Research Methods in Social Relations ch. 9 Sampling Methods –
G. Maruyama & C.S. Ryan (2014)
Selecting a sample from a population can affect the validity of generalisations from the sample to the
population. Basic definitions:
Population: The entire group defined by specific criteria, such as all US residents.
Stratum: A subset of the population defined by additional specifications, like women aged
21-40.
Population Element: A single member of the population.
Census: A complete count and characteristic assessment of all elements in a population.
Sample: A subset of the population selected for study, used to make inferences about the
entire population.
Sampling Plan: A strategy to ensure the sample accurately reflects the population, including
measures for accuracy and confidence.
Probability Sampling: Each element has a known chance of being selected, allowing
estimation of how closely sample findings match the population and accurate generalisation.
Nonprobability Sampling: The probability of each element being included is unknown, often
used for convenience but less reliable for generalizations.
Margin of Error: The range within which the true population value is expected to fall, based
on the sample estimate. For example, a margin of error of ±5% means the sample estimate is
expected to be within 5 percentage points of the true population value.
Confidence Level: The probability that the sample estimate is within the margin of error
(e.g., 90% confidence).
Representative Sampling Plan: A plan designed to produce a sample likely to accurately
reflect the population, though it cannot guarantee it.
Each method in nonprobability sampling has challenges with evaluating bias and calculating margins
of error or confidence levels due to the lack of random sampling.
Haphazard/Convenience/Accidental Samples: Selects individuals randomly or those who are
easily accessible until the desired size is reached; introduces potential bias.
Quota/Representative Samples: Ensures diverse representation by including cases from
each stratum (e.g., age, gender), but individual strata are still selected haphazardly, so
potential bias remains.
Purposive Samples: Handpicks individuals based on specific criteria; not generalisable and
may introduce selection bias.
Snowball Samples: Initial participants recruit others from hard-to-reach populations; grows
through referrals, which are haphazard and may introduce bias.
Probability sampling ensures that each element of the population has a known chance of being
included, improving accuracy and reliability. It helps by allowing us to estimate the likelihood that
sample results reflect the true population values and ensures sufficient representation from different
strata for accurate estimates.
Simple Random Samples: Every element has an equal chance of being chosen using a
complete sampling frame and random number generators. The sampling frame is a
comprehensive list or method specifying the population from which samples are drawn. In
G. Maruyama & C.S. Ryan (2014)
Selecting a sample from a population can affect the validity of generalisations from the sample to the
population. Basic definitions:
Population: The entire group defined by specific criteria, such as all US residents.
Stratum: A subset of the population defined by additional specifications, like women aged
21-40.
Population Element: A single member of the population.
Census: A complete count and characteristic assessment of all elements in a population.
Sample: A subset of the population selected for study, used to make inferences about the
entire population.
Sampling Plan: A strategy to ensure the sample accurately reflects the population, including
measures for accuracy and confidence.
Probability Sampling: Each element has a known chance of being selected, allowing
estimation of how closely sample findings match the population and accurate generalisation.
Nonprobability Sampling: The probability of each element being included is unknown, often
used for convenience but less reliable for generalizations.
Margin of Error: The range within which the true population value is expected to fall, based
on the sample estimate. For example, a margin of error of ±5% means the sample estimate is
expected to be within 5 percentage points of the true population value.
Confidence Level: The probability that the sample estimate is within the margin of error
(e.g., 90% confidence).
Representative Sampling Plan: A plan designed to produce a sample likely to accurately
reflect the population, though it cannot guarantee it.
Each method in nonprobability sampling has challenges with evaluating bias and calculating margins
of error or confidence levels due to the lack of random sampling.
Haphazard/Convenience/Accidental Samples: Selects individuals randomly or those who are
easily accessible until the desired size is reached; introduces potential bias.
Quota/Representative Samples: Ensures diverse representation by including cases from
each stratum (e.g., age, gender), but individual strata are still selected haphazardly, so
potential bias remains.
Purposive Samples: Handpicks individuals based on specific criteria; not generalisable and
may introduce selection bias.
Snowball Samples: Initial participants recruit others from hard-to-reach populations; grows
through referrals, which are haphazard and may introduce bias.
Probability sampling ensures that each element of the population has a known chance of being
included, improving accuracy and reliability. It helps by allowing us to estimate the likelihood that
sample results reflect the true population values and ensures sufficient representation from different
strata for accurate estimates.
Simple Random Samples: Every element has an equal chance of being chosen using a
complete sampling frame and random number generators. The sampling frame is a
comprehensive list or method specifying the population from which samples are drawn. In
