Surveying Hard-to-Reach Populations in Comparative Perspective
*Tom W Smith, NORC
Keywords: cross-national, total surevy error, undercoverage
First this paper examines the the topic of hard-to-reach populations from the total-survey-error (TSE) perspective. This problem of undercoverage error is considered along with all components of TSE. Second, it puts the issue into a hierarchy examining the attributes of a) countries, b) groups, and c) individuals that makes them hard or easy to reach. For example, countries will be undercovered if they are authoritaraian and/or underdeveloped. Undercovered groups include primitive tribes, nomads, and the homeless. Undercovered individuals would include misanthropes and the mentally impaired. Altogether about three dozen factors across the levels of countries, groups, and individuals are identified. Third, the different levels of the hard-to-reach populations across societies are considered. It is argued that both the level and composition of hard-to-reach populations varies across societies and as a result creates differential bias and sub-optimal comparability across cross-national surveys.