The strategic voting across families of various electoral systems is often depicted as "settling for lesser evil" (e.g. Cox 1997, Gschwend 2004) in order to prevent a worse outcome. In this paper we experimentally (13 sessions with 18-30 participants) explore the degree of strategic voting (as compared to non-voting) in situations with (non-compulsory) costly voting where voters face outcomes with positive payoffs ("choosing the greater of several goods") and/or with negative payoffs ("settling for lesser evil"). We explore voters´ decisions in rather complex environment of the two-round electoral system, with symmetric amount of private information about the preferences of electorate available and show the behavior does differ systematically in both situations. The results have implications for the representation (extremism, polarization, presidential races).
