Given its enormous social and psychological importance, gender can
play an important role in political cognition. Just as the racialization
of welfare leads racial conservatives to oppose welfare programs and
racial liberals to support them (Gilens 1999),
political issues can likewise become gendered. This process, which I call
“gender implication,” leads to polarization, not necessarily
between men and women (i.e., a gender gap) but between supporters of
traditional and egalitarian gender arrangements. We know that this occurs
for issues that impinge directly on gender equality, or for
issues—such as abortion or child care—that the public believes
affect gender roles or the fortunes of women and men differently (Sapiro 2003, 618–19). I argue that another sort
of gendering can occur. An issue can be gender-implicated if it is framed
in ways that symbolically evoke people's ideas about gender.
An issue can be gender-implicated if it is framed in ways that
symbolically evoke people's ideas about gender, leading people to
view the issue through an unconscious “lens of gender” (Bem 1993).
In this article, I argue that the 1993–94 American health care
reform debate demonstrates gender implication at work. Before 1993, mass
opinion on health care reform was not linked with gender ideology. The
politics and rhetoric deployed during 1993–94 brought a change, by
linking health care with gender in new ways. These linkages were subtle
and symbolic, and they unconsciously associated people's feelings
about gender relations with their thinking about health care reform. After
reform efforts died, these linkages faded among the public.
This analysis is important for several reasons. First, it expands our
understanding of how gender ideology can structure public opinion, even
without explicit references to gender. This adds to our understanding of
the role of issue frames in shaping opinion. The analysis also draws
attention to one of the ways that the “lens of gender”
structures political and social cognition. Most broadly, this approach
treats gender implication within a general framework that allows analysis
of implication among both men and women and that holds the prospect for
comparing gendering and racialization.
I begin with a theoretical account of the conditions under which
political rhetoric will engage people's ideas about gender, even
without any explicit gender references. Next, drawing on secondary
accounts of the Clinton administration's 1993–94 health care
reform effort, I explore the set of rhetorical frames deployed by
supporters and opponents of reform. In this discussion, I demonstrate how
these frames were consistent with implicit gender coding. Then I employ
survey data from the National Election Studies (NES) to demonstrate that
public opinion did become gender-implicated in response to these frames.
Finally, I conclude with some observations about the significance of the
findings for health care reform specifically, as well as for our
understanding of the role of gender in political cognition and
politics.
THEORY
“Gender implication” is the process by which opinion on
political issues becomes entwined—in political discourse and in
citizens' minds—with considerations of gender. The term
“implication” makes clear that the process need not be
explicit: The rhetoric need not refer explicitly to gender and individuals
may be unaware of gender's impact on their opinions.
Mechanisms of and Conditions Governing Group
Implication
The interaction between psychological schemas and rhetorical
frames governs this process. Schemas are “cognitive
structure[s] that represent knowledge about a concept”
(Fiske and Taylor 1991, 98). They are invoked
when we encounter ambiguous phenomena, including political issues, and
they play an active role in perception by filling in missing information
and by suggesting bases for evaluation. When a person encounters a
political issue, some schema is invoked to understand it; that schema then
influences the basis for evaluating the issue (Conover
and Feldman 1984; Smith 1998).
An issue frame, on the other hand, is “a central organizing idea
or story line that provides meaning to an unfolding strip of events,
weaving a connection among them. The frame suggests what the controversy
is about, the essence of the issue” (Gamson and
Modigliani 1987, 143). Frames often suggest explicitly how best to
view an issue. I argue that frames can also work unconsciously to evoke a
particular schema, which then influences issue perception and evaluation.
A frame will evoke a particular schema when the frame is congruent with
the schema. Schemas, then, are the psychological analog of rhetorical
frames.
Schemas contain attributes that describe the category in question, as
well as a structure that positions those attributes in relation to each
other and provides a basis for judgments. For example, white
Americans' schemas for understanding race contain attributes
drawn from common cultural stereotypes: that whites are rich, that blacks
are athletic, that discrimination occurs against blacks, that whites are
hardworking, and so on.1
Patricia Devine
shows that high-prejudiced and low-prejudiced people are equally aware of
cultural stereotypes (1989).
