cite, search and arrest.  Relevant post-stop outcomes include, but are not limited to: citation rates, stop durations, search rates, and search hit rates.  Rand indicates that analysis of data over time is essential to reliable analysis of the influence of racial bias in discretionary traffic stop decisions and outcomes.  

In its first evaluation report, “Bias-free Policing and Officer Accountability,” RAND reviewed the internal systems and processes the Cincinnati Police Department used in order to “identify and better understand the reasons for statistical differences, especially related to arrests, traffic stops and pedestrian stops,” and “identify potential “at-risk” police officers – those likely to engage in damaging behavior – before the behavior occurs.”

The following factors, cited by RAND as relevant as regards effective assessment, are instructive:

• Data quality is of paramount importance for accurate reliable analysis. A robust data platform and qualified dedicated team are essential. Failure to accurately and consistently record all relevant stop data subjects all analyses to potential biases in reporting.

• Officers whose stop decisions are racially biased are much more likely to use equipment violations as a pretext for initiating vehicle stops.

• Citation rates, stop duration, search rates and search ‘hit rates’ are important indicators.

• Search ‘hit rates’ reflect the percentage of searches that yield contraband. Distinctly lower hit rates for a given racial group as compared to the hit rates of other racial groups indicate excessive searches of the racial group(s) with the lower hit rate evincing racial bias in the decision to search.

• Stops that are racially biased have a tendency to be longer in duration than non-biased stops.

• Stops that are racially biased tend to yield higher search rates than non-biased stops.

• Racially biased searches tend to be more thorough and longer in duration than non-biased searches.

What did the January 11, 2014 DOJ Settlement Agreement require SCPD to do?

SCPD will implement a revised Chapter 13, Section 9, "Traffic Stop Data Collection," as previously approved by the United States.


One year after the Effective Date and annually thereafter throughout the pendency of this Agreement, SCPD will provide to the United States a report analyzing the collected traffic stop data and explaining what measures, if any, SCPD will take as a result of the analysis. 

SCPD will check its website for accuracy, formatting, and ease of comprehension within 90 days of the Effective Date and then annually thereafter throughout the pendency of this Agreement.

What did the March 12, 2021 Report and Recommendations of the Eastern District Court say?

The SCPD targets Latinos for stops, establishing a pattern and practice of discriminatory policing. 

Plaintiffs allege that SCPD officers have stopped, ticketed, searched, and/or arrested Latino motorists on the basis of their ethnicity . . . In support of this claim, they proffer an expert report,16 which analyzed the SCPD’s available traffic stop data and found that Latinos are disproportionately mistreated after being stopped. Ex. T, “Smith Report.”17Defendants reject these findings without much elaboration . . . (“[T]heir expert reports are based upon speculative allegations not supported by the record, or statistical formulas that show nothing more than the fact that Hispanic persons experienced traffic stops in Suffolk County.”) Plaintiffs’ expert report was based on an audit of SCPD’s own traffic stop data. If the data shows no more than the fact that Latinos were stopped by the SCPD, that is likely due to the insufficiency of the collected data, as found by the DOJ’s compliance reports and discussed below. 

Michael R. Smith, J.D. PH.D. is a Professor and Chair in the Department of Criminology and Criminal Justice at the University of Texas at San Antonio with over 25 years of experience as a researcher on police behavior and the criminal justice system. See Smith Report at 3. 

”Notwithstanding the limitations on the data, we conducted bivariate post-stop analysis on the traffic stop data, which found statistically significant, higher percentages of Latino motorists who were arrested and ticketed compared to Whites across all years examined and lower percentages who were warned or released with no further action taken....in my opinion these results are ... consistent with ethnicity-based enforcement.” 

The SCPD has failed to collect reliable data and has failed to assess its data to prevent biased policing. 

Plaintiffs allege that the data collected by the SCPD, including data collected since the DOJ Agreement, is unreliable and that if the data had been collected and audited properly, the SCPD would be on notice that their biased policing continues. Smith Report at 50–51.

It is well established that the data collection is unreliable and that an assessment of the data was not timely made by the SCPD. The Settlement Agreement mandated that SCPD implement a system to collect traffic stop data so that the SCPD could properly analyze that data. Following the Agreement, The United States issued periodic reports that assessed the SCPD’s compliance with the terms of the Agreement. These reports reveal that the SCPD has never been in full compliance with the Agreement’s data requirements. Both the United States and the SCPD acknowledge that the data that has been collected is unreliable. Further, the SCPD has admitted that, at the time the instant motion was filed, the SCPD never analyzed its stop data as required by the DOJ Agreement.

