Introduction

There are no universally accepted definitions of such terms as “hard-to-reach,” “elusive,” and even “mobile” populations, although all are well recognized as populations that are difficult to sample (see Kish, 1991, for a taxonomy of elusive populations). A common but not universal characteristic of these populations is that they are rare populations for which no separate sampling frames exist. The prime focus of this chapter is on sampling these types of populations using probability sampling methods. Rather than attempt a formal classification, I will present a range of illustrative examples:

• Blacks, Hispanics, American Indians/Alaska Natives, Vietnamese immigrants;

• Religious groups;

• Abused children, abused adults;

• Persons with AIDS, rare forms of cancer, dementia;

• Street prostitutes;

• Men who have sex with men (MSM), lesbians;

• Users of illicit drugs;

• Disabled scientists;

• Pregnant women;

• Children in poverty, the very rich;

• Homeless persons, nomads;

• International travelers, visitors to museums or shopping malls; and

• Drivers and passengers of cars in transit.

These examples are mostly drawn from other reviews such as Christman (2009), Kalton (1991, 2009), Kalton and Anderson (1986), Kish (1965a, 1991), Sudman and Freeman (1988), and Sudman and Kalton (1986). Flores Cervantes and Kalton (2008) present a review of the sampling methods for such populations in the context of telephone surveys and Elliott, Finch, Klein, Ma, Do, Beckett et al. (2008) review sampling methods for the widely encountered need to sample rare race/ethnic populations.

There are different analytic objectives for surveys of this diverse set of populations. With static populations, a main distinction is between surveys aiming to estimate the size of the population and its prevalence in the general population and surveys aiming to study the characteristics of the population. For example, is the main survey objective to estimate the proportion of children in poverty or the health, education, and housing conditions of poor children? With mobile populations, a key issue is whether the population of inference is the “visit” or the “visitor.” For example, is the population of inference the trips made by international travelers or the persons who engage in international travel?