if you can say something nice
TRANSCRIPT
8/7/2019 If you can say something nice.
http://slidepdf.com/reader/full/if-you-can-say-something-nice 1/39
1
If You Can’t Say Something Nice: The Gender Dynamics of Negative Advertising
Kjersten NelsonDepartment of Political Science
University of [email protected]
Women running for Congress face a double-bind. They can adhere to genderexpectations, which suggest they be compassionate and conciliatory. Or they can exhibit traits
associated with competent leaders, which tend to run in opposition to gender expectations. Onemanifestation of this tension is the decision of whether or not to adopt aggressive negative
advertising tactics in their campaigns. Analyzing campaign commercials from 2000, 2002, and2004, the author concludes that the candidate’s party, the gender context of the race, and
citizens’ familiarization with women in elected office can work together to influence candidates’decisions to use negative advertising.
8/7/2019 If you can say something nice.
http://slidepdf.com/reader/full/if-you-can-say-something-nice 2/39
2
With every election cycle comes discussion of different campaigns’ negative strategies.
Coverage of the 2006 Senate race in Minnesota, for instance, paid significant attention to the
increasingly negative ads leveled by Congressman Mark Kennedy (R) against Hennepin County
prosecutor Amy Klobuchar (D). Similarly, many in the state were riveted by the acrimonious
attacks involved in the race for Minnesota Congressional District 6, between candidates Patty
Wetterling (D) and State Senator Michele Bachmann (R) (e.g., Page 2006; Kirkpatrick 2006;
Goodman 2006). As the number of women running for office grows, and as the negative
campaign ad appears to be here to stay, a systematic investigation of the differences between
men’s and women’s negative messages is vital. Do men and women go negative against their
opponents at different rates? What are the conditions that lead to gender differences in negative
messages?
These questions are inspired, in part, by the relatively recent explosion – both in number
and stature – of women in elected office. For instance, there are 79 female representatives1
and
16 female senators in the 110th
Congress; this is more than there have ever been (Center for
American Women and Politics (CAWP) 2007a). Similarly, the United States is experiencing a
record number of female governors – nine – matched only in 2004 (Center for American Women
and Politics (CAWP) 2007b). Beyond sheer numbers, women are reaching new heights in
politics as well. Representative Nancy Pelosi serves as the first female Speaker of the House,
while Senator Hillary Clinton mounted the first viable presidential campaign by a woman.
Despite this excitement, women’s representation remains low, particularly in higher
elective offices. What is more, women may win races at rates equal to male candidates,but
gender-neutral results do not imply a gender-neutral process. The increasing presence of women
1This number includes 3 delegates from U.S. territories.
8/7/2019 If you can say something nice.
http://slidepdf.com/reader/full/if-you-can-say-something-nice 3/39
3
at all levels of politics spawned an interest in what type of obstacles female candidates may face,
as well as constraints that may lead women and men to pursue different strategies as candidates.
In this paper, I investigate whether the use of negative advertising is an area where
female candidates experience different constraints than male candidates. There is reason to
believe that gender expectations may play a particularly prominent role in this regard. Women
face the double-bind of proving they are tough enough to live up to stereotypic expectations of
what constitutes a good leader while avoiding implications that they violate prescriptive gender
norms of communality and “niceness” (Rudman and Glick 1999, 1004). A negative message
that attacks an opponent as unfit for office certainly reveals a certain level of grit but may also
signal the candidate does not conform to prescriptive feminine expectations.
This is timely, as negative messages are a tactic that candidates have increasingly relied
upon in order to highlight the shortcomings of their opponents and, ostensibly, increase their
likelihood of electoral victory (Geer 2006). While there has been extensive debate amongst
political scientists as to whether negative advertising has positive consequences – both for the
sponsoring candidates and democracy (e.g., Geer 2006; Brooks 2006; Lau, et. al. 1999;
Ansolabere and Iyengar 1995) – the popularity of negative advertising amongst candidates
implies that candidates believe them to be successful tools for electoral victory.
This paper is part of a larger project to determine whether these seemingly invaluable
campaign tools can be used to the same effect by female candidates as they are by male
candidates. All candidates behave strategically in order to maximize their likelihood of winning.
The outcomes of elections ultimately come down to voters’ choices. Thus, the double-bind faced
by female candidates is ultimately dependent upon the degree to which voters ascribe to and vote
based upon stereotypes of gender and leadership. In order to uncover what role the gender
8/7/2019 If you can say something nice.
http://slidepdf.com/reader/full/if-you-can-say-something-nice 4/39
4
expectations of voters play in candidates’ strategies, I must first determine whether and how
candidates vary their negative campaign strategies by gender and by the gender context of a
given race in order to anticipate the voters’ expectations of gender and leadership.
Building on past work on gender expectations and negative advertising, I look at
televised advertisements in the 2000, 2002, and 2004 congressional election cycles to determine
whether the gender context of a campaign affects its level of negativity. In other words, does it
matter whether a male is running against another male versus a female? Does it matter if a
woman is running against a man versus another woman? Additionally, I re-examine variables
that have been established as consequential, such as the incumbency status of the candidate, the
competitiveness of the race, and party affiliation of the candidate to examine whether these
established predictors have endured (e.g., Kahn and Kenney 1999; Sapiro and Walsh 2002; Lau
and Pomper 2001).
I proceed with a review of the literature relating to gender and party stereotypes and
negative advertising. I follow with hypotheses and a description of my dataset and methods.
Finally, I present my findings and discuss the implications both for the role of gender in the
political process and for women’s representation in its elected institutions.
Gender Expectations and Other Stereotypes
What role does gender play in a candidate’s decision of whether and how to go negative?
The theoretical answer to this question depends, first, upon the extent to which voters ascribe to
and vote based upon gender stereotypes. Second, it depends on whether candidates perceive that
their particular voters are ascribing to and voting based upon these stereotypes and then to what
extent these candidates play to (or against) the established stereotypes – whether consciously or
not.
8/7/2019 If you can say something nice.
http://slidepdf.com/reader/full/if-you-can-say-something-nice 5/39
8/7/2019 If you can say something nice.
http://slidepdf.com/reader/full/if-you-can-say-something-nice 6/39
6
with the issue expectations of her gender or her party? A similar potential disconnect faces male
Democrats.
