In 2 experiments, the generation effect, which occurs when individuals
remember materials they have generated (and, thus, targeted) better t
han materials generated (and targeted) by others, was investigated. Th
e effect has not generally been found in natural settings when total t
est performance was examined. The examination of total performance rat
her than just targeted items may mask the effect because items not stu
died (nontargets) are included. In Experiment 1, groups were compared
that generated their own outlines or study questions with groups that
received experimenter-generated materials. Test scores showed generati
on effects for target items. In Experiment 2, students who generated q
uestions were compared with yoked students who used those student-gene
rated questions. Results were again significant, indicating that the g
eneration effect and not potential confounding variables produced the
results. The effect does occur in a natural setting, but only for test
items targeted by generating students.