Two experiments were conducted to examine correction for perceived bia
s in persuasion situations. Study 1 showed that, although a manipulati
on of source likability had an impact on attitudes when no instruction
to remove bias was present, when people were asked to remove any bias
from their judgments, the effect of the source likability manipulatio
n disappeared. The fact that the correction instruction did not increa
se the impact of an argument quality manipulation on attitudes suggest
ed that effort aimed at correction is conceptually distinct from effor
t aimed at processing a message in general. Study 2 showed that a corr
ection for source likability took place under low elaboration conditio
ns-where a manipulation of source likability had an impact when no cor
rection instructions were provided, and under high elaboration conditi
ons-where a manipulation of source likability had no impact when no co
rrection instructions were provided. In the high elaboration condition
s, correcting for an impact that was not actually present led a dislik
able source to be more influential than a likable source.