The idea that voters' decisions are largely determined by the issues at stake in the election. Before survey research, most writers assumed that issue voting was the norm in democratic elections (although there have always been sceptics, from Condorcet to Schumpeter, who denied this—Schumpeter even denying that it ought to be). The first surveys of the determinants of voting showed that habit and party identification played leading roles, and issue voting almost none, as most voters knew nothing at all about many of the issues discussed by politicians and journalists during elections. Since the late 1960s issue voting has enjoyed a modest revival. This is due partly to the influence of rational‐choice theory on election studies, and partly to the recognition that the politics in the United States in the 1950s—when the most influential surveys of the party identification school were done—were unusually bland and consensual. When ‘issues’ are restricted to ‘issues which are salient to the electorate’, it can be shown that issue voting plays quite a prominent role in a typical election. The other main determinant is retrospective voting, or evaluation of the party (team) currently in office. Here voters must compare the performance of the government with the promises of the opposition, a comparison which most voters are well aware is lopsided.