As is evident, these stereotype studies have largely focused on the implications of these
stereotypes for issue competencies. However, if voters typically expect women to have
competence on “compassion” issues because female candidates are considered to be more
compassionate – and men to be more competent on issues like national defense because men are
considered to be more agentic and assertive – might these same expectations spill over into the
type of campaign these candidates are expected to run? In other words, setting the substance of
the campaign aside for a moment, might certain tactics – most specifically, negative attacks on
one’s opponent – more neatly fit with the expectations of men and male candidates than their
female counterparts?
Reactions to Unexpected Behavior. If this is the case – that attack advertising does not comport
with gender expectations for women – how might a female candidate assume that voters would
respond to this behavior? In general, social psychology has concluded that individuals acting
counter to stereotypical expectations will be punished (Cialdini and Trost 1998). This is
particularly true for expectations that women behave communally (Rudman and Glick 1999).
Women who adopt masculine leadership styles are also assessed negatively, particularly those in
traditionally male-dominated fields (Eagly, Makhijani, and Klonsky 1992). Negative messaging
is not a communal tactic, and it represents an aggressive campaign tactic in a traditionally male-
dominated field. Thus, we could expect savvy female candidates to anticipate negative feedback
from voters for adopting these techniques and avoid campaign tactics that include negative
messaging.
8/7/2019 If you can say something nice.
http://slidepdf.com/reader/full/if-you-can-say-something-nice 7/39
7
However, female candidates and office holders face the infamous double-bind of politics.
On the one hand, there is the possibility that behavior deemed unfeminine will hurt a woman’s
electoral prospects. On the other hand, like women breaking into other male-dominated fields,
adhering too closely to gender expectations can lead others to doubt whether the female
contender has the qualities it would take – often stereotypically “male” qualities – to be an
effective elected leader. Eagly and Karau (2002) describe this disconnect, as characteristics
associated with successful individuals in male-dominated fields – such as “leadership” and “self-
confidence” – run counter to characteristics typically used to describe women – such as being
“communal” and having a “willingness to compromise.” Although Eagly and Karau do not
directly test the level of masculine expectations tied to elective office, it seems reasonable to
assume for now – even after decades of increased women’s participation in elected office – that
national elected leaders are expected to exhibit stereotypically masculine traits, particularly in a
time of heightened national fear and concern about homeland security (e.g., Lawless 2004; Falk
and Kenski 2006). These expectations may establish incentives for women to adopt more
stereotypically masculine campaign tactics in order to compensate for their expected
shortcomings. As Sapiro and Walsh (2002) explain, “Women may need to ‘go negative’ earlier
to show they are fighters and serious candidates” (6). Or, in the words of House Speaker Nancy
Pelosi, “The minute you go into this arena, especially at this altitude, you have to prove you can
breathe the air” (Goodman 2006).
However, this tension sets up a potentially stark trade-off for female candidates. A
female candidate who chooses an aggressively negative campaign may prove she meets
expectations for a competent elected official, but may inadvertently stoke negative assessments
of her, due to her counterstereotypical gender behavior. A female candidate who reinforces
8/7/2019 If you can say something nice.
http://slidepdf.com/reader/full/if-you-can-say-something-nice 8/39
8
gender expectations with a nice campaign, however, may get voters to like her – but not vote for
her, if she has not proven her competence as a candidate.
Of course, it is not just women who face gender expectations during a campaign. Men
are also bound by norms and expectations of behavior, which are typically the mirror-opposite of
those expected from women. Preciselybecause men do not face the same double-bind as
women, should they be expected to go negative more often, in order to prove they are both tough
enough for the job and comport with male gender expectations? Of course, men face campaign
gender minefields of their own, particularly owing to the gender of the opponent. Certainly, men
are expected to be assertive, as are candidates (Eagly and Karau 2002). What, though, of a male
candidate running against a female opponent? In such an instance, we might expect norms of
chivalry to take over, depressing levels of negativity from male candidates directed at female
opposition.
In general, based on the review of gender and leader expectations, I lay out the following
candidate-level hypotheses:2
• A male candidate – who is running against another man – will have the highest levels of
negative advertising of any candidates. These are the instances where expectations of
gender coincide with expectations of effective elected officials – assertiveness is a
desirable characteristic in both, a characteristic exemplified by negative messaging.
• A male candidate – who is running against a woman – will have the lowest levels of
negative advertising of any candidates. Although the expectations of elective office
remain, chivalrous norms should dampen the negativity.
2Note that these hypotheses relate to the level of negativity expected from one of the candidates in the race. In other
words, these are not predictions for levels of negativity in a given campaign, but for levels of negativity sponsored
by one of the candidates in a given campaign, contingent on the gender context of the race.
8/7/2019 If you can say something nice.
http://slidepdf.com/reader/full/if-you-can-say-something-nice 9/39
9
• A female candidate – who is running against a man or a woman – will have moderate
levels of negativity as compared to male candidates. The stereotypical expectations of
women – to be compassionate and conciliatory – may depress levels of negativity as
women attempt to comport with gender stereotypes. At the same time, the desire to
prove she comports with the expectations of elected office holders will counterbalance
any tendency to reduce levels of negativity. These competing factors – which are not
experienced by male candidates in any gender context – will moderate the negativity
levels of women running against men.
Past studies of gender and negative advertising have come to mixed conclusions as to its
effect (e.g., Lau and Pomper 2001; Kahn and Kenney1999; Hitchon and Chang 1995; Bystrom
and Kaid 2002). This can partially be attributed to the fact that candidate gender alone, and not
gender context, was the variable of interest. Subsequent studies have incorporated whether the
race is two men, a woman and a man, or two women, although the small number of women
running against each other has hampered in-depth analysis of this group (Sapiro and Walsh
2002).
