Negative Political Advertising and It Effect 

onVoters' Behavior

By Amanda Fischer

Relevant Links
Email me at


    How does negative political advertising affect voters' perceptions of candidates? There is literature to suggest that "attack politics" has an effect on how voters' view the candidate doing the attack and the candidate being attacked. The literature says that voters use the information and impressions they receive from negative advertising to make their final decision at the polls (Roese & Sande, 1993). There is also literature on the boomerang effect tate demonstrates that a candidate who uses insults in his or her campaign receives fewer votes while the target receives more (Jasperson & Fan, 2002). Fifty students (14 males, 36 females), ranging in ages from 18-22 were used as participants A Multivariate analysis of variance (MANOVA) [3(condition)X 2(candidate)] was performed on each individual trait. Significant differences in candidate perceptions between the control group and the two experimental groups were found for the characteristics of intelligence, likeability, credibility, charisma, and integrity. An index for all of the personality traits of each candidate was calculated and once again, a significant difference was found between the control condition and the experimental conditions in which the conservative candidate attacked the liberal candidate. One final MANOVA was done on the index of each candidate's job related characteristics. No significant difference was found, however a trend towards significance was found with those in the experimental condition, finding the liberal candidate to be more overall competent. These results supported the hypothesis that negative political ads would effect how voters view candidates. The second part of the hypothesis stated that when presented with negative political advertising, voters would tend to vote for the target of the attack. Results did not show this. However, it was found that voters were more decisive when there were not negative advertising, which shows implications for future research.

