1. McDonald (2011 McDonald, I. 2011. Migration and sorting in the American electorate: Evidence from the 2006 Cooperative Congressional Election Study. American Politics Research, 39(3): 512–33. [Web of Science ®] , [Google Scholar]) finds support for this idea at the level of congressional districts with survey evidence that movers' destinations are a closer ideological match to their own preferences than their place of origin.

2. At this point a question arises as to whether we should evaluate individual movement or household movement, because many voters move together with family members (e.g., spouses, adult children), as a part of a household unit, not as individuals, suggesting clustering or dependence across individual observations. But after estimating alternative models, it quickly became clear that with the large number of cases in this research there was no substantively significant difference in estimates resulting from using households as observations or from clustering standard errors by household. We therefore present straightforward models of individual movement, although the alternatives are available.

3. ZIP codes appear to be good proxies for neighborhoods. They are geographically compact for purposes of mail delivery, and they are typically constructed at the appropriate scale at which a socioeconomic or political environment might be evaluated by someone looking to relocate. ZIP codes are similar in size to census tracts, which sociologists have long used to measure residential poverty, concentration, and segregation (Jencks and Mayer 1990 Jencks, C. and Mayer, S. E. 1990. “The social consequences of growing up in a poor neighborhood”. In Inner-city poverty in the United States, Edited by: Lynn, L. E. Jr and McGeary, M. 111–86. Washington, DC: National Academy Press.  [Google Scholar]; Massey and Denton 1993 Massey, D. S. and Denton, N. 1993. American apartheid: Segregation and the making of the underclass., Cambridge, MA: Harvard University Press.  [Google Scholar]). ZIP code data are also commonly reported on the Web sites of realtors and in real estate marketing efforts, suggesting that they are sometimes directly relevant to relocation decisions.

4. The principal exception is New Jersey, where a peculiar registration law encourages voters to register as unaffiliated but then declare a political party once they vote in their first primary election.

5. Our model includes an intercept (that can vary by ZIP code and state), coefficients for the Level 1 independent variables, and a stochastic term (, where the indexes i, j, and k denote individuals, ZIP codes, and states, respectively).

6. These types of party switches are more common than one might think largely because individuals do not commonly reregister with a different party at their place of residence if their official registration does not preclude them from casting a ballot for the candidate of their choice. Migrating and the necessity to reregister, however, lessens the cost of registering with their favored party or new allegiance. In this way, migration offers a voter an opportunity to reveal updated and more true party preferences that have evolved over time.