The gender context of a race does not occur in a vacuum, however, and considerations of
party are also fundamental (Sapiro and Walsh 2002). Recall from above that Democratic
stereotypes largely overlap with the stereotypes of women – that is, that Democrats will be more
competent on “compassion” issues, while Republicans will be more competent on issues like the
military. For female Democrats, then, we have two sets of stereotypes that would suggest that a
softer image satisfies the largest set of expectations – in other words, less reliance on negative
messages. For Republican women, though, we see again the double-bind. Gender would
8/7/2019 If you can say something nice.
http://slidepdf.com/reader/full/if-you-can-say-something-nice 10/39
10
suggest an emphasis on compassion – thus, avoiding negative messages – while the toughness
associated with the Republican Party might lead to higher levels of negative messaging.
This intersection of party and gender is present in all gender contexts. With the
intersection of party, candidate gender, and gender context, I hypothesize the following:
• Male Republicans will have the highest levels of negativity, due to the
convergence of gender and party expectations to be more aggressive. However,
due to considerations of chivalry, Republican men will be less negative towards
female opponents than male opponents.
•
Male Democrats will be less negative than male Republicans, due to the
competing pressures they feel between gender expectations (more negative) and
party expectations (less negative). Like their Republican counterparts, however,
they will still be less negative towards female opponents than male opponents,
due to constraints of expectations of chivalry.
• Democratic females should exhibit some of the lowest levels of negativity, due to
the overlap of gender and party expectations that point towards more
compassionate, and therefore less negative, campaign strategies. Negativity
levels should be relatively higher when Democratic females run against males,
however, as it becomes more necessary to prove competence for office against a
male opponent.
There is a potentially interesting dynamic present for Republican woman running against
a Democratic female. For the Republican women, this is the context with theleast cross-
pressures. The Republican woman can expect an opponent with two forces – gender and party –
pushing towards less negativity. At the same time, although the Republican woman may
8/7/2019 If you can say something nice.
http://slidepdf.com/reader/full/if-you-can-say-something-nice 11/39
11
experience some conflict between gender (less negativity) and party (more negativity)
expectations, the fact that she does not face the same prohibitions as her male counterparts
(chivalry), may create the perfect storm for increased negativity against her opponent – who, by
all accounts, will be less willing to strike back on her own, due to a convergence of gender and
party expectations.
Predictors of Negative Advertising
Scholars studying negative advertising have established some reliable predictors of
campaign negativity. The most prominent of these is the status of the candidate (e.g., incumbent
versus challenger), the status of the race (e.g., whether it is an open seat), and the
competitiveness of the race. The general consensus has been that challengers are more negative
than incumbents, controlling for the amount of campaign spending and regardless of the measure
used for campaign negativity (Kahn and Kenney 1999; Lau and Pomper 2001). This empirical
finding is intuitive. On the one hand, incumbents have a public record to attack, whereas
challengers, depending on their current occupations, may not. The incumbent, on the other hand,
has less of an incentive to attack a challenger. Most incumbents enjoy a significant name-
recognition advantage over their opponents, one they are unlikely to endanger by providing
publicity – albeit negative – to the otherwise unknown opponent.
Along these same lines of reasoning, open races – ones in which there is no incumbent –
should be more negative, as neither candidate necessarily enjoys the same overwhelming name
recognition advantage that incumbents might. Studies bear this expectation out, showing that
open races are, on average, more negative than those in which there is an incumbent (Kahn and
Kenney 1999; Lau and Pomper 2001). Finally, past studies have consistently found that
competitive races are more negative than non-competitive races (Kahn and Kenney 1999; Sapiro
8/7/2019 If you can say something nice.
http://slidepdf.com/reader/full/if-you-can-say-something-nice 12/39
12
and Walsh 2002; although, see Lau and Pomper 2001 for results that suggest competitive races
are not statistically significantly different from non-competitive races).
Still More Considerations
Party-sponsored Messages. Certainly, candidates are not the only agents strategizing about
televised messages in a campaign season. Parties can play an important role in the most
competitive races by producing and airing ads for their preferred candidates. Due to campaign
finance regulations, these expenditures are done independently, without consulting the
candidate’s campaign. Thus, while there is theoretically no communication about how the
candidate him or herself wishes to reinforce or compensate for gender and party stereotypes, at
the same time, party officials are making the same decisions about how to navigate these
stereotypes, presumably with the same constraints and intersections of conflicting expectations.
The difference is that the parties have the luxury of going negative against the opponent,
while providing some plausible deniability for the candidate her or himself. Certainly, it may be
problematic to assume that voters knowingly delineate between party and candidate-sponsored
spots. However, relatively recent changes in campaign finance law have required more explicit
identification of ad sponsorship (e.g, the McCain Feingold Bipartisan Campaign Finance Reform
Act of 2002). While these changes were made in an effort to increase accountability, it may
have had the effect of increasing the parties’ (and other third parties’) ability to go negative for
the candidate, without the potential negative fallout.3
Assuming this advantage now exists, it
would be most critical for candidates who, in facing the intersection of gender, party, and gender
context expectations, cannot go negative as much as they might hope. Thus, we should see
3This, of course, is an empirical question, one to which I do not have an answer at this point. For now, I will make
the logical assumption that requirements for more explicit acknowledgement of ad sponsorship will correspond to a
higher likelihood that viewers will delineate between party and candidate-sponsored ads. Or, at least, that campaign
strategists will assume this relationship exists and act accordingly.
8/7/2019 If you can say something nice.
http://slidepdf.com/reader/full/if-you-can-say-something-nice 13/39
8/7/2019 If you can say something nice.
http://slidepdf.com/reader/full/if-you-can-say-something-nice 14/39
14
in state office may signal less rigid gender expectations which then suggests that women and
men in these contexts may behave similarly when it comes to their campaign strategies. In those
states where gender expectations, particularly for candidates and elected officials, are more rigid
– i.e., in states with fewer women in the state legislature – I expect larger gender context
differences in negative advertising behavior, as candidates attempt to optimize their chances of
winning in an environment constrained by gender expectations (Von Baeyer, et. al. 1981).