Back to the top


    When looking at how advertising affects voter habits, several factors must be examined. How voters form an impression, what form of communication they use to attain the information, and the techniques candidates use to persuade voters all contribute to an individual making a decision as he or she comes to the polls. These factors also play a role in how voters behave and help to understand their habits when negative advertising of candidates serves as a stimulus.
    In this paper negative advertising is also referred to as attack politics and negative campaign advertising. Attack politics refers to campaigns that degrade and try to discredit opponents. The working definition of negative political advertising will be any negative campaign advertising attempting to discredit the opponent. Negative political advertising is a source of considerable interest. Political attacks in the past have been so severe, that many psychologists and political analysis have dedicated a good deal of time researching this topic.
    To understand the effect of negative advertising, one must first look at how voters form their opinions of candidates in general. Voters form impressions of where candidates stand on specific issues. To relieve cognitive inconsistencies, they adjust candidates’ positions to make those candidates that they like, agree with them and candidates they dislike, disagree with them (Kenney, 1993). If this is true, then voters may vote more for candidates they like personally than candidates with whom they share common opinions. What is interesting is they may or may not realize it. How voters form these personal impressions of candidates is a crucial aspect of how they make their decisions. This raises the question of whether it is more important for a candidate to promote his or her image or promote his or her position on important issues.
     The Yale Attitude Change Approach is also an integral part of campaign advertising. First studied by Hovland, Janis, & Kelley (1953), it focuses on "who said what to whom," using the factors of who the communicator is, what he or she says, and who the audience is.
    Another more recent model of persuasion is the elaboration likelihood model (ELM) researched by Petty & Cacioppo (1986). It uses two routes to understand attitude change. The first of these is the central route, which takes place when people pay close attention to the arguments. The second is the peripheral route, which occurs when people are not paying close attention to the arguments, and are easily persuaded by surface characteristics such as characteristics of the speaker. The ELM is an integral part to understanding the effects of campaign advertising.
    The ELM and Yale Attitude Change Approach both play a role in a study done by Johnston & Kaid (2002). They looked at the types of advertisements that are used by politicians during election campaigns. They grouped them into the two categories of "image ads" which were intended to humanize the candidate for the voters, and "issues ads" which intended to show where a candidate stood on specific key issues. After studying 1,213 television ads from 13 US presidential elections, they found that in issue advertisements candidates spoke for themselves and in image ads there was an anonymous announcer speaking favorably about the candidate. Here the Yale Attitude Change Approach and the ELM are utilized. The Yale Attitude Change Approach is used in both instances. When the candidate speaks for him or herself in issue ads he or she is saying what he or she personally believes to the voters. When the candidate is spoken for in the image ads rather than just by him or herself, it conveys that someone else believes he or she is a good person and/or candidate so the audience should as well. The use of the different narrator (the “who said it” part of the Yale Attitude Change Approach) combined with the two different types of messages (the “said what” part of the Yale Attitude Change Approach), is what makes these commercials effective. Had a candidate boasted about what a great man or woman he or she is, the voter would not be impacted nearly as much, passing the advertisement off as merely a scheme for the candidate to stroke his or her own ego. Had an unknown narrator discussed what the candidate’s issues were, this may have instilled doubt in the voter who may come to the conclusion that those views on the issues stated were dictated to the candidate rather than decided upon by him or herself. The ELM can also used to explain what is happening. By having someone else proclaim how good a candidate is, the peripheral route may take place by making it seem as though he or she is a good candidate. The narration is a way of marketing the candidate in these instances just as a spokesman marketing a breakfast cereal.  If the voter is taken in by the strong voice and tone of the narrator, whether it be the candidate or an unknown, the voter is making an impression on the glamour of the commercial rather than by facts. Had the voter been persuaded by facts, he or she would have formed an opinion by means of the central route. The ELM is also important when creating negative political ad. The factor of the narrator may be taken into account in order to create the right quality of ad. By having an anonymous narrator attack the opponent, it could shield the candidate doing the attacking from any backlash effect.
    How candidates use their image in advertising is still an interesting factor to look at. To show voters an important aspect of themselves, politicians often use topics dealing with rights and morals. Domke, Shah, & Wackman (2000) found that emphasis by politicians and news media on rights and morals in political debates do have an influence on citizens’ evaluations of issues and candidates in political campaigning. They also found that citizens tend to make attributions about candidates’ morality in terms of issues that are ethically charged. For example, abortion and gun control are morally and ethically charged while health care, though still an important issue, does not bring about the same intensity of a moral or ethical questioning from voters (Domke, Shah & Wackman, 2000). This concept of deciding what is ethical is important because citizens also tend to make distinctions between candidates on the impression of their integrity (Domke, Shah & Wackman, 2000). Furthermore, when such issues are discussed within a context of what is moral and right, rather than in an ambiguous context, individuals are more likely to interpret that issue in an ethical manner. (Domke, Shah, & Wackman, 2000) This type of advertising can inform voters on both what to think and how to think. Also, when issues are presented in a context of what is moral and right, individuals are more likely to make morality attributions about the candidates. This information can be critical in predicting how a voter will make his or her decision which shows how important advertising campaigns can be. Negative political advertising often focuses on issues surrounding morality (Domke, Shah, and Wackman, 2000).
    Negative political advertising can effect voters. Voters may use the information and impressions they receive from negative advertising to make their final decision at the polls (Roese & Sande, 1993). Voters also feel negative political campaigning is less useful than both comparative advertising (advertisements that compare two candidates) and non-negative advertising (advertising that does not incorporate attack politics) (Roese & Sande, 1993). In short, the more negative the advertising, the less useful voters find it. They also view negative advertisement as hard to trust, and the more negative a campaign is, the more negativity a viewer has towards it (Pinkleton, Um, & Austin, 2002).
    If a voter is turned off by a campaign, he or she will like the candidate less and therefore be less likely to vote for him or her. Negative advertising are interesting when examining how a voter feels about campaigning while it is taking place. Usually a voter  views attack politics negatively (Schlenker & Britt, 1999). This is not surprising, as people are likely to engage in impression management to make themselves look as good as they possibly can in a situation when they want to impress a large amount of people (Schlenker & Britt, 1999). In this situation, to say that they enjoy watching conflict and seeing what is typically dubbed as petty tactics is to make themselves seem immature and mean. People who claim to be unaffected and impervious to such advertisements may very well in fact behave and vote to the contrary. The bottom line is that no matter how much a voter may try to give the impression that he or she wants nothing to do with attack political campaigning, he or she may very well participate in behavior that is affected by it (Meirick, 2002).
    Attack Politics can also affect voter turnout. There are three different views on this effect. The first by Ansolabehere, Iyengar, Simon, & Valentino (1994) suggests that negative political advertising is a deterrent for voters to come to the polls and demobilizes the electorate. In contrast Goldstein & Freedman (2002) argue that attack political campaigning stimulates voter turnout if anything, sparking interest in the campaign. Lau & Pomper’s (2001) data say that these two ideas meet half way, that whether negative political advertising affects voter turnout depends on the level of negativity and aggression used.
    Roese & Sande’s (1993) studied the effects of negative political campaigning. Their results demonstrated that a candidate who used insults in his campaigning received fewer votes while the target received more.
    The second study done by Jasperson & Fan (2002) concerned the 1996 U.S. Senate race between Boschwitz and Wellstone. They also found a boomerang effect on the attacker. Jasperson & Fan (2002) point out a well known persuasion principle in that the more trustworthy or credible a source is, the more persuasive the message the source is trying to explain.
    Finally a study done by Kaid (1997) showed much different results from the previous two studies. Kaid cited the 1996 Presidential race to be the most negative presidential campaign in the history of the presidency. The topic of technologically manipulating television ads is specifically discussed (Kaid, 1997). These distorted ads were used in 84% of Clinton’s ads and 49% of Dole’s ads. Kaid found that most of these distortions occurred in negative ads as reported by participants. The vote likelihood increased significantly when using distorted ads as opposed to undistorted. The results show that “the most costly, most negative, and most technologically misleading political campaign in history" (p. 1090) succeeded in getting enough votes to be in office. Her point here is that despite using negative advertising, Clinton was still reelected into office, so negative advertising did not hurt him.
       To examine some of these effects in more detail, this study was designed to address the effects negative political advertisements have on voters’ attitudes towards candidates and if it effects voter decision. According to the research, it is hypothesized that attack politics should have an effect on voters’ attitudes towards both the candidate doing the attack and the target. The candidate doing the attacking will most likely be viewed more unfavorably than the candidate being attacked. It is also hypothesized that the variable of voter decision will yield results of participants voting for the candidate being attacked rather than the candidate doing the attacking.

