Sie sind auf Seite 1von 19

ARTICLE

10.1177/1081180X03253508

Press/Politics
Gordon
et al.8(3)
/ Negative
SummerAdvertising
2003
and Female Candidates

Is Negative Advertising
Effective for Female Candidates?

An Experiment in Voters
Uses of Gender Stereotypes
Ann Gordon, David M. Shafie, and Ann N. Crigler

Female candidates sometimes are discouraged from negative campaigning because


they would risk voter backlash by defying gender stereotypes. In this experiment, a
negative television campaign advertisement was manipulated to control for the candidates gender, issue content, and the character traits emphasized in the narration. The
female candidates support was not diminished because of the attack upon her opponent. The findings show that in low-information contests, female candidates are
assumed to be weaker on male issues. However, when the female candidate used a
male issue as the basis of an attack upon her male opponent, she was judged more
competent on the issue. This suggests that negative advertising could be an effective
tool to neutralize the disadvantages caused by gender stereotypes.
Keywords: gender; stereotypes; negative advertising; political campaign; women and
politics

In the year 2003, the number of women in the U.S. Congress reached seventythree. Fourteen of them were in the Senate and fifty-nine were in the House of
Representatives. While these numbers reflect significant gains in recent years,
women still are underrepresented in Congress as well as in other elective offices.
This lack of parity has been attributed to incumbency advantage (Burrell 1994)
and other structural barriers such as the eligibility pool and the electoral system
itself (Darcy et al. 1994). Another impediment to womens electoral success is
believed to be the role of gender stereotypes in voter decision making (Sapiro
1981-82; Huddy and Terkildsen 1993b; Leeper 1991; Kahn 1996). Even as
women run for office and win elections in greater numbers, the nature of
Press/Politics 8(3):35-53
DOI: 10.1177/1081180X03253508
2003 by the President and the Fellows of Harvard College

35

36

Press/Politics 8(3) Summer 2003

political campaigns is undergoing a transformation that, according to conventional wisdom, may put women candidates at a disadvantage, threatening to offset some of their electoral gains.
Political campaigns have become increasingly negative (Lau and Sigelman
2000). Moreover, there is a belief by consultants and other practitioners that
attack advertising is highly effective, while scholars have noted that negative ads
are more easily recalled by voters (Kahn and Kenney 2000). Though negative
political advertising is on the rise overall, campaign consultants have advised
women that negative advertising may damage their chances of electoral success.
A guidebook titled 101 Campaign Tips for Women Candidates and Their Staffs
warns that female candidates who use negative campaign tactics are especially
susceptible to voter backlash (Lansing 1991). Such warnings come from the
experience of campaign consultants and candidates who believe they lost elections because the public was not ready to accept negative campaigning from
women because of gender stereotypes (Trent and Friedenberg 1995). Some
research supports the view that gender stereotypes can work against women
(Leeper 1991; Fox and Smith 1998), while other research has demonstrated that
gender stereotypes actually work to the advantage of female candidates (Kaid et
al. 1984) or at least do not hurt them (Thompson and Steckenrider 1997).
These mixed results may indicate that gender stereotypes are activated in
some circumstances and not in others. For example, a female candidate may risk
activating stereotypes through negative advertising. If attacking ones opponent
is seen as violating stereotypical expectations of womens behavior, then stereotypes of appropriate behavior for women may be activated when a female candidate defies those behavioral norms. Thus, the public may rely heavily on gender
stereotypes when evaluating a female candidate who goes on the attack. Moreover, according to Eagly and Karau (2002:573), the incompatibility of beliefs
about gender roles and leadership roles leads to a prejudice against women leaders, who are viewed less favorably than their male counterparts, especially in
situations that heighten perceptions of incongruity between the female gender
role and leadership roles.
Despite the potential for a voter backlash, women have been using negative
ads. Female candidates tend to be challengers and, as such, tend to include negative advertisements in their strategy (Procter et al. 1994; Williams 1998). In
recent years, female candidates have been more willing to use attack advertisements than male candidates in U.S. Senate races (Kahn 1996) and in gubernatorial races (Procter et al. 1994). Other research has found that women use negative ads at the same rates as men, suggesting that much of the uncertainty about
womens campaign strategies may be due to the sampling used in earlier studies
or differences in what is considered negative (Williams 1998). While there is
disagreement over the frequency of negative advertising by women and men,it is
clear that women do employ negative advertisements.

Gordon et al. / Negative Advertising and Female Candidates

37

If negative advertising is an integral part of campaign strategy, are women


improving their chances for electoral success by turning to negative advertisements? Or,as the conventional wisdom warns,are women more likely to be punished because of the interaction between gender stereotypes and a negative campaign strategy? Further, can women overcome the potentially damaging effects
of stereotypes by making strategic choices to emphasize particular issues and
character traits in their negative advertisements? This study reports the results of
an experiment in which participants viewed negative advertisements that systematically varied candidate gender, issues, and character traits.
Gender Stereotypes and Campaigns