Data and Methods
The ability to analyze campaign communications has increased exponentially since
tracking of campaign advertisements done by the Campaign Media Analysis Group (CMAG)
have been made available for academics through the Wisconsin Advertising Project (WiscAds).4
CMAG advises political clients. In an effort to do this more effectively, CMAG developed a
technology to capture and record all political advertisements aired on the major networks, 25
cable networks, and local advertising in the largest media markets in the country. These markets
initially encompassed the 75 largest media markets in the country. Beginning in 2002, the
CMAG technology monitors the 100 largest media markets in the nation (WiscAds web-site).
The monitoring includes screen shots of every fourth second of the advertisement, as well as a
complete transcript of the audio portion. The technology also creates a dataset with an
observation for each airing of any political commercial. Once the advertisements are compiled
by CMAG, coders with WiscAds code the ads for a multitude of attributes, including whether
and how the ad is negative.5
4The Wisconsin Advertising Project was sponsored by a grant from The Pew Charitable Trusts. The opinions
expressed in this article are those of the author(s) and do not necessarily reflect the views of the Wisconsin
Advertising Project, Professor Goldstein, Joel Rivlin, or The Pew Charitable Trusts.5
For all three election cycles, ads were coded as either “attack,” “promote,” or “contrast.” For my purposes, I have
counted both attack and contrast ads in my calculation of negativity, as they both represent an assertive campaign
strategy.
8/7/2019 If you can say something nice.
http://slidepdf.com/reader/full/if-you-can-say-something-nice 15/39
15
Beginning with the spot-based dataset, I compiled a dataset with the Congressional
candidate as the unit of analysis for the 2000, 2002, and 2004 election cycles. I calculated what
percentage of each candidate’s total airings were negative in tone, as well as what percentage of
the spots run for candidates by the party were negative. To this initial dataset, I also added
dummy variables accounting for the party of the candidate and gender context of the race. A
man running against another man receives a one for the “man versus man” variable; a man
running against a woman receives a one for the “man versus woman” variable; a woman running
against a man receives a one for the “woman versus man” variable; and a woman receives a one
for the “woman versus woman” variable if she is running against another woman.
6
Additionally,
I included several measures to control for factors that have consistently proven to be influential.
In particular, I accounted for whether the race was competitive; the campaign expenditures of the
candidates; and whether the candidate was an incumbent, challenger, or if it was an open seat.
The competitiveness of the race was determined using the competitiveness rankings assigned by
Congressional Quarterly Weekly. These are prospective competitiveness measures, typically
issued two to four weeks before the elections, and they represent the conventional wisdom upon
which the candidates and campaigns are basing their strategies. In this sense, these prospective
measures should more accurately assess the environment in which candidates are operating, as
opposed to retrospective measures – such as final vote margin – which, one could argue, more
accurately represent the outcome of candidates’ strategies and may or may not reflect how close
candidates felt the race was.7
6I largely coded these races based on the names of the candidates. For names that were not obviously either male or
female, I undertook an Internet and LexisNexis search.7
CQ Weekly ranks the competitiveness of the race on a four-point scale, ranging from “Safe” on one extreme to
“Toss-Up” on the other. For parsimony’s sake, I have converted this ranking into a dummy variable, where one
represents that the race is competitive and that CQ Weekly considered this race a toss-up. All other rankings are
considered non-competitive races.
8/7/2019 If you can say something nice.
http://slidepdf.com/reader/full/if-you-can-say-something-nice 16/39
8/7/2019 If you can say something nice.
http://slidepdf.com/reader/full/if-you-can-say-something-nice 17/39
17
restricting my analysis to candidates who chose to pursue this strategy, meaning that these are
campaigns that are likely better-funded and more competitive. At the same time, broadcast
strategies are more feasible – and probably more effective – in districts where television
advertising is not astronomically expensive and where media markets more closely overlap with
congressional districts. For instance, candidates in New York City will not only pay the high
advertising rates of that area, but will also reach many citizens with the advertising who are not
even in their districts. What is more, there are other means by which candidates can convey
negative messages to their constituencies – for instance, direct mail and campaign web-sites
(Druckman, Kifer, and Parkin 2007).
That said, broadcast advertising remains an important means by which campaigns
communicate with voters (Sapiro and Walsh 2002). Moreover, a quick check of the data reveal
even amongst campaigns that choose broadcast advertising, almost 47 percent of them are
considered non-competitive races. Additionally, the gender composition of the sample closely
resembles that of the universe of candidates. While 84 percent of all congressional candidates
running in 2000, 2002, and 2004 were male, 81 percent of candidates who chose broadcast
advertising were male, suggesting the sample and population are similar in terms of gender.
INSERT TABLE 1 ABOUT HERE.
Women continue to run for office at rates much lower than men, and the data,
summarized in Table 1, reveal this trend. The dataset contains 560 observations where the
candidate is a man running against another man, while there are only 30 instances where the unit
of analysis is a female candidate who is running against another woman. In 135 observations,
the candidate is a man running against a woman, and in 127 observations, the candidate is a
8/7/2019 If you can say something nice.
http://slidepdf.com/reader/full/if-you-can-say-something-nice 18/39
18
woman running against a man.10
Similarly, the continuing underrepresentation of women in
elective office is reflected in the distribution of candidate status by gender, reported in Table 2.
Because there are fewer female incumbents, the percentage of female candidates who aired
advertising and who are incumbents is only 32 percent, as opposed to 49 percent of male
candidates who are similarly situated. This distribution is reversed for challengers who choose
to air campaign spots. Forty percent of female candidates who broadcast were challengers, while
only 27 percent of male candidates were. Candidates running in open seats were more equally
distributed: 24 percent of broadcasting male candidates were candidates in open seat races, while
28 percent of broadcasting female candidates were. A chi-square test reveals this relationship
between gender and candidate status to be statistically significant (Pr=.000).
INSERT TABLE 2 ABOUT HERE.
