Abstract
American Jews have long understood the vitality of the Jewish community by its numbers: initially its population size and later its characteristics. Acquiring such data, though, has always been complicated by the exclusion of questions on religion from the U.S. Census of Population and Households. In the absence of official counts of the Jewish population, American Jews have sought other ways of counting themselves. The methods used to this end have, however, varied considerably across the decades, as have the quality of the methods used. The long-awaited National Jewish Population Survey of 2000-01 was the culmination of these trends, with a vast cost of $6 million and multiple methodological flaws.
Phillips examines a variety of responses to the challenges of the present survey research environment. Use of expanded list frames drawn from the 10 or so largest Jewish organizations in a community in combination with random digit dialing appears to be the most efficient sampling scheme. Phillips concludes that cost-variance optimization for estimating Jewish incidence leads to dramatic increases in accuracy compared to non-optimized distribution of the sample across frames and strata.