that they were relatively certain about whom they were going to vote for and almost all indicated
that they would definitively or probably vote in the election.
Both the Post/Univision/TRPI and National Exit Poll estimates could certainly be
accurate, if a large number of Mexican American or Latino voters shifted preferences in the
weeks between the poll and the election or if a large number of likely Kerry voters stayed home
or didn’t vote in the presidential race. The first explanation seems unlikely; this poll agreed with
other surveys conducted in Latino communities in the 2004 election (including a companion poll
conducted by the same organizations in July 2004). Nothing happened in the final weeks of the
election, either nationally or in any of the states with sizeable Latino electorates to shift
candidate preference dramatically. The second explanation also seems unlikely. Latino turnout
increased at rates higher than the customary increase in Latino turnout between presidential
elections. Since 1976, the Latino electorate has increased by an average of 20 percent between
each presidential election (largely due to the growth in the Latino population). Between 2000
and 2004, the Latino electorate grew by nearly 28 percent (DeSipio and de la Garza 2004: Table
1.1; U.S. Bureau of the Census 2005: Table 1). A significant drop off in Democratic Latino
votes would have led to a decline or slowed growth in Latino voting.
Literature Review
Partisanship
We expect partisanship to have a large impact upon the vote choices of Mexican
Americans given its importance for the general population. It remains to be seen, below, just
how large an impact that is and whether or not it outweighs other factors. Given our expectation
that the effect will be non-negligible, we also examine the factors associated with Mexican
3
Convention