Finally, Table 3 compares the competitiveness of the race with the gender of the
candidate. The data here suggest that, at least amongst candidates who choose to air ads, female
candidates are more than likely to be in competitive races. Of the female candidates who aired
commercials, 41 percent are in non-competitive races. Of the male candidates in the dataset, 48
percent of them are in non-competitive races. The chi-square test suggests that, amongst those
candidates that air ads, being a male is associated with being in a non-competitive race
(Pr=.09).11
INSERT TABLE 3 ABOUT HERE
Bivariate Results. I turn first to a comparison of levels of negativity between male and female
candidates. Table 4 displays the bivariate comparison of the means of these groups for
10While it is tempting to divide these figures by two to determine how many races are included in each category,
this would be misleading. There are several instances in the data where only one candidate in a race ran ads.11
To be clear, this comparison cannot reveal whether incumbent women are more likely to be challenged in the first
place or whether those challenges are likely to be deemed competitive.
8/7/2019 If you can say something nice.
http://slidepdf.com/reader/full/if-you-can-say-something-nice 19/39
8/7/2019 If you can say something nice.
http://slidepdf.com/reader/full/if-you-can-say-something-nice 20/39
8/7/2019 If you can say something nice.
http://slidepdf.com/reader/full/if-you-can-say-something-nice 21/39
21
Democrat; dummy variables for the election cycle, with 2002 serving as the omitted category; a
variable coded one if the race was considered competitive, zero otherwise; candidate spending in
the race, in thousands; and, finally, the percentage of the candidate’s state legislature that is
female in that year.
First and foremost, unlike the bivariate results, the candidate-sponsored negativity model
does not reveal any statistically significant differences between any of the four gender context
groups. Regardless of the sex of the candidate and the sex of his or her opponent, levels of
negativity in candidate-sponsored ads do not vary. The model in Table 5 reveals the p-values to
be well above standard levels of significance when compared to men running against men.
Further statistical tests revealed similarly high p-values for comparisons amongst the gender
context groups (in an analysis not shown here.)
Results for the candidate-sponsored ads do reinforce some statistically significant
bivariate findings, as well as past research. For instance, challengers run ads that are, on
average, 25 percentage points more negative than incumbents. Candidates in open seat races are
also more negative than incumbents, broadcasting ads that are, on average, 11 percentage points
more negative. Of all three types of candidates, challengers have the most negative ad
campaigns, with negativity proportions that are 15 percentage points higher than open seat
candidates’ (p=.001, in a separate analysis not shown here). Additionally, reduced negativity in
the campaign following 9/11 holds in the multiple regression analysis, showing that in 2002,
candidate-sponsored ad campaigns were, on average, seven percentage points less negative than
2000 and 10 percentage points less negative than 2004.
INSERT TABLE 5 ABOUT HERE
8/7/2019 If you can say something nice.
http://slidepdf.com/reader/full/if-you-can-say-something-nice 22/39
22
The party of the candidate also matters according to this model in predicting levels of
candidate-sponsored negativity. Democrats, on average, are less negative by 5 percentage points
(p<.03). This result reinforces the notion that, perhaps, because Democrats are stereotypically
expected to be more competent on compassion issues, they are also expected to run more
“compassionate” campaigns – or, at the very least, that Democratic candidates believe it is in
their interests to run more “compassionate” – i.e., less negative – campaigns.
The model examining party-sponsored ads reveals even fewer statistically significant
results. Gender-context remains insignificant, although the context where a woman is running
against a man approaches statistical significant (p=.12), suggesting that women running against
men have more negative campaigns, by seven percentage points (as compared to a man running
against another man). Party support for challengers tends to be more negative than party support
for incumbents by an average of nine percentage points (p=.04). As with candidate-sponsored
ads, the Democratic Party is less negative by an average of nine percentage points (p=.01).
Finally, the 9/11-effect emerges for party-sponsored ads as well, with campaigns, on average, 16
percentage points less negative in 2002 than 2000 and 19 percentage points less negative in 2002
than in 2004 (p<.001 for both).
Interaction Models. The main effects models are instructive, but in order to better represent my
theory and hypotheses, interaction terms between the gender contexts and female percentage of
state legislatures and gender contexts and party should be added. My theory does not stipulate
that the percentage of a state’s legislature that is female should directly affect the level of
campaign negativity. Instead, the degree to which a state’s citizens are accustomed to women in
politics should moderate the effects of gender context. As described in more detail previously,
states with higher percentages of women in the state legislature should translate into less rigid
8/7/2019 If you can say something nice.
http://slidepdf.com/reader/full/if-you-can-say-something-nice 23/39
23
campaign-related gender expectations. Thus, I would expect differences in gender contexts
amongst states with lower levels of female legislators, where male and female candidates must
contend with gender expectations that have not been frequently challenged or overcome.
Similarly, I hypothesized a potential interaction between gender context and party. The overlap
of party-based and gender-based stereotypes may increase the effect of campaign negativity in
some cases and mitigate it in others.
Context Matters for Candidates
Columns three and four of Table 5 display results of the multiple regression model with
added interaction terms. The model now suggests that, amongst candidates who purchase air
time, women running against women are, on average, 42 percentage points more negative than
men running against men (p=.05). What is more, as women-only races are interacted with party,
a clearer picture emerges. The large, positive coefficient on the woman versus woman variable
appears to be due to the fact that Republican women running against other women are more
negative, while Democratic women running against other women are only moderately more
negative than men running against men. In more concrete terms, Democratic women running
against women have an average level of negativity of 25 percent. Republican women running
against other women have an average level of negativity of 48 percent. (The level of
significance for the interaction term between a woman running against another woman and party
is p=.1)
These findings suggest interesting dynamics in the intersection between party and gender
expectations. As mentioned before, I suggest that expectations about issue competencies may
also influence expectations about campaigns. In other words, if women and Democrats are
expected to be competent on “compassion” issues, perhaps they are also expected to demonstrate
8/7/2019 If you can say something nice.
http://slidepdf.com/reader/full/if-you-can-say-something-nice 24/39
8/7/2019 If you can say something nice.
http://slidepdf.com/reader/full/if-you-can-say-something-nice 25/39
25
Recall that the explanatory theory was that the increasing female proportion of state
legislatures signals a populace that is more familiar with women in elected office and one that
may not adhere as strongly to rigid gender expectations. If this explanation holds in these cases,
it would suggest that as women become unconstrained by voters’ gender expectations, they are
less inclined to negative advertising, at least against their female opponents. In this instance, it
suggests that women no longer feel it is necessary to prove they have the grit to be an effective
member of Congress.
