218-220

Sampling in research can be customized to meet various needs.
Important components of sampling:
- Tailoring components to specific purposes similar to modifying research design principles.
- When suitable sample frames are unavailable, multistage cluster sampling is employed.
Multistage Cluster Sampling:
- Moves from aggregate sample units to actual sample elements.
- Stratification can ensure representative samples based on key variables.
- Samples can be designed for proportionate or disproportionate representation.

Lack of a national list of households in the U.S. necessitates multistage cluster sampling.
Primary Sampling Units (PSUs):
- Defined as metropolitan areas or nonmetropolitan counties.
- Used to sample households and residents.

National Crime Victimization Survey (NCVS) and Crime Survey for England and Wales (CSEW) implement multistage cluster sampling but differ in strategies.
- Each survey adapts building blocks for necessary respondent counts across categories.

Largest 93 PSUs classified as self-representing are included in the first sampling stage.
Remaining PSUs are stratified by size, population density, reported crimes, etc.
A two-stage sampling approach:
- First stage includes 93 self-representing PSUs.
- Additional 152 non-self-representing PSUs are selected based on population.

For example, if Indiana has a population of 5,000 and Rancid, Missouri has 3,000:
- Selected probability for Bugtussle (7,000) is 7/15, Rancid is 3/15, Punkinseed is 5/15.
Since 1972, NCVS has adapted sampling strategies with significant changes to sample size and the method of telephone interviewing.

Four different sampling frames designated within each PSU:
1. Housing unit frame: Lists addresses of housing units.
2. Group quarters frame: Lists group living conditions from census records.
3. Building permit frame: Lists newly constructed housing units.
4. Area frame: Census blocks are used to generate address lists.
Important to note: Comprehensive lists of residential addresses are typically not available in the U.S.

For the 2014 NCVS:
- Completed interviews from 158,090 individuals in 90,380 households.
- Highlights the relationship between sample size and population variation as serious crime rates are low, necessitating larger samples in multiple stages.

CSEW benefits from a national list of addresses (Postcode Address File).
Uses postcode sectors as defined clusters for sampling.
Adaptive Sampling for minority respondents:
- Special sampling methods used to target respondents from ethnic minorities, increasing representation in the sample.
Example: Interviewers identify adjacent housing units for minority selection following initial sampling.

Respondents aged 16 or over are randomly selected for household information, including children aged 10-15 since 2009.
2014-2015 CSEW final sample size was about 33,350 with a response rate of 70%.
Sampling designs differ between NCVS and CSEW due to the availability of suitable sampling frames.

Advantages of Probability Sampling:
1. Reduces bias in element selection, increasing representativeness of the sample.
2. Allows for estimates of sampling error associated with selected samples.
Despite its benefits, in some cases, standard probability sampling cannot or should not be employed, leading to the use of nonprobability sampling methods.

Nonprobability sampling is used increasingly due to its practicality in research settings.
Types of Nonprobability Sampling:
1. Purposive or Judgmental Sampling: based on the researcher's knowledge of the population, used for specific aims.
2. Quota Sampling: ensures specific characteristics in the sample.
3. Reliance on Available Subjects: allows the use of individuals easily accessible to researchers.
4. Snowball Sampling: participants help recruit others for the study.