Media critics often complain about too much coverage of ''horse-race''
polls. A common assumption sustaining this criticism is that such cov
erage competes with issue coverage. We propose the opposite, that hors
e-race polls might increase voters' attention to other election messag
es, including issue information, which in turn leads to a better under
standing of public policies. To test these competing theories, we cond
ucted three studies during the 1992 presidential election: a controlle
d experiment, a statewide one-time survey, and a three-county two-wave
panel survey. Each of the studies reports a positive relationship bet
ween horse-race polls and issue knowledge. The two surveys also report
positive correlations between exposure or attention to polls and expo
sure or attention to general election coverage, which in turn is posit
ively correlated with issue knowledge.