When you use a sample of a population to find out information, you must be very careful.  For instance, say I send out 1000 surveys asking what the family car’s value is to a certain suburb in a city.

This is not a good sample, as the suburb I select may have a higher or lower living standard than the city average.  The results I get from the survey would not represent the entire population, and therefore would be said to be biased.

Another example would be running a survey on whether guns should be banned the day after a gang shooting.  This too would give biased results.