Understanding fertility trends in a cohesive high-fertility population within a low-fertility environment

Project Summary
Western populations collectively responded to historically contingent, society-wide developments in the 19th and 20th
centuries by reducing births to record-low levels. However, a minority of Westerners, largely ethnic sectarian religions,
have maintained pre-transition fertility levels despite exposure to contexts that motivate fertility decline. How have these
populations—conspicuous among them being the Amish—maintained high birth rates? Decades of sporadic research
suggest that high birth and low attrition rates explain the unabated 20-25 year Amish population doubling time. Yet, it is
less clear why Amish women maintain high birth rates and, perhaps more revealing, why births vary so widely among
them, with community-level completed birth rates ranging from approximately 4 to 12. Occasional cross-sectional studies
suggest that multiple variables—including household head occupation, religious differences, and individual status—
predict Amish birth rates, but what these results mean, exactly, remains educated guesswork. This project overcomes past
barriers and delivers new knowledge by making two ground-breaking contributions to Amish demography research. First,
this project develops a large population-wide, time-sensitive dataset that will permanently transition Amish demography
research from limited, single-site, cross-sectional analyses to full-scale, continent-wide, time-sensitive analyses. This
transition will substantially advance the scope and interpretability of statistical predictors of births. This longitudinal
database pulls information from Amish-produced population record books that provide extensive documentation of vital
events, sociocultural information, and geocodable addresses at the household level. These geographically referenced data
will then be linked to publically available contextual data. The launch-point for this database is my cross-sectional Amish
population dataset, which incorporates 71 current record books with information about 54,731 unique Amish households
(estimated at about 270,000 individuals), covering approximately 89.6% of all Amish in North America. The proposed
longitudinal database will use all editions of three major directory sets, piloting the process of eventually incorporating all
editions of all Amish directories. Second, this project will validate the data by testing population growth explanations,
beginning with articulation of population structure trends over time. This project will then test the relative influence of
competing fertility theories by disentangling the effects of Amish-internal versus external structural-cultural forces.
Namely, are birth differences more the product of Amish-internal differences in spite of contextual socioeconomic
development or is birth variation more the product of relative isolation from society-wide structures in spite of Amish
internal demarcations? To test these competing theories, I will employ this project’s internally linked, geographically
referenced pilot longitudinal database in time-sensitive analyses. To achieve these project goals, I propose a research and
training program that will enable me to apply conventional and recent advancements in time-sensitive statistical analysis,
fertility theory, and relational database design and management. In so doing, I will quickly achieve researcher
independence in preparation for career-long management, ongoing development, and analysis of a population database
that has sprawling potential to address many demography and population health questions.

Grant Number: 5K99HD111665-02
NIH Institute/Center: NIH
Principal Investigator: Cory Anderson