James Davis et al. continued Gagne and Smith’s studies and divided the variable of thinking aloud into two variables: thinking aloud and moderator presence. The other variable of searching for a general strategy was left as it was. With the groups that worked alone without the moderator, the moderator instructed the subjects and helped them familiarize themselves with the situation, but left the room before the actual test problems. Before leaving the room, the moderator told that he would be in the next room and observe the subject through one-way-mirror.
The results of the Davis et al. study were somewhat conflicting, since the results with Five-tile Tower of Hanoi indicated that the moderator presence caused most significant statistical difference between the subject groups, but with the final six-tile tower that every group solved silently, the ones that had thought aloud previously, performed significantly better, as in Gagne and Smith’s experiment. The experimenter effect was not detected any more on the six-tile tower. Davis et al. conclude that the mere presence of others [in a test setting] is a complicated variable, and that talking while working apparently improved performance potential that was detectable when working silently.