Incumbency has long been recognized as one of the primary barriers
limiting the election of women to the U.S. House of Representatives (see,
for example, Andersen and Thorson 1984; Burrell
1988, 1994; Carroll 1994; Darcy and Choike
1986; Darcy, Welch, and Clark 1994; Palmer and Simon 2001; Seltzer,
Newman, and Leighton 1997; Welch and Studlar
1996). Once candidates win an election and become members of the
House, they accrue considerable advantages, such as name recognition, use
of the franking privilege, and wider access to campaign funds.
Consequently, incumbents are virtually assured of reelection (see, for
example, Herrnson 1998; Jacobson 1997; Ornstein, Mann and
Malbin 1998). Thus, women have a hard time winning seats in the
House not because they are women but because of incumbency, and most
incumbents are men.
For this reason, open seats are seen as one of the principle avenues
of opportunity for women to increase their numbers. For example, the
dramatic increase in the number of women in Congress in 1992, “The
Year of the Woman,” is largely attributed to the unusually high
number of open seats in that election cycle (Burrell
1994; Carroll 1994; Chaney and Sinclair 1994; Cook and
Wilcox 1995; Duerst-Lahti 1998; Gaddie and Bullock 1995; Wilcox
1994; see also Berch 1996). In addition,
a substantial subset of analyses within the literature uses open seat
races as the unit of analysis to study the success of women candidates
(Bernstein 1997; Burrell
1992; Gaddie and Bullock 1995; Green 1998; Herrick 1995;
Hoffman, Palmer and Gaddie 2001).
While the focus on open seats is understandable and appropriate, there
are relatively few of these opportunities in a given election cycle. On
average, just over 9% (40 seats) of all House races are open, but it has
been as low as 6% (27 seats) in recent elections (Palmer and Simon 2001, 66).1
The count of open seats includes those vacated by an incumbent
and new seats created by the process of reapportionment and
redistricting.
Even in the peak year of 1992, only 22% (96) of
House seats were open. In 2002, the most recent redistricting year, 11.5%
(50) of the seats fell into this category. The downside to the emphasis on
open seats is that a majority of districts—those with incumbents
seeking reelection—have received considerably less attention (but
see
Bernstein 1997;
Fox
1997). As a result, a great deal of what goes on in House
elections, including the participation of female candidates and where they
choose to run, has not been examined very extensively. In particular,
there is little research on the question of whether the presence of a
female incumbent influences the competitive environment in a congressional
district.
Our research focuses on electoral competition in districts where
incumbents are running for reelection and, in particular, districts with
female incumbents. Using the literature on the impact of sex-role
stereotypes in campaigns, we formulate two hypotheses. The first, the
Competition Hypothesis, asserts that compared to their male counterparts,
female incumbents will face a more competitive electoral environment in
their districts. The second, the Gender Effect Hypothesis, predicts that a
female incumbent seeking reelection will induce other women to enter the
race. Using all elections to the U.S. House of Representatives from 1956
through 2002, we develop seven indicators of the competitive environment
to test these hypotheses. The data provide considerable support for each
hypothesis and lead us to conclude that female incumbents have a
“hidden influence” in House elections: Female incumbents are
more likely to face competition than their male counterparts and are more
likely to be challenged by other women.
EQUAL, BUT DIFFERENT
One reason for the relative inattention given to female incumbents is
the evidence that women candidates have reached parity with men in terms
of electoral success rates. Women no longer run as “sacrificial
lambs” in races that they have no hope of winning any more often
than men do (Darcy, Welch, and Clark 1994; Deber 1982; Fox 1997; Gertzog and Simard 1991; Lamson
1968; Seligman 1961). When women run for
office, they win at rates comparable to men (Burrell 1992, 1994; Darcy and Schramm 1977; Darcy,
Welch, and Clark 1994; Fox and Oxley
2003; Gaddie and Bullock 1995; Palmer and Simon 2001; Seltzer,
Newman and Leighton 1997; Welch et al.
1985; but see Fox, Lawless, and Feeley
2001). Female incumbents are reelected at the same rates as men
(Palmer and Simon 2001; Seltzer, Newman, and Leighton 1997). Female
challengers are as likely to win (or lose) as male challengers (Seltzer, Newman, and Leighton 1997). Women candidates
are as effective at raising money as men, and in some election cycles
have, on average, done slightly better than men, particularly in primaries
(Francia 2001; Gaddie and
Bullock 1995; Herrick 1996; Seltzer, Newman, and Leighton 1997; Theilmann and Wilhite 1991; Uhlaner
and Schlozman 1986).
Despite this parity, studies consistently demonstrate that campaigns
with women candidates are fundamentally different from those where only
men compete for nomination and election. In particular, there are
differences in 1) how the media cover a campaign that includes a female
candidate, 2) how voters perceive and evaluate male and female candidates,
and 3) how candidates formulate campaign strategy in light of the
stereotypes present in media coverage and voter perceptions. As a whole,
this research suggests that the road to elective office is more hazardous
for women than for men.