See also Smith Report at. (“The primary takeaway from this review of the data is that SCPD data collection protocols changed every 2–3 years and never consistently captured the fields necessary to determine whether Latino motorists were treated disparately from White motorists in key post-stop outcomes.”)

There are seven Compliance Assessments from the DOJ. 

The most recent assessment states that the SCPD is in “partial compliance” with the Agreement’s requirements for stop data. Ex. J, “October 2018 Assessment.” The report defines partial compliance as “the County has achieved compliance on some of the components of the relevant provisions of the Agreement, but significant work remains.” 

Other recent assessments state: “[a]s set forth in our last Report, the data collected by SCPD omits critical variables that are necessary for meaningful analysis of bias-free policing,” such as “why a stop was initiated,” and the results of conducted stops, “including whether a particular search revealed contraband or not.” Ex. QQ, “January 2017 Compliance Assessment,” 6–7; “[The SCPD] remains in partial compliance...due to the continued failure to implement an adequate data collections system.” Ex. DDD, “March 2018 Compliance Assessment.”  In 2017, “after months of preparation” leading up to the launch of a new data system, “SCPD discontinued using the system the very day it launched it” due to implementation issues. Id.; “We renew our recommendation that SCPD supervisors develop specific protocols for the substantive review of traffic stop data as a part of supervisors’ regular supervisory activities and that SCPD provide renewed training for supervisors, many of whom have not received supervisor-specific training since attaining the rank of sergeant.” October 2018 Assessment at 7. 

A 2016 Report issued by the SCPD noted that after launching a new system in October 2014, due to a “computer glitch” it did not detect approximately “7,000 incomplete entries,” rendering data unreliable from October 2014 until July 2015. Ex. X, “SCPD’s February 2016 Report.” The same report indicated 139 incomplete traffic stop entries for the month of December 2015 alone. Id. See also Ex. Z, “SCPD’s March 2017 Report,” 4 (the SCPD agreed with the DOJ’s conclusion that “the existing capture fields require more specificity” to conduct a “meaningful analysis of the data”); Ex. R, Love Deposition I at 182 (Q: What steps are taken to ensure that the officer is inputting the information accurately, correctly? A: There would be none [...]. Q: (same regarding the input of the correct race of the individual stopped) A: There are none). 

Sergeant Christopher Love is an attorney assigned to the SCPD Legal Bureau. Ex. 3, “Affidavit of Christopher Love.” He was designated as the County’s Rule 30(b)(6) witness on policies related to complaints, the Settlement Agreement, and data collection. 

In addition to the traffic stops and post-stop searches, plaintiffs allege that the SCPD uses traffic checkpoints in an unlawful, discriminatory manner to target Latinos . . . No checkpoint data was collected until 2018 and the SCPD still does not collect data on post-stop outcomes. Plaintiffs allege this data field is necessary at minimum to assess these stops for bias. See Smith Report. at 33–35. Similar to the traffic stop data, the SCPD has never analyzed checkpoint data to assess whether checkpoints were being disproportionately set up in Latino neighborhoods nor to otherwise assess its use of checkpoints. In summary, plaintiffs argue that by failing to collect accurate traffic stop and checkpoint data, the SCPD has employed a successful strategy to escape a finding of discrimination. Plaintiffs allege that, had data been accurately collected and timely assessed as required by the agreement, the SCPD would have been alerted to concerning trends regarding ongoing discriminatory policing. 

Defendants have filed a report commissioned from the John F. Finn Institute for Public Safety, Inc. as an exhibit with their pending motion for summary judgment. The Court notes that this report was completed in late September 2020, on the eve of defendants’ summary judgment motion deadline, nearly seven years after defendants entered the DOJ Agreement, and nine years after the SCPD was formally put on notice regarding its insufficient data collection in the Technical Assistance Letter.

See Love Deposition I at 149–50 (Q: So R&D has not conducted an analysis of the traffic stop data[?] A: No. That’s correct, they have not.); Id. at 115 (Q: So the [traffic stop] data has never been reviewed for atypical traffic stop activity? A: No. [...] Q: “Commanding officers shall review the annual report” [...] So that also has not been done? A: Correct).

