Using Family Trees to Identify Turning Points in U.S. Migration: 1789 to 1926

Caglar Koylu, University of Iowa
Alice B. Kasakoff, University of South Carolina

In this paper we compare and evaluate different methods for partitioning data on family migration within the U.S. into periods, introducing a methodology for creating migration rates from family trees. The data is from a set of user contributed, cleaned, geocoded and connected trees from Rootsweb.com. containing information on about 80 million individuals who were born between 1630 and 1930 (Koylu et al., 2020). These trees contain information on where individuals were born over several generations. The child – ladder approach provides a sequence of locations where families lived derived from changes in birth states of consecutive siblings, which we use in this paper to estimate a migration rate. The number of moves that can be detected using the child-ladder method depends upon the characteristics of the child bearing period. If birth intervals or the length of the period changed over time, these will affect such a migration rate. We explore ways to adjust the child-ladder migration rate for changes in these components of fertility. Since fertility was declining during this period, this is important. We use different methods for dividing the adjusted rate into periods: fixed lengths, quartiles, natural breaks, standard historical periods and periods based upon transportation changes. We compare these using a goodness of fit measure to evaluate the results and to see which provide the “best” description of the flows. We illustrate changes in the flows using a series of maps summarizing the most “efficient” partitions.

No extended abstract or paper available

 Presented in Session 70. New Perspectives on Migration and Mobility