The promise of high-quality, integrated data sets for policy analysis is great. However, significant impediments remain that make the creation of such data sets very difficult. The identification of processes and algorithms for carrying out approximate string matching on personal-level information while at the same time keeping that information unknown (private linkage) is quickly growing in importance. Without these processes and algorithms, the creation of high-quality data sets integrated from multiple public agencies remains difficult. In this paper, a method for carrying out secure, private string-matching between data sets using standard string similarity measures is presented.