CANCELED: Merging Data Sources: Record Linkage Techniques and Analysis of Linked Datasets

Different systems create vast amount of personal information. Opportunities to use this information are often missed because this information cannot be combined due to privacy concerns. Record linkage methods link data from multiple sources when unique identifiers are unavailable. These techniques harness the power of advanced algorithms, predictive and probabilistic modelling to link records from disparate sources, even when the data suffers from inconsistencies and discrepancies. Recent computational advances resulted in an explosion of record linkage methods. Record linkage methods work well when there are informative linking variables, but even advanced methods suffer from linkage errors. Incorrect linkage could lead to biases in estimating associations between variables that are exclusive to one dataset. Multiple statistical methods have been proposed to adjust for linkage errors. This short course describes available linkage methods, methods to analyze linked datasets and it illustrates both sets of methods using the R software. In an era where data quality is paramount, record linkage methods, coupled with rigorous error adjustments, are important to ensure that insights drawn from linked datasets are robust and actionable. The course is intended for applied statisticians who are interested in using record linkage methods and estimating relationships with linked datasets.