Over the last thirty years, there has been ever-increasing growth in the volume of poll data on voting intentions and political attitudes in advanced democracies. This trend is most apparent near election time, with large numbers of polls conducted and published during the election campaign. This increasing volume of polls has been met with a parallel increase in the proportion of media coverage devoted to them, and now their increasing dissection in social and political fora online – across the blogosphere, on twitter and the like. Ever since the infamous ‘Dewey Defeats Truman’ headline of 1948, polls have come under scrutiny for their ‘failure’ to accurately predict the results of election outcomes – over time and across a range of political systems. Whereas once upon a time debates over polling methodology concerned the merits of quota sampling, more recently the increasing use of internet polling has been much debated. Often, however, discussion of the polls has lacked grounding in evidence-based research, relying on anecdote or misdirection and often reflecting the partisan disposition of their proponents or critics.
The purpose of the Polling Observatory, then, is to provide analysis and discussion of the vote intention polls and their media coverage. It also acts to disseminate the latest research on polling accuracy.