Hamilton and Gifford called this pattern an illusory correlation, because the impression students formed of Group B members was incorrectly associated with more undesirable behaviors. What could be responsible for this illusory correlation?

It seems to be related to the fact that fewer undesirable behaviors were exhibited overall (so undesirable behaviors were unusual) and that Group B had fewer members than Group A (so Group B was a "minority group").

"Group B" and "undesirable behaviors" were both less frequent than the alternatives, so their rare occurrence together (highlighted in the table, using the actual values from our rating exercise) caught people's attention, leading them to later overestimate the frequency of their joint occurrence.