Fowler: Turnout in a small world

Fowler. 2005. Turnout in a small world. In Social Logic of Politics, ed. Alan Zuckerman. Temple University Press.

In Brief

We've all heard that there are "six degrees of separation" between everybody on earth. Using similar logic, Fowler suggests that we live in a "small world" of social networks; we have friends, who have friends, who have friends. What's more, our behavior influences our friends, which influences their friends, which influences their friends. Thus, turnout has a contagion effect: By turning out to vote, I influence many other people to do likewise. And since my political views are similar to those of my friends (and their friends, etc), then my decision to turn out will improve my favored candidate's margin of victory--by anywhere from 2 to 9 votes. Fowler calls this contagion effect a "turnout cascade."

Method

Fowler runs a simulation. He plugs in certain parameters--how many friends each person has, how likely their friends are to imitate a turnout decision, how many of your friends' friends you know ("clustering"), how frequently you interact with your friends, population size, and average path length between any two people (e.g. six degrees of separation). He obtains his parameter values from a small-scale networking study conducted by Huckfeldt and Sprague; thus, his estimates are realistic (and on the conservative side).

Fowler then manipulates a single (simulated) person's decision of whether to vote and runs the simulation thousands of times to see how changing one person's turnout decision affects other (simulated) people. On average (under conservative assumptions), this one person's decision to vote increase total turnout by about 4 votes; it increases the one person's favored candidate's margin of victory by 2 or 3 votes.

Figure 2 shows how varying the other parameters affects the simulation. The effect of one person deciding to turn out rises if

• Friends discuss politics more frequently;
• Imitation (of a friend's turnout decision) is more likely;
• One's friends are more likely to know one another (to a point; this reverses if groups are overly clustered);
• One has more friends;
• There are fewer degrees of separation between any two people.

Interestingly, population size has no effect at all on the number of voters affected by a single person's decision to turn out (although the meaning of increasing turnout by 4 votes would diminish with population growth).

Comments and Criticism

Once again, Fowler has produced an interesting and novel study that is hard to complain about. I would be curious whether the model changes when different people have different numbers of friends--that is, when there are opinion leaders. Also, it seems that the probability of imitation wouldn't be uniform--I'm more likely to imitate some people than others, even though I talk politics with all of them. But I doubt these changes would dramatically change the model's results.

On another note, I'm not sure where this work leaves us with regard to the rational voting paradox. It certainly would increase the probability that one's vote is decisive, but the probability would still be nearly zero in a large population. [[]]