Behavioral Analysis of Background Investigations – IN PROGRESS
Insider risk begins the moment an organization hires and onboards a new employee. MITRE is conducting behavioral analysis of derogatory background investigation data to identify data-driven patterns in critical flags, develop indicators, and identify the best data sources for screening and vetting decisions. This effort includes:
- Structuring and aggregation of 30,000 investigative files
- Analyzing data (e.g., ML, logistic regressions, latent class analyses, other feature level analyses, etc.) to identify statistical patterns in when (pre, post) and how (flags, data source) negative adjudicative decisions were made
- Providing data-driven recommendations regarding which flags and indicators are most critical, at which stages of the vetting process the criteria should be investigated, and the most valid and reliable means to collect the data (e.g., automated vs. human investigator)
The analyses will provide data-driven input in the debate and value proposition of all automated background investigation data collection, versus human investigator data collection. With additional research and funding, the outcomes from the proposed research could be used to create risk-scoring algorithms.