There are two other interesting results in this model that deserve note. First is the
interaction between legislative composition and the man versus woman gender context. The
statistical significance (p=.08) suggests that as state legislatures obtain higher proportions of
female legislators, men running against women in those states will become less negative. The
size of the coefficient mirrors that of the interaction between a female-only race and legislative
composition – .01. Also interesting here, however, is the fact that the coefficient runs in the
opposite direction from what I expected. I surmised that gender expectations would constrain
men from going negative against women – a norm of chivalry. However, this finding suggests
that as female representation in the state legislature increases – and, ostensibly, as the constraints
of gender expectations decrease – men will reduce their levels of negativity towards female
opponents. This unexpected finding suggests an interesting relationship that should be further
studied.
Finally, compared to the main effects model, the coefficient on the man versus woman
gender context approaches statistical significance in this interaction model (p=.13). If this
finding were to hold – perhaps with a larger sample – it suggests that men running against
8/7/2019 If you can say something nice.
http://slidepdf.com/reader/full/if-you-can-say-something-nice 26/39
8/7/2019 If you can say something nice.
http://slidepdf.com/reader/full/if-you-can-say-something-nice 27/39
27
be vital for Democratic women facing female opponents, particularly if they feel constrained by
the intersection of party and gender expectations from matching their opponents with negative
messages.
Moreover, the model verifies that parties sponsor, on average, ads that are 11 percentage
points more negative when the candidate is a challenger (p=.02) than when the candidate is an
incumbent. The 9/11 effect also persists in this model, with ads that are 15 percentage points
less negative, on average, than 2000 and 18 percentage points less negative, on average, than
2004 (p<.001).
Conclusion
The paper provides preliminary evidence that campaign context plays a role in the
decisions of men and women to go negative. More specifically, the gender of the candidate, the
gender of his or her opponent, the party of the candidate, and women’s previous successes in
state elected office all influence how often candidates go negative and, to a lesser extent, when
the parties go negative on behalf of the candidates. In an examination of broadcast advertising
from the 2000, 2002, and 2004 election cycles, the data suggest that these dynamics manifest
themselves most obviously in races where women are running against a female opponent. Here,
Republican women are most likely – or, perhaps, even compelled – to launch negative attacks
against their female opponents. Democratic women, although somewhat meeting the challenge,
do not match their Republican opponents in negativity. Instead, the Democratic Party attempts
to compensate for this by sponsoring higher proportions of negative advertising on behalf of their
female candidates running against female opponents. From a methodological standpoint, it is
particularly striking that the significant results come from this subgroup, as they represent, by
far, the smallest subgroup in the data (n=30).
8/7/2019 If you can say something nice.
http://slidepdf.com/reader/full/if-you-can-say-something-nice 28/39
8/7/2019 If you can say something nice.
http://slidepdf.com/reader/full/if-you-can-say-something-nice 29/39
8/7/2019 If you can say something nice.
http://slidepdf.com/reader/full/if-you-can-say-something-nice 30/39
30
campaigns are undertaken. A better understanding of how and when campaigns are gendered
will provide one more piece to the puzzle of women’s electoral underrepresentation.
8/7/2019 If you can say something nice.
http://slidepdf.com/reader/full/if-you-can-say-something-nice 31/39
31
References
Alexander, Deborah and Kristi Andersen. 1993. “Gender as a Factor in the Attributions of Leadership Traits.” Political Research Quarterly 46: 527-45.
Ansolabahere, Stephen and Shanto Iyengar. 1995.Going Negative: How PoliticalAdvertisements Shrink and Polarize the Electorate. New York: The Free Press.
Brooks, Deborah Jordan. 2006. “The Resilient Voter: Moving Toward Closure in the Debatecaover Negative Campaigning and Turnout.” Journal of Politics 68: 684-696.
Burrell, Barbara. 1994.A Woman’s Place Is in the House: Campaigning for Congress in the
Feminist Era. Ann Arbor, MI: University of Michigan Press.
Bystrom, Dianne and Lynda Lee Kaid. 2002. “Are Women Candidates Transforming CampaignCommunications? A Comparison of Advertising Videostyles in the 1990s.” In C.S.
Rosenthal (Ed.), Women Transforming Congress (pp 146-169). Norman: University of Oklahoma Press.
Center for American Women and Politics (CAWP). 2008. “Women in the U.S. Congress 2008.”Eagleton Institute of Politics, Rutgers University.
http://www.cawp.rutgers.edu/fast_facts/levels_of_office/documents/cong.pdf
Center for American Women and Politics (CAWP). 2007a. “Women in the U.S. Congress2007.” National Information Bank on Women in Public Office. Eagleton Institute of
Politics, Rutgers University.
Center for American Women and Politics (CAWP). 2007b. “Statewide Elective ExecutiveWomen 2007.” National Information Bank on Women in Public Office. Eagleton
Institute of Politics, Rutgers University.
Center for American Women and Politics. 2004a. “Numbers of Women in Office: StateSummaries.” Eagleton Institute of Politics, Rutgers University.
http://www.cawp.rutgers.edu/Facts/StLegHistory/stleghist.pdf .
Center for American Women and Politics. 2004b. “Summary of Candidates for Selected Offices1970-2006 (Major Party Nominees)” Eagleton Institute of Politics, Rutgers University.
http://www.cawp.rutgers.edu/fast_facts/levels_of_office/documents/can_histsum.pdf
Center for Responsive Politics. 2008.www.opensecrets.org. Accessed during July 2008.
Cialdini, Robert and M. R. Trost. 1998. “Social Influence: Social Norms, Conformity, andCompliance.” In D. T. Gilbert, S. T. Fiske, and G. Linzey (Eds.),The Handbook of
Social Psychology, Boston: McGraw-Hill.
Druckman, James N., Martin J. Kifer, and Michael Parkin. 2007. “Going Negative in a NewMedia Age: Congressional Campaign Websites, 2002-2006.” Prepared for delivery at
8/7/2019 If you can say something nice.
http://slidepdf.com/reader/full/if-you-can-say-something-nice 32/39
32
the 2007 annual meeting of the Midwest Political Science Association, Chicago, IL,April 12-14.
