Locke & Latham (1990a) report that specific, difficult goals lead to better
performance than "do-your-best" instructions, whether the goals are self-s
et or are set by an external source. However, in Experiment 1, as well as i
n previous research (White, Kjelgaard, & Harkins, 1995), we did not find se
lf-set goal effects. A meta-analysis showed that self-set goal effects can
be produced if two conditions are mel: Prior to setting their goals, partic
ipants take part in a pretest that is equal in duration to the experimental
task; and the experimenter has access to the participants' goals and the p
erformances. Experiment 2 confirmed these meta-analytic findings by showing
that when these two conditions were met, goals stringent enough to produce
the effects were set, and goal-setting effects were obtained.