Institutional Affiliation: University of Michigan
|How Well Do Automated Linking Methods Perform? Lessons from U.S. Historical Data|
with , , : w24019
This paper reviews the literature in historical record linkage in the U.S. and examines the performance of widely-used automated record linking algorithms in two high-quality historical datasets and one synthetic ground truth. Focusing on algorithms in current practice, our findings highlight the important effects of linking methods on data quality. We find that (1) no method (including hand-linking) consistently produces representative samples; (2) 15 to 37 percent of links chosen by prominent machine linking algorithms are identified as false links by human reviewers; and (3) these false links are systematically related to baseline sample characteristics, suggesting that machine algorithms may introduce complicated forms of bias into analyses. We find that prominent linking algorithms at...
|Do Grandparents and Great-Grandparents Matter? Multigenerational Mobility in the US, 1910-2013|
with , : w22635
Studies of US intergenerational mobility focus almost exclusively on the transmission of (dis)advantage from parents to children. Until very recently, the influence of earlier generations could not be assessed even in long-running longitudinal studies such as the Panel Study of Income Dynamics (PSID). We directly link family lines across data spanning 1910 to 2013 and find a substantial “grandparent effect” for cohorts born since 1920, as well as some evidence of a “great-grandparent effect.” Although these may be due to measurement error, we conclude that estimates from only two generations of data understate persistence by about 20 percent.