Media coverage
Campaign reporting has been found to vary substantially depending on
the sex of the candidate. With respect to the amount of coverage, there is
little research on House elections, although studies of campaigns for U.S.
Senate show that women receive less media coverage than their male
counterparts (Kahn and Goldenberg 1991; see also
Kropf and Boiney 2001, Norris
1997). Beyond this, it is usually the content of coverage that
makes the media “the bane of the political woman's
existence” (Witt, Paget, and Matthews
1995, 184). Reporting tends to reinforce sex-role stereotypes and
traditional attitudes about women's roles, particularly in campaigns
for higher-level offices (Niven and Zilber 2001;
Witt, Paget and Matthews 1995; see also Clift and Brazaitis 2000; Kahn 1995, 1996; Schroeder 1999). Women candidates have long complained
about the “soft news” focus in which their wardrobe,
hairstyles, femininity, and family relationships receive more emphasis
than their political experience or issue positions (Braden 1996; Fox 1997; Jamieson 1995; Niven and Zilber
2001).2
Witt, Paget, and Matthews
(1995, 181) provide numerous examples. Consider
how the New York Times compared Carol Moseley-Braun (D) and
Richard Williamson (R), the Illinois nominees for the Senate in 1992:
“She is commanding and ebullient, a den mother with a
cheerleader's smile; he, by comparison, is all business, like the
corporate lawyer he is” (181). It should be noted that Moseley-Braun
was also a lawyer and had served as a U.S. attorney.
The tendency
to “discuss political women in intimate, almost Playboy-like
detail,” according to Witt, Paget, and Matthews, “is an
artifact of this country's age-old but unresolved debate over the
women citizens' proper roles versus ‘proper women's’
place” (
1995, 182).
Gender stereotypes
The manner in which the media frames campaigns involving female
candidates serves to reinforce gender stereotypes held by voters. A
substantial body of research demonstrates that male and female candidates
are perceived as having different leadership traits and different
abilities or competence in handling issues. Women are viewed as being more
compassionate, trustworthy, and willing to compromise. Men are seen as
more assertive, aggressive and self-confident (Alexander and Andersen 1993; Brown,
Heighberger, and Shocket 1993; Burrell
1994; Huddy and Terkildsen 1993a, 1993b; Leeper 1991; Niven 1998; Sanbonmatsu
2002a). In addition to personality traits, there are perceived
differences in “issue ownership,” the issues on which men and
women are viewed as more competent (Petrocik
1996; see also Koch 2002). Women
candidates are typically seen as more competent on issues such as
education, health care, rights issues, the environment, and welfare,
whereas men are seen as more competent on issues such as taxes, budgets,
crime, national defense, and foreign policy (Alexander
and Andersen 1993; Carroll 1994; Delli Carpini and Fuchs 1993; Dolan
2001; Fox and Oxley 2003; Huddy and
Terkildsen 1993a, 1993b; Kaufman and Petrocik
1999; Leeper 1991; Sanbonmatsu 2002a; Shapiro and
Mahajan 1986; but see Dolan 2004).
Finally, voter perceptions of a particular candidate's ideology
are strongly related to the gender of that candidate. Compassion issues
such as education, health care, and welfare are largely associated with
the Democratic Party and liberal policy positions. In contrast, the
Republican Party is generally considered more competent to deal with
issues like taxes, national defense, and crime (Petrocik 1996). These general party associations
interact with gender. Female Democrats are perceived as more liberal than
they actually are, and female Republicans are perceived as less
conservative than they actually are (Alexander and
Andersen 1993; Huddy and Terkildsen
1993b; Koch 2000).
Thus, like a political party label, the sex of the candidate, which
can usually be inferred from the name on the ballot, acts as a cue for
voters; it provides a great deal of information associated with gender
expectations (Atkeson 2003; Koch 2000, 2002; McDermott 1997, 1998; Sanbonmatsu 2002a; but see Thompson
and Steckenrider 1997): “By knowing this one bit of
information, respondents … make inferences about a candidate's
issue positions, policy competencies, ideological leanings, and character
traits” (Koch 2000, 414). It provides a
shortcut to help “estimate the views of candidates”
(McDermott 1997, 271). Because women running for
office, especially statewide office, is still a rare event, voters are
more likely to rely on sex as a cue (Koch 2002,
460; see also Atkeson 2003). These cues are
especially salient when women are a “contextual novelty,” such
as in a primary election with a woman candidate running against several
male competitors (Koch 2002, 455).