For some,
this schema includes a structural linkage that suggests that blacks are
poor
due to discrimination; others' racial schemas include a
different structural link that suggests whites are rich because they work
harder than blacks (
Wittenbrink, Gist, and Hilton
1997). This variation in structure leads to variation evaluations
of situations perceived in terms of the race schema.
Bernd Wittenbrink and colleagues conducted an intriguing experiment
that shows how schematic structure can drive evaluations of situations
that are—on their surface—quite removed from the schema (1997). Subjects watched a series of animated videos
involving the interaction of a single fish with a larger group of fish.
These videos involved conflict between the fish and the group, but were
ambiguous as to the individual fish's and the group's
motivations. The crucial finding was that subjects' racial schemas
affected their interpretations of the videos. Subjects who believe blacks
are poor because they do not work hard tended to hold the individual fish
responsible for the interactions; those who believe blacks' position
is due to discrimination held the group responsible. The structural fit
between schema and video was key: Racial schemas did not
influence interpretation of a different video that did not involve
conflict between unequal groups of fish.
This research by Wittenbrink and colleagues demonstrates that a schema
can influence evaluation of a situation that bears little or no surface
resemblance to the contents of the schema. I extend this argument to
gender schemas and to the evaluation of political issues. For a gender
schema to drive issue evaluation, the issue rhetoric must frame the issue
in a way that is congruent with the structure of the gender schema. That
is, the schema and frame must share the same structure;2
when they do, that schema can govern
the perception and evaluation of the issue.
Structure of the Gender Schema
The structure of Americans' gender schemas is therefore crucial,
because the match—or lack of match—between structure and
rhetoric governs gender implication. In this analysis, I focus on three
central aspects of the gender schema: the centrality of difference, the
importance of power, and variation in the evaluation of difference and
power.
First, the idea of difference has been central to theoretical
understandings of gender for centuries. The idea of gender difference
gives rise to beliefs about appropriate roles and spheres of activity for
men and women and ultimately underlies the distinction between public and
private (e.g., Epstein 1988). Most importantly
for my purposes, the mass public understands gender in terms of
difference. Children learn very young about sex differences and are
socialized early and often to understand and respect gender differences
(Stockard 1999), and the power of the idea of
fundamental gender difference is evident in the resilience of the claim of
biological bases for all manner of observed sex differences (e.g., Fausto-Sterling 1992).
Second, a central point of feminist theorizing is that gender is more
than mere difference; it is fundamentally also about power and dominance.
Catharine MacKinnon argues that “construing gender as a difference,
termed simply the gender difference, obscures and legitimizes the way
gender is imposed by force…. [T]he idea of gender
difference helps keep the reality of male dominance in place” (1987, 3). Gender is “deeply embedded in the
politics of family relations” (Goldner et al.
1998, 556)—it defines appropriate roles, behavior, and power
within the family sphere, and also between the public and private spheres.
In turn, dominance relationships in the family sphere both reflect and
support dominance relationships in politics and society (e.g., Phillips 1991, 102–4). This dominance is
codified and enforced in laws and in the design and implementation of
public policy (Epstein 1988, Chapter 6; Fraser 1989; Mettler 1998;
Skocpol 1992; Tolleson
Rinehart and Josephson 2000, Section II). Moreover, people's
lived experience of gender serves to normalize structures of power,
dominance, and inequality (e.g., Goffman 1977;
West and Zimmerman 1987).
