Sampling may be performed from both real and hypothetical populations.
Examples of sampling from a real population are opinion polls (when a finite number of individuals is chosen from a much bigger group, say, a population of a city or a whole country), or quality control sampling, when a relatively small number of units is chosen periodically from a much larger set to judge whether the process is remaining stable. In the latter case, the population is actually the process that continues over time.
An example of sampling from a hypothetical population is tossing a coin, say, 100 times. This sample, consisting of 100 head-or-tail values, reflects properties of the hypothetical infinite population of head/tail values which could have been obtained if the process of tossing were infinite.
In theoretical statistics, sampling is usually assumed to be random (see random sampling ). In practical statistics, the goal is usually to make the real sampling process as close to the random sampling as possible.