Back to the top



    Participants were students taking an introductory course in psychology for course credit. They were from a small, Catholic, Liberal Arts college in New England. Fifty participants ranging in ages 18 to 22 were used. Fourteen were male while thirty-four were female.


     A standard informed consent form which can be found in Appendix A, was used.  Participants were given a questionnaire to assess their political background. It is titled the “Political Background Questionnaire”. It asks questions about voting history, political ideology, and means of acquiring political information.
Four political advertisements were used as stimuli. The advertisements were created using information attained from the websites and pamphlets of the Republican and Democratic candidates from the 2004 Senatorial race in Pennsylvania and were made to look like advertisements that could be found in a newspaper. All ads centered around the topic of education. The first advertisement was for the democratic candidate, Joe Hoeffel who is the challenger. It is a neutral ad in which he strictly discusses his views on the topic. A second ad for Hoeffel was made in which he attacked Arlen Specter, the republican incumbent, by comparing records. A third ad was a neutral ad in which Specter’s views on the topic were laid out. A fourth ad was made to be a negative ad in which Specter attacks Hoeffel through comparison. In short each candidate has two ads: one neutral ad in which he address his views on education and one negative ad in which he attacks the other candidate through comparison. Each candidate’s neutral ad was created to look similar to their opponent as were each candidate’s negative ad. A post-stimuli questionnaire to assess any impressions they may have made due to the advertisements was also used. A series of Likert scales were used to assess impressions made in characteristics, for example, competency and leadership abilities. Five of these were considered to be personality characteristics and seven were considered to job related characteristics. This questionnaire also asked about questions regarding the participants impressions of the advertisements in general, which candidate they like more, and for whom they would vote. The participants were also given a post data questionnaire in which a manipulation check on the independent variable is assessed. It used a Likert scale to assess just how negative participants found the advertisement to be.

