An interparty gender gap developed over a 40-year period resulting in a
pattern where men are sightly more likely to be Republicans while women are
slightly more likely to be Democrats. A gender gap in the opposite direction
existed in the 1950s when women were more likely to support the Republican
party and men the Democratic party (Campbell et al.
1960). This older gender gap began to be replaced with the
current pattern in the early 1960s as men, most dramatically white southern
men, altered their partisan preferences (Abramowitz and
Saunders 1998; Kaufmann and Petrocik
1999; Norrander 1999a; Wirls 1986). Some women, primarily southern white
women, also moved into the Republican party but at a slower pace than men.
The new interparty gender gap first reached statistical significance in the
early 1970s and stabilized around 5 percentage points in the 1980s. During
the Clinton years, the gender gap grew to 15 percentage points. This time
women moved in greater numbers with growing allegiance to the Democratic
party while men's preferences remained more stable. In the 2000 election,
the gender gap fell back to 9 percentage points.
Gender differences occur in issue preferences and ideological identification,
as well. The oldest and largest of these differences exist for use-of-force
issues, with women less supportive of military intervention abroad and more
supportive of gun control at home. Women are more likely to favor using the
government to help individuals in need, on what are called the compassion
issues. On the other hand, women may be more conservative on some issues
involving moral questions. By the 1980s, women and men also were adopting
more distinctive ideological identifications that reflected some of these
issue differences (Shapiro and Mahajan 1986; Norrander 1999b). Thus, by the end of the 20th
century it appeared that men and women had sorted through their issue and
ideological differences and matched these to two more ideologically
distinctive parties. Yet, the match between issues and parties has never
been perfect among the American public. An imperfect sorting process may
still leave differences between men and women within each party, creating an
intraparty, as well as an interparty, gender gap. This article briefly
explores the intraparty gender gap by examining men's and women's
preferences in the 1980–2000 presidential primaries.
An intraparty gender gap could take on a variety of magnitudes. In general,
the intraparty gender gap might be expected to be smaller than the
interparty gender gap, if many Americans have a more or less good fit
between their issue preferences and partisan identities. A second
possibility is that the intraparty gender gap is more varied than the
interparty gap. Presidential primaries are characterized by a larger number
of candidates, which often change even over the course of one election year.
Finally, single candidates or combinations of candidates might evoke more
gender differences in support than other combinations of presidential
nominee seekers. However, several differences exist between men and women
that may produce, at least at times, an intraparty gender gap.
Intraparty Gender Gaps in Voters' Identities and Concerns
One difference between men and women is that women are slightly more
partisan than men. The independence gap finds men about 5 percentage
points more likely to view themselves as leaning independents while
women are more likely to view themselves as weak partisans (Norrander 1997). Women's more partisan
identities might lead them to be more supportive of incumbent presidents
seeking renomination, vice presidents hoping to move up to the top post,
and other party frontrunners. Likewise, candidates who have a greater
appeal among independent rather than partisan primary voters also should
experience a gender gap in support.
Table 1 examines partisan differences between
men and women participating in the 1980–1996 Democratic and Republican
presidential primaries.1
In most of the
primary exit polls, respondents are only asked to place
themselves as Democrats, independents, or Republicans. The
question in 1980, 1984, and 1988 asked “Do you usually think of
yourself as a: Republican, Democrat, independent.” In 1992 and
1996 the question was expanded to “No matter how you voted
today, do you usually think of yourself as a:…” In 1992, the
number of response categories was enlarged to strong Democrat,
not so strong Democrat, strong Republican, not so strong
Republican, independent, something else. The 1996 exit polls
used the 1992 question wording but reduced the number of
responses back to Democrat, Republican, independent, something
else. None of the exit polls solicited a response that would
separate leaning from pure independents. For the analysis,
responses were coded into the three categories and difference in
proportion tests were employed to identify primaries with gender
gaps. A. 05 level is reported as significant.
First, in
both parties' primaries on average three-quarters of the voters identify
with the party. About one-quarter view themselves as independent voters,
and only a scant 4% are crossover voters. The independence gap occurs
more clearly among Democratic primary voters. In 43% of the primaries,
the Democratic electorate has statistically a significantly larger
proportion of male than female independents. On the other hand, in 51%
of the Democratic primaries, women are a larger proportion of the
partisan voters. Generally, the few crossover voters are not
differentiated along gender lines, but in about one in five primaries
more men than women are crossover voters. The patterns are not as
distinct among voters in the Republican primaries. In two-thirds to
three-quarters of the Republican primaries, no differences exist along
gender lines for partisan, independent, or crossover voters. Still, a
weak independence gap does occur. Somewhat less than a third of the
time, men are more likely to outnumber women among independent
voters.
Issue or ideological differences also could separate men from women in
each party's primary electorate. Table 1
illustrates that a gender difference on ideology is quite prevalent in
the Republican primary electorate.2
The
ideology question in all surveys was, “On most political matters
do you consider yourself:”. In 1980 to 1988, three options were
given: liberal, moderate, conservative. In 1992 and 1996, five
options were listed: very liberal, somewhat liberal, moderate,
somewhat conservative, very conservative.
