The authors use the 1993 municipal election in Atlanta, Georgia, and a
special election to fill two vacancies on the Fulton County (Georgia)
Commission to test ballot-confusion, saliency-of-the-contest, and vot
er-fatigue explanations for roll-off (ballot noncompletion). No eviden
ce of the ballot-confusion explanation is found, but some evidence of
voter fatigue and contest visibility is apparent. Racial differences i
n roll-off have implications for the responsiveness of public official
s and the creation of single-member districts.