The best example in my stats class on why sampling was used: working for the phone company he had to determine if poles were rotting. Practically speaking he could not use a random sample: he sent a pole climber to an area and while there, climb a few or several in the area. It would not be practical to climb one here, another 3 miles away, etc..