Generally,
about half of the Republican primary electorate view themselves as
conservatives, one-third as moderates, and 10% as liberals. Men
outnumber women among the conservative voters in three-quarters of the
primaries, while women outnumber men among moderate voters in two-thirds
of the Republican primaries. The Democratic primary electorate is more
broadly divided among the three ideological identifications but the
differences between men and women are more muted. About half of the
Democratic primary electorate view themselves as moderates, and in about
20% of the primaries women outnumber men in this category. Nearly
one-third of Democratic voters describe themselves as liberals, and once
again, about 20% of the time women outnumber men in this category as
well. Finally, in the quarter of the Democratic electorate who view
themselves as conservatives, men outnumber women about one-third of the
time. Similar intraparty ideological gender gaps occur among caucus
participants and delegates to the presidential nomination conventions
(
McMullen and Norrander 2000;
Rapoport, Stone, and Abramowitz 1990).
If primary voters are retrospective voters (Monardi
1994; Wattier 1990), women's more
pessimistic view of the economy might lead them to be less supportive of
incumbents (Chaney, Alvarez, and Nagler 1998)
even during the nomination stage. Among voters in the 1992 Republican
primaries, women were more pessimistic than men about the economy in 45%
of the states while men were more pessimistic than women in 9%. Thus,
while women's stronger partisanship could lead them to give greater
support to incumbent presidents, women's more pessimistic evaluations of
the economy could push their support away from incumbents.
A Split Decision. According to Norrander, men and women
simply appear to be looking for different qualitites in a
candidate, with differences in partisanship, ideology, and more
leading both groups to support different candidates in
presidential primaries. Photo: istockphoto.com/Andrea
Gingerich.
Men and women appear to be looking for different qualities in candidates.
Women often are more reluctant to vote for “maverick” candidates. In
general election voting, men have been greater supporters of third-party
candidates (Alvarez and Nagler 1995; Lacy and Burden 1999; Lacy
and Monson 2002). Likewise, in response to the 1992 primary
exit polls, men were more likely than women to pick a candidate who
would “shake things up.” Women focused more on qualities such as caring,
courage, ability to handle a crisis, or experience. Men, on the other
hand, also were more likely to be looking for a candidate who could win
the fall election (McMullen and Norrander
2000). Thus, a number of differences in partisanship, ideology,
evaluations of the economy, and preferred candidate qualities could lead
men and women within the same party to support different candidates for
their party's presidential nomination.
Intraparty Gender Gaps in Candidate Preferences
To measure the intraparty gender gap in candidate support during the
1980–2000 presidential primaries, the proportion of male supporters is
subtracted from the proportion of female supporters and a difference in
proportions statistical significance test is applied. A positive gender
gap number indicates more support from women and a negative gender gap
figure denotes more support from men. When only two candidates contest
an election, a gender gap in preferences for one candidate is reflected
in the opposite direction for the other candidate. However, in a
multicandidate setting a number of combinations could occur. Pairs of
candidates could have inversed gender gaps, one favored by women and the
other by men. One candidate could garner stronger support from one of
the sexes, but voters of the other sex might distribute their
preferences across a number of candidates such that none of the
candidates has disproportionate support from these voters. It is even
possible that a single candidate might be preferred by men in some
electoral settings and supported more by women in other primaries.
Table 2 lists the intraparty gender gaps for
Democratic candidates. The 1980 contest between President Jimmy Carter
and Senator Ted Kennedy produced a significant gender gap only one-third
of the time. Contrary to the incumbency support argument due to women's
stronger partisan identification but in support of the economic vote
hypothesis, women were slightly more likely to support Kennedy (an 8.5
percentage point gap in the statistically significant instances) while
more men supported Carter.
The 1984 Democratic contest had some candidates, such as Jesse Jackson,
producing a more frequent gender gap, and some odd combinations where
the same candidate (Gary Hart, Jackson, Walter Mondale) at times had
greater support from women and in other primaries greater support from
men. Much of the explanation for these gender gaps in the 1984
Democratic contest has to do with racial voting patterns in that
election. In 1984, the support for Jesse Jackson came almost entirely
from black voters. In primaries with exit poll data, Jackson on average
received the support of only 5.5% of white voters while attaining the
votes from 77% of African American voters. Black women are more likely
to turn out to vote in elections than are black men (McClain and Stewart
1998, 83). This was true in many of the 1984 primaries, where women
comprised 59.3% of the black electorate in comparison to 51.8% of the
white electorate (a difference statistically significant in eight of the
13 primaries). Thus, the greater presence of black female voters led to
an overall gender gap of women supporting Jesse Jackson.
When the 1984 election choices are controlled for race, the gender gap
for Jackson among white voters occurred in only two states, and in these
two states (California and Illinois) men gave Jackson higher support.
