The votes are in! Asked to complete the same exercise, my class of fifteen students (each voting as a “ten-person bloc”) voted as follows:
Round 1 (no polling data given):
- Charmander: 32 (21.3%)
- Squirtle: 33 (22.0%)
- Bulbasaur: 85 (56.7%)
Round 2 (with polls showing C = 40%, S = 35%, B = 10%, and Undecided = 15%):
- Charmander: 57 (38.0%)
- Squirtle: 47 (31.3%)
- Bulbasaur: 46 (30.7%)
Round 3 (with polls showing C = 30%, S = 30%, B = 27%, and Undecided = 13%):
- Charmander: 41 (27.3%)
- Squirtle: 44 (29.3%)
- Bulbasaur: 65 (43.3%)
These results cannot be directly and unproblematically compared with the results from Amanda’s class, since the differing class sizes forced me to distribute the voter profiles in slightly different proportions. HOWEVER…there are some striking similarities, especially between the respective sets of results from Rounds 1 and 2: The first round, “baseline,” results were nearly identical, and except for an exchange of Charmander and Squirtle, the Round 2 results were nearly identical as well, indicating a similar willingness on the part of Bulbasaur boosters to engage in strategic voting in an attempt to avoid a Charmander/Squirtle victory after they saw the dire straits their favored candidate was foundering in.
The greatest difference arose in the results from Round 3: in Amanda’s class, the “voters” flocked to Bulbasaur more strongly, giving him a 51.8% share of the votes, versus 43.3% in my class, on seeing that how competitive his candidacy was. This is interesting, especially since only 4 out of 11 (36.4%) of Amanda’s class were given decidedly pro-Bulbasaur profiles, while 6 out of 15 (40%) of the students in my class were given such profiles. I might have to chalk this one up to the small sample sizes.
While I didn’t keep tabs on the behavior of individual voters in Amanda’s class, I did think to do this in my own, marking the ballots so that I could distinguish each student’s votes on successive rounds of voting. What can we tell from looking at this information?
Focusing on Bulbasaur’s performance, 11 out of 15 voters ranked Bulbasaur lower in Round 2 than they did in Round 2. This group of 11 voters includes 5 of the 7 voters who gave Bulbasaur more votes than any other candidate in Round 1; the remaining two Bulbasaur boosters gave their favorite 10 points in all three rounds! All but one of the 7 Bulbasaur boosters gave at least some support back to Bulbasaur in Round 3, giving him more points then than in Round 2 (and in some cases, more than they’d given him in Round 1). The behavior suggests, again, the efficacy of positive polling data in bolstering a minor-party candidate’s electoral success.
The only two voters not considered above include a clear Charmander supporter who was also okay with Bulbasaur, whose votes were Charmander 7/6/9 and Bulbasaur 3/4/1, suggesting galvanization around the favored candidate when the race tightened in the final available poll. The remaining voter clearly despised Charmander, initially voting for Squirtle over Bulbasaur 7-3, and shifting to Bulbasaur over Squirtle, 9 to 1, in both of the next rounds. I have to admit I’m not sure what profile this behavior suggests.
In any case, the data, whether considered aggregately or individually, suggest yet again a fairly strong impact of polling data on electoral behavior.
How frightened should we be? Given that West Coast states often exhibit depressed voter turnout relative to their more easterly sister states (Hawai’i consistently wins the dubious honor of having the lowest voter turnout among all fifty states), perhaps the early release of election returns on election night is enough to keep a substantial number of people at home.