Assessing empirically to what extent citizens’ preferences are considered during the elaboration of public policy remains a difficult issue in political science research, although it concerns one of the central pillars of modern democracies. While most scholars agree that public opinion has some influence on public policy, their findings vary from a small influence to a very important influence (Stimson 2007; Manza &amp; Cook 2002; Page 2002; Glynn et al. 2004, Burstein 2010). The ability to capture trends in public opinion on important political issues is a prerogative for the study of democratic responsiveness. The traditional approach has been to measure public opinion through the preferences expressed in representative surveys, notably through the use of items such as the ‘most-important-problem’ (Binzer Hobolt &amp; Klemmensen 2008), questions about preferences for ‘more’ or ‘less’ spending in certain policy areas (Soroka &amp; Wlezien 2010), or a combination of survey indicators about specific policy preferences (Manza, Cook 2002a &amp; Page 2002). One major disadvantage, however, is the dependence on the availability of survey data to characterize public policy preferences. Scholars are becoming increasingly aware of the possible bias introduced by the availability of survey data (Barabas 2016; Burstein, 2003). Several authors underscore the fact that this could lead to an excessive focus on issues to which governments are especially responsive as polling agencies ask questions on issues they consider to be salient. Yet, salience seems to favor responsiveness (Page 2002; Burstein 2014). On the other hand, and in line with the critical literature on public opinion, we could imagine that governments try to avoid survey coverage of issues they consider to be too salient. Jacobs and Shapiro (2004) report, for instance, that the Johnson’s White House did not include the Vietnam War in the list of possible answers to the Most-Important-Problem question, fearing this issue could drown out all other issues. Mobilizing data of surveys directly commissioned by the German government (2013- 2021), we analyze the importance of salience for the likelihood of issues to be covered by surveyed public opinion. In a first step, we test whether salient issues are overrepresented among survey questions commissioned by the government compared to issues on the political agenda of the government and the parliament. The changing nature of salience (Geer 1996) not only implies variation in attention across policy domains but also in attention within domains over time. Based on the observation of responsiveness studies that the change of the degree of salience has a bigger impact on the reactivity of governments than the actual level of salience (Brettschneider, 1990), we test in a second step whether an issue has a higher likelihood of being covered by survey questions when it gains quickly in salience among citizens.We therefore distinguish salience of issues in the media as measured by a media-indicator developed by the Bundespresseamt and salience of issues in the population as measured by the questions.< Réduire