SAMPLING SCHEMES
Systematic sampling
-
if sampling fraction is k, one takes every (1/k)th element
-
if population members are listed in random order, then it is equivalent
to simple random sampling
Simple random sampling
-
every population member has an equal chance of being selected
-
can be unrepresentative of the overall population
Stratified sampling
-
if the population profile has key qualitative attributes (e.g., sex,
social class, age) that split the population into strata (subpopulations),
representativeness can be enhanced by constructing overall sample from
the amalgamation of simple random samples within each stratum
-
e.g., with each age stratum
Multistage sampling
-
used where a hierarchy of units exists
-
e.g., with herds of cattle, herds would be primary sampling units
and individual cattle are secondary sampling units
-
any number of levels is possible
-
the lower-level sampling units are the focal point
-
cluster sampling - a special case of two-stage approach (where
a complete census is required of each of the primaries sampled)
Quote sampling
-
particularly associated with street interviewing for public opinion
and market research surveys
-
elements include (a) stratification (e.g., age-by-sex), and (b) assigning
interviewers quotas for interview