See Ex. W, “Love Deposition II” at 667–68.


How has SCPD complied with DOJ’s guidance and the provisions of the Settlement Agreement?

Independent analysis of SCPD traffic stop data reveals the following:

The integrity of SCPD raw traffic stop data is critically compromised.

Blank fields, inconsistent criteria, duplicate entries, missing records and anachronistic entries compromise analysis.  SCPD Historical Traffic Stop Data is so replete with blank fields, duplicate and anachronistic entries and changes in reporting criteria that it is incapable of supporting meaningful analysis by the average community member with consumer grade computing resources.  It also calls into question whether the data enables the Department itself to fulfill its commitments to DOJ and the community - or whether the Department has access to a separate set of data.  The inadequacies include, but are not limited to the following:   

• Anachronistic entries:

  • Q1 2017 and Q3 2018 files – all records not in chronological order

  • Q4 2019 file contains 10,029 records from January 2020 (Q1 2020)

• Blank fields: all files are replete with blank fields (fields containing no data) – as distinguished from fields that commonly contain “NULL” entries

• Missing records

  • Q2 2014 – no records of activity prior to April 3rd

• Approximately 112,000 non-random, bulk duplicate traffic stop records.

  The persistence, scope and scale of these potential discrepancies are manifestly evident.  Two aspects of the apparent duplicate record entries are particularly troubling:  In the first instance, they occurred in blocks of 826 records per discrete CCNo (one unique record followed by 825 duplicate records), 827 records per discrete CCNO (one unique record followed by 826 duplicate records), blocks of 1,652 records per discrete CCNo (one unique record followed by 1,651 duplicate records) and 2,478 records (one unique record followed by 2,477 duplicate records).  If the ensuing duplicate records were identical – if each of the values in each of the fields was identical to those of the original record - that would be indicative of a strong possibility that the duplicate entries were the product of inadvertent human error or of a programming glitch.  But that is not the case.  The values (designations) in the Precinct fields of the duplicate record entries were consistently applied pursuant to an apparently predetermined percentage scheme that allocated the duplicates among the various precincts, although the majority were allocated to no precinct – the field was left blank.  Many of the duplicates’ “Unit” and “Precinct” fields are blank.  Beyond that, the overwhelming majority (more than 74%) of the duplicate records reflected interactions with whites, thereby distorting the true level of discretionary enforcement interactions between the respective demographic groups – particularly among minority groups. The appearance of the potential for mischief is indisputable. 

  Ambiguous undefined terms and codes, by field, include, but are not limited to:

  • Field: “ReasonTStopDesc”

    • OTHER VTL | OTHER MOVING VIOLATION

  • Field: “VehicleSearchReason

    • CONSENT | CONSENT SEARCH

  • Field: “VehicleSearchType

    • SEARCH WITHOUT CONSENT | SEARCH WITHOUT SIGNED CONSENT 

  • Field: “PersonSearchOutcome”

    • pso0000 | pso0001 | pso0010 | pso0011 | pso0100 | pso0101 | pso0110 | pso1000 | pso1001 | pso1010

  • Field: “DurationOfStop”

    • “1” | “2” | “3”

  • Field: “Stop Disposition”

    • “Other”

  • Field: “EquipViolations”

    • “1” | “2” | “3” | “4” | “5”

  • Data integrity was further compromised by

    • Unreliable location data

    • Unreliable sector data

    • The absence of any information regarding DWI enforcement activity

    • The absence of any charge information related to arrest activity

    • The absence of a glossary of terms and codes

Analysis of the data by the criteria set forth by Rand revealed striking racial disparities in discretionary decisions and post-stop outcomes that indicate inequity. 

Equally troubling is the engagement performed by Finn Institute.  The Department was obligated to begin performing and reporting on annual analysis of traffic stop data in 2015 yet failed to produce a report until October 2020.  Notwithstanding the dubious methods Finn employed, the questionable research that it relied upon and its inability to adequately justify its finding of no inequity, a critical question about the data set that it analyzed has gone unanswered since October 2020.  Finn reports reviewing a total of 146,320 records that were generated between March 5, 2018 and March 4, 2019.  Those records are contained within SCPD traffic stop data files TStop_Web_2018_Q1.csv through TStop_Web_2019_Q1.csv.  Isolating the records generated between March 5, 2018 and March 2019 yields a total of 150,992 total records. Examination of those records reveals no fewer than 4,956 non-random bulk records that were attributed to five discrete CC numbers (18-0144377; 18-0152001; 18-152034; 18-0153869; and 18-0154612).  

