In democracies, policies are expected to be responsive to public opinion. Extant research showed that responsiveness is selective. It varies across issues, time and countries. Yet, how come policies vary in their responsiveness has not received a satisfying answer. POLEVPOP formulates and examines a novel answer to the puzzle why policy responsiveness varies. Its core argument holds that politicians evaluate public opinion and let their actions—in line with public opinion or going against it—depend on their appraisal. When public opinion is evaluated negatively, it has no effect on what politicians do; that it is evaluated positively increases the chance that politicians act congruently. Politicians’ appraisal of public opinion has been completely overlooked as a mechanism bringing about responsive representation. Considering it a core factor POLEVPOP examines three matters: (1) which criteria politicians use to appraise public opinion; (2) how, depending on the opinion content of the message, the channel through which the opinion is conveyed and the group from which it comes, concrete public opinion signals are evaluated; and, (3) which effect these evaluations have on politicians’ political action. The central expectation is that public opinion is evaluated by politicians based on a consistent and common scoreboard. For instance, opinion signals are rated based on their representativity and underlying public opinion is evaluated on its quality and its intensity. The project tackles these matters drawing on a comparative study in eight different countries (Australia, Belgium, Canada, Czech Republic, Israel, Portugal, Switzerland, and Sweden). In two consecutive rounds of data gathering, a large sample of politicians is surveyed and interviewed, and they are subjected to a series of survey-embedded experiments. To put politicians’ behavior in perspective, their answers are compared to parallel citizen surveys in all countries.