Sampling error

Sampling error is the difference between a statistic calculated from a sample and the population value it estimates, caused by observing only part of the population.

Sampling distribution: The probability distribution of a statistic across repeated samples drawn under the same design. It shows the possible sampling errors as variation in a statistic across repeated samples.
Population parameter: A numerical characteristic of an entire population, such as its mean or proportion. Sampling error is defined relative to the population value the statistic aims to estimate.
Margin of error: A stated bound on sampling uncertainty around an estimate, commonly tied to a confidence level. It translates estimated sampling variability into a range around a reported estimate.
Total survey error: The combined set of errors that can affect a survey estimate, including sampling and nonsampling sources. It places sampling error within the broader inventory of survey inaccuracies.
Standard error: The standard deviation of a statistic's sampling distribution, estimated from data or a model. It measures the typical size of sampling variation for an estimator.
Sample statistic: A numerical summary calculated from observations in a sample. The statistic is the sample-based quantity whose difference from the parameter is sampling error.
Confidence interval: An interval procedure that, under repeated sampling, covers a fixed parameter at a specified rate. Its width reflects sampling uncertainty under the assumed model or design.
Nonresponse bias: Bias arising when people who do not respond differ systematically from those who do. A large sample can still be distorted if nonrespondents differ from respondents.
Law of large numbers: A theorem stating that sample averages converge toward their expected value as sample size increases. It explains why sample averages tend to stabilize near the population mean as samples grow.
Probability sampling: A sampling design in which population units have known, nonzero chances of selection. Known selection probabilities make sampling variation quantifiable under the design.

Linked from 19 pages