Eagley, Alice H. and Steven J. Karau. 2002. “Role Congruity Theory of Prejudice Toward
Female Leaders.” Psychological Review 109: 573-98.
Eagly, Alice H., Mona G. Makhijani, and Bruce G. Klonsky. 1992. “Gender and the evaluationof leaders: A meta-analysis.” Psychological Bulletin 111: 3-22.
Elazar, Daniel. 1966.American Federalism: A View from the States. New York: Thomas Y.
Crowell.
Falk, Erika and Kate Kenski. 2006. “Issue Saliency and Gender Stereotypes: Support forWomen as Presidents in Times of War and Terrorism.” Social Science Quarterly 87: 1-
18.
Giroux, Gregory L. and Sandra Basu. 2000. “House Control Teeters.”Congressional QuarterlyWeekly Report, October 21, page 2456.
Giroux, Gregory L. 2002. “House Races: An Election Bucking Traditions.”Congressional
Quarterly Weekly Report , October 26, 2002, page 2794.
Geer, John. 2006.In Defense of Negativity: Attack Ads in Presidential Campaigns. Chicago:University of Chicago Press.
Goldstein, Kenneth, Michael Franz, and Travis Ridout. 2002. “Political Advertising in 2000.”Combined File [dataset]. Final release. Madison, WI: The Department of Political
Science at The University of Wisconsin-Madison and the The Brennan Center for Justiceat New York University.
Goldstein, Kenneth, and Joel Rivlin. 2005. “Political Advertising in 2002.” Combined File
[dataset]. Final release. Madison, WI: The Wisconsin Advertising Project, TheDepartment of Political Science at The University of Wisconsin-Madison.
Goldstein, Kenneth, and Joel Rivlin. 2007. ““Congressional and gubernatorial advertising, 2003-
2004” Combined File [dataset]. Final release. Madison, WI: The University of WisconsinAdvertising Project, The Department of Political Science at The University of
Wisconsin-Madison.
Goodman, Ellen. 2006. “In Today’s Politics, Nice Gals Finish Last.”Boston Globe November3, 2006 A17.
“GOP Confident of Holding the House.” 2004.Congressional Quarterly Weekly, October 23,
2004, page 2508.
8/7/2019 If you can say something nice.
http://slidepdf.com/reader/full/if-you-can-say-something-nice 33/39
33
Gordon, Craig and Tom Brune. 2008. “Memos Show Clinton Turmoil.”Newsday August 13,2008. Accessed at http://www.newsday.com/news/nationworld/nation/ny-
uscamp135799024aug13,0,3877096.story.
Hill, David. 1981. “Political Culture and Female Political Representation.”Journal of Politics
43: 159-168.
Hitchon, Jacqueline C., and Chingching Chang. 1995. “Effects of Gender Schematic Processing
on the Reception of Political Commercials for Men and Women Candidates.”Communication Research 22: 430-58.
Huddy, Leonie, and Nayda Terkildsen. 1993. “Gender Stereotypes and the Perception of Male
and Female Candidates.” American Journal of Political Science 37: 119-47.
Jalonick, Mary Clare. 2002. “Senate Races: Melting in the Middle.”Congressional Quarterly
Weekly Report, October 26, 2002, page 2790.
Kahn, Kim Fridkin and Patrick Kenney. 1999.The Spectacle of U.S. Senate Campaigns.
Princeton, NJ: Princeton University Press.
Kirkpatrick, David D. 2006. “Race: Amy Klobuchar/Mark Kennedy.”New York Times
September 19, 2006, A5.
Koch, Jeffrey W. 2000. “Do Citizens Apply Gender Stereotypes to Infer Candidates’
Ideological Orientations?” Journal of Politics 62: 414-29.
Langer, Gary. 2008. “Exit Polls: Negative Campaign Tarnishes Clinton, Obama.”ABC News
April 22, 2008. Accessed at
http://abcnews.go.com/PollingUnit/Politics/Story?id=4703379&page=1.
Lau, Richard R. and Gerald M. Pomper. 2001. “Negative Campaigning by U.S. SenateCandidates.” Party Politics 7: 69-87.
Lau, Richard R., Lee Sigelman, Caroline Heldman, and Paul Babbitt. 1999. “The Effects of
Negative Political Advertisements: A Meta-Analytic Assessment.” American Political
Science Review 93: 851-75.
Lawless, Jennifer L. 2004. “Women, War, and Winning Elections: Gender Stereotyping in thePost-September 11
thEra.” Political Research Quarterly 57: 479-90.
McDermott, Monika I. 1997. “Voting Cues in Low-Information Elections: Candidate Gender as
a Social Information Variable in Contemporary US Elections.” American Journal of
Political Science 41: 270-83.
McDermott, Monika I. 1998. “Race and Gender Cues in Low-Information Elections.”Political
Research Quarterly 51: 895-918.
8/7/2019 If you can say something nice.
http://slidepdf.com/reader/full/if-you-can-say-something-nice 34/39
34
Newton-Small, Jay. 2007. “Will Clinton’s Obama Attacks Backfire?”Time Magazine
December 4, 2007. Accessed at
http://www.time.com/time/politics/article/0,8599,1690519,00.html.
Page, Susan. 2006. “Fall TV Lineup Filled with Political Spots; A Day’s Viewing in One State
Makes Clear that Broadcast Ads Remain ‘the Weapon of Choice’.” USA Today October19, 2006.
Petrocik, John R. 1996. “Issue Ownership in Presidential Elections, with a 1980 Case Study.”American Journal of Political Science 40: 825-850.
Pierce, Emily and Barbara Murray. 2000. “GOP Likely to Hold Senate.”Congressional
Quarterly Weekly Report , October 21, page 2462.
Powers, Rebecca S., J. Jill Suitor, Susana Guerra, Monisa Shackelford, Dorothy Mecom, andKim Gusman. 2003. “Regional Differences in Gender-Role Attitudes: Variations by
Gender and Race.” Gender Issues 21: 40-54.