The third element of the gender schema turns on an evaluation of the
first two: (1) the centrality of individual differences and the
articulation of these differences into appropriate spheres of conduct, and
(2) the power relationships that operate within and between these
differences. Analysts have long noted the centrality of prescription in
gender beliefs: These are beliefs not just about how men and women differ
but how they should differ (Fiske and Stevens
1993). For supporters of traditional gender arrangements, the
difference between men and women is fundamental (whether its basis is
divine or biological), and gender hierarchy is a natural and necessary
outgrowth of that difference. In contrast, gender progressives believe
that “the artificial division [of gender] is neither fair
nor functional and that it promotes an unfair and unjust system”
(Sigel 1996, 15). MacKinnon (1987, 21) characterizes these opposing interpretations
of gender difference in contrasting her perspective with Phyllis
Schlafly's:
We both see substantial differences between the situations of
women and of men. She interprets the distinctions as natural or
individual. I see them as fundamentally social. She sees them as
inevitable or just—or perhaps inevitable therefore
just—either as good and to be accepted or as individually
overcomeable with enough will and application. I see women's
situation as unjust, contingent, and imposed.
The public is also divided along this axis. Despite liberalization in
gender norms, there is still considerable public debate about gender
equality and especially about changes in actual gender arrangements (Huddy, Neely, and LaFay 2000; Sanbonmatsu 2002). Moreover, this axis of disagreement
structures political conflict over explicit gender issues, including the
Equal Rights Amendment (Mansbridge 1986) and
abortion (Luker 1984), and serves as an
organizing principle for both liberal and conservative women's
organizations (Dworkin 1983).
There are several reasons to expect men and women to share a common
gender schema structure. Both undergo similar socialization (boys and
girls are taught to assume different positions in the gender
system, but they are socialized to the same system), are immersed in
essentially the same culture, and watch largely the same media. Moreover,
the fact that the social structure puts men and women in close and
intimate contact with each other should limit the degree to which they
develop radically different understandings of what gender is.3
This is in contrast to modern race relations,
where spatial and task segregation leads blacks and whites to different
understandings of race (Jackman 1994; Sigelman and Welch 1991).
Men and women may
differ in their average location on the evaluative continuum, but they
should share the same basic schematic structure.
4 This construction of gender develops in important ways out of
the structural relationship between white women and men (Collins 1990; Higginbotham
1992; Hurtado 1989). Unfortunately, the
limited number of nonwhite respondents in the data prevent me from
exploring racial differences in this analysis.
To summarize, the gender schema consists of three interconnected
elements: (1) beliefs about the centrality of individual differences and
the articulation of these differences into appropriate spheres of conduct;
(2) beliefs about the power relationships and hierarchy; and (3) a
dimension of evaluation of the first two. Gender traditionalists fall at
one end of this dimension; they believe that the differences are natural
and the hierarchy is appropriate, and they therefore oppose change. Gender
egalitarians fall at the other end; they believe that the differences are
socially constructed and that the hierarchy is inappropriate. Others fall
somewhere in the middle.
People may draw on this schema to understand political
issues—even issues that have nothing to do with gender—when
those issues are framed to fit the gender schema. The key is not an
explicit reference to gender; it is in the structure of the
appeal: the invocation of difference, of power relations, and of
appropriate roles within and across spheres.
Relationship with Other Work on Gender and Political
Behavior
This article differs in important ways from two major lines of
research on gender and opinion. The first is work on the gender gap (Sapiro 2003, 605–10). The idea of gender
implication arises from a fundamentally different assumption, namely, that
gender can influence public opinion for both men and women, and that it
can operate similarly for both. Of course, insofar as men and women differ
in their average support for traditional or egalitarian gender
arrangements, gender implication can give rise to a gender gap, but this
need not be the case.
A second major approach has focused on people's understanding of
their own gender and its impact on political beliefs and
behavior. Much of this work has explored the roles played by gender
identification and consciousness among women (Conover
and Sapiro 1993; Gurin, Miller, and Gurin
1980; Tolleson Rinehart 1992). Though
important for opinion, especially among women, identification and
consciousness are theoretically orthogonal to beliefs about
appropriate gender arrangements, although they are likely to be related
empirically, with identified or conscious women likely to fall at one
extreme or the other of the gender ideology scale (Tolleson Rinehart 1992, Chapter 4; for an overview of
work in this vein, see Sapiro 2003). My approach
to gender implication differs in that it allows for the analysis of
women's and men's opinion in a single framework. Whereas
identification and consciousness are clearly very different theoretical
constructs among women and men (Fiske and Stevens
1993), cognitive beliefs about proper gender roles may
operate similarly among men and among women. As a theoretical approach,
gender implication lets us explore gender-opinion connections among both
men and women, and also see how beliefs about gender can serve as a
symbolic template for interpreting political issues far from the domain of
gender itself. I argue in the following section that this occurred during
the health care reform debate in the United States.
