國立臺灣大學政治科學論叢

Abstract / Chinese PDF Download

Negative campaign ads have become an indispensable technique in modernday elections. Particularly for one-to-one elections, it is very common to seevarious ads attacking the other party, broadcast throughout the media. Althoughmuch research has been undertaken examining the effects of negative ads, theresults are inconclusive and there are still many important issues worthy of furtherexploration. This paper investigates the effects of these ads on the incumbent’sreputation. It also looks at which types of ads are important but were previouslyneglected by previous empirical research into negative campaign advertising.Based on the expectancy disconfirmation model and the theory of persuasiveknowledge, we infer the hypotheses and conduct two experiments to explore theeffects of these variables under different contexts: the ad sponsor being theincumbent or the challenger. Results indicate that when the sponsor is theincumbent, it is not beneficial for the incumbent to use attacking ads, as this leads toan increase in self imposed damage to reputation and a decrease in voters’ votingintention to a greater degree than for those with lower reputation. Using emotionalappeals would decrease the amount of damage done. As for low reputationincumbents, appealing types of advertisement do not make a significant differenceto voter responses. When the ad sponsor was the challenger, attacking the high-reputation incumbent generated boomerang effect, causing greater damage to thesponsor, whilst attacking the low-reputation one would generate the victimsyndromeeffect, improving voters’ target evaluation, attitude, and voting intentiontoward the target. Thus it is not suitable for the challenger to use negativecampaign ads. If he still decides to go with negative ads, rational appeals are morebeneficial for him. Besides deceasing potential for self-inflicted damage, rationalappeals would either make voters feel more unfavorable towards the targetedcandidate or would maintain voters’ political participation intent, depending on theincumbent’s reputation. The implications of these and other findings are discussed.