| dc.description.abstract |
<p>Geographic clustering has been linked to contemporary political polarization by
jour- nalists and other researchers in recent years, most recently and notably by
Bishop and Cushing (2008). In these accounts, clustering is motivated, in part, by
shared tastes for combinations of place attributes that attract individuals with interrelated
values and similar characteristics or skillsets. In order to test whether political
pref- erences aligns with location choice, this paper proposes a sorting model based
on the composition of migrants' political preferences. </p><p>Sorting is defined as
the increase in the variation of a parameter of preference distributions of different
location, in the absence of individual preference change. The model estimates the
separate prob- abilities of party identification in U.S. congressional districts among
migrants and non-migrants.</p><p>Based on an empirical application using the 2006
Cooperative Congressional Elec- tion Study, I find that a significant number of district
satisfy the sorting condition. Aa multinomial logit model predicts that individual
ideology is significant explana- tory variable in the partisanship of destination
districts among migrants, even after controlling for the partisanship of originating
districts.</p><p>The final chapter evaluates sorting and polarization in U.S. congressional
districts based on intra-decade changes to population size. I show that overall polarization
in high growth districts exceeds sorting, and suggest this results from an increase
in electoral bias that could result from heavy migration into districts that begin
the decade as very homogenous.</p>
|
|
