Sampling bias occurs when a survey or series of observations deals with a sample that is, based on something inherent in the methodology, biased in some way. (The Wikipedia article does a much better job explaining it than I ever can.)
This can be seen in everyday life, without conducting systemic studies. I’ve compiled a few examples below.
- I always seem to travel on the most congested lane. It turns out that this is not just bad luck—the most congested lane has more cars, so basic probability dictates that I’m more likely to be on that lane.
- How long does the average relationship last? Not as long as you think—you’re more likely to witness a relationship if it lasts longer, so we all have a distorted view of the average relationship.
- Consider the average number of friends each of your friends has. Because the people you are likely to make friends with are more sociable than the average person, this number is higher than the number of friends the average person has. In effect, this means that most people have fewer friends than their friends do. This is called the friendship paradox.
These fun examples illustrate how observations we make can be biased in subtle ways. Understanding sample bias is not just for the professional statistician, but is in fact important for everyday life.