Sampling Flashcards
Macro-level
Countries, cultures, societies
Meso-level
Organizations, agencies, communities and social network
Micro-level
Individuals, small group behaviours and interactions
Population/target population
Total number of cases for which the
quantitative study seeks to develop insights
Full coverage
Includes all relevant cases of a population
Sample
Only includes a selection of cases of a population
Sampling method
The process that determines which cases from the population are selected into the sample. There are non-probability (non-random) and probability (random) sampling methods.
Sampling size
The number of cases in a sample
Characteristic-specific representative sample
The composition of the
sample corresponds to the composition of the population with regard to some
relevant characteristics. This is typically achieved through non-probability quota
sampling.
Globally representative sample
The composition of the sample corresponds
to the composition of the population in all characteristics and combinations of
characteristics. This can only be ensured by probabilistic sampling methods,
provided that a minimum sample size is observed at the same time
Ad-hoc sampling
Non-probability sampling. Cases are randomly selected for a sample because they
are currently available or easily accessible (e.g., street survey).
Snowball sampling
Non-probability sampling. Individual members of the population are asked to recruit further study subjects via their personal social networks. It is suitable for
populations that are difficult to reach, but whose members are well connected
to each other.
Quota sampling
Non-probability sampling. Sampling of cases with certain sociodemographic
characteristics. It cannot claim global representativeness. If the composition of
some characteristics in the population is known and replicated, it can claim characteristic-specific representativeness.
Simple random sampling
Probability sampling. Every case in a population has an equal chance of
being selected (e.g., according to a “blind” statistical random principle). The
target population has to be finite (e.g., members of a political party).
Stratified random sampling
Probability sampling. The target population is divided into sub-populations (layers) based on one or several characteristics and a simple
random sample is taken from each of these layers.
Self-selection/non-response bias
Sampling bias. Decision to participate may be correlated with traits that affect the measurements. Potentially concerns all studies with voluntary participation (e.g., high-level officials and leadership styles)
Exclusion bias
Sampling bias. Results from the exclusion of particular groups. Potentially
results from communication channels (e.g., online vs. paper-based surveys).
Survivor bias
Sampling bias. Only “surviving” subjects are selected while ignoring those that
fell out of view. Potentially concerns studies with an unobserved selection
process in the past (e.g., employee turnover)
Noise
Noise refers to random errors or variability in the data that do not reflect any underlying trend or pattern. Noise is essentially the “background” variability that can obscure the true signal in the data.
Noise is random and can lead to variability that may mask true effects. A noisy sample may show variability around the true population parameter without a consistent direction.
Bias
Bias refers to systematic errors that lead to an incorrect estimate of a population parameter. It occurs when the sample is not representative of the population, which can result in consistent overestimation or underestimation of the true value.
Bias is systematic and leads to consistent errors in one direction and will consistently misrepresent the population.
Bias can lead to conclusions that do not accurately reflect the true characteristics of the population, which can affect decision-making and policy formulation based on the study results