HEALTH CARE REFORM
After making comprehensive national health care reform a major
campaign issue in 1992, the Clinton administration organized a large task
force to construct and promote a plan for health care reform. Led by
Hillary Rodham Clinton, the task force put together a complex and
comprehensive plan, which sought to guarantee universal coverage and
contain costs. In September 1993, the White House launched the Health
Security proposal; after a year of intense debate, comprehensive health
care reform was essentially politically dead by September 1994.
Rather than work closely with cabinet officials, interest groups, and
Congress, the administration developed the policy within the White House
and engaged in a campaign to sell the plan to the public, in order to
create pressure for its passage.5
On the
genesis of the administration strategy and the ensuing political struggle,
see Jacobs and Shapiro (2000) and Skocpol (1997). For a more policy-oriented discussion of the
genesis of the reform plan itself, see Hacker (1997); broader accounts include Navarro (1994), which sets 1993 in the context of other reform
efforts, and Oberlander (2003), which lays out
the larger political context of the federal government's involvement
in health policy administration.
In response to the
administration's “public opinion” strategy, a wide range
of players who had been closed out of the task force process also tried to
shape opinion, including various interest groups, Democrats and
Republicans in Congress, and others. All sides of the debate focused on
crafting and disseminating appeals to the public, which meant that the
public was awash in communications campaigns relating to health care
reform, creating good conditions for changes in framing to influence the
structure of public opinion. In the sections that follow, I note the
gendered character of health care policy, and then review the frames that
both sides deployed during 1993–94, focusing on the ways that these
frames—unlike those that came earlier—should have engaged the
public's gender schemas.
Health Care as a Gendered Policy Domain
Social policy generally is built on gendered assumptions about the
roles of service providers and recipients (Sapiro
1986). In the medical realm, this is reinforced by the fact that
women and men have different medical needs—some due to biological
difference, many more due to the effects of gendered differences in
socialization, insurance coverage, poverty, and other social and economic
resources (e.g., Tolleson Rinehart and Josephson
2000). Health care is symbolically gendered as well. Linda
Gordon argues that “in establishing themselves as professionals with
cooptive authority to admit or exclude others, doctors made particular use
of their power over women” (Gordon 1990,
157).6
See Luker (1984, 27–39) for a similar argument in the
context of abortion policy.
This symbolic gendering continues
today. As Mary Ellen Guy describes:
Gender power relations in medicine are an exaggeration of
[gendered] power relations embedded in the political culture.
Patients spend more time with nurses but pay physicians…. Most
reimbursement schedules are predicated on whether the physician orders the
services of the ancillary professional. (1995,
243)7
This presumes, of course, that doctors
are men and nurses women. This is symbolically true, and was literally the
case during the nineteenth century. Even in 2004, 92% of nurses are women,
and 71% of doctors are men (U.S. Department of Labor, Bureau of Labor
Statistics 2005).
This symbolic gendering extends, finally, to the doctor–patient
relationship itself, as doctors maintain a sort of paternalistic control,
as the only professional in the system qualified to assess the
patient's best interest.
None of this guarantees that health care opinion will be
gender-implicated among the public without frames that engage the gender
schema. As we shall see, the gendered character of health care policy and
delivery provide fertile ground for these sorts of frames.
Health Care During the 1992 Presidential Campaign
During 1992, the Clinton campaign emphasized universal health coverage
and cost limitations. The Bush campaign emphasized free-market principles,
with tax incentives to expand coverage and increased efficiency to cut
costs. In his July 3 radio address, George H. W. Bush said: “We
would lower costs for patients and providers alike by keeping high taxes,
costly litigation, and big bureaucracies off their backs…. The
biggest story of our time is the failure of socialism and all its empty
promises, including nationalized health care and government
price-setting.”