    Upon arrival, participants were given an informed consent to sign. They were told that this study had to do with thoughts on the senatorial race in Pennsylvania. Once each participant had finished reading and signing the informed consent form, the Political Background Questionnaire to assess political ideology was distributed and participants were instructed that all the questions were straight forward, but should they have a question to ask. They were also instructed to flip the questionnaire over upon completion and wait patiently for everyone else to complete it. Once all participants finished, they were each presented with a two of the ads depending on which condition they were in. The control condition showed both candidates’ neutral ads. The two experimental conditions are counterbalanced so that one ad was one of the candidates’ neutral ad and the other ad was the other candidates’ negative ad and reversed in the other experimental condition. The participants were placed into their condition based on the time for which they signed up. They had five minutes to read through the two advertisements, and then they were collected. They were then given a questionnaire to assess any impressions that may have formed by viewing the stimuli. Finally, they were debriefed, given a chance to ask any questions, given their credit and thanked for their participation.

Back to the top


    Three Multivariate analysis of variances (MANOVA) [3(condition- liberal attacks on conservative; both neutral; conservative attacks on liberal)X 2(candidate)] were conducted on the mean of the five traits having to do with personality characteristics (attraction, likeability, trustworthiness, open-mindedness, and charisma); the mean of the six traits having to do with job related characteristics (competency, intelligence, leadership, verbal ability credibility, and integrity); and the mean of all eleven characteristics. Internal consistency was tested for the personality characteristics, the job characteristics, and for all the characteristics. The alpha value for both candidates personality characteristics was alpha=.8387. The alpha value for both candidate’s job traits was alpha=.9011. The alpha value for all eleven traits for both candidates was alpha=.9358. These three scores show that the internal consistency of these variables was very good. Then a series of MANOVAs  were conducted to assess the influence of each of the eleven individual traits, making up the dependent variables  from the post-stimuli questionnaire. A Chi Square was performed for the dependent variable, for who the participants would vote.
    To conduct a manipulation check, a test of within-subjects contrasts measure was used. The manipulation check showed that ads meant to be negative were perceived as more negative than neutral ads and ads meant to be neutral were perceived to be less negative than negative ads (F(1,47)=7.63, p=.008).
    The MANOVA performed on the index for the personality traits of each candidate revealed that a significant effect was found for conditions (F(2,46)=2.08, p=.04). Post Hoc Tukey tests revealed that the main effects were found between the control condition and the experimental conditions. Another MANOVA was done on the index of each candidate's job related characteristics. No significant effect was found for condition, however a trend towards significance was found (F(1,43)=3.51, p=.06). Consistently, all of the means of the condition in which Specter attacked Hoeffel were higher than those of the condition in which both candidates presented neutral advertisements.
    A MANOVA was performed on each individual trait. Significant differences in candidate perceptions between the control group and the two experimental groups were found for the characteristics of intelligence (F(2,47)=3.75, p=.03), likeability (F(2,47)=6.96, p=.02), credibility (F(2,47)=8.77p=.03), charisma (F(1,4)=9.55, p=.02), and integrity (F(2,47=7.04, p=.04). Part of the hypothesis that participants would create impressions of each candidate due to the negativity of the advertisement was supported. The hypothesis also stated that participants would rate the candidate more negatively when they were doing the attacking than when they were not doing the attacking. This was also supported because the means show participants tended to rate the candidates higher when the candidate is not attacking than when the candidates were doing the attacking.
    The Chi Square on voter decision showed that there is no significance. The hypothesis about this variable was that the candidate being attacked would be voted for due to distrust of the candidate doing the attacking. The hypothesis included that they would have more certainty in the attack conditions than they would in the neutral conditions. The frequencies, however, seem to show that voters had the most certainty as to who to vote for when they were presented with both candidate’s neutral ads. This does not support the hypothesis that the candidate being attacked would be more likely to receive more votes than a candidate either not being attacked or doing the attacking.

