Suppose we want to know the proportion of a certain population that has certain trait. For example, we might want to know what proportion of Americans would answer yes to a particular public opinion question.
If we take a random sample from the population and measure the proportion of our sample with the trait of interest, we can't expect that the sample proportion will always equal the population proportion (unless the popoulation proportion is either 0 or 1). But we might hope that it is at least close, especially if the sample size is large.
This page will let you experiment with this idea and "see" the results of a sampling from a population. Typically, you will not know the true proportion for the population and will only have the result from one sample. This page lets us simulate a large number of samples from a population with known or unknown proportion and displays the results of the sample proportions in a histogram.