This post could be lifted from an introductory statistics text. But in my experience working with researchers the important distinction between a sample statistic and a population parameter gets a little muddled.
A population parameter is an value that describes a population, such as the mean BMI of all American’s or the mean rent in NYC. A sample statistic is a value from a sample, such as the mean BMI of a sample of 100 Americans or the mean rent of a sample of 20 apartments in an NYC neighborhood. A sample statistic is often used to estimate a population parameter, but they are not the same thing.
Hypothesis tests are used to test values of population parameters. For example, suppose I create a weight-loss program I believe is effective for the general population. My interest is not if it works in any one random sample, but rather if it works in the general population. I want to test if the mean weight loss in the population is greater than zero. In symbols, if 
(assuming that weight loss is denoted as a positive value). In practice we will never know the true value of 
(sorry).
But we can test (alternative hypothesis) vs 
(null hypothesis). Suppose I find a random sample of 100 people, weight them before and after my weight-loss program, and find that the mean of the before-after differences within each subject is 
(i.e., the subjects lose an average of three pounds on the program). Statistical theory tells us that 
is an unbiased estimator of , i.e., a fair guess is that 
but we can’t say that 
. It could be that is actually -10, or 10, or 0 regardless of whatever value assumes, or whatever p-value is found from the test.
For example, suppose 
for the test described above. This would supply “strong evidence” that 
is incorrect and that is acceptable, but we still don’t know the true value of .