By surveying only grades 1, 2, and 3, she is not taking a representation of the entire school. It is possible that the popular color or the color the students like would change over time and the opinion of the younger kids could be different from that of the older kids. Therefore, the sample is biased unless she surveys students from all grades.
