QUALITATIVE AND QUANTITATIVE SAMPLINGspeakers, yet as of 2007, roughly 5 percent of U.S.
households were “linguistically isolated” (no one
over 14 spoke English very well (U.S. Census
Bureau, 2007).
To draw a probability sample we start with a
population, but populationis an abstract concept.
We must conceptualize and define it more precisely
in a process similar to conceptualization in the mea-
surement process, for example, all people in Tampa,
Florida, or all college students in the state of
Nevada. A target populationis the specific collec-
tion of elements we will study (e.g., noninstitution-
alized persons 18 years of age and older with legal
residences with the city limits of Tampa on May 15,
2011; students enrolled full-time in an accredited
two- or four-year postsecondary educational facil-
ity in the state of Nevada in October 2010). In some
ways, the target population is analogous to our use
of a conceptual definition of the measurement pro-
cess.
Populations are in constant motion, so we need
a temporal boundary. For example, in a city at any
given moment, people are dying, boarding or get-
ting off airplanes, and driving across city bound-
aries in cars. Whom should we count? Do we
exclude a long-time city resident who happens to be
on vacation when the time is fixed? A population
(e.g., persons over the age of 18 who are in the city
limits of Milwaukee, Wisconsin, at 12:01 A.M.on
March 1, 2011), is an abstract idea. It exists in the
mind but is difficult to pinpoint concretely (see
Example Box 2, Examples of Populations).
After we conceptualize our population, we
need to create an operational definition for the
abstract population idea in a way that is analogous
to operationalization in the measurement process.
We turn the abstract idea into anempirically
concrete specific list that closely approximates all
population elements. This is our sampling frame.
There are many types of sampling frames: tele-
phone directories, tax records, driver’s license
records, and so on. Listing the elements in a popu-
lation sounds simple, but it is often difficult because
often there is no accurate, up-to-date list of all
elements in a population.
A good sampling frame is crucial for accu-
rate sampling. Any mismatch between a sampling
frame and the conceptually defined population
can create errors. Just as a mismatch between our
theoretical and operational definitions of a variable
weakens measurement validity, a mismatch between
the abstract population and the sampling frame
weakens our sampling validity. The most famous
case in the history of sampling involved an issue
of sampling frames.^3 (See Expansion Box 1,
Sampling Frames and the History of Sampling.)
Let us say that our population is all adult resi-
dents in the Pacific coast region of the United States
in 2010. We contact state departments of trans-
portation to obtain lists of everyone with a driver’sSampling frame A list of cases in a population, or
the best approximation of them.Target population The concretely specified large
group of many cases from which a researcher draws
a sample and to which results from the sample are
generalized.EXAMPLE BOX 2
Examples of Populations- All persons ages 16 or older living in Australia on
December 2, 2009, who were not incarcerated in
prison, asylums, and similar institutions - All business establishments employing more than
100 persons in Ontario Province, Canada, that oper-
ated in the month of July 2005 - All admissions to public or private hospitals in the
state of New Jersey between August 1, 1988, and
July 31, 1993 - All television commercials aired between 7:00 A.M.
and 11:00 P.M. Eastern Standard Time on three
major U.S. networks between November 1 and
November 25, 2004 - All currently practicing physicians in the United States
who received medical degrees between January 1,
1960, and the present - All African American male heroin addicts in the
Vancouver, British Columbia, or Seattle, Washington,
metropolitan areas during 2004