Abstract
Previous explanations of bandwagons from election polls have exclusively emphasized conformity causes. We propose, in addition, “indirect' causes, in which election predictions first affect key actors, influencing their decisions concerning financial support, volunteer work, or endorsements. These decisions then produce major campaign alterations that influence the voters and alter the election outcome. Our addition clarifies anomalous bandwagon research findings and directs attention to the possibility of bandwagon feedback on subsequent elections. If the same forecasters create frequent bandwagon effects, their credibility should increase as a result of enhanced accuracy. But increased credibility should in turn increase the self-fulfilling tendency of their subsequent forecasts. Such deviation-amplifying feedback would permit polls to produce a highly significant, and expanding, influence on elections.