Campaign strategy
All of this suggests that successful female candidates must adapt
their campaign strategies to account for gender stereotypes about traits,
issue competence, and ideology, as well as the media coverage that
reinforces these stereotypes (see for example Witt,
Paget, and Matthews 1995). In essence, women face particular
challenges in their “presentation of self” (see Fenno 1977, 898). This is the fundamental act of
campaigning in which candidates place themselves in the “immediate
physical presence of others” and “make a presentation of
themselves” (Fenno 1977, 898). The
presentation of self is both verbal and nonverbal (Fenno 1977, 898). The nonverbal is critical,
particularly for women, since it may enhance or undermine the credibility
given to verbal presentations and the level of trust that audiences place
in the candidate.3
An excellent example of
this is when Lynn Yeakel, the Democratic nominee for a Senate seat from
Pennsylvania in 1992, was described in the Washington Post as
“a feisty and feminine fifty-year-old with the unmistakable Dorothy
Hamill wedge of gray hair” (26 August 1992, sec. C). In one
sentence, the reporter was able to link Yeakel's credibility as a
candidate with her haircut.
It is from this presentation of self,
for example, that voters draw inferences about the leadership traits of
candidates. For the woman who seeks elective office, the challenge is to
“craft a message and a public persona” establishing that
“she can be as clear and independent a decision maker as any man,
but more caring and trustworthy” (
Witt, Paget and
Matthews 1995, 214). As one political consultant explained, in
appearing before the public, women candidates “‘can't
afford … not to be nice, [or they will] immediately be
branded as a bitch. But the men won't play by tea party
rules’” (quoted from
Witt, Paget, and
Matthews 1995, 214). “[T]he woman candidate has to
maintain some level of the traditional altruistic and apolitical
above-it-all demeanor expected of a lady, all the while beating her
opponents in what sometimes seems the closest thing to blood sport that is
still legal” (
Witt, Paget, and Matthews
1995, 214).
Women candidates must also account for the “political
mood” or temper of the times, both nationally and locally, in
formulating their campaign message and issue agendas. There are two
important ways political mood can affect women's success. The first
pertains to the problems and issues deemed most important by their
constituency and the degree to which these concerns mesh with voter
perceptions of issue competency. If the focus rests on compassion issues,
female candidates will be advantaged. In such circumstances, when women
candidates use sex-role expectations to their advantage, run on compassion
issues, and target women voters, they are substantially more likely to win
(Herrnson, Lay, and Stokes 2003; Iyengar et al. 1997; Sanbonmatsu
2002a; Williams 1998; Witt, Paget, and Matthews 1995). To the extent that
the political mood and agenda focus on foreign and defense policy or
budgets and economic policy, women must formulate strategies to weaken the
stereotypes and establish perceptions of issue competency on these
traditional male issues. Second, there are times when the political mood
is especially restive toward politics and officeholders. Women can take
advantage of being perceived as “outsiders” and as more honest
during election cycles when events and scandals call into the question the
trustworthiness of politicians (Burrell 1994;
Kahn 1996; Sanbonmatsu
2002a).
The election of 1992 provides an example where the temper of the times
favored female candidates. The nation had turned from the first Gulf War
to focus on domestic concerns;4
According to
a Gallup poll released on 17 September 1992, 27% of the issues and
concerns mentioned by respondents when asked about the most important
problem facing the country were “compassion issues” (poverty,
homelessness, health care, education, drug abuse, environment, abortion,
and AIDS). Additionally, 7% of the responses cited dissatisfaction with
government, ethics, and moral decline, and 15% mentioned unemployment and
jobs. See the annual volume, The Gallup Poll (1992, 160). Because Gallup permits multiple replies to
the “most important problem” question, the percentages we
report are based upon the total number of responses, rather than the total
number of respondents.
the controversial nomination of Clarence
Thomas triggered widespread discussions of sexual harassment; and a
scandal involving checks drawn on the House bank not only led numerous
incumbents to retire but also heightened the advantage enjoyed by female
candidates as honest outsiders. Many women ran “as women” and
were successful, such as Patty Murray, who ran as the “mom in tennis
shoes” (Dolan
1998,
2001;
Kahn 1996;
Witt, Paget, and Matthews 1995). An exit poll that
year indicated that voters actually preferred female candidates to male
candidates (
Cook 1998, 59).
A final consideration in the formulation of strategy involves the
character of the constituency. Research demonstrates that the ideology of
a constituency is related to its demographic character and that there is
considerable variation in the ideological leanings of congressional
districts (see Ardoin and Garand 2003).5
Ardoin and Garand (2003) create simulated ideology scores that are an
econometric function of the following district-level characteristics: home
ownership, blue-collar workers, urban population, location in the Deep
South, and the Democratic vote for president.
Thus, particular
districts, because of their demographic characteristics, are more
receptive to an agenda emphasizing compassion issues (see
Koch 2000).