Rahn, Wendy. 1993. “The Role of Partisan Stereotypes in Information Processing aboutPolitical Candidates.” American Journal of Political Science 37: 472-496.
Rudman, Laurie and Peter Glick. 1999. “Feminized Management and Backlash Toward Agentic
Women: The Hidden Costs of a Kinder, Gentler Image of Middle Managers.” Journal of
Personality and Social Psychology 77: 1004-1010.
Sapiro, Virginia and Katherine Cramer Walsh. 2002. “Doing Gender in Congressional
Campaign Advertisments.” Prepared for delivery at the Annual Meeting of theInternational Society for Political Psychology, Berlin, July, 2002.
Scammon, Richard M., Alice V. McGillivray, and Rhodes Cook. 2001.America Votes #24.
Washington, DC: CQ Press.
Scammon, Richard M., Alice V. McGillivray, and Rhodes Cook. 2003.America Votes #25.
Washington, DC: CQ Press.
Scammon, Richard M., Alice V. McGillivray, and Rhodes Cook. 2006.America Votes #26.
Washington, DC: CQ Press.
Sides, John. 2006. “The Origins of Campaign Agendas.”British Journal of Political Science
36: 407-436.
University of Wisconsin Advertising Project. 2008. “An Overview of the Wisconsin
Advertising Project.” Accessed at wisadproject.wisc.edu.
8/7/2019 If you can say something nice.
http://slidepdf.com/reader/full/if-you-can-say-something-nice 35/39
8/7/2019 If you can say something nice.
http://slidepdf.com/reader/full/if-you-can-say-something-nice 36/39
36
TABLES
Table 1 – Distribution of the Gender Context of Races in 2000, 2002, and 2004 of those
Candidates Running TV Advertisements
Table 2 – Distribution of Candidates in 2000, 2002, and 2004, by Gender and Candidate
Status (percentage) of those Candidates Running TV Advertisements
Women Men
Incumbent 32% 49%
Challenger 40% 27%
Open Seat 28% 24%Total 100% 100%
Table 3 – Distribution of Candidates in 2000, 2002, and 2004, by Gender and Race
Competitiveness (percentages) of those Candidates Running TV Advertisements
Women Men
Competitive 59% 52%
Non-Competitive 41% 48%Total 100% 100%
Men Women
Men running against… 560 135
Women running against… 127 30
8/7/2019 If you can say something nice.
http://slidepdf.com/reader/full/if-you-can-say-something-nice 37/39
37
Table 4 – Bivariate Comparisons of Mean Levels of Negativity
Note: Between each set of bold lines, matching superscript letters indicate that the two associated means are
statistically significantly different from one another at the .05 level.
Mean Level of
Negativity,Candidate-Sponsored
Mean Level of
Negativity, Party-Sponsored
Overall .34(.01)
.79(.02)
Men .33a
(.01)
.77
(.02)
Women .39 a
(.03).82(.04)
Man versus Man .33 a
(.01).77 a
(.02)
Man versus Woman .31b
(.03)
.80
(.04)
Woman versus Man .41 a b
(.03)
.85 a
(.04)Woman versus Woman .31
(.06)
.71
(.10)
Incumbent .21 a b
(.01)
.74
(.04)
Challenger .49 b c
(.02).83(.03)
Open Seat .39 a c
(.02).79(.02)
Competitive .41a
(.03)
.67
(.06)Non-Competitive .20a
(.02)
.76
(.12)
Democrat .34
(.02)
.75a
(.03)
Republican .34(.02)
.82a
(.02)
Percentage of State Legislature, Female (AboveMean)
.33(.02)
.79(.02)
Percentage of State Legislature Female (Mean and
Below)
.35
(.02)
.78
(.03)
Year 2000 .36 a
(.02).84 a
(.03)
Year 2002(compared to 2000 and 2004)
.29 a b
(.02).68 a b
(.03)
Year 2004 .37 b
(.02).88 b
(.03)
8/7/2019 If you can say something nice.
http://slidepdf.com/reader/full/if-you-can-say-something-nice 38/39
38
Table 5 – OLS Regression Results
(DV is the proportion of televised advertisements that are negative)
Model 1 –
Candidate-
Sponsored Ads
as DV
Model 2 –
Party-
Sponsored Ads
as DV
Model 3 –
Candidate-
Sponsored Ads
as DV
Model 4 –
Party-
Sponsored Ads
as DV
Man versus Woman -.02
(.03)
.02
(.05)
.15
(.10)
.12
(.17)
Woman versus Man .02
(.03)
.07
(.05)
.10
(.10)
.17
(.17)
Woman versus Woman -.04(.06)
-.06(.09)
.42**(.21)
.18(.33)
Open Seat Race .11**(.03)
.05(.04)
.10**(.03)
.05(.04)
Challenger .25**(.02)
.09**(.05)
.25**(.02)
.11**(.05)
Competitive Race .24**(.02)
.07(.09)
.24**(.02)
.06(.09)
Democrat -.05**(.02)
-.09**(.03)
-.04(.03)
-.06(.04)
Expenditures (in thousands) .02(.3)
-.09(.4)
.02(.3)
-.05(.4)
2000 .08**(.03)
.16**(.04)
.07**(.03)
.15**(.04)
2004 .10**
(.02)
.19**
(.04)
.10**
(.02)
.18**
(.04)
Percentage of StateLegislature, Female -.001(.001) .0004(.002) .0008(.002) .001(.003)
Perc of F State Legislators
X Woman versus Woman
-.01**
(.007)
-.02
(.01)
Perc of F State Legislators
X Man versus Woman
-.007*
(.004)
-.002
(.01)
Perc of F State Legislators
X Woman versus Man
-.003
(.004)
-.001
(.01)
Party X Woman versus
Woman
-.19*
(.11)
.29*
(.18)
Party X Man versus Woman -.02
(.06)
-.11
(.1)Party X Woman versus Man -.02
(.06)-.14(.09)
Constant .106**(.04)
.59**(.11)
.06(.05)
.57**(.11)
Note: **=p<.05, *=p<.1
8/7/2019 If you can say something nice.
http://slidepdf.com/reader/full/if-you-can-say-something-nice 39/39