This study tests the hypothesis that poll data have a weaker influence on voter judgment of the distribution of opinions than do exemplars, i.e. media presentations of statements by individual voters. In an experiment, 274 student subjects read a newspaper story containing poll results and voter statements varied in three different ways (pro-government coalition, anti-coalition, no information, 3 x 3 design). The dependent variables were the perceived climate of opinion, personal opinion, and voting intentions. Findings show that for climate of opinion and personal opinion, the influence of exemplars was considerable stronger, to the point of overriding the effects of poll results, but neither type of information had any influence on voting intentions. The results are discussed with regard to their implications for survey research, media effects research, and journalism.