If nothing else, democratic politics requires compromise. Mass polarization, where citizens disagree strongly and those disagreements magnify over time, presents obvious threats to democratic well-being. The overwhelming presumption is that if polarization is occurring, a substantial portion of it is attributable to fragmentation attendant an increasingly choice-laden media environment where individuals expose themselves only to opinion-reinforcing information. Under what conditions does mass opinion polarization occur? Through two over-time, laboratory experiments involving information choice behavior, this paper considers, first, the effects of slant in one's information environment on over-time opinion dynamics and, second, the moderating role of attitude importance on those effects. The experiments reveal that, despite similar information search behavior, those with strong attitudes are dogmatic, resisting even substantial contrary evidence; those with weak attitudes by contrast hear opposing arguments and develop moderate opinions regardless of the prevalence of those arguments in their environment. Evaluations of information, rather than information search behavior per se, explain why individuals with strong attitudes polarize and those with weak attitudes do not. Polarization therefore seems to require more than media fragmentation and, in fact, a more important factor may be the strength of citizens' prior attitudes on particular issues.If nothing else, democratic politics requires compromise. Mass polarization, where citizens disagree strongly and those disagreements magnify over time, therefore seems to present obvious threats to democratic well-being. Political science has demonstrated a modulating, though frequently uneasy, view of political polarization. This ambivalence reflects the ap- 1 party contrast without mass polarization.1 Just a decade later, McClosky, Hoffmann, and O'Hara (1960) demonstrated that the two-party system offered clear differences between parties without undesirable polarization.2 In more recent times, McClosky, Hoffmann, and O'Hara's assessment of American politics seems to resonate much more with the conclusion drawn by Fiorina and Abrams (2008) that "the American public as a whole is no more polarized today than it was a generation ago" (584) than with the contemporary observation by Hetherington (2001) that "Greater ideological polarization in Congress has clarified public perceptions of party ideology, which has produced a more partisan electorate" (629), but the polarization literature is still growing (see, for example, Levendusky 2009Levendusky , 2010Levendusky , 2011Stroud 2011;Carmines and Ensley 2004).The overwhelming presumption is that if polarization is occurring, a substantial portion of that polarization is attributable to the fragmentation attendant an increasingly choice- which is necessary to avoid faulty ecological inferences about opinion dynamics (Druckman 1 "If the two parties do not develop alternative programs that can be executed, the voter's frustrat...