Sex stereotypes are a structured set of beliefs about personal attributes of


women and men (Ashmore and Del Boca 1979:222). Individuals rely upon such
stereotypes to simplify the world around them. Men, for example, are thought
to be tough,aggressive,and independent,while women are seen as honest,sensitive, and compassionate (Huddy and Terkildsen 1993b; Kahn 1996).
Political scientists have recognized that stereotyping may have an important
impact on the electoral process (e.g., Sapiro 1981-82; Huddy and Terkildsen
1993b; Kahn 1996; Iyengar et al. 1997). One effect of stereotyping is voters
inferences about candidates on the basis of gender. In a seminal study, respondents were more likely to infer that a female candidate was more competent to
improve the educational system and health care and was thought to be better able
to maintain honesty and integrity in government, while a male candidate was
considered more competent to deal with military issues and farm issues (Sapiro
1981-82). Subsequent experimental research has found that when a candidate
stresses stereotypically masculine traits, respondents are more likely to infer
other masculine traits, regardless of the candidates gender (Huddy and
Terkildsen 1993b). Further, male and female candidates were thought better
able to deal with military issues if they emphasized masculine traits such as
ambition and toughness rather than feminine traits such as compassion. Similarly, Leeper (1991) found that when respondents read a masculine policy
speech attributed to either a male or female candidate, the female candidate was
still thought to possess feminine traits.
Many studies of the effects of gender stereotypes rely on written descriptions
of candidates or speeches as the experimental stimulus (e.g., Sapiro 1981-82;
Huddy and Terkildsen 1993b; Fox and Smith 1998). A few have used videotape
campaign advertisements as the stimulus (Kaid et al. 1984; Ansolabehere and
Iyengar 1995; Iyengar et al. 1997). One notable study involved a set of experimentsone that used a single advertisement that varied by issue and gender of
the candidateand a second that paired advertisements that focused on male
or female issues (Iyengar et al. 1997). The results suggested that women had an
advantage when they campaigned on womens issues. These experiments had a

38

Press/Politics 8(3) Summer 2003

great deal of external validity because the advertisements featured well-known


candidates and the experiments were conducted during actual campaigns. However, conducting the experiments in the context of actual campaigns made it
more difficult to control for the impressions of the selected candidates that people had already formed (Iyengar et al. 1997).
While gender stereotypes may provide the basis for inferences about candidates, there are a variety of other important cues for voters. One potential cue is
ideology. While Kinder (1983) argued that by and large, Americans were innocent of ideology,other studies have shown that voters use ideology as a heuristic.
Brady and Sniderman (1985) demonstrated that even if voters do not understand
what liberalism and conservatism mean in the abstract, they can still make fairly
reliable inferences based on a kind of ideological constraint. If a voter knows
where a candidate or group stands on a single issue, the voter can make a judgment about other issues as well. Likewise, Conover (1981) argues that ideological cues are important in inferences about traits. Voters use both issues and character traits when they evaluate candidates and decide whom to support (Popkin
1991). These are especially important in the absence of partisan cues and may
interact with gender stereotypes to produce an overall assessment of the
candidate.
Data and Method

This study presents participants with television spot advertisements, thus


enabling an investigation of the relative success of certain campaign strategies for
male and female candidates. Each condition of the experiment features a manipulated negative advertisement. The major expectation is that a negative advertisement is more likely to provoke a backlash if it features a female candidate than
a similar ad featuring a male candidate. Because an aggressive and negative strategy runs counter to gender stereotypes, the negative ad is expected to be less
effective for the female candidate.It is possible to detect a backlash by comparing
participants responses to the female and male versions of the ads. A comparison
of the number of comments about the tone of the ads should reveal differences in
the way the negative ad is received by participants.
A second theoretical expectation is that female candidates may be disadvantaged by the issues of the campaign. Because of gender stereotypes, male candidates are assumed to be stronger in certain issue areas than female candidates.
Thus, respondents are likely to infer that the male candidate is more competent
to deal with male issues, while the female candidate is more likely to be judged
competent to handle female issues. Further, candidates with gender-consistent
messages (female candidate, female issue, feminine traits) may be judged
more competent and be more popular than candidates with messages that contradict prevailing stereotypes. The impact of these stereotypes may be

Gordon et al. / Negative Advertising and Female Candidates

39

heightened in the context of a negative campaign because a candidates decision


to attack her opponent is likely to be perceived as aggressive, an act that would
also appear inconsistent with stereotypical gender norms.
A final expectation is that people will use gender stereotypes to make inferences about the candidates issue positions, even those not mentioned in the candidates own messages. For example, a female candidate may be perceived as
more competent on education than a male candidate, even if education is not
addressed in the campaign. Candidates may be able to use these stereotypes to
their advantage, encouraging voters to draw these inferences by providing such
cues as the issues they choose to address and the character traits they emphasize
in their messages. Regardless of gender, a candidate who emphasizes protecting
the environment (a female issue) may also be thought of as competent in other
stereotypical female issue areas. Moreover, a candidate whose ad mentions the
candidates compassion is thus more likely to be perceived as proenvironment
and less likely to be perceived as antitaxes than a candidate whose ad stresses such
masculine traits as strength or independence.
The Experimental Design

The experiment featured a two-by-two-by-two factorial design, which is displayed in Table 1. Participants were assigned randomly to eight groups, and each
group viewed a video presentation consisting of a manipulated campaign advertisement and news programming. Each presentation featured a thirty-second
negative ad with a unique combination of the three treatment variables: candidate gender, candidate trait, and issue content.1 The participants were 169 students from four undergraduate classes in Southern California.2 Between groups,
there were no statistically significant differences on demographic variables (gender, ethnicity, religion, ideology, and party identification).
The campaign advertisements were nested between two distracter news
features. The first feature was a CNN health news package about new snack
foods fortified with herbal supplements. Appearing after the advertisement was
a light news feature from a local television station, offering the viewer advice on
avoiding speeding tickets. The manipulated ad appeared in between the two
news features. A television news producer reedited a campaign advertisement to
create the eight versions. New scripts were written to manipulate the name of
the candidate and the message in each version of the ad. For authentic sound, the
narration was digitally rerecorded,using the voice of a network news correspondent. The ads were carefully manipulated so that no evidence of tampering was
visible.3
The spots used in the stimulus were manufactured primarily from actual
attack ads that had been used in other states. To allow for the interaction between
gender of the candidate and gender stereotypes about issues, we examined two
gender-linked political issues: the environment, traditionally seen as a female