In each instance, an ostensibly unique record is followed by consecutive records with identical values in each field – with the exception of the Precinct field.  That is, the CCNo., Date, Time, Unit, Plate, State, Age, Gender, Race, Latitude and Longitude fields contain identical values.

If each of the values in each of the fields was identical to those of the original record - that would be indicative of a strong possibility that the consecutive entries were the result of inadvertent human error or of a programming glitch. But that is not the case. The values (designations) in the Precinct fields of the duplicate record entries vary; in fact, in each instance, the duplicate records were allocated among the precincts according to a consistent percentage scheme: the Precinct field is blank in 36.6% of the duplicate records; 11.5% of the duplicate records are allocated to the 1st Pct.; 6.2% to the 2nd Pct.; 7.0% to the 3rd Pct.; 9.1% to the 4th Pct.; 9.0% to the 5th Pct.; 10.0% to the 6th Pct.; 0.5% to the 7th Pct.; and 10.2% to the HWY Bureau.  Presuming that the initial record for the five CC No.s in question are, in fact, unique, by definition, the remaining 4,951 respective consecutive records are apparent duplicates.  

Removing the 4,951 apparently duplicate records from the 150,992 records generated between March 5, 2018 and March 4, 2019 yields a total of 146,041 discrete records – 279 fewer records than the 146,320 that Finn reports it examined in its review.  In total, the 4,951 apparent duplicate records represent 3.4% of the records in the record set that Finn examined.  The overwhelming majority (83%) of the 4,951 duplicate records were attributed to interactions with Whites – a level of disparity that has the potential to distort the true level of discretionary enforcement interactions between the respective demographic groups – particularly among minority groups. This leads to the questions: Were these records included in or excluded from Finn’s review?  Inasmuch as they were generated between the start and end dates reported by Finn, if they were not included, why did Finn’s report not mention their existence and exclusion?  If they were included in Finn’s review, why did Finn not report on their repetitive nature?  Finn did note that 48 stops had records for two drivers – a discrepancy that involved as few as 96 records – a mere 0.07% of the data set.  These apparent discrepancies are orders of magnitude greater than that; it seems implausible that they would have escaped Finn’s notice. 

Reality Check

SCPD attempts to rationalize its failure to comply with the recording, reporting and analysis provisions of the DOJ Settlement Agreement with the glib generalization “The initial processes for data collection was incredibly challenging for the Department, as it required overhauling an antiquated technology system to collect the level of data that was called for by the Department of Justice. In addition, internal data analytics platforms needed to be developed in order to synthesize and review traffic stop records.”

It goes on to gild the lily by speciously claiming, “The working group concluded that in order to properly address community concerns, and to grant SCPD command staff the ability to proactively address disparity in traffic stops, an entirely new toolkit would be needed.”  As the Federal District Court in the Eastern District of New York noted in its March 12, 2021 Report and Recommendation (cited above), SCPD has repeatedly used the transfer to upgraded data platforms as an excuse for its continued delay and failure to comply with its obligations under the Agreement.  The March 2018 starting date of the data set that Finn reviewed was reportedly coincident with the go-live of SCPD’s third system upgrade.  In 2021, seven years after the Settlement Agreement was executed, SCPD continued to defer accountability by once again claiming they need yet another new data platform. That claim can no longer be credited.  

SCPD launched a Transparency Dashboard in Q1 2022 that contains a Traffic Stops Dashboard.

The following claim from the plan merits reconciliation with its obligations pursuant to the Settlement Agreement, including DOJ-approved Department Procedure Chapter 13 Section 9, and revelations contained with the Eastern District’s March 12, 2021 Report and Recommendation:  
“The working group created an interfacing dashboard to monitor Department statistics, creating the opportunity for leadership to recognize atypical traffic stops, and accordingly provide individual officers with retraining and open dialogue to address the concern.
The internal dashboard will allow for multiple supervisory levels of review on traffic stop data.” 

DOJ-Approved Department Procedure Chapter 13 Section 9 VI. PROCEDURES B. Supervisors’ Responsibility and D. Responsibility of the Police Commissioner’s Office enumerates specific oversight protocols from line level supervisors up to command staff and the Commissioner’s office that ensures ‘the opportunity for leadership to recognize atypical traffic stops, and accordingly provide individual officers with retraining and open dialogue to address the concern.” 

