Campaigns and the Mitigation of Framing Effects on Voting Behavior: A Natural and Field Experiment

There is a widely held belief that the way a ballot proposition title is framed (the way the issue is presented) has an effect on its eventual success or failure on Election Day. Prior to California’s 2010 General Election, the ballot language for two propositions was disputed and challenged in court. For both Proposition 22, intended to change the allocation of state and local tax revenues, and Proposition 23, intended to overturn California’s landmark global-warming law, Fresno County failed to make the court ordered changes. The election proceeded with the unchanged ballot language in Fresno County and the newly adjusted ballot language throughout the rest of the state. This paper takes advantage of this natural experiment to evaluate the potency of framing effects in direct democracy elections, as well as, the role that campaigns (high salience versus low salience) can play in limiting those effects. As a further test, survey data using identical language collected in an area where neither issue was of high salience is used as a comparison. This additional test serves as a mechanism to isolate any potential framing effect of campaigns. We conclude that the way a ballot measure is framed has an impact on its Election Day success so long as it is a relatively low-salience measure. For initiatives with vigorous campaign activity, such as Proposition 23, framing effects are less effective and not statistically significant in this instance.