40

Press/Politics 8(3) Summer 2003

Table 1

Design of the experiment


Condition 1
Condition 2
Condition 3
Condition 4
Condition 5
Condition 6
Condition 7
Condition 8

Female candidate
Female candidate
Female candidate
Female candidate
Male candidate
Male candidate
Male candidate
Male candidate

Feminine character trait


Masculine character trait
Feminine character trait
Masculine character trait
Feminine character trait
Masculine character trait
Feminine character trait
Masculine character trait

Environment spot
Environment spot
Taxes spot
Taxes spot
Environment spot
Environment spot
Taxes spot
Taxes spot

issue, and taxes, a traditional male domain (Kahn 1996). The narrator of the
environment spot attacks an unseen opponent for his record on the
environment:
Ever notice how some politicians change their colors in the fall? Michael Bernhart
now claims hell protect our environment. But hes voted against legislation protecting rivers and streams. Against a tough water quality act. And against
strengthening Act 250, our most important environmental law. So, has he turned
over a new leaf? Or is this just a seasonal change?

The other spot, attacking an opponent on taxes, was originally used in an Idaho
campaign. In the edited versions, the narration and on-screen text were changed
to correspond to the new candidate:
Here in Idaho, we pay tax after tax: federal income taxes, state income taxes, sale
taxes, property taxes. And now candidate for Senate Mel Richardson has come up
with still another tax. A national sales tax. Its all in black and white in his campaign brochure.Hes proposing a national sales tax that experts say would hurt the
economy. Dont take that risk.

In an additional manipulation, there are two variations on the ads endings. Half
of the ads scripts emphasized stereotypical positive masculine traits about the
candidate, such as strong leadership and independence, and the other half
emphasized positive feminine traits, such as compassion and trust. Viewers are
asked to vote for a compassionate Senator you can trust in the feminine version
or strong, independent leadership in the masculine version. To establish the
gender of the sponsoring candidate, a picture of either Jeff or Louise accompanies the slogans. To ensure an authentic look, footage from actual congressional campaign ads was inserted into the manipulated spots. The footage was chosen because of the similarities in the video footage of the two candidates. Both
Jeff and Louise are seen photographed from the same angle, dressed in business
attire and seated behind a desk. Because candidates rarely mention party in their

Gordon et al. / Negative Advertising and Female Candidates

41

advertisements (Just et al. 1996; Vavreck 2001), no information regarding the


party of the candidate or the opponent was included. Visually, the modified ads
were identical to each other and to the original ads, except for the inserted footage of either a male candidate or a female candidate at the end.
After viewing the videotape, participants were asked to complete a brief
questionnaire, which included open- and closed-ended questions about what
they had just viewed. The open-ended questions asked the participants to
describe what they had seen in general, how they would describe the ad, and
finally what their impressions of the candidate were.The closed-ended questions
asked participants to rate the candidate on a feeling thermometer, to assess how
well various traits and issue positions described the candidate, and to estimate
the candidates partisan and ideological affiliation. Finally, the questionnaire
asked for demographic information, party identification, and the ideology of the
participants.
Results and Discussion

A reading of the open-ended comments revealed that participants felt free to


discuss and make inferences about what they had seen. Although 14 percent of
the comments about Louise or Jeff indicated that participants did not have
enough information or did not form an impression of the candidates,the remaining 86 percent covered a wide range of topics.
The negative tone of the ads was one of the themes mentioned by participants
in open-ended comments, although not as frequently as other themes. Some of
the participants expressed displeasure with the fact that the advertisements were
negative, but they were not significantly more likely to punish Louise for the
negativity than Jeff. Table 2 lists the themes most frequently mentioned in the
open-ended comments. Of the eighty-nine respondents who saw the Louise
ads, sixteen made references to their negativity, while nine of the eighty participants who viewed Jeff ads commented on the ads negativity. The number of
respondents mentioning the negative tone was small in both conditions (18 percent compared to 11 percent),and the difference was not statistically significant.
Overall, participants were most likely to discuss three topics: the candidates
appearance, competence, and character. Fewer respondents volunteered comments about the issues raised in the ads and the candidates partisan or ideological leanings. As Table 2 indicates, participants spoke differently about the two
candidates. Sixty percent of participants who saw the Louise ad mentioned her
appearance, 22 percent mentioned her competence as a politician or candidate,
and 61 percent mentioned her character. In contrast, participants viewing the
Jeff ad concentrated primarily on his competence, while very few made mention of his appearance or character. Participants were significantly more likely to
mention his competence (45 percent) than those who saw the equivalent

42

Press/Politics 8(3) Summer 2003

Table 2

Volunteered comments, by theme and candidate gender


Female Candidate

Negativity
Appearance
Competence
Character/personality
Position on taxes
Position on environment
Party/ideology
Other topics
Total

Male Candidate

16
53
20
54
7
10
9
72
89

18
60
22
61
8
11
10
81

9
34
36
26
11
10
10
47
80

11
43*
45*
33*
14
13
13
59*

Note: Other topics include impressions of the candidates opponent, comments expressing no
impression of the candidate, and statements suggesting that the respondent did not have enough
information to form an impression. An asterisk indicates a pair with a significant difference.
*p < .05.

