Article Preview
Top1. Introduction
With the advent of new technologies and the arrival of e-participation dashboards in smart cities (Bertot et al., 2010, IRPP, 2016, Kumar & Vragov, 2009, Li & Gregor, 2011, Matteus et al., 2018, Xenos & May, 2007), more flexible referendum alternatives have been proposed and investigated in previous literature (Vragov & Kumar, 2007, Attiyeh et al., 2000, Cassella et al., 2005, Freixas et al., 2014, Lorenco & Costa, 2007, Posner & Weyl, 2017). The purpose of these mechanisms is to mitigate some of the deficiencies of simple majority yes-no voting. A simple example can illustrate some of these deficiencies. Suppose that citizens A, B, and C are considering building a public stadium in their small community, and they all start with the same initial endowment of 101. The endowment can represent income, property and anything else of value that the individual possesses. Suppose that the net financial effect of building the stadium is 2 to A, 2 to B, and -10 to C. If the three citizens vote to build a stadium using majority voting, A & B will both vote “Yes” because they gain from the proposal and C will vote “No” because s/he loses. The stadium will be built and that will result in A gaining 2 for a total final endowment of 12; B gaining 2 for a total final endowment of 12, and C losing 10 for a total final endowment of 0. The net social gain: 2+2-10=-6 is negative and so the outcome does not increase the total social value. One could also argue that the outcome is unjust since the citizens start on an equal financial footing but end up with a skewed income distribution.
Five classes of voting mechanisms have been proposed in the literature to deal with this issue: weighted voting (Freixas & Zwicker, 2009), voting with compensation (Oprea et al., 2007), pivotal voting (Attiyeh et al., 2000), quadratic voting (Posner & Weyl, 2017), and voting with storable votes (Casella et al., 2005). Some of these mechanisms have been tested in the laboratory with small groups of people with promising results. Our purpose here is to test how moving one of those mechanisms from the laboratory to an online setting, where voters are able to submit votes asynchronously over an extended period of time, influences an individual’s voting behavior.
So far the mechanism performing best in laboratory tests is the one described in Oprea et al. (2007). Under this mechanism, decision-makers are allowed to express their preference intensities in a referendum and then get compensated in case the alternative they vote for is not selected according to a pre-specified mathematical formula. As Oprea et al. (2007) state, there is currently no widely acceptable strategic behavioral model that can predict human voting behavior under such a setting, except when there are only two voters. Surprisingly enough, however, laboratory tests of the mechanism in small groups with synchronous voting have shown that decision-makers do tend to reveal a fraction of their preference intensity in voting. Thus the usage of this mechanism has resulted in improved social value and in more equitable voting outcomes in the laboratory as compared to simple majority voting.
The applicability of this voting mechanism to decision-making in the field is not yet clear and, as already mentioned above, a solid theoretical model that can predict individual voting behavior does not currently exist. Thus we follow the principles of design-science and conduct an exploratory semi-field experiment to see if voters follow the same voting strategies as in the laboratory. This is seen as a necessary intermediate step before this electronic voting mechanism is tested in the field.