Political scientist Bob Stein of Rice University offers readers of “The Monkey Cage” the following analysis of the link between changing voter preferences (the daily change in early voters’ support for McCain in Harris County, TX) and the change in the Dow-Jones Industrial Average.
bq. The 2008 election witnessed a significant increase in the proportion of voters who balloted before Election Day. In 2004 20 percent of all votes cast for President were cast before Election Day. This percent is expected to increase to nearly 30%. In 32 states, voters have the opportunity to cast their ballots up to three weeks before Election Day at in-person voting places in their voting jurisdiction. Since Texas first adopted this method of voting in 1988 its been widely demonstrated that early voting does not significantly effect either overall turnout or election outcomes. Early voters are more partisan and ideological but there is no evidence that the alleged convenience of that early voting has significantly increased voting among new or infrequent voters (Berinsky 2006).
bq. One hypothesis about early voting that has not garnered much empirical study is whether early voters might miss information important to their vote choices by casting their ballot before the end of the campaign (i.e., Election Day). The figure below plots the difference in the proportion of vote cast for McCain on each day of early voting (October 20-October 31, 2008) in Harris County, Texas against the closing Dow Jones daily average for the preceding day. The expectation is that voter support for McCain will covary positively with the Dow Jones daily average. (The exit poll was conducted with 1,055 voters at 14 of 32 early voting sites between October 20 –31, 2008.) There is some evidence that this expected relationship is observed. That is, support for McCain tended to rise on the same day, or the day after, the Down headed upwards, and to decline according to the same pattern.
MY COMMENT: Whether the same pattern can be observed in other locales is an open question at this point, and it should also be noted that the daily swings in vote support aren’t really all that large. Even so, this constitutes some provocative first-cut evidence about the dynamics of early voting.