In most experiments, data about the full population is not available or too expensive to obtain. A sample is therefore taken to make inferences about the population and the sample distribution includes only the available data from that sample of the population. The larger the sample the more likely its distribution is to correlate with the frequency distribution of the entire population.
For the validity of an experiment, it is imperative that samples are as representative as possible. Otherwise there is a greater risk that the sample distribution is not representative of the population.