Personal interviews were conducted with 353 eligible voters in Madison, Wisconsin, in October 1976, and repeated with all but 30 after election day to examine the correlation between voters' age, paper reading (especially public affairs), and television watching (especially public affairs), and the amount of behavioral volatility (voter abstention, unstable party affiliation, and openness to alternatives) and subjective volatility (contingent voting, antiparty sentiment, conflict and indecision, and compromise). The study specifically measured the contribution of media (especially television) to voter volatility, the unidimensionality of volatility, and the various normative aspects of volatility (single subjective dimension or several dimensions). Results indicated that both behavioral and subjective volatility are multidimensional rather than unidimensional. Education did not predict either kind of volatility very well; political interest had more consistent nonvolatile effects, except as reflected in conflict and indecision; and both television viewing time and public affairs viewing had little effect on volatility. Among the younger respondents there was a stronger correlation between public affairs viewing time and volatility than among the older subjects. (TJ)