Sampling by Ethnic Surnames: The Case of American Jews

It is often difficult and costly to locate members of numericallysmall minorities using standard probability sampling. Consequently,nonprobability sampling techniques of various sorts are commonlyused. This paper analyzes the differences between samples chosenby two such techniques—the use of ethnic surnames andthe use of organization lists—tand compares them withprobability samples. Using data from the National Jewish PopulationStudy, we find that Jews with one of 35 so-called distinctiveJewish names do not differ substantially from the general populationof Jews in demographic characteristics or indicators of Jewishidentification, and that this technique produces a sample whichis more similar to the general population of Jews than doesthe organization list sample technique.