Date of Award


Degree Name

Doctor of Philosophy


Political Science

First Advisor

Dr. J. Kevin Corder

Second Advisor

Dr. John Clark

Third Advisor

Dr. Gunther Hega

Fourth Advisor

Dr. Matthew Higgins


Low turnout remains a persistent problem in American politics. The decline in turnout has been studied in various ways. In some cases scholars analyze aggregate turnout data and compare turnout in election districts with high and low concentrations o f particular ,social groups (Neimi and Weisberg, 1993). In other cases, surveys provide an opportunity to examine the causes and correlates o f turnout at the individual level. Various researchers find that socio-economic factors are related to turnout. People with more education vote at much higher rates than those with less education, higher income and middle class people are more likely to vote than lower income people.

Based on various surveys, it has been widely accepted that lower class poople turnout out at low rates and contribute disproportionately to the decline in overall turnout in American presidential elections. However, other scholars argue that the class differences between voters and nonvoters in presidential elections remain the same from 1964 through 1988. This research examines whether lower class turnout at lower rates than non lower class. This research question starts from the problem of accuracy o f survey research. As Neimi and Weisberg (1993) argue, surveys always obtain a higher turnout rate than official statistics reveal. They argue that misreporting turnout is related to demographics, with more highly educated people most likely to claim they voted when they did not. To determine how accurate individual-level surveys are, I will use the method of ecological inference to examine voting behavior.

This study is expected to contribute to the study of voting behavior in several ways. First, using ecological inference, we do not have to rely solely on the survey data to study individual voting behavior. Secondly, as we are able to use aggregate level data, we can locate behavior within its economic context.

The results confirm that the level o f the participation o f the lower class was lower in presidential elections than that o f non lower class, and the lower class contributed to the decline of turnout more than non lower class did (contrary to claims of Leighley and Nagler). The estimates also indicate that non lower class turnout was stimulated by economic context to a greater degree than was the lower class turnout. Specifically, in most states and years, as the unemployment rate increases, the probability o f the turnout by non lower class decreases at greater degree than the probability o f the turnout by the lower class. Reproduced

Access Setting

Dissertation-Open Access