Others have shown that Clinton's emphasis on costs and universal
coverage evoked considerations of equality among the public (Jacobs and Shapiro 2000; Koch
1998); we would also expect that the Republican framing would evoke
concerns about the scope of government. In short, during 1992, health care
was framed in terms of the traditional post–New Deal alignment, with
Democrats calling for greater government effort to promote equality and
Republicans championing a more limited government role.
The Clinton Administration's “Health
Security” Frame
This changed in 1993. The administration feared that discussing cost
controls would frighten middle-class voters who had health coverage, and
that emphasizing universal coverage would draw attention to the poor
(Skocpol 1997, 117–20). Therefore, they
focused on two different themes: security and personal
impact. Their consulting team advised that in discussing the plan,
“the dominant goal should be health security….
[T]here is also an emotion in security (lacking in cost) that
empowers our rationale for bold change.” They advised that
discussion of the plan should focus on “personal, human
impact,” and on “you and your family” (quoted in Skocpol 1997, 117; emphasis in original). Thus,
“security” was the first of five principles that President
Clinton articulated in his September 1993 speech that launched reform, and
that speech included frequent references to the health care woes of
ordinary families.
Opponents' Frames: Big Government and Private
Decision Making
Opponents focused on two frames: that the plan would create giant new
government bureaucracies and that it would project the government into the
private realm of health care provision. Opponents believed “that
support for Clinton's plan could be eroded by accentuating and
arousing Americans' dread of government and the personal costs of
health reform” (Jacobs and Shapiro 2000,
130). For example, Representative Dick Armey (R-TX) suggested in an
October 1993 letter to the Wall Street Journal that the
“Clinton health plan would create 59 new federal programs or
bureaucracies, expand 20 others, impose 79 new federal mandates….
[T]he Clinton plan is a bureaucratic nightmare that will
ultimately result in higher taxes, reduced efficiency, restricted choice,
longer lines, and a much, much bigger federal government” (quoted in
Skocpol 1997, 144–45).
Opponents coupled these standard invocations of bureaucracy run amok
with claims that those bureaucrats would intrude in the private health
care realm. Images of intrusion built upon existing images of health care
provision; the implicitly private “doctor–patient
relationship” has been an icon of health care discussion since the
American Medical Association (AMA) worked to kill “socialized”
health care in the 1930s (e.g., Patel and Rushefsky
1995, 21–22). More recently, in the aftermath of failed
health care reform, for example, the AMA described that relationship this
way:
The patient-physician relationship must ultimately be one of
trust, but all too often trusting relationships are disrupted not because
of dissatisfaction between patient and physician but because of choices
made by the patient's employer, a health insurance plan, or both.
(Dickey and McMenamin 1999)
Kathleen Jamieson and Joseph Capella (1994)
found that bureaucratic control and diminished doctor choice were two of
the major themes that appeared in commercials that opposed reform. The
most famous example were the “Harry and Louise” spots, which
portrayed a fictitious 40-something couple discussing their concerns about
the administration's plans. One major theme was the impending
intrusion of the federal government into a traditionally private domain:
“‘There's got to be a better way’ Harry and Louise
opined for the cameras, as they discovered the horrible possibilities of
bureaucrats choosing their health care plan” (Skocpol 1997, 137). Although they received only
moderate airplay, their influence was magnified by extensive media
coverage.
Conservative activists also saw the debate as an opportunity to
mobilize opinion against Democratic social programs generally. Republican
operative William Kristol warned in 1994 that the administration plan
would “relegitimize middle class dependence for
‘security’ on government spending.” He argued that
Republicans should oppose any reform and should advance a broader
antigovernment agenda (Skocpol 1997, 145).
Kristol advocated exploiting this opportunity by focusing on personal
fears and the intrusion of the government into the private
sphere.
