Data Sharing Case Study Checkpoint Effectiveness: Models and Metrics Jay Nunamaker & Elyse Golob 19 March 2014 Background • Project initiated by CBP- Office of Border Patrol (OBP) to assess the effectiveness of traffic checkpoint operations for the public good. – Purpose: To address recommendations noted in GAO Report No. GAO-09-824: “Checkpoints Contribute to Border Patrol's Mission, but More Consistent Data Collection and Performance Measurement Could Improve Effectiveness” • Key goals were to evaluate and address: – – – – Checkpoint data integrity, consistency and accuracy Measures of checkpoint impacts on local communities Effectiveness metrics and models Development of a managerial tool to inform resource allocation • Overall Timeline – Project initialization – October 2010 – Final report – December 2012 2 Data Collection • Site visits to five different sectors • Meetings with OBP HQ for feedback • Group Collaboration sessions and brainstorming with BP agents to better understand the data collection process • Requested electronic data sets – e3: Apprehension Information – CAR: Checkpoint Activity Report 3 Data Collection • OBP representatives provided full support with 3 data collection processes – Site visits, HQ meetings, Brainstorming sessions • Challenges encountered in obtaining e3 and CAR data – Data was determined to be Sensitive – Time spent by DHS to determine appropriate solution for data handling delayed related project milestones 4 Explored Processes for Data Handling • Explored: – Use of CRADA – Altering Cooperative agreement to allow for SBU data • Solutions: – OBP determined amount of SBU data was not significant to project and did not make it available – OBP would incorporate the SBU data in after work is completed – Cooperative agreement unaltered – Unclassified data provided; No SBU data provided to BORDERS 5 Data Acquired • November 2011, BORDERS received a open source data set • The data set was a subsample of the data that had been requested – e3 data set was comprised of apprehension data spanning from FY2006 - May 31 of FY2011. – Set contained only data that did not require an NDA, amendment to the cooperative agreement, 6 Data Quality • BORDERS researchers identified multiple challenges with this data set – For example: a percentage of latitude and longitude values for the ‘ARREST_LATUTUDE’ and ‘ARREST_LONGITUDE’ data points were not accurate, resulting in arrest locations outside of the United States 7 Observations • • Time consumption spent requesting/reviewing data received Limited data set received – – • As a result of identifying the data quality issues, BORDERS was able to make recommendations to OBP on how to improve their data collection practices The restricted data set potentially limited the findings and potential outcomes of this project as data quality issues present in the complete data set were not accessible to researchers, and could thus not be identified Challenges – – Large amount of time attempting to acquire data Unclear decision process and criteria as to what sensitive data could be sent made it challenging for BORDERS to be proactive in providing clarification 8