Partners provide Watson and Opportunity Insights with data on programs and their participants, which is then merged into federal administrative records within secure government servers. This data-sharing is governed by a Data Use Agreement to ensure the data are handled securely and in accordance with partner expectations.  Our approach works best when partners can share data on a broad sample of participants (not only those in a particular program) and going back as far as possible historically, to obtain the largest sample sizes.

Shared data are then linked (through Social Security Numbers or other identifying information) to a Protected Identification Key (PIK), used to handle data in the secure environment in a deidentified way, and then (through the PIK) matched to earnings records and other data within the secure federal data enclave at the US Census Bureau. We then apply our matching approach to estimate program returns.

All results must go through disclosure review, to ensure that no confidential information is included. Once approved, we share these results back with partners in a non-public technical memo — this allows for early iteration and interpretation with data contributors and in some instances, leads to subsequent analyses. Eventually the results will be incorporated into publicly available research papers and policy reports.