Skocpol shows how this strategy turned into a veritable blizzard of
media coverage and grassroots mobilization against the plan. For example,
in December 1994, the conservative Heritage Foundation's Policy
Review warned that “we [will be] forced to purchase
health care insurance through our regional alliances,” and that
“a basic concern is whether they will be able to keep their own
doctors under the Clinton plan.” This emphasis on large government
bureaucracy and private intrusion spread to the popular media. For
example, a March 1994 Reader's Digest article emphasized
that “they are taking away our choice of doctor” (both cited
in Skocpol 1996, 147–49).
Other interest groups also employed these two frames. On their
Website, Washington group Patient Advocacy put it this way: “What
qualifies a bureaucrat—whether it be a federal one or a private
sector one—to make medical decisions? These decisions should be left
to the patient and his or her doctor” (1999).
Of course, criticism of government bureaucracy is nothing new. As I
discussed earlier, the Bush campaign employed this frame in 1992, and as
Skocpol notes, “1994 is hardly the first time that political
conservatives and business groups have used lurid antistatist rhetoric to
attack Democratic-sponsored social security initiatives” (1997, 164). What was new to the health care debate was
the way this frame was combined with the focus on personal, private-realm
interference. Health care was gender-implicated by the prospect of
vigorous government effort to meddle with private health decisions and
disrupt established power relationships within health provision.
Hillary Rodham Clinton as a Gendered Image
Hillary Rodham Clinton's close association with reform further
reinforced the gendering effect of these frames. Of course, as a woman,
Clinton would have raised the salience of the gender schema among the
public due to her prominent participation (Huddy and
Terkildsen 1993). More importantly, as head of the administration
task force, “violate[d] the traditional separation of the
masculine sphere and the feminine domestic sphere that ha[d]
previously defined the role of First Lady” (Burrell 1997, 18). Consequently, she became the focus
of public debate on changing gender roles in 1993 and 1994. Moreover, her
role put a woman in charge of reforming the traditionally male-controlled
health care industry (Burden and Mughan 1999;
Burrell 1997). As Skocpol argues:
Hillary Rodham Clinton could easily appear “too
strong” in relation to a husband many thought was “too
weak.” She also symbolized the increasing presence and assertiveness
of career women, whom many people—including men in elite,
professional positions—secretly or not so secretly fear and
hate…. Cartoonists and talk radio hosts could ridicule the Clinton
plan for its alleged governmental overweeningness—and in the process
subliminally remind people how much they resent strong women. (1997, 152–53)
Her association thus served to reinforce the gender implication
inherent in the issue rhetoric over reform.
Hypotheses
My expectation is that the frames deployed during the reform debate
influenced opinion on health care reform. Specifically, I anticipate that
the 1994 debate made Americans much more likely to evaluate health care
reform through a gender schema. I hypothesize that opinion on health care
was more strongly associated with gender ideology in 1994 than in other
years. I expect that in 1994, Americans who held traditional gender views
would oppose reform more (or support it less) than otherwise similar
gender egalitarians. In addition, for the theoretical reasons discussed
above, I hypothesize that these patterns of gender implication operated
similarly among men and women. While men and women may differ in their
average position on the evaluative dimension of the gender
schema, both men and women should have applied the gender schema to their
evaluation of health care reform in 1994.
DATA AND MEASURES
I use data from the National Election Studies to test these
hypotheses. The NES includes a question about respondents' support
for a government insurance plan to address rising health costs.8
Cumulative file variable CF0806. With minor
variations, this item reads: “There is much concern about the rapid
rise in medical and hospital costs. Some people feel there should be a
government insurance plan which would cover all medical and hospital
expenses for everyone. Others feel that all medical expenses should be
paid by individuals, and through private insurance plans like Blue Cross
or other company paid plans. Where would you place yourself on this
[7-point] scale, or haven't you thought much about
this?”
This general measure has several advantages,
compared with questions that focus specifically on the Clinton plan.
First, because I seek to compare gender implication over time, I need a
consistent measure, rather than one tailored specifically to any
particular year. Second, this measure represents a somewhat conservative
test of gender implication. If the 1993–94 debates engendered
opinion on the administration's plan
and nothing else, that
would not say much for the scope of gender implication generally. I am
precisely interested in seeing whether a wide-ranging and symbolically
rich debate had effects on opinion within the domain of government action
and health care more generally. Finally, there is precedent for the use of
this standard NES measure in analyses of the effects of health care reform
on opinion (
Koch 1998).
