Dean is an expert on record linkage methodology and software development. He has developed innovative approaches to enhancing the Fellegi-Sunter record linkage model and has led developed of a proprietary record linkage software package (NORCLink) that applies enhanced methods and includes the use of the Expectation-Maximization algorithm to estimate linkage parameters. Dean is also an expert on the use of administrative records for policy analysis and survey evaluation and highly proficient implementing survey analysis methods such as weighting, imputation, and variance estimation.
He has developed methods and code that are used in the linkage program for the National Center for Health Statistics. This work includes linkages of the National Hospital Care Survey (NHCS) to the National Death Index (NDI), Medicare and Medicaid administrative records, and HUD and VA administrative records. He also developed an imputation model for race and ethnicity and synthetic data to allow users access to data that protected the privacy of represented patients and their data. He has also conducted research on privacy preserving record linkage (PPRL). With the software he has developed he conducts linkages of AmeriSpeak data to vendor data bases to provide enhanced characteristics to the survey records. He conducted iterative proportional fitting (using software he developed) to provide representative weighting for patient health claims sample files for the Agency for Healthcare Research and Quality.
Prior to coming to NORC, Dean worked in the U.S. Census Bureau’s administrative record area. Here he developed software that is still being used to run the Person Identification Validation System (PVS) that conducts linkages of survey and administrative record data involving hundreds of millions of records. He also performed imputation modeling of personal characteristics and evaluated survey quality (particularly related to public program participation) by record check analyses. He also worked in the health policy area of the Urban Institute applying data science methods to health policy analysis.
University of Pennsylvania
Franklin & Marshall College
Appointments & Affiliations
American Statistical Association
Founder and First Chairman
Record Linkage Interest Group
opens in new tabA Methodological Assessment of Privacy Preserving Record Linkage using Survey and Administrative Data.
Journal Article | June 7, 2022
opens in new tab"Using National Synthetic Data to Conduct Health Services Research."
Journal Article | February 16, 2022
opens in new tab“Measurement of Type I and Type II Record Linkage Error.”
Presentation | August 4, 2021
opens in new tab“Using Supervised Machine Learning to Identify Efficient Blocking Schemes for Record Linkage.”
Journal Article | August 4, 2021
opens in new tab"Using Synthetic Data to Replace Linkage Derived Elements: A Case Study."
Journal Article | March 26, 2021
opens in new tab"Linking Survey Data with Commercial or Administrative Data for Data Quality Assessment."
Book | September 30, 2020
opens in new tab"Simulation Approach to Assess the Precision of Estimates Derived from Linking Survey and Administrative Records."
Journal Article | November 9, 2018
opens in new tab"Quality of Linked Data: Linking the National Hospital Care Survey Data to the National Death Index."
Journal Article | September 6, 2018