Louise ad. Compared to participants who viewed the Louise ad, they were less
likely to comment on his appearance (43 percent) or his character (33 percent).
The different rates of mentions between the male and female candidates in all
three of these areas were statistically significant.
Not only were participants who viewed the female candidates ad more likely
to mention her appearance, those comments were more likely to be negative. To
be sure, most of the comments about the candidates appearance were neutral.
As Table 3 indicates, twenty-four of the eighty-nine participants who saw the
female candidates ad made neutral comments and eight were positive, while
twenty-two of the eighty participants viewing the male candidate made neutral
comments about his appearance, and seven were positive. These included mentions of hair color, clothing, age, race, and gender. Despite the fact that candidates were pictured for less than ten seconds at the end of each ad, some very
strong opinions were expressed about their appearance. The female candidate
was more than three times more likely to receive negative comments about her
appearance as the male candidate. The difference in the rate of negative comments is striking: while five out of eighty participants made negative comments
about the male candidates appearance, twenty-one out of eighty-nine participants made negative comments about the female candidates appearance. Participants did not like her hair and thought she was dowdy and not very powerful
looking.
Comments about Jeffs appearance were far more neutral or positive, emphasizing what he was doing as well as his appearance. He is working at a desk late at

Gordon et al. / Negative Advertising and Female Candidates

43

Table 3

Tone of volunteered comments, by theme and candidate gender


Appearance
Tone of Comments

Competence

Character/
Personality

Female

Male

Female

Male

Female

Male

Positive
Neutral
Negative

8
24
21

7
22
5

7
3
10

11
3
22

34
3
17

13
1
12

6.14 (p < .05)

0.9 (n.s.)

4.48 (n.s.)

night (Participant 5003). He was clean-cut (Participant 5009) and young,


well-kempt and full of promises (Participant 5002). Jeff was even compared to
Al Gore: [He is] clean cut, All-American, Al Gore look alike, suit and tie in an
office of some sort, I believe sitting at a desk. [He] smiled, looked friendly, honest, trustworthy (Participant 7011).
The negative comments about Louises appearance do not translate into negative assessments of her competence or character. She was seen as hardworking . . . and concerned about saving taxpayers money (Participant 4024).
Another participant wrote,She appears friendly,and work-oriented,in her limited display in the ad. This imagery at the end of the ad reinforced her as being a
serious politician and not like her opponent (Participant 3003). In fact, participants made more than twice as many positive comments about Louises character than they did about Jeffs character.While the female candidate was criticized
for her appearance, the male candidate was criticized with regard to his competence. Twelve percent of the comments made about Jeff expressed concern
about his competence as a candidate. As one participant surmised, There are no
clear positions that he has taken. I didnt see why he would be competent (Participant 6022). He was called lackadaisical, unmotivated and unknowledgeable
(Participant 8019) and a typical politician working on a lot of nothing (Participant 6021). One participant concluded,
The guy seems weak-minded and liberal; seems like he was raised to be a politician. [He] probably has [a] strong family, but he is a follower of those who lead his
campaign. He probably does not know the slightest details about taxes, except
what those working for him tell him. (Subject 8012)

The male candidate received about the same number of negative comments
about his character as did the female candidate. However, he received far fewer
positive character comments than she did.

44

Press/Politics 8(3) Summer 2003

Do these comments about appearance, competence, and character matter to


participants evaluations of the candidates? Yes and no. As Table 4 shows, the participants thermometer ratings, the likelihood of voting for the candidate, and
assessment of the candidates chances of winning were significantly related to
participants comments about competence and character but not appearance.
For these experimental participants, the highly negative comments about Louises appearance did not produce significantly lower thermometer scores, voting
preferences, or assessments of winning than the more balanced comments about
Jeffs appearance. This is not the case for comments about the candidates competence. For both candidates, negative competence assessments significantly
lowered participants mean thermometer ratings, likelihood of voting for the
candidate, and assessments of the candidates overall chances of winning, while
positive assessments were associated with these measures. The same was true for
comments assessing the candidates character. Regardless of the gender of the
candidate, competence and character assessments played a significant part in
participants ratings of vote preference. This finding extends past research,
which has found that character and competence are key factors in voters assessments of presidential candidates (Just et al. 1996).
Responses to closed-ended questions reveal a similar pattern. Participants
judged the degree to which the candidate possessed typically masculine and
feminine traits and competencies in issue areas associated with gender.4 Analysis of variance (ANOVA) results, in which manipulated candidate gender, traits,
and issues were between-subject factors, revealed several candidate evaluations
that were significantly related to issue content. Participants relied most frequently on the issue content of the advertisement to make inferences about the
candidates positions on issues and character traits. However, they based other
inferences on candidate gender and, to a lesser extent, the character traits that
were emphasized in the ad.
Issue Effects

Of the three experimental stimuli, issue content of the ad yielded the largest
number of main effects. To determine the effects of the manipulated issue content, separate ANOVAs were conducted on several assessments of the candidate
(F-ratios are reported in the third column of Table 5). Controlling for the effects
of candidate gender and traits, the ads issue content was significantly related to
four inferred issue positions and one of four personality traits, as well as to ideology and party.
Participants made several inferences about the candidates on the basis of the
issue content of the advertisement they saw. Accordingly, those who saw the
environmentad thought that proenvironmentdescribed their candidate well.
The means reported in Table 6 provide an illustration. Participants gave their
respective candidates a mean rating of 3.06 on the proenvironment 4-point

