Samples, in the context of the population, allow statisticians to measure - as accurately as possible- a specific demographic for trends that may affect that category of people.
Demographic data includes, among other things, the size of a population, its density, spatial distribution, age, and gender.
It would be impossible to obtain any meaningful information from millions of people. Apart from the paperwork involved and man-hours, it would be a very impractical exercise. Therefore, "samples" are used to gather information and, due to strict guidelines (when the result is critical) the results are usually characteristic and accurate within the category or demographic measured.
Due to the fact that the distribution of people around the world is uneven - large tracts of land remain uninhabited and uninhabitable whereas others are densely populated such as
China (which) contains 21 percent of the world's population (at least)
it is necessary to use samples within countries and communities for the establishment of "norms."
It is important, if information is to serve any purpose and be of any use, that the sample being measured is representative of the population and is relevant. It would be no use doing a survey about a country's favorite beer if the sample consists of mostly under 18s or 21s (depending on the legal age).
The size of the sample is also crucial. A popular claim of manufacturers of face creams is the anti-ageing properties of the ingredients. The sample would be useless if companies made their claims based on having tested the products on ten or twenty participants. The length of time of the study would also be relevant.
Therefore for population and sample to have meaning in the same context the sample must be a reflection of the population towards which the research is directed.