Back to the top


    When presented with the negative advertisements, participants were hypothesized to view the candidates in the attacking conditions differently than how they viewed candidates in the neutral condition. The results of this study confirmed this hypothesis. Pinkleton, Um, & Austin (2002) found that voters find negative advertising hard to trust, and the more negative a campaign is, the more negatively they will see the candidate. In this study, it was expected that voters would view the candidates doing the attacking more negatively and participants would rate them accordingly in the post stimuli questionnaire, and that is what happened. Participants as a whole rated the candidate doing the attacking more negatively than the candidate being attacked.
    A second hypothesis was that the candidate being attacked would be more likely to receive votes than the candidate doing the attacking. This hypothesis was not confirmed and, in fact just the opposite effect was found. Participants tended to have the most certainty about their decision when presented with neutral ads from both of the candidates. Masterson & Biggers (1986) showed that when confronted with attack political advertising, voters would be less likely to vote for the attacker. Roese & Sande (1993) found that a candidate who uses insults in his or her campaigning receive fewer votes while that target receives more. Jasperson & Fan (1993) also suggested that voters would tend to vote for the candidate being attacked due to the boomerang effect. These three studies suggest that participants who viewed the negative advertising conditions would have been more likely to vote for the candidate being attacked over the candidate doing the attacking. The results of this study, however, showed that participants had the most certainty as for whom to vote in the neutral condition than either of the negative conditions. This may be due to the demobilization effect which states that negative political advertising discourages voters, especially Independents from participating in elections (Ansolabehere, Iyengar, Simon, & Valentino, 1994). Participants may have been so turned off by the negativity of the advertisements, that they chose not to make a decision as a voter in a real life election would choose to not vote.
    Implications of this study can suggests applications to real life political advertising campaigns. This study showed that negative political advertising does effect voter behavior and did so using a small sample of participants. This in itself shows that more research may and probably will yield more knowledge about the impact of this ever growing type of advertising on voters. The past few presidential campaigns have incorporated attack politics in their advertising and the last two elections have been extremely close. If attack politics continue to show an effect on voters and if the precise ways in which this effect works can be found, it is within the realm of possibility that it may change the outcome of an election. The literature says that voters are more likely to vote for candidates being attacked and less likely to vote for candidates who do the attacking (Masterson & Briggers, 1986; Roese & Sande, 1993; Jasperson & Fan, 2002).
    Future research on this topic should examine the impact of attack politics on voter turnout. Due to the time frame and logistics of this project, voter turnout was not possible to study. The most that could have been done in this area was to ask the participants if they would be likely to vote in this election. The researcher decided not to include this variable in this project due to the weak external validity of the study. The researcher would want to measure if a participant voted and not if the participant says he or she would vote. An interesting future study would be if researchers conducted a longitudinal design and monitored voters’ attitudes and opinions from the very beginning of a real life negative political campaign through the election and the final outcome. Then they would know for sure if attack political advertising would affect voter turnout. Voter turnout in the United States would be an interesting variable even if attack politics were not involved. Though it has gone up due to the perceived importance of the most recent election, voter turnout is still very poor in a country whose very basis is on freedom for all.           Future studies could also examine approval ratings of candidates who won political office and see if there is a difference between a candidate who won a campaign using attacking politics and a candidate who did not.
    The methodology of this study was not perfect. Participants were taken from a pool that was somewhat scarce. This pool also held very little diversity as far as ethnic background and all participants were from the same region of the country. Also, this study was conducted in five sessions and used three different rooms. Two of the rooms were set up like a standard classroom using desks in rows. One of the rooms was set up with a conference table and crowded the ten participants more so than in the two classrooms. This may have presented a confounding variable in the presentation of the stimuli. Due to the overcrowding in this one session, participants may have become agitated more so than in the other conditions which could have affected their ratings.
    Also when the researcher was creating the printed advertisements, an original template of one of the candidates was used and was the groundwork for constructing the other candidate’s ad. When participants were asked to rate how well done they found each ad to be, Hoeffel’s advertisements, the ones that were found as an original template, were consistently found to be more well done than the other in all three conditions. This may have created a confounding variable in that viewing an advertisement that was done better facilitated information to be processed better and may have been the cause of the decision instead of the negativity of the ad. Using this study as a pilot study and replicating it in future research without the confounding variables would probably yield interesting results.