Gordon et al. / Negative Advertising and Female Candidates

45

Table 4

Thermometer score, voting likelihood, and likelihood of winning, by tone of comments about
appearance, competence, and character
Thermometer
Score
Tone of Comments
Appearance
Positive
Neutral/ambivalent
Negative
Competence
Positive
Neutral/ambivalent
Negative
Character
Positive
Neutral/ambivalent
Negative

Voting
Likelihood

Likelihood of
Winning

Female

Male

Female

Male

Female

Male

45.4
41.4
38.6

57.5
43.1
37.5

2.2
2.2
1.9

2.6
2.1
2.5

2.6
2.5
2.4

2.6
2.5
2.8

59.2
40.1
37.5

54.2
42.6
42.9

3.2
2.1
2.1

2.3
2.2
2.1

3.0
2.5
2.3

3.0
2.5
2.5

53.0
40.0
27.8

58.3
42.9
37.8

2.7
2.1
1.8

2.3
2.2
2.1

2.9
2.5
2.0

3.0
2.5
2.3

Note:Entries are means. For each participant, the number of negative comments was subtracted
from the number of positive comments to arrive at a single tone of comments score. Positive
indicates more positive than negative comments; neutral/ambivalent indicates an equal number (possibly zero) of positive and negative comments; negative indicates more negative than
positive comments. Thermometer score is on a scale of 0 to 100; voting likelihood is on a scale
from 1 (not at all likely to vote for the candidate) to 4 (very likely to vote for the candidate); likelihood of
winning is on a scale from 1 (not at all likely that the candidate will win) to 4 (very likely that the candidate will win). No main effects or interactions are significant for any of the three dependent variables. The main effect of tone of comments regarding competence is significant for all three
dependent variables: thermometer, F(2, 157) = 5.7, p < .01; voting likelihood, F(2, 160) = 4.4,
p < .05; likelihood of winning, F(2, 160) = 3.4, p < .05. The main effect of tone of comments
regarding character is significant for all three dependent variables: thermometer, F(2, 157) =
14.9, p < .001; voting likelihood, F(2, 160) = 6.7, p < .01; likelihood of winning, F(2, 160) =
9.1, p < .001.

scale, which was higher than any other issue position. Similarly, participants who
saw the taxes ad gave their candidate a mean rating of 3.04 on the antitaxes
item, which was also higher than any other issue position.
More important, the issue positions were used to draw inferences about the
candidates views on issues not mentioned in the advertisements. Controlling
for gender and character traits, a candidate appearing in the environment ad
was more likely to be perceived as liberal and a Democrat and less likely to be
perceived as probusiness, promilitary, or patriotic than a candidate appearing in
the taxes ad. Consistent with the findings of Brady and Sniderman (1985),

46

Press/Politics 8(3) Summer 2003

Table 5

Impact of candidate gender, traits, and issue content of ads


Candidate Candidate
Gender
Traits
Issue positions
Probusiness (df = 1,153)
Proenvironment (df = 1,155)
Promilitary (df = 1,153)
Antitaxes (df = 1,156)
Personality traits
Patriotic (df = 1,154)
Caring (df = 1,154)
Trustworthy (df = 1,155)
Knowledgeable (df = 1,153)
Political variables
Ideology (df = 1,153)
Party ID (df = 1,135)

Ad Issue
Content

Gender
Traits

Gender
Issue

6.41**
0.07
5.86*
0.29

0.15
5.11*
0.27
3.11

26.88**
66.79**
8.89**
65.91**

4.15*
0.57
2.39
0.23

0.20
3.29
0.32
7.73**

0.20
0.61
0.01
1.08

2.99
5.58*
0.73
4.10*

5.25*
1.09
0.47
0.08

1.13
0.53
4.58*
7.27**

0.46
0.10
0.14
0.00

1.93
0.33

1.91
0.71

18.70**
9.69**

2.33
0.66

0.14
0.21

Note: None of the two-way interactions between traits and issues were significant, so they were
not discussed. No three-way interactions were significant.
*p < .05. **p < .01.

participants reached beyond the information presented to them to make inferences about the candidates political views.
Gender Effects

Participants were most likely to use the issues stressed in the ads as cues about
the candidates; however, there were significant main effects of gender on the
issue positions attributed to the candidates. Holding constant the effects of the
advertisements issue content and character traits, the male candidate was more
likely to be perceived as probusiness and promilitary. The perceived differences
between female and male candidates on these stereotypically male issues are
illustrated by comparing the means for each group, which are reported in Table
6. On the question, How well does probusiness describe the candidate? the
mean response for the female candidate was 1.89, significantly lower than the
2.18 mean rating of the male candidate. Similarly, on the corresponding item,
How well does promilitary describe the candidate? the mean response for the
female candidate was 1.55, compared to the 1.81 mean rating of the male candidate (see Table 6). Participants were more likely to attribute these
stereotypically male positions to the male candidate, even though no information regarding either candidates views in these areas was provided. Contrary to
expectations, participants were no more likely to attribute issue positions consistent with female stereotypes to Louise than to Jeff.