6 While not
extensive, there is also research that examines the demographic
characteristics of districts in which women run (Burrell, 1994; Darcy and Schramm, 1977; Welch and Studlar
1996).
With respect to the perceived ideology of
candidates, we know that voters are more likely to use stereotypes in
evaluating female candidates; women of both parties are perceived as more
liberal than their male counterparts (
Koch 2000;
Sanbonmatsu 2002a). Unlike their male
counterparts, women must work to neutralize these ideological stereotypes
if they seek election in districts where those stereotypes are a detriment
to electoral success.
Implications for the competitive environment
Two important implications emerge from the research on leadership
traits, issue competence, ideology, and campaign strategy. The first
pertains to electoral competition. The research suggests that female
candidates, including female incumbents, will be perceived as more
vulnerable, in the electoral sense, than male candidates. Perceived
vulnerability flows from the gender stereotypes and media coverage that
reinforce those stereotypes. In addition, despite the increasing presence
of women in the electoral arena, a female nominee or incumbent remains a
novelty, and the research demonstrates that reliance on stereotypes is
stronger in exactly these circumstances (Koch, 2002).7
In the five
election cycles from 1992 to 2000, for example, a woman won a Democratic
nomination for the House in 176 districts (40.5%); voters in 259 districts
(59.5%) saw no female Democratic nominee in that 10-year period. The
nomination of a female Republican is even more of a rare event. For the
same period, there were female Republican nominees in 113 districts
(26.0%), with no female nominee in 322 districts (74.0%).
As much as a candidate's gender serves as a cue for voters, it
can serve as a cue for potential opponents. As Sharon Rodine (1990), former president of the National Women's
Political Caucus observed, “We know that women are targeted more
often.” Male candidates typically reformulate their campaign
strategies when they run against women, and many plan campaign activities
that target women voters (Fox 1997, 49). They
are often advised to “steal the rainbow,” by quickly and
specifically raising women's issues or compassion issues in order to
“[b]eat your opponent to her strongest issue”
(quoted by Witt, Paget, and Matthews 1995, 226;
see also Chaney 1998; Fox
1997).8
The quotation is drawn from an
article entitled “How to Defeat Women and Blacks” written by
David Beiler and published in Campaigns and Elections (August-September 1990), a widely read trade magazine.
The next issue of the magazine (October-November 1990) featured a piece by
Sharon Rodine entitled “How to Beat Bubba.”
If women
“run as women,” this can be used against them. If a woman
candidate builds her campaign around stereotypes in order to win, this may
actually limit her strategic choices and the types of responses she can
use effectively (
Witt, Paget, and Matthews 1995,
226). Thus, even though women win elective office as often as men do,
women candidates, especially incumbents, may be initially
perceived as easier to defeat and may face a more competitive
environment.
The second implication is that this increased competition will not be
gender neutral. Women candidates may also foster competition in a
different way, as a role model for other women. For example, in states
with competitive female candidates, women citizens are more likely to
discuss politics, have higher levels of political knowledge, and are more
likely to feel politically efficacious; viable women candidates
“represent symbolic and substantive cues to women citizens that
increase their political engagement” (Atkeson
2003, 1042; see also Hansen 1997). The
logical extension of this is that successful women candidates inspire
other women to run.
Beyond the role-model effect, running against another woman is a
strategic decision (see for, example, Cooperman and
Oppenheimer 2001; Palmer and Simon 2001,
2003; Rule 1981).
Against the backdrop of gender stereotypes, it is important to consider
what the success of a woman winning a House seat signifies. It
demonstrates that the woman was able to neutralize the stereotypes or make
them work to her advantage. Her victory serves as a cue signaling that a
woman can overcome the hurdles and compete successfully in that
district.
As with voters, stereotypes may work in a number of ways to stimulate
competition. The gender stereotypes and the novelty of female candidates
may suggest vulnerability. Female candidates as role models may inspire
more women to run. A female incumbent may provide a “strategic
signal” about the probability of winning a district. Accordingly, we
suggest the following:
Competition Hypothesis: Women incumbents will face more
competition for their House seats than their male counterparts.
Gender Effect Hypothesis: Women incumbents will face more
competition for their House seats from female candidates than from their
male counterparts.
DATA
To evaluate these hypotheses, we have developed a data set that
includes all elections to the U.S. House of Representatives from 1956
through 2002. Our major source is the series America Votes. The unit of analysis is the
congressional district. For each district in each election year, we
gathered the following information: the number of female candidates
running for the Democratic and Republican nominations, respectively, the
total number of candidates seeking each party's nomination, whether a
woman won the Democratic or Republican nomination, and the outcome of the
general election.9
In gathering these data,
there are several special cases. The states of Connecticut, Utah, and
Virginia employ a mixed system of conventions and primaries to nominate
their congressional candidates. The nominating conventions are held first,
with primaries scheduled only if there is a significant challenge to the
designated convention nominee. In instances where there is no primary, we
coded the gender of the nominees only because the number of candidates
seeking the nomination at the convention is unknown.