This specious claim raises reinventing the wheel to an art form. The following excerpts, previously cited on page 20 above, from the Eastern District’s May 12, 2021 Report and Recommendation place it in appropriate context:

In 2017, “after months of preparation” leading up to the launch of a new data system, “SCPD discontinued using the system the very day it launched it” due to implementation issues. “We renew our recommendation that SCPD supervisors develop specific protocols for the substantive review of traffic stop data as a part of supervisors’ regular supervisory activities and that SCPD provide renewed training for supervisors, many of whom have not received supervisor-specific training since attaining the rank of sergeant.” October 2018 Assessment at 7.  

A 2016 Report issued by the SCPD noted that after launching a new system in October 2014, due to a “computer glitch” it did not detect approximately “7,000 incomplete entries,” rendering data unreliable from October 2014 until July 2015. Ex. X, “SCPD’s February 2016 Report.” The same report indicated 139 incomplete traffic stop entries for the month of December 2015 alone. “SCPD’s March 2017 Report,” 4 (the SCPD agreed with the DOJ’s conclusion that “the existing capture fields require more specificity” to conduct a “meaningful analysis of the data”); R, Love Deposition I at 182 (Q: What steps are taken to ensure that the officer is inputting the information accurately, correctly? A: There would be none [...]. Q: (same regarding the input of the correct race of the individual stopped) A:There are none). 

See also Smith Report at 27. (“The primary takeaway from this review of the data is that SCPD data collection protocols changed every 2–3 years and never consistently captured the fields necessary to determine whether Latino motorists were treated disparately from White motorists in key post-stop outcomes.”) 

Other recent assessments state: “[a]s set forth in our last Report, the data collected by SCPD omits critical variables that are necessary for meaningful analysis of bias-free policing,” such as “why a stop was initiated,” and the results of conducted stops, “including whether a particular search revealed contraband or not.” Ex. QQ, “January 2017 Compliance Assessment;”  “[The SCPD] remains in partial compliance...due to the continued failure to implement an adequate data collections system.” Ex. DDD, “March 2018 Compliance Assessment.”  

Sergeant Christopher Love is an attorney assigned to the SCPD Legal Bureau. Ex. 3, “Affidavit of Christopher Love.” He was designated as the County’s Rule 30(b)(6) witness on policies related to complaints, the Settlement Agreement, and data collection. 

In addition to the traffic stops and post-stop searches, plaintiffs allege that the SCPD uses traffic checkpoints in an unlawful, discriminatory manner to target Latinos . . . No checkpoint data was collected until 2018 and the SCPD still does not collect data on post-stop outcomes. Plaintiffs allege this data field is necessary at minimum to assess these stops for bias. See Smith Report. at 33–35. Similar to the traffic stop data, the SCPD has never analyzed checkpoint data to assess whether checkpoints were being disproportionately set up in Latino neighborhoods nor to otherwise assess its use of checkpoints.

See Love Deposition I at 149–50

• Q: So R&D has not conducted an analysis of the traffic stop data[?] A: No. That’s correct, they have not.

• Q: So the [traffic stop] data has never been reviewed for atypical traffic stop activity? A: No. [...] 

• Q: “Commanding officers shall review the annual report” [...] So that also has not been done? A: Correct).

In summary, plaintiffs argue that by failing to collect accurate traffic stop and checkpoint data, the SCPD has employed a successful strategy to escape a finding of discrimination. Plaintiffs allege that, had data been accurately collected and timely assessed as required by the agreement, the SCPD would have been alerted to concerning trends regarding ongoing discriminatory policing. 

Defendants have filed a report commissioned from the John F. Finn Institute for Public Safety, Inc. as an exhibit with their pending motion for summary judgment. The Court notes that this report was completed in late September 2020, on the eve of defendants’ summary judgment motion deadline, nearly seven years after defendants entered the DOJ Agreement, and nine years after the SCPD was formally put on notice regarding its insufficient data collection in the Technical Assistance Letter.