Gordon et al. / Negative Advertising and Female Candidates

47

Table 6

Mean candidate evaluations by candidate gender, traits, and issue content of ad


Candidate
Gender

Candidate Traits

Ad Issue Content

Female Male Feminine Masculine Environment Taxes


Issue positions
Probusiness (range = 1-4)
Proenvironment
(range = 1-4)
Promilitary (range = 1-4)
Antitaxes (range = 1-4)
Personality traits
Patriotic (range = 1-4)
Caring (range = 1-4)
Trustworthy (range = 1-4)
Knowledgeable
(range = 1-4)
Political variables
Ideology (range = 1-7)
Party ID (range = 0-1)

1.89 2.18**

2.00

2.06

1.67

2.38**

2.43 2.57
1.55 1.81*
2.49 2.32

2.32
1.70
2.28

2.66*
1.65
2.55

3.06
1.51
1.78

1.93**
1.84**
3.04**

2.19 2.24
2.29 2.18
2.10 2.10

2.08
2.08
2.04

2.35
2.40*
2.17

2.05
2.16
2.15

2.38*
2.32
2.06

2.15 2.29

2.08

2.35*

2.20

2.24

3.79 3.99
0.64 0.61

4.03
0.59

3.75
0.67

3.41
0.76

4.36**
0.51**

Note: For issue positions and personality traits, participants were asked how well each of the
descriptions applied to the candidate, where 1 = not at all well and 4 = very well. Participants also
were asked to judge the candidates ideology on a 7-point scale, where 1 = very liberal and 7 = very
conservative. Their estimate of the candidates party affiliation was coded 0 for Republican and 1
for Democrat. Entries are means. Each value marked with an asterisk is significantly different
from the value to its left, as tested by ANOVA. Comparable F-ratios are presented in Table 5.
*p < .05. **p < .01.

Candidate gender also appeared to matter when the campaign message defied
a gender stereotype. The female candidate, attacking her opponent as weak on
taxes, actually received a greater boost from the attack than the male candidate
using the same message (The F-ratio is presented in column 5, Table 5). In this
interaction, the combined effect of candidate gender and the issue content of the
ad was significantly related to the perception that a candidate was antitaxes.
The mean antitaxes score for the female candidate appearing in the environment ad was just 1.60, compared to 1.95 for the version of the ad featuring the
male candidate. In the absence of any information about their views on taxes, the
female candidate was less likely to be perceived as having an antitax position than
the male candidate. However, this gap closed when taxes were the subject of the
ad. Both the male and female candidates were assumed to hold strong antitax
views. Not only did the female candidate overcome the perception that she was
softer on taxes than the male candidate, but she actually received a higher mean
score on the antitaxes item. By attacking her opponent on the tax issue, the

48

Press/Politics 8(3) Summer 2003

female candidate received a higher rating on the issue than would normally be
attributed to a female candidate. Curiously, the male candidates attack over the
environment did not yield a corresponding result. This finding is consistent with
Eagly and Karaus (2002) contention that
given very convincing evidence of a womans competence in a situation that
would seem to be extremely unfavorable to her because of its strongly masculine
definition, perceivers may view her performance as even more competent than
that of the equivalent man. (P. 583)

Thus, the female candidate received a boost for assuming an offensive posture in
a traditionally male issue area, but the male candidate did not.
Trait Effects

The third experimental stimulusmanipulating the candidates character


traitsyielded several interaction effects but few main effects. Controlling for
gender and issue content, the candidate who emphasized strength and independent leadership was more likely to be rated as knowledgeable, caring, and
proenvironment than the candidate who emphasized compassion and trust. The
absence of main effects is not surprising, considering that the traits emphasized
in the narration were mentioned briefly in the tagline at the end of the ad. Thus,
candidate traits were extremely subtle compared to the primary stimuli of the
candidates gender and the issue content of the ads.
One unexpected finding is the tendency for participants to rate the candidate
who emphasized masculine traits higher on all four of the personality traits. Even
though two of the four measures were not significant at the .05 level, the mean
value for the candidate who stressed masculine traits was higher in all of the criteria, even on such personality traits as caring and trustworthy that are typically considered feminine traits. A possible explanation for this pattern is suggested by a prior research finding that people tend to associate masculine
personality traits with politicians (Huddy and Terkildsen 1993a). However subtle, the brief description of the candidate as strong and independent might
have confirmed participants expectations that a politician should have masculine personality traits. If masculine traits are generally associated with politicians, but feminine traits are considered desirable, then it may help explain why
participants attributed a variety of traits, including caring, to those candidates.
Such a conclusion should be made with caution, however, because few of these
differences were statistically significant.
Despite the absence of main effects, there were significant interactions
between candidate traits and candidate gender. These interaction effects suggest that the emphasis on certain character traits appears to make some difference, but only for the female candidate. There were also interactions between

Gordon et al. / Negative Advertising and Female Candidates

49

candidate gender and issue content of the ads. The fourth column of Table 5
shows how likely the respondents were to attribute the qualities of probusiness,
trustworthy, and knowledgeable to the female and male candidates, controlling for whether the ads emphasized traditionally feminine or masculine
traits. The candidate considered least likely to support business was the female
candidate, Louise, when feminine traits were emphasized. However, when
Louise emphasized the masculine traits of independence and strong leadership
in her advertisement, her rating as probusiness was on par with Jeffs. It seems
that if a female candidate wants to build credibility on economic issues, one useful strategy would be to emphasize stereotypical masculine traits. According to
the respondents, the least knowledgeable and least trustworthy of the candidates
was Louise, when she emphasized feminine character traits in her advertisement. Jeffs ratings changed little when the traits he emphasized were varied.
Conclusion