Gender Ideology
There are two measures in the NES that capture elements of the gender
schema. The first is the women's equal role item, which asks whether
women should mainly stay at home or be equal with men in all sectors of
society.9
Cumulative file variable CF0834.
With minor variations, this item reads: “Recently there has been a
lot of talk about women's rights. Some people feel that women should
have an equal role with men in running business, industry and government.
Others feel that a women's place is in the home. Where would you
place yourself on this [7-point] scale or haven't you
thought much about this?”
This item is ideal, in that it
addresses the intersection of gender and social roles and duties, and
focuses on what roles men and women
should have in society. It
captures the public/private distinction, it focuses on the cognitive,
and it avoids gender identity and the details of current political
conflict over gender. The only disadvantages with this measure are that it
is a single item, and that it is somewhat skewed toward the progressive
response.
The second measure comes from the thermometer score battery, in which
respondents were asked to rate their feelings about the women's
movement and/or feminists on a zero to 100 scale.10
A rating for “the women's movement” was
included in 1992, 1994, 1996, and 2000; “feminists” in 1988,
1992, and 2000.
Both feminists and the women's movement are
closely associated with efforts to make gender arrangements more
egalitarian, and so people's positive or negative evaluations of them
should relate closely to their own beliefs about proper gender
arrangements (
Huddy, Neely, and LaFay 2000). My
strategy was to build a single composite measure by averaging the equal
role item and whichever thermometer score is available in a given
year.
11 The items are reasonably highly
correlated with each other. Cronbach's alpha for the combined scale
is 0.49 (0.51 among women, 0.46 among men). This measure has a mean of
0.64 and a standard deviation of 0.23. The measure correlates quite highly
with alternate gender measures from the NES in the few years they are
available, and predicts opinion strongly on gendered issues such as
abortion. A complete reliability and validity analysis is available from
http://falcon.arts.cornell.edu/nw53.
Control Variables
Because the debate over health policy included calls to principles of
egalitarianism and limited government, I include measures of these two
political principles. For egalitarianism, I use the six-item NES scale
(Feldman 1988). For limited government, I
construct a scale from two items that assess support for government effort
in specific programmatic areas: The first asks respondents to indicate the
degree to which the government should see to it that all Americans have a
job and a good standard of living, and the second asks respondents to
evaluate the trade-off between the government supplying more services
versus cutting spending.12
The NES sometimes
includes a three-item scale that measures support for limited government
(Markus 2001); unfortunately, these items did
not appear in the crucial 1994 study. The two items I employ are less
abstract than those in the Markus scale, and it could be argued that they
represent a policy opinion dependent variable, rather than a
predisposition. However, there is precedent for using them as a
predisposition (Kinder and Sanders 1996).
Moreover, entering this measure of support for government action is
conservative. Insofar as the scale picks up social policy preferences
above and beyond “principled” feelings about the role of
government, its explanatory power in the model will be biased away from
zero. The additional variance that it explains will come at the expense,
potentially, of the measures of gender predispositions. In any case,
substituting the abstract measure, when available, does not change the
substantive gender-implication findings.
In addition, the model includes measures of partisan and ideological
predispositions (entered as a series of indicator variables), and a set of
social location variables, including class (indicators for income and
education categories), gender, race, age and age over 65, permanently
disabled status, and being unmarried. All variables in the model are
scaled to run from zero to one, with one representing the liberal response
for nonindicator variables.
I ran a series of regressions, one per year, of support for government
health insurance on gender ideology and the control variables.13
As the dependent variable is measured on a
seven-point scale, regression is a reasonable estimation strategy that
makes interpretation particularly easy. In any case, and as usual, the
substantive results are identical when estimated by ordered
probit.
I ran this model for presidential years from 1988 through
2000, and for 1994, providing two years on either side of the crucial 1994
study for comparison.
