Early and Absentee Voting and Unrecorded Votes in the 2002 Midterm Election
David C. Kimball, Martha E. Kropf · 2004
Abstract We examine the growing use of early and absentee voting methods and their impact on unrecorded votes in the United States. While several studies assess the impact of early voting and mail-in voting on voter turnout, we are unaware of any published research on the impact of early voting on unrecorded votes. Further, while many studies consider the effects of voting equipment on unrecorded votes, the studies fail to differentiate between those occurring at the polling places, and those occurring during early or absentee voting. On the one hand, early voting may reduce unrecorded votes since voters may have more time to engage in the voting process on a day that is most convenient for them. In addition, if absentee voters tend to be higher in socio-economic status than Election Day voters, then unrecorded votes may be less common for early and absentee voters. On the other hand, many absentee and early voting methods do not include mechanisms that notify voters of ballot errors. We create a dataset from six states in the 2002 general elections that considers factors of ballot design, voting technology and demographic factors. The unit of analysis for this study is the county, since voting technology and ballot-design decisions are made at the county level in almost all states. We calculate the unrecorded vote rate for early or absentee voters and Election Day voters in each county. Ballots from counties in six states are collected (Iowa, Kansas, Georgia, Florida, Tennessee and Illinois). Three of the states in our sample (Florida, Kansas and Tennessee) allow in-person early voting. We also code paper-based (including optical scan) and pencil ballots in terms of several graphic design elements. The results presented here confirm what we know about voting technology and demographic factors and their effect on unrecorded votes. However, this paper takes a step further and examines the differences among early, absentee and precinct unrecorded votes. We find that there is very little average difference between precinct unrecorded votes and early/absentee unrecorded votes. Moreover, this paper confirms that many of the same factors that affect unrecorded votes at the precinct also affect early and absentee unrecorded votes. For example, “good” ballot features reduce the incidence of unrecorded votes for all three types of voting. Our results suggest that states should not adopt early voting as an answer to the problems of unrecorded votes that we have experienced. It appears that the shift to early and absentee voting will not reduce unrecorded votes, and thus will not lessen the chances of another vote counting controversy. On the other hand, our results seem to suggest that election officials (and campaign staffs) should not worry that an increase in early voting will make the problem of unrecorded votes any worse than it has been in the past