These data do not support the conventional wisdom that female candidates
will be penalized more than male candidates for using negative advertisements.
Many participants objected to the negativity of the advertisements they viewed.
However, those who saw the female candidates advertisements were no more
likely to comment on their negativity than those who saw advertisements featuring the male candidate. While these results found no relative backlash against a
female candidate for going negative, they do not suggest that she would be
viewed as favorably as if she had avoided negative campaign tactics altogether.
Future research could use an expanded experimental design to examine the risks
for women of using negative instead of positive campaign ads. A study with twice
as many conditions as the research design used here could incorporate positive
versions of the same ads (and possibly include issues and traits not tested here) to
gauge the absolute, as well as relative, effects of negative campaign messages by
female candidates.
Participants in this experiment made inferences about unfamiliar candidates
on the basis of all three of the stimuli manipulated in the advertisements. The
issue content of the ads appeared to be used most often by participants to guess
the candidates other issue positions, traits, and political views. Participants
were less likely to make evaluations on the basis of candidate gender or on the
character traits emphasized in the narration. Thus, even as viewers of campaign
ads make judgments about candidates on the basis of gender stereotypes, the
message itself matters most of all.
Participants frequently made assumptions about the candidates on the basis of
the issue that was emphasized in the attack. For example, participants who saw
the taxes ad were more likely to infer that the candidate was probusiness and
patriotic as well as antitaxes. Participants who saw the environment ad were

50

Press/Politics 8(3) Summer 2003

more likely to consider the candidate to be a Democrat and liberal as well as


proenvironment. Because the candidates were unfamiliar to the participants,
and other political information such as party were omitted from the ads (to make
the ads consistent with real-world political commercials), participants seemed
to rely most heavily on the issues on which the opponents were attacked for
other clues about the candidates. These findings are not surprising considering
the large body of work on the relevance of issues to candidate evaluations (Nie et
al. 1976; for a review, see Popkin 1991).
Because issue effects dominated both gender and character trait effects, it
would be useful for future research to investigate the interaction of gender stereotypes with issues. Additional experiments could further our understanding
of gender stereotypes by exploring a broader range of issues than the two
employed here. Furthermore, it is not clear from these data if some traits and
issues are more gendered than others. Traits and issues found to be gendered in
past research were not significantly related to gender in this study. This may have
been a result of the particular experimental stimulus used; however, it may also
be that the nature of gender stereotypes changes over time. Another possibility is
that we should not treat traits and issues as dichotomous, either male or female.
Instead, there may be a kind of continuum along which traits and issues are
aligned. Future research could help explain why some issues are more likely to
activate gender stereotypes than others. Although candidate gender and character traits were not as important as ad issue content in predicting how the candidates would be perceived, there were some interesting exceptions that merit
discussion. The male candidate in this experiment had an advantage over the
female candidate when it came to certain issue domains. With respect to economic and military issues, the presumption of competence is with the male candidate,consistent with gender stereotypes.The findings reported here show that
in low-information contests, female candidates are assumed to be weaker on
male issues, simply on the basis of gender. For example, Louise was given
poorer ratings on taxes than Jeff when the advertisement contained no mention
of taxes. However, when Louise attacked her opponent on the issue of taxes, she
closed the gap. This suggests that one strategy to combat gender prejudice may
be for female candidates to attack their opponents on these male issues. In this
way, negative advertisements can be effective for female candidates.
Acknowledgments

The authors wish to thank Todd Belt, Gabriela Cowperthwaite, Lori Cox
Han, Tom Hogen-Esch, Steven Greene, Stephanie Hszieh, Phillip Richards, and
Randy Riddle for their invaluable help with this experiment.

Gordon et al. / Negative Advertising and Female Candidates

51

Notes
1.

2.

3.
4.

Our stimulus is a negative ad because it is predominately an attack on the opponent, with


the sponsoring candidate identified on screen for only a few seconds at the end of the ad.
This was necessary to establish the gender and traits of the candidate. This type of ad has traditionally been recognized as negative (Lau and Sigelman 2000). But see Jamieson et al.
(2000) for a discussion of an alternative classification scheme.
Useful generalizations can be made from experiments that use undergraduates as participants, even though it limits the external validity of results. Dolan (1997) makes several
arguments to support this contention. First, much important data has been collected this
way (Fox and Smith 1998; Dolan 1997; Nelson et al. 1997; Huddy and Terkildsen 1993b;
Leeper 1991; Sapiro 1981-82). Second, Dolan points out that gender-related studies can
benefit from using undergraduate participants because students have been socialized to be
more egalitarian. Thus, any observed effects may be greater for the general population,
making the use of students a more difficult test of hypotheses.
In open-ended responses, none of the participants expressed familiarity with either of the
candidates; nor did they suspect the ads had been manipulated.
Participants evaluations of candidate traits and issue competencies were measured with 4point scales. The question wording of the probusiness item was, How well does probusiness describe the candidate? Values range from 1 = not at all well to 4 = very well.

References
Ansolabehere, Stephen, and Shanto Iyengar. 1995. Going Negative: How Political Advertisements
Shrink and Polarize the Electorate. New York: Free Press.
Ashmore, Richard, and Frances Del Boca. 1979. Sex Stereotypes and Implicit Personality Theory: Towards a Cognitive-Social Psychological Conceptualization. Sex Roles 5:21948.
Brady, Henry, and Paul Sniderman. 1985. Attitude Attribution: A Group Basis for Political Reasoning. American Political Science Review 79:106178.
Burrell, Barbara. 1994. A Womans Place Is in the House Campaigning for Congress in the Feminist Era.
Ann Arbor: University of Michigan Press.
Conover, Pamela Johnston. 1981. Political Cues and the Perception of Candidates. American
Politics Quarterly 9:42748.
Darcy, Robert, Susan Welch, and Janet Clark. 1994. Women,Elections,and Representation. Lincoln:
University of Nebraska Press.
Dolan, Kathleen. 1997. Gender Differences in Support for Women Candidates: Is There a Glass
Ceiling in American Politics? Women & Politics 17(2):2741.
Eagly, Alice, and Steven Karau. 2002. Role Congruity Theory of Prejudice toward Female
Leaders. Psychological Review 109:57398.
Fox, Richard, and Eric R. A. N. Smith. 1998. The Role of Candidate Sex in Voter DecisionMaking. Political Psychology 19:40515.
Huddy, Leonie, and Nayda Terkildsen. 1993a. The Consequences of Gender Stereotypes for
Women Candidates at Different Levels and Types of Office. Political Research Quarterly
46:50325.
Huddy, Leonie, and Nayda Terkildsen. 1993b. Gender Stereotypes and Perceptions of Male and
Female Candidates. American Journal of Political Science 37:11947.

