Polling, Voting and Information Cascades
It’s voting day! Given all the discussion about election outcomes, I am curious about networks in the context of public opinion formation and voting patterns. And no, this isn’t (just) an excuse to write my post on the day it is due.
In researching this, I stumbled across a study by Blais, Gidengil and Nevitte (2006). Their guiding research question is whether or not polling influences voting. This made me think of information cascades, because people’s own decisions are affected by their expectations for what others will do. Information cascades are arguably more about early behavior influencing later behavior, but it is still a related concept. Blais et al study the 1988 Canadian election to assess this question. They choose Canada because it is a first-past-the-post (FPP or FPTP) election in which the candidate with the largest share of votes “wins the election”. This is opposed to a system in which each political party’s representation is proportional to the number of votes they get, rather than a “winer takes it all” situation as in FPP. Blais et al argue that polls are especially important in the context of FPP elections because if a voter thinks a preferred party is unlikely to win they may vote for a more likely party even though it is less liked. This might not be true to the same extent in a proportional representation election system, since the vote would still give political power to the first-choice party.
Blais et al consider two potential reasons for how polls could affect voting patterns. The first is through strategic voting (described above), which is when voting is intended to influence the end-result of the election, rather than to vote for the party which one prefers. Blais et al mention that strategic voting might take the form of voting for a second or third choice party which polls indicate have a larger likelihood of winning, rather than voting for the most preferred party. In the case of the dominance of two parties, such as in the United States, this effect can clearly be bad for voter engagement more generally, as there are few “second choice parties” to choose from. Another option is the contagion effect, where people see parties that are doing well in polls more positively.
So, what did this study show? Blais et al considered if polls 1) affected expectations about the outcome of the election and 2) affected voting patterns. In both instances, the researchers concluded there was evidence that polls affected both expectations and voting intentions. Moreover, they identified strategic voting as a better causal mechanism because their models suggested that vote intentions were affected by polls independently from how voters actually felt about the parties or candidates themselves, meaning a contagion effect was unlikely and that the polls themselves influenced voting.
Finally, how is this related to course content? From Chapter 16 of the textbook, we learn that information cascades can occur when people make decisions in a sequence and the prior behavior or actions of earlier people influences decisions by people further down the line. I think there are two key differences between elections and the information cascades we have discussed in class. Firstly, voters aren’t influenced by what they think others know, but rather by what they think their ability to influence election outcomes are based on how others have voted or are likely to vote. Secondly, polls are not sequential decisions. However, they do indicate expectations of what is going to occur. My point of comparison with information cascades is that people may see polls as an indication of what people are likely to do in the future and base their decisions off of that information. If sufficiently many “prior” or “polled” people behave in the same way, many “subsequent” decisions might be the same. For example, if many people who affiliate with a small opposition party see that the ruling party has strong electoral support based on polling and is likely to win the election and that their own party is doing poorly, they may feel less inclined to vote for the opposition party because they think it is unlikely to win. Rather, they may vote for a more popular opposition party which they like less but feel is more probable to win, especially in a FPP system. They could also feel less motivated to vote at all. That tipping point, based on what people predict will happen based on prior knowledge of people’s behavior (or in this case political attitiudes) is very similar to an information cascade.
Works Cited:
- Blais, André, Gidengil, Elisabeth & Nevitte, Neil. ”Do Polls Influence the Vote?” Capturing Campaign Effects, The University of Michigan Press, 2006, https://www.press.umich.edu/pdf/0472099213-ch11.pdf.
- Easley, David & Kleinberg, Jon. “Networks, Crowds, and Markets.” Cambridge University Press, June 2012, https://doi.org/10.1017/CBO9780511761942.