Here the District Court grants judicial notice of SCPD’s failure to comply with its obligations under the DOJ Agreement regarding traffic stop data recording, analysis and reporting and of its use of dishonest methods and delay tactics.  These include repeated claims of data systems glitches and upgrades, incomplete, inaccurate and inadequate stop data recording and failure to comply with the provisions of the DOJ Agreement.  In spite of repeated DOJ reminders, requests and recommendations (as recently as October 2018) as regards its obligations to ensure supervisor training and substantive supervisory review of traffic stop, and ensure data integrity and reliability, it is noted that, as recently as 2018, following two apparent SCPD traffic data system upgrades, DOJ cited SCPD for its “continued failure to implement an adequate data collections system” and attributed those continuing failures to the inability to adequately assess traffic stop data for bias.  The court noted that the Finn report was submitted on the eve of SCPD’s deadline for submission of its summary judgment motion deadline and that the analysis was provided nearly seven years after entering the DOJ Agreement and nine years after being formally notified by DOJ of its insufficient data collection.  Most damning is the testimony of SCPD’s own witness, Sergeant Love, whose testimony verified the following flagrant violations of the terms of the Agreement: 

• No steps were taken to ensure that the officer is inputting the information accurately, correctly

• No steps were taken to ensure the input of the correct race of the individual stopped

• R & D had not conducted an analysis of the traffic stop data

• Traffic stop data had never been reviewed for atypical traffic stop activity

• Commanding officers had not reviewed the annual report

Suffolk’s reform plan claim that it needs yet another data system upgrade and more time for supervisors to perform oversight is, thus, discredited.  In light of the failure of the County Executive, Chairman of the Public Safety Committee and Police Commissioner to acknowledge receipt of or respond to written requests for clarification of the impact of non-random, racially-skewed, bulk-generated duplicate traffic stop records on Finn’s analysis, the claim in the Suffolk County reform plan regarding the “integrity of the data that was used for this [the Finn] study” is contemptuous and deceitful.

Commitment to record pedestrian stop data is a positive step forward. The failure to fully adopt the Right to Know and STAT Act provisions of The People’s Plan is unfortunate.  

Unless and until voters hold elected officials politically accountable, SCPD will continue to resist meaningful sustainable reform and the residents of Suffolk County will continue to endure the resultant harms and costs. 

UJPLI written requests to executive, legislature and police commissioner for written clarification

In October 2020 UJPLI wrote to the office of County Executive Steve Bellone and then Public Safety Committee Chairman Tom Donnelly regarding the non-random bulk-generated duplicate traffic stop records. The correspondence included comprehensive assessment of the duplicates including their CC Numbers, their allocation among the precincts, the attribution by racial / ethnic cohort, and their statistical impact on digital analyses.  Specifically, it explained how the duplicates mitigated the striking racial disparities in discretionary enforcement actions between racial / ethnic cohorts.  It also identified nearly 5,000 duplicate records that were likely included in the data set that was analyzed by Finn.  The correspondence requested written clarification including how and why the duplicates were generated and what impact duplicate records had on Finn’s analysis.  Neither the Executive nor Chairman Donnelly has acknowledged receipt of or provided a response to the correspondence.

During a January 2021 Zoom presentation, then Police Commissioner Hart acknowledged that the duplicate records had been deliberately generated.  She claimed, implausibly, that they were “a misguided attempt to make the data more digestible” and said that they had been removed. She noted that UJPLI’s count of duplicate records was ‘inaccurate’ and refused to provide any other relevant information. 

In February 2021, UJPLI wrote to the Police Commissioner regarding the non-random bulk-generated duplicate traffic stop records. The correspondence included a copy of the October 2020 correspondence to the Office of County Executive Steve Bellone and Public Safety Committee Chairman Tom Donnelly.  The correspondence requested clarification of the phrase “misguided attempt to make the data more digestible” and explanation of the processes that led to the generation of the duplicate records and their subsequent removal from the raw traffic stop data sets.   The Commissioner never acknowledged receipt of or responded to that correspondence.  

What does the County’s policing reform plan provide?

Review of Traffic Stop Data Collection 

In compliance with a Department of Justice Settlement Agreement, Suffolk County has been collecting traffic stop data since 2014. The initial processes for data collection was incredibly challenging for the Department, as it required overhauling an antiquated technology system to collect the level of data that was called for by the Department of Justice. In addition, internal data analytics platforms needed to be developed in order to synthesize and review traffic stop records. 