52

Press/Politics 8(3) Summer 2003

Iyengar, Shanto, Nicholas Valentino, Stephen Ansolabehere, and Adam Simon. 1997. Running
as a Woman: Gender Stereotyping in Womens Campaigns. In Women,Media,and Politics, ed.
Pippa Norris. New York: Oxford University Press.
Jamieson, Kathleen Hall, Paul Waldman, and Susan Sherr. 2000. Eliminate the Negative? Categories of Analysis for Political Advertisements. In Crowded Airwaves, eds. James Thurber,
Candice Nelson, and David Dulio. Washington, D.C.: Brookings Institution.
Just,Marion,Ann Crigler,Dean Alger,Timothy Cook,Montague Kern,and Darrell West.1996.
Crosstalk: Citizens, Candidates, and the Media in a Presidential Campaign. Chicago: University of
Chicago Press.
Kahn, Kim Fridkin. 1996. The Political Consequences of Being a Woman. New York: Columbia University Press.
Kahn, Kim, and Patrick Kenney. 2000. Negative Campaigning and Knowledge of Senate Elections. In Crowded Airwaves, eds. James Thurber, Candice Nelson, and David Dulio. Washington, D.C.: Brookings Institution.
Kaid, Lynda Lee, Sandra L. Myers, Val Pipps, and Jan Hunter. 1984. Sex Role Perceptions and
Televised Political Advertising: Comparing Male and Female Candidates. Women and Politics
4(4):4153.
Kinder, Donald. 1983. Diversity and Complexity in American Public Opinion. In Political Science:The State of the Discipline, ed. Ada Finifter. Washington, D.C.: American Political Science
Association.
Lansing, Jewell. 1991. 101 Campaign Tips for Women Candidates and Their Staffs. Saratoga, CA: R &
B Publishers.
Lau, Richard R., and Lee Sigelman. 2000. Effectiveness of Negative Political Advertising. In
Crowded Airwaves, eds. James Thurber, Candice Nelson, and David Dulio. Washington, D.C.:
Brookings Institution.
Leeper, Mark. 1991. The Impact of Prejudice on Female Candidates: An Experimental Look at
Voter Inference. American Politics Quarterly 19:24861.
Nelson, Thomas E., Zoe M. Oxley, and Rosalee A. Clawson. 1997. Toward a Psychology of
Framing Effects. Political Behavior 19(3):22146.
Nie, Norman, Sidney Verba, and John R. Petrocik. 1976. The Changing American Voter. Cambridge, MA: Harvard University Press.
Popkin, Samuel. 1991. The Reasoning Voter. Chicago: University of Chicago Press.
Procter, David E., William Schenck-Hamlin, and Karen Haase. 1994. Exploring the Role of
Gender in the Development of Negative Political Advertisements. Women and Politics
14(2):122.
Sapiro, Virginia. 1981-82. If Senator Baker Were a Woman: An Experimental Study of Candidate Images. Political Psychology 2(1):6183.
Thompson, Seth and Janie Steckenrider. 1997. The Relative Irrelevance of Candidate Sex.
Women and Politics 17(4):7192.
Trent, Judith, and Robert Friedenberg. 1995. Political Campaign Communication. Westport, CT:
Praeger.
Vavreck,Lynn.2001.The reasoning voter meets the strategic candidate:Signals and specificity in
campaign advertising, 1998. American Politics Research 29(5):50729.
Williams, Leonard. 1998. Voter Reaction to Women Candidates. In Women and Elective Office:
Past, Present, and Future, eds. Sue Thomas and Clyde Wilcox. New York: Oxford University
Press.

Gordon et al. / Negative Advertising and Female Candidates

53

Biographical Notes
Ann Gordon is an assistant professor of political science at Ohio University. She is the coeditor of
Anticipating Madam President and coauthor of When Stereotypes Collide:Race,Gender,and Videostyle in
Congressional Campaigns (forthcoming).
Address: Political Science Bentley Hall, Ohio University, Athens, OH 45701; phone: (740) 5934377; fax: (740) 593-0394; e-mail: gordonc@ohiou.edu.
David M.Shafie is an assistant professor of political science at Ohio University.He is the author of
Rethinking California: Politics and Policy in the Golden State and recipient of the Morris K. Udall
fellowship.
Ann N. Crigler is an associate professor of political science at the University of Southern California and director of the Jesse M. Unruh Institute of Politics at USC. She is the coauthor of Common
Knowledge: News and the Construction of Political Meaning and Crosstalk: Citizens, Candidates and the
Media in a Presidential Campaign and the editor of The Psychology of Political Communication. She is
coeditor of a volume on election reform titled Rethinking the Vote:The Politics and Prospects of American Election Reform.

Das könnte Ihnen auch gefallen