Random sampling is a statistical technique in which individuals are randomly selected from the target group (usually a sub-set of the population). More precisely, everyone has a known probability of selection. Generally the aim of this method is to remove bias in selection of samples, though it may not be suitable for all cases of statistical analysis, depending upon the circumstances.
The reliability of a poll is a different dimension and refers to the amount of random error still present. The larger the poll, the smaller the random error. (Increasing the size of a poll by X reduces the error by the square root of x. A poll with 4 times as many people will have half the random error.)
An example is all eligible voters being equally likely to be included in a poll. This method does not ensure accuracy, but increases its likelihood.
In statistical studies, random sampling refers to the extraction of values (sampling) of a designated input probability distribution for ensuing statistical calculations, such that the relative probability of extraction is that indicated by the input probability density function in the case of a continuous distribution or the probability mass function in the case of a discrete distribution.
- Graham Kalton. Introduction to Survey Sampling (Quantitative Applications in the Social Sciences) (1983) excerpt and text search