There was also a significant main effect for the pair-share/no pair-share variable, F(1, 186) = 4.73, p < .05. However, exam performance was better when lectures did not include pair-share experiences. The estimate of the magnitude of effect indicated this effect was small, h2 = .02. The instructor self-disclosure/no disclosure by pair-share/no pair-share interaction was not statistically significant, F(1, 186) = .09, p > .05.

The instructor's providing personal examples of concepts presented in lecture improved retention of lecture material. Exam questions from lectures with instructor self-disclosure were answered correctly more often than were exam questions from lectures that had not included such examples. This result agrees with those of Wollen, Quackenbush, and Hamlin (1985) for textbook material: Providing examples improves student understanding and later exam performance. On the other hand, my primary focus in doing this study was to look at self-referencing. I added the variable of instructor self-disclosure primarily as a way to model self-referencing for students.

Thus, I did not evaluate the parameters of examples an instructor might provide. I did not include in this study whether these examples work best because they are humorous, for example, or whether personal examples work better than simple illustrative examples, though these questions may be important to look at in future studies. Given the results of Goldstein and Benassi (1994), I thought instructor self-disclosure might help establish a positive classroom atmosphere, but was surprised to find its effect on retention was as prominent as it was.