Among white voters, John Glenn had greater support from male rather than
female voters in only one state, Hart garnered more support from women
rather than men in only one state, while Mondale drew greater support
from women in two states and men in one state. Overall, the number of
significant gender gaps among white voters shrunk from 16 states to
seven states. (The discrepancy in total number of gender gap states with
those listed in Table 1 occurs because exit
polls in several states did not include a race question.) Among black
voters, Jackson had only one significant gender gap, drawing the support
of more female voters in New York. Black male voters in New York were
sightly more likely to support Mondale. Finally, black male voters were
more likely than black female voters to support Hart in three states
(New Jersey, Pennsylvania, and Tennessee).
In 1988, Jackson's gender gap is once again primarily explained by
differential turnout rates among black voters. In three of the four
states where Jackson had greater support from female voters, black women
participated at higher rates than white women. In the one state where
Jackson drew greater support from male voters, black men participated at
a higher rate than white men.3
When
controlling for race in these five states, no gender gap occurs
for Jackson among white voters. Only twice does a gender gap
occur for Jackson among black voters, but the direction is split
with greater male support in Texas and greater female support in
Mississippi.
Al Gore had the most frequent gender gap
in 1988. In six mostly southern states, Gore received more support from
men than women, perhaps tapping into the greater number of conservative
male voters. In four states, women gave greater support to Michael
Dukakis, while the votes for Richard Gephardt, Hart, and Paul Simon
never produced a gender gap. In 1992, a gender gap occurred only about
one-fourth of the time. Men gave greater support to Jerry Brown, hoping
to shake things up. Women gave greater support to Clinton because he
cared or could handle a crisis and to Paul Tsongas because he was
courageous (
McMullen and Norrander 2000). In
the 2000 Democratic contest, Gore experienced a more consistent gender
gap than he received in 1988, but in 2000 female voters gave him greater
support than male voters in 11 of the 20 states. This may be the case of
women's greater partisan leanings producing added support for a vice
president seeking to move up to the top post. In general, a gender gap
occurred about one-fourth of the time for candidates in the 1980– 2000
Democratic primaries.
Table 3 presents evidence on the gender gaps
for Republican candidates. These differences between male and female
voters appear to be slightly more pronounced in the Republican primary
electorate. Overall, about one-third of the time a Republican
presidential nominee seeker experienced a gender gap in support. The
1980 primaries are a notable exception, with only one instance of a
gender gap. The most frequent gender gap in the 1988 Republican
primaries occurred in support of Pat Robertson, who drew greater support
from women than from men. Women are more religious than men, which
appears to explain women's greater support for Robertson. Robertson also
had more women among his financial supporters than did the other
Republican candidates (Wilcox 1992).
While George H. W. Bush drew greater support from men than women in a
small number of the 1980 and 1988 primaries, in 1992 President Bush was
consistently and strongly favored by female voters. The 1992 Republican
contest pitted an incumbent president against a maverick-style opponent.
Men supported Pat Buchanan to shake up politics and because of their
concern for the deficit, trade issues, and affirmative action. Women's
stronger partisanship and search for candidates with experience led them
to support President Bush's renomination, despite their more negative
evaluation of the economy. Women who supported Buchanan did so due to
concerns about family values while men who supported Bush cited him as a
winner (McMullen and Norrander 2000). The
reasons and patterns of support for Buchanan and Robertson hint that the
gender gap might partially overlap with divisions between economic and
social conservatives within the Republican party.
The 1996 Republican primaries saw a repeat of Buchanan's wider support
from male voters, although in fewer primaries than in 1992. When the
vote for Dole reflected a gender gap, he was favored more by women than
men—a pattern that also occurred in the 1988 primaries. For the two main
candidates in the 2000 Republican primaries, George W. Bush drew broader
support from women while John McCain attracted greater support from men.
Bush's compassionate conservatism versus McCain's maverick style and
support from independent voters may underlie the gender gap in the 2000
Republican primaries.
A number of instances of gender gaps exist within as well as across the
two major American parties. About one-fourth of the time, candidates
seeking the Democratic nomination experience a statistically significant
difference in support levels from male versus female voters. A gender
gap is more likely to occur for Republican candidates, about one-third
of the time. Within the Democratic primary electorate, women are more
likely to view themselves as Democrats while men are more likely to
describe themselves as independents. This pattern is less strong in the
Republican primary electorate. Yet, the Republican primary electorate is
more divided along gender lines when it comes to ideology. Men in
three-quarters of the Republican primaries are more likely to describe
themselves as conservatives while women see themselves more often as
moderates in two-thirds of the primaries. Men and women also appear to
be searching for different qualities in a presidential nominee. The
gender gap for two candidates (George H. W. Bush and Al Gore) reversed
itself when these candidates changed from being challengers, when they
were favored more by men, to incumbents, when women gave them more
support. While the gender gaps within each party are often small and
somewhat varied in their appearance, victory in the primaries is often
attained by small vote margins. A greater awareness of gender
differences within each party should serve to better our understanding
of the dynamics of support during the nomination phase of the
presidential selection process.