Louisiana is yet another special case. The state employs an open
primary system in which candidates, regardless of party, run in a single
primary. If a candidate wins an absolute majority of the primary vote, the
candidate is elected to the House and there is no general election. For
Louisiana, we coded the number of Democrats and Republicans (women and
total) running in the initial primary. In instances where there was a
general election, we followed the same conventions used with other states,
noting, of course, instances in which the general election involved two
candidates from the same party.
Finally, there are states that have a primary runoff system. In these
states, a candidate must win an absolute majority of the primary vote to
obtain the party nomination. If no candidate wins a majority, there is a
runoff primary between the top two finishers. The winner of the primary
then becomes the party nominee. Our coding records the number of
candidates (women and total) in the initial primary and the gender of the
ultimate nominees.
For each district, we also recorded the party
and sex of the incumbent, whether the incumbent was seeking reelection,
and the incumbent's share of the two-party vote in the prior
election.
Identifying the sex of candidates was done by first examining the
names listed in each district for each primary and general election. In
the case of questionable names (e.g., Pat Lear), we consulted relevant
editions of the Almanac of American Politics and
Congressional Quarterly Weekly Report. Quite often, the coverage
in these sources provides information about the sex of the party nominees.
Second, for the more recent period (approximately 1974 forward), we
conducted a Nexis search of newspaper coverage. In almost every
case, we were able to find media coverage that revealed the sex of these
candidates. Finally, if our methods of searching did not provide any
information, the name was excluded from our count of candidates. The total
number of exclusions was less than 2% of all candidate names examined. In
applying these procedures to electoral data from 1956 through 2002, we
coded 10,431 House elections involving more than 32,000 candidate
names.
ANALYSIS
On the surface, electoral outcomes for the U.S. House do indicate
parity between male and female candidates. Among incumbents, 94.8% of men
(8,330 of 8,783) running in the general election were reelected; the
comparable figure for female incumbents was 95.3% (506 of 531). Across all
elections, the average share of the two-party vote won by male incumbents
was 70.2% and 70.1% for female incumbents. In addition, only 1.2% (107 of
8,890) of male incumbents and 0.9% (5 of 536) of female incumbents lost
their bids for renomination.10
A difference
in means test on the average two-party vote won by male and female
incumbents is insignificant (t = 0.240); similarly, there are no
significant male-female differences between the proportions of males and
females winning their general election race (t = −0.246) or losing
in the party primary (t = 0.562).
There are, however, additional
aspects of the competitive environment that suggest something other than
gender parity. One indicator is whether an incumbent faces any opposition
in the campaign to retain the seat. Here, there are three possibilities:
no opponent in the primary, no major party opponent in the general
election, and the “free pass” in which the incumbent has no
opposition in both the primary and general elections. Our Competition
Hypothesis leads to the expectation that the frequency of uncontested
races will vary with the gender of the incumbent.
Table
1 shows, for male and female incumbents, the proportions of
uncontested general elections, contested primaries, and “free
passes.” The first set of rows presents a systemic or macro view and
includes all incumbents; the second and third sets of rows present the
results for Democrat and Republican incumbents respectively.
Uncontested general and primary election races among incumbent
candidates for the U.S. House of Representatives,
1956–2002
The table reveals that female incumbents are less likely to enjoy the
luxury of having no opponent. In districts where women stand for
reelection, there are fewer uncontested general elections (8.8% versus
16.3% for males), more contested primaries (34.0% versus 29.7% for males)
and fewer instances of no competition in the entire election cycle (5.9%
versus 12.1% for males). The differences between the male and female
proportions reported in the first row of the table are statistically
significant according to the standard difference in proportions test.
The table also reveals that there are partisan differences. Among
Democratic incumbents, women are substantially more likely to face major
party opposition; the proportion of uncontested general elections (9.7%)
is less than half the rate for men (21.3%). A similar result holds for the
“free pass.” Only 5.7% of Democratic female incumbents avoid
competition throughout the election cycle, compared to a rate of 14.8% for
Democratic men. Male and female incumbents do face primary challenges at
roughly comparable rates. The pattern is different among Republicans.
While women experience fewer uncontested races in the general election and
“free passes,” the differences are not as sizable as those for
the Democrats. There is, however, a significant difference in the case of
contested primary elections. Republican female incumbents are more likely
to be challenged for renomination than their male counterparts (27.3%
versus 19.7% for males).11
Note that, in
general, Democrat incumbents face a more contentious primary arena than
their Republican counterparts. Democrat incumbents face primary challenges
in 36.8% of the cases, whereas the rate for Republicans is
20.1%.
