If two polling firms asked identical questions of the same group of 1,000 people, yet reported different results, this difference would likely be caused by