Back to the top


Ansolabehere, S., Iyengar, S., Simon, A., & Valentino, N. (1994). Does attack advertising
    demobilize the electorate? American Political Science Review, 88(4), 829-838.
Brookfield, S. (1986). Media power and the development of media literacy: an adult educational
    interpretation. Harvard Educational Review, 56(2), 151-170.
Domke, D., Shah, D. V., & Wackerman, D. B. (2000). Rights and morals, issues, and candidate
    integrity: insights into the role of the news media. Political Psychology 21(4), 441-665.
Goldstein, K., & Freedman, P. (2002). Campaign advertising and voter turnout: new evidence for
    a stimulation effect. The Journal of Politics, 64(3) 721-740.
Hovland, C.I., Janis, I.J. & Kelley, H.H. (1953) Communication and persuasion: Psychological
    studies of opinion change. New Haven, Ct: Yale University.
Jasperson, A. E. & Fan, D. P. (2002). An aggregate examination of the backlash effect in
    political advertising: the case of the 1996 U.S. Senate race in Minnesota. Journal of Advertising, 31(1), 1-12.
Johnston, A., & Kaid, L. L. (2002) Image Ads and Issue Ads in U.S. Presidential Advertising:
    Using Video style to Explore Stylistic Differences in Televised Political Ads from 1952 to 2000. Journal of
    Communication, 52(2), 281-300.
Kaid, L. L. (1997), Effects of the Television Spots on Images of Dole and Clinton. American
    Behavioral Scientist, 40(4), 1085-1094.
Kenney, P. J. (1993). An examination of how voters form impressions of candidates’ issue
    positions during the nomination campaign. Political Behavior, 15(3), 265-288.
Lau, R. R., & Pomper, G. M. (2001). Effects of negative campaigning on turnout in U.S. Senate
    Elections, 1988-1998. Journal of Politics, 63(3), 804-820.
Masterson, J. T. & Biggers, T. (1986). Emotion-eliciting qualities of television campaign
    advertising as a predictor of voting behavior. Psychology, A Quarterly Journal of Human Behavior, 23(1), 13-19.
Meirick, P. (2002). Cognitive responses to negative and comparative political advertising.
    Journal of Advertising, 31(1), 49-62.
Petty, R. E., & Cacioppo, J.T. (1986). Communication and persuasion: Central and peripheral
    routes to attitude change. New York: Springer-Verlag.
Pfau, M., Cho, J., & Chong, K. (2001). Communication forms in U.S. Presidential Campaigns
    influence on candidate perceptions and the democratic process. Harvard International Journal of Press and Politics,
    6(4), 88-105.
Pinkleton, B. E., Um, N.H., & Austin, E. W. (2002) An exploration of the effects of negative
    political decision making. Journal of Advertising, 31(1), 13-25.
Richardson, G. W. (2001). Looking for meaning in all the wrong places: why negative
    advertising is a suspect category. Journal of Communication, 51(4), 775-800.
Roese, N. J. & Sande, G. N. (1993). Backlash effects in attack politics. Journal of Applied Social
    Psychology, 23(8), 632-653.
Schlenker, B. R., & Britt, T. W. (1999). Beneficial impression management: Strategically
    controlling information to help friends. Journal of Personality and Social Psychology, 76, 559-573.
Whitney, D. C., & Goldman, S. B. (1985). Media use and time of vote decision: a study of the
    1980 Presidential election. Communication Research 12(4), 511-529.
Zhao, X., & C., Steven H. (1995). Campaign advertisements versus television news as sources of
    political issue information. Public Opinion Quarterly, 59, 41-65.

Back to the top

Key Words: political advertising, negative advertising, social psychology, political psychology

Relevant Links

Effectiveness of Negative Political Advertising
Public Perceptions of Negative Political Campaigns
The :30 Second Candidate
Eastern Psychological Association
Social Psychology Network

Back to the top