Reddit reviews The American Voter

In political science most of the literature on vote choice, at least in contexts with stable party systems, builds out of the loyalties people have to political parties. Partisanship creates what the authors of the seminal work The American Voter call a "perceptual screen" which filters information in ways that reinforce these ties. In other words, people first identify with a political party, then interpret the world in ways that support these views (think confirmation bias and motivated reasoning). This identification, moreover, typically [comes from parents](http://press.princeton.edu/titles/654.html] or other early social experiences.

Vote choice and candidate preference then follows from these loyalties. Loyalties to a political party is symbolically and psychologically meaningfully, much like supporting a sports team or adhering to given religious tenets. That's why you'll see people sticking by candidates regardless of information, among many other political phenomena.

See this in the NYTimes for a quick overview.

Typically you vote the way your parents voted. National leadership of the Republican Party worked to integrate libertarians into the party throughout the mid-20th century, taking advantage of growing concerns about corruption stemming from the New Deal and from the economic alternatives to Keynesianism that began to blossom following the then-shocking success of a little-known, mostly-theoretical economist named F.A. Hayek (the link is to a really fun podcast detailing the rise of Hayekian thought in the US). Modern American libertarianism largely coalesced in its infancy around critiques of the theory and practice of the public policies put into place by the Democratic Party, which enjoyed uninterrupted rule for twenty years prior.

Libertarianism had been around in the United States a long time before this but not as a serious part of party politics until the mid-20th century. So essentially you get a generation of libertarians welcomed with open arms into the Republican Party. They came to conflict with pretty much every other wing of the party soon ( though they were mobilized as Republicans a whole generation before, say, Evangelicals ), but mostly I would say the momentum starts from there. The first generation of modern Libertarians were Republicans, and so their kids naturally will be, too.

Legislatures polarize. People do not.


And yes, legislatures (and the congress) have polarized substantially. DW-NOMINATE data on legislative voting behavior demonstrates a recent trend towards a highly polarized congress. Not all state legislatures are as polarized (see Shor-McCarty) but nationally this is certainly the case. DW-NOMINATE has put out this great video (you can see it on youtube here) that demonstrates the ideological movement of legislators in congress over time (from the 1st to the 111th legislature).

What we see amongst the voter however is maybe more of a faux-dealignment. Which is to say that people claim to identify less with either political party thusly, reporting themselves as being "independent". We can see this trend clearly in the surveys (see 2015 Gallup Poll).

However the prevailing models of electoral behavior cast a great deal of doubt on this being anything more than the electorate signally displeasure (perhaps over legislative polarization) while otherwise doing what they've always done. Campbell et al.'s The American Voter (which is the seminal electoral behavior work in contemporary american political science) argues that partisan identification is so stable that it is essentially inherited via a process of socialization and from one's own parents. They then go on to point out that all political preferences and decisions are then viewed through that inherited partisan lens. So while we see people self-reporting less affiliation with either party than we did before we don't see people behaving any differently. In fact, when we consider partisan leanings amongst independents (meaning: whether independents "lean democrat" or "lean republican") we don't see any added likelihood of the voters "switching" parties. In fact, we see most independents consistently vote for one party or the other based on their leanings in precisely the way Campbell's model suggested they would.


Sources:

DW-NOMINATE - national polarization data

Shor-McCarty - state polarization data

Campbell et al. 1960. The American Voter

Additional reading:

V.O. Key. 1966. Responsible Electorate

Are you positing the premise that most voters select a candidate based on the relationship between their views and the policies the candidate espouses? In fact, only a tiny percentage of people vote this way. Most people vote strictly along party lines, even if they claim to be "independent," and during primaries respond only to name recognition and one or two valence issues, which are typically very abstract (e.g., small government, gun control reform, lower taxes). In addition, they are often wrong about their chosen candidate's actual opinions on these valence issues.

Academic sources:

Baldassarri & Gelman, "Partisans Without Constraint: Political Polarization and Trends in American Public Opinion"
Fenno, Senators on the Campaign Trail: The Politics of Representation
Jacobson, A Divider, Not a Uniter: George W. Bush and the American People
Mair, Party System Change: Approaches and Interpretations

This view, that issues have very little impact on the decisions of individual voters absent a rare systemic shock (such as the Great Depression, the Civil War, etc.), is called the Michigan Model, after its origin in the National Election Studies at the University of Michigan and the seminal text The American Voter. Some modern scholars have attempted to criticize this model, but statistically, it has generally held true. For a look at some of these attempts, one decent source is "Choice, Context, and Consequence: Beaten and Unbeaten Paths Toward a Science of Electoral Behavior" by Paul Allen Beck, excerpts from which are available on Google Books here.