This is much more representative; if you look at a country's population and at their politicians you will see large differences in age, education, wealth etc. This can be used to overcome that by selecting, as much as possible, a random representative group of people, much as you would in a scientific study. It also avoids party agendas and things like that.