In 2019, the County procured the services of the John F. Finn Institute for Public Safety, Inc. (FINN) to analyze the Department's traffic stop records collected between March 5, 2018 and March 4, 2019. Out of 146,320 traffic stop records collected between March 2018 and 2019, 86 collected were incomplete, all occurring in March 2018, the first month of collection. This accounted for .0005% of all records collected and speaks to the integrity of the data that was used for this study. 

The Study Brought to Light the Following: 

Overall, the FINN analysis confirmed that Black and Hispanic drivers are overrepresented in police traffic stops relative to their share of the Suffolk County population, while White drivers are underrepresented.  The recorded reasons for stops vary across racial and ethnic demographics. However, Black and Hispanic drivers represent a higher percentage of individuals who receive tickets for equipment violations. 

Suffolk County Task Force Reinvention Plan 


1) Create a Public Traffic Stop Data Dashboard 
The Department will address disparities in traffic stop data by leveraging an online data dashboard to internally overhaul policing oversight and externally provide data to the 
In order to address the disparity in traffic stop data; or, “atypical” traffic stops, the Department utilized the findings of the FINN Report to launch its re-envisioning of traffic stop practices: 


In response to the FINN analysis, the SCPD established a working group dedicated to analyzing and developing new policing strategies as they relate to traffic stops in Suffolk County. This team consisted of PD command staff, researchers, and the SCPD Information Technology team. 

The working group concluded that in order to properly address community concerns, and to grant SCPD command staff the ability to proactively address disparity in traffic stops, an entirely new toolkit would be needed. From there, the working group developed a first-of-its-kind on Long Island, an online traffic stop data dashboard with two distinct facets: an outward-facing arm for public transparency, and an inward-facing arm for Department oversight, analysis, and correctional response. 

The working group presented this traffic stop data dashboard to Task Force members, where additional input led to the final version. The dashboard aims to serve three main functions: 

1.     Provide the public with transparent and easily-accessible data about traffic stops; and 


2.     Utilize a proactive, Business Intelligence tool to produce the best service for the people 
of Suffolk County. 


3.     Create oversight of traffic stop data to ensure that disparities seen in the FINN report are 
addressed and eliminated. 


2) Internal Traffic Stop Data Review Dashboard 

The working group created an interfacing dashboard to monitor Department statistics, creating the opportunity for leadership to recognize atypical traffic stops, and accordingly provide individual officers with retraining and open dialogue to address the concern.
The internal dashboard will allow for multiple supervisory levels of review on traffic stop data. 

Currently, the FINN analysis gave us a picture of traffic stops for the entire police Department for 2018 to 2019. The next step in the deeper dive is giving command staff the ability to see and act upon substantial statistical differences in traffic stops by having tools to monitor Precincts, Squads, Zones and individual officers, in real-time. 

The goal of these new data driven oversight tools is to identify disparities and correct them. This tool is specifically designed to highlight disparities so that they can be immediately addressed. The precinct command staff will be tasked with reviewing the data on a consistent basis to ensure the disparities seen in the FINN report are eliminated. 

4) Early Warning Procedure 

SCPD Commanding Officers will identify and amend atypical patterns of traffic stops and/or enforcement activity by reviewing summary analyses generated by the internal traffic stop data portal. The goal is to address substantial statistical disparities highlighted in the FINN report. 

This review process will take the shape of an Early Warning Intervention System (EIS), which will generate reviews of traffic stop and enforcement activity patterns to identify substantial statistical differences and ensure data integrity. By granting supervisors the ability to access these data reports on a daily basis, the Department will take advantage of robust and transparent data to sustainably monitor, amend, and ensure excellence in individual and broader patterns of policing for the general public. 

Reconciling provisions and assertions of Suffolk’s policing reform plan with relevant provisions of the DOJ 2011 Technical Assistance Letter, the 2014 DOJ Settlement Agreement, other correspondence, relevant findings and rulings in civil actions and prior assertions of SCPD reveals often-used disingenuous rationalizations intended to deflect attention from the persistent failure to meet established commitments in a perpetual ‘run the clock’ strategy.  It is a regrettable attempt to justify and normalize persistent non-compliance.  

Consistent with the Department’s history, the pattern is clear: Deny the allegations, demand more proof and relentlessly resist all appeals for transparency.  Commit to reform only when compelled to and even then, do so only in word, and then backslide in deed.  Notably, the plan contains little language regarding internal controls and includes no enforcement mechanisms.  

CAUTION: In the absence of competent independent oversight, without adequate internal controls and enforcement mechanisms, the future will look much like the past.