The relationship between the gender of the incumbent and contested
elections should not be confined to intraparty competition. If, as we
hypothesize, female incumbents are perceived as vulnerable, then it is
expected that there will be greater competition for the nomination within
the opposition party as well. Table 2 shows, for
male and female incumbents, the proportion of contested primary elections
within the opposition party. Competition for the opposition party
nomination is significantly greater in districts with a female incumbent.
This holds across all incumbents and within each party as well. When a
female Republican holds the House seat, 54.3% of the Democratic nominees
are chosen in contested primaries, compared to 46.3% when the incumbent is
a Republican male. Similarly, when a female Democrat is the incumbent,
46.0% of Republican primaries are contested, compared to 38.7% in
districts where the Democrat incumbent is male. Together, Tables 1 and 2 provide support for
our Competition Hypothesis. Women incumbents are associated with a more
competitive electoral environment; they face proportionally more contested
races than their male colleagues and, at the same time, foster more
contested primary races within the opposition party.
Contested primary races for the U.S. House of Representatives within
the opposition party, 1956–2002
We turn next to the Gender Effect Hypothesis and the question of
whether the presence of a female incumbent will draw other females into
the campaign. In Table 3, we first calculated
the proportion of contested primaries in which a woman challenged an
incumbent of her own party. The first row of Table
3 displays the results for all incumbents and then a breakdown for
Democrat and Republican incumbents separately. The second row shows the
proportion of primaries in which a woman sought the nomination in the
opposition party. Finally, in the third row of the table, we combined the
first and second rows, incumbent and opposition party, to obtain an
aggregate picture of female presence in all districts where an incumbent
pursued reelection.
Female competition in contested primary races for the U.S. House of
Representatives, 1956–2002
In general, the results support our hypothesis and show that the
presence of a female incumbent leads to additional female candidacies in a
district. Among all incumbents (row 1), the proportion of female
officeholders being challenged by a female in the party primary exceeds
the rate at which females challenge male incumbents (15.4% versus 10.5%).
This intraparty gender effect is more pronounced among Democrats than
Republicans. Female Democrats face challenges from another woman in 17.3%
of contested primaries, while male Democrats are challenged by women in
11.3% of the contests. Among Republicans, the difference is smaller. Women
challenge female Republicans in 10.4% of the contests, while male
incumbents face a female opponent in 8.4% of the contested Republican
primaries.
Table 3 also reveals that a female incumbent
seeking reelection influences the gender distribution of candidates
seeking the nomination within the opposition party. The proportion of
opportunities in which a woman seeks the nomination of the opposition
party is greater in districts with a female incumbent (23.5%) than in
districts with male incumbents (13.2%).12
Additionally, within the opposition party, the incidence of
two or more females contending for the nomination is greater in districts
with female incumbents (3.5%) than in districts with male incumbents
(1.3%).
The relationship is especially strong for Republican
women running in Democratic districts; Republican women seek the
nomination in 24.1% of the elections with a female incumbent, compared to
10.8% of the primaries in districts held by men. And it holds for
Democrats running in Republican districts as well; female Democrats seek
the nomination in 22.4% of the primary elections in districts with a
female Republican incumbent and in only 16.6% of the contests in districts
held by male Republicans.
The aggregate picture of primary elections for the U.S. House of
Representatives presented in the third row of Table
3 is clear. The presence of female candidates is significantly
greater in districts where a female incumbent holds the seat. This
suggests then that female incumbents do provide, as role models or as
testaments to the “win-ability” of the district, a signal that
leads other women to run for the House seat.
Moreover, within the opposition, not only do women seek the nomination
more frequently in districts held by female incumbents but they win the
nomination more frequently in these districts as well. Table 4 presents the proportion of nomination
opportunities won by female candidates.13
The denominator used in calculating the percentages in Table 4 is based upon all nomination opportunities,
including uncontested primary elections, contested primaries, and
convention nominations.
Across all of these opportunities, the
success rate for women in districts with a female incumbent (13.1%)
exceeds the rate in districts with male incumbents (6.9%). The
relationship holds for both Democrats and Republicans. The Republican
difference in success rates (12.7% versus 5.1%) is larger than the rate
among female Democrats seeking the nomination in Republican-held districts
(13.7% versus 9.4%).
14 At the suggestion of
one reviewer, we divided the time frame of our study into two periods:
1956–80 and 1982–2002. We then examined our indicators of the
competitive environment within each period. Among Democrats, the
differences between male and female incumbents are similar in both
periods. There are no departures from the patterns reported in Tables
1, 2, 3, and 4. For Republicans,
three statistically significant relationships emerge in the
1982–2002 period: 1) the proportion of uncontested general elections
is greater for male (15.2%) than for female (8.4%) Republican incumbents,
2) male Republicans (13.5%) enjoy more free passes than females (7.3%),
and 3) the proportion of contested primary elections within the opposition
party is substantially greater in districts with a female Republican
incumbent (57.9% versus 42.7% for male Republicans).