RESULTS: GENDERING OF HEALTH CARE
OPINION
Table 1 presents the results from this
model. The first row gives the effect of gender ideology on health care
opinion in each year. In years other than 1994, health care opinion is
slightly gender-implicated. The coefficients vary around an average of
0.066 and are on the edge of statistical significance. This is a small
effect—compared to gender traditionalists, the most egalitarian
respondents are 0.066 more supportive of government health care, which is
less than half of the distance between two points on the seven-point
scale.
Gendering of health care opinion 1988–2000
The impact of gender ideology on health care opinion is three times
larger in 1994 (b = 0.198, p < 0.01). Now the most egalitarian
respondents support government health care by just over one point on the
seven-point scale, compared with the most traditionalist respondents. This
supports the hypothesis that the frames deployed in the debate
gender-implicated health care opinion in 1994. These results are
illustrated in Figure 1, which depicts
graphically how health care opinion varies with gender ideology, for an
otherwise-average respondent.
Impact of gender ideology on health care opinion, 1992–1996.
(Source: National Election Studies. Figures show predicted
opinion [with 95% confidence interval shaded], based on the
models presented in Table 1. Gender ideology
varies from zero to one; other variables set at their sample
means.)
It is interesting to note that although support for limited government
is substantially related to health care opinion, the impact of limited
government is not noticeably larger in 1994 (average coefficient
−0.476, p < 0.01 in all years). Despite opponents' emphasis
on the specter of government bureaucracy, citizens' feelings about
the appropriate size and scope of the federal role played no stronger a
role in 1994 than they played throughout the late 1980s and 1990s. This
provides additional indirect evidence that the frames deployed in
1993–94 did not resonate particularly with fear of the federal
government in the abstract. Rather, this rhetoric—combined with
claims that the plan would interfere in the private realm of health care
and family—resonated with the gender schema, and thereby increased
the association between gender ideology and opinion.
Several other results are interesting. Controlling for the other
factors in the model, partisan differences sharpened slightly in 1994, as
we might expect given the partisan nature of the debate and of the 1994
congressional campaigns. Republican identifiers were 0.104 more
conservative on health care in 1994, compared with independents. In 1988
and 1992, on the other hand, they were only 0.066 more conservative on
average. This partisan difference has faded slowly since 1994, with the
estimated coefficient back to −0.063 in 2000 (p < 0.01 in all
years).
The results for egalitarianism parallel those of Jeffrey Koch (1998): It is strongly associated with opinion in 1992
as a result of the egalitarian frames deployed during the campaign. When
the debate shifted away from egalitarian frames in 1994, Americans became
less likely to view health care through an egalitarian lens.
One last finding bears mention: Once all the other factors are taken
into account, differences between men and women are negligible. The
relatively modest gender gap on health care noted here operates through
the various other predispositions included in the model.
Feelings Toward Hillary Rodham Clinton
I suggested that Hillary Clinton's role in health care reform
should have operated symbiotically with the gendering rhetoric to solidify
the gender implication of health care. Another possibility is that the
apparent increase in gendering merely reflects the association of Hillary
Rodham Clinton, who is herself gendered, with health care policy. To
assess this possibility, I ran a model of health care opinion that adds
the thermometer score rating of Hillary Rodham Clinton in the years
following 1992, when it is available. Table 2
presents the coefficients of interest from this model.14
Gendering of health care opinion 1992–2000, model with Hillary
Rodham Clinton evaluation
These results confirm that feelings about health care did become
associated with feelings about Clinton in 1994. In 1992, one's rating
of Hillary Clinton is barely related to health care opinion (b = 0.059, p
> 0.10); by 1994, that rating is substantially related to opinion (b =
0.130, p < 0.01). The association fades to about half that in 1996 (b =
0.076, p < 0.05) and falls further in 2000. The inclusion of Hillary
Clinton ratings reduces the estimates of gender implication by about a
quarter from 1992 through 1996, but if anything, sharpens the central
finding that gender implication was stronger in 1994 than in