The Network Effects of Polls
https://www.press.umich.edu/pdf/0472099213-ch11.pdf
There have been many studies on the impact of prediction polls on voter behavior and election results. This paper, Do Polls Influence Votes, highlights evidence in support of polls changing voter behavior and concludes that the polls “affected voters’ perception of the various parties’ chances of winning” and “strategic voting as some voters became less inclined to support a party whose chances of winning appeared slim” (Blais et. al.). An example of this phenomenon in the first past the post system of the United States is the dominance of the republican and democratic parties which prevent smaller parties from winning elections. The final conclusion made by the study is that polls “did not have a contagion effect, since voters did not come to evaluate the parties and leaders who were doing well in the polls more positively” or in other words, polls effect the behavior of voters and the outcomes of elections but do not necessarily change how voters think about candidates (Blais et. al.).
The effects of polls on voters can be further explained using network effects, particularly through information cascades and the “rich-get-richer” phenomenon. Polls provide information that allow voters who are unsure of their preferences to follow the wisdom of the crows and voters who are sure of their preferences to strategize their vote based on how everyone else is voting. The “rich-get-richer” phenomenon is clear in that voters are less likely to vote for candidates or parties with a slim chance of winning, in the same way that no one will buy a product if they believe everyone else will also not buy it. In some cases, if it is clear that one candidate or party will win, some people may also think their vote does not matter and be less inclined to vote. Network effects also have explanatory power here as they show us that once a certain level of popularity is reached the graph or the people voting will converge to a high value.