Women winning the nomination of the opposition party in districts
where an incumbent is seeking reelection, 1956–2002
The results in Table 4 lend further credence
to our earlier observation that the presence of a female incumbent may
serve as a signal about the electoral prospects for women in a district.
Given that more women seek nominations and win nominations in districts
with a female incumbent, the presence of a female incumbent is likely to
be a component in the strategic calculus in decisions women make about
where to run. An additional question that arises from our analysis is
whether the effect of female incumbents on the competitive environment
varies with the electoral security of the incumbent. Table 5 presents the seven indicators used in our
analysis for safe and marginal male and female incumbents. In this
analysis, we rely upon the conventional definition of safe and marginal
districts: A marginal district is one in which the incumbent won with less
than 55% of the two-party vote in the previous election.
Indicators of electoral competition and electoral security of
incumbents seeking reelection, 1956–2002
Three sets of tests are included in the table; we use a difference of
proportions test to compare male and female incumbents from safe
districts, male and female incumbents from marginal districts and, as
reported in the last column of the table, females from safe and marginal
districts. Table 5 demonstrates, first of all,
that there is significant difference between safe female incumbents and
their male counterparts on all seven indicators. Safe female incumbents
face a more competitive electoral environment; there are fewer uncontested
general elections (10.3% versus 18.9% for males), more contested primaries
(34.5% versus 30.0% for men), and fewer “free passes” (6.9%
versus 13.9% for men). In this sense, women enjoy less of the electoral
security that is conventionally attributed to holding a safe seat. Note as
well that women from safe districts face more primary challenges from
women than do their male counterparts (15.8% versus 10.5% for men) and
stimulate more competition within the opposition party and greater
participation by women within the opposition party.
Turning to the comparison of male and female incumbents from marginal
districts, the table reveals that there are no significant differences on
the first four indicators; men and women running for reelection are
equally likely to face major party competition in the general election,
challenges in the primary election, and female competition in those
primaries. The major differences between men and women from marginal
districts are found in the activity of the opposition party. Marginal
districts with female incumbents have a greater proportion of contested
primaries (69.9% versus 55.2% for men), draw a larger proportion of female
candidates (38.0% versus 16.5% for men), and nominate a larger proportion
of women (17.7% versus 7.7% for men).
Finally, consider the extent to which safe and marginal female
incumbents differ on the indicators.15
We
also tested to determine whether there are significant differences in the
competitive environment faced by safe and marginal male incumbents. The
statistically significant differences are identical to our comparison of
safe and marginal females. Compared to their safe male counterparts,
marginal male incumbents enjoy fewer uncontested general elections and
“free passes.” Within the opposition party, there are more
contested primary elections and a higher incidence of female candidates
seeking the nomination.
Consistent with conventional expectations
about electoral security, there are proportionally fewer uncontested
general elections and “free passes” in the marginal districts;
in fact, no female incumbent from a marginal district has enjoyed an
uncontested general election or a “free pass,” while 30 men
have. In marginal districts held by women, there are more contested
primaries within the opposition party (69.9% versus 45.0% in safe
districts held by women), and more of these primaries feature women
candidates (38.0% versus 21.0% in safe districts held by women).
16 We also tested whether there are significant
differences in the competitive environment faced by safe and marginal male
incumbents. The pattern of significance is identical to what we report in
the last column of Table 5 for marginal and safe
female incumbents. Compared to safe male incumbents, marginal male
incumbents have proportionately fewer uncontested general elections and
“free passes.” Further, districts with marginal male
incumbents have a larger proportion of contested primaries within the
opposition party, as well as a larger proportion of female aspirants among
opposition candidates.
As Table 5 shows, the Competition and Gender
Effect Hypotheses are still supported even after controlling for electoral
security.17
As an additional step, we
treated the seven measures examined in our analysis as dependent variables
and estimated a multivariate equation for each. Specifically, we cast each
dependent variable as a function of three binary measures: party of the
incumbent (coded one if Democrat), electoral security of the incumbent
(coded one if safe), and gender of the incumbent (coded one if female).
Because the dependent variables are themselves binary and their
distribution is skewed, the equations were estimated using a “rare
events” logit procedure (Tomz, King, and Zeng
1999). The results mirror those reported in Tables 1, 2, 3, 4, and 5. For six of the seven dependent variables, the
coefficient for the gender of the incumbent was properly signed and
statistically significant. The exception was the measure of whether an
incumbent was challenged in the primary election. In this case, we found
that the probability of a primary challenge significantly increases when a
female Republican holds the seat. This is consistent with the results
reported in Table 2. Overall, the exercise shows
that the relationships reported in Tables 1,
2, 3, 4, and 5 hold when we
control simultaneously for the political party and electoral security of
the incumbent.
It is important to note that there are two difficulties with refining
or expanding these models. First, the distribution of the binary dependent
variables are skewed or unbalanced. For example, only 15.9% of all
incumbents face no opposition in general elections; 11.7% of incumbents
face no opposition in both the primary and general election. Further,
while increasing over time (Palmer and Simon
2001), the presence of women in House elections remains, from a
statistical viewpoint, a rare event. For example, only 10.8% of sitting
incumbents face a challenge from a female candidate in their party
primary; women as a proportion of all candidates seeking the nomination
within the opposition party is 13.8% over the time frame of our study. The
upshot of working with such unbalanced distributions is that while the
estimated coefficients can be used to make reliable inferences about the
impact of the independent variables, the estimated models are not terribly
powerful in classifying the “positive cases” (for a similar
difficulty in predicting military conflict in the international system,
see Beck, King, and Zeng 2000, 2004).
Expanding the list of control or independent variables in these models
presents a second difficulty. Suppose, for example, we modified each
equation to include prominent demographic characteristics of the
constituency (e.g., presidential vote in the district, urban dwellers,
median income). In such an exercise, the objective is to determine whether
the relationship between gender of the incumbent and the dependent
variable holds in the face of such controls. Our most recent research
(Simon and Palmer, 2005), however, provides
strong evidence that the gender of the incumbents elected to the House is
not independent of such constituency characteristics. There are distinct
demographic profiles that make a district “woman friendly” and
more likely to nominate and elect a female candidate. The statistical
implication of this finding is that an equation that includes demographic
variables and the gender of the incumbent will violate a fundamental
assumption of the estimation technique—the explanatory variables are
not independently distributed.
The hidden influence of female
incumbents persists. Among safe incumbents, there are distinct differences
in the competitive environment of districts held by men and women. Within
marginal districts, female incumbents, compared to their marginal male
counterparts, stimulate more competition within the opposition party and
draw more women of the opposition party into the fray.
CONCLUSION
The point of departure for this analysis is the apparent equality
between male and female incumbents and the resulting inattention given to
women running for reelection. Consideration of the broader literature on
women and campaigns, however, implies that there should be significant
differences in races involving male and female incumbents. Using seven
indicators of the competitive environment in House districts, our
empirical analysis produces considerable support for both the Competition
and Gender Effect Hypotheses. Women running for reelection face a more
competitive environment as measured by uncontested general elections,
contested primaries, and “free passes.” Female incumbents also
face more challenges from female candidates in primary elections. This
relationship extends to the opposition party in House districts. Within
the opposition party in districts with female incumbents, there are more
contested primaries and these primaries feature more women as candidates
and ultimately as the chosen nominees. As a result, in spite of the parity
that exists between male and female candidates in terms of electoral
outcomes, female incumbents face a more rigorous electoral environment and
incur higher “costs” in achieving reelection. Male and female
incumbents may have the same success rates, but women have to work harder
to retain their seats.
Our results thus reveal that female incumbents have a hidden
influence: Their presence increases the entry and participation of female
candidates in House elections. This may be due to party recruitment
efforts. Although party chairs, who are still overwhelmingly male, might
prefer candidates more like themselves (i.e., other men), there is an
incentive to “neutralize” the advantages of a female opponent
by persuading another women to run against her (see Niven 1998; Sanbonmatsu
2002b). The increase in female candidates could also be a function
of a “role model effect.” The electoral success enjoyed by a
female incumbent might, in and of itself, inspire more women to run. It
also sends a strategic signal to groups active in candidate recruitment
that women can be successful in that district. This would seem to enhance
the symbolic representation of women in electoral politics (see Pitkin 1972; Mansbridge
1999); electoral success by women encourages more women to run. In
addition, it also increases awareness and activity among female voters
(Atkeson 2003; Hansen
1997).
On the other hand, given where women are more likely to
run—in districts with female incumbents—the overall number of
women in the House is not necessarily going to increase under these
circumstances. Female candidacies are disproportionately concentrated in
districts already represented by women. Once a woman is elected, she faces
higher probabilities of being challenged for renomination by a woman and
facing a female opponent in the general election. In our data, for
example, there are 70 instances of female challengers running against
female incumbents; well over half (46) of those have occurred since 1992.
In these contests, incumbency maintained its supremacy; female incumbents
lost to a female challenger in only five of those seventy elections
(7.1%).18
For comparative purposes, it is
worth noting that female challengers defeated male incumbents in 23 of 607
opportunities for a victory rate of 3.9%. This rate does not differ
significantly from the 7.1% rate in races that feature women as both
incumbents and challengers.
As a result, this increase in
competition does not trigger changes in the gender composition of the
House. While the presence of female incumbents encourages more women to
run, the power of incumbency continues to act as a “political glass
ceiling,” impeding the increase in the number of women who serve in
the U.S. House of Representatives.
