In Sunday's New York Times Magazine, Matt Bai made the data-free assertion that "the American public doesn’t seem to move very much in its basic attitudes about government":
The cautionary note here, for jubilant Democrats, is that there is little reason to believe that the electoral trend in their favor actually reflects any widespread ideological shift. If you only look at numerical majorities, it might well seem that the story of the last 20 years in American politics is one in which voters have swerved erratically from one ideological pole to the next, embracing a harsh kind of conservatism in 1994 and then a resurgent liberalism in 2006. In reality, though, the American public doesn’t seem to move very much in its basic attitudes about government, which have remained mostly pragmatic and predictable; simply put, people tend to want a little more government when times are tough and a little less when things are going well. The number of voters who identified themselves in exit polls as conservative, liberal or moderate remained virtually unchanged between 2004 and 2008 — and in fact, those numbers have been more or less steady for decades.
But as I've pointed out before, UNC's Jim Stimson has shown using his "policy mood" measure (Excel file) that the American public's demand for more or less government is (a) not static and (b) doesn't simply respond to the state of the economy:
Instead, the public tends to move in the opposite direction from policy (after some lag), acting like a thermostat of sorts against government overreach in a liberal or conservative direction. Stimson has only updated his measure through 2006, but I'd guess that the public has continued to move in a liberal direction as part of a continued backlash against the Bush administration.
Bai's second point is also uninformed by knowledge of the relevant research. Decades of political science scholarship have shown that individual self-identification as liberal or conservative is a weak indicator of actual policy preferences for most of the population. In fact, Stimson shows in later work that a significant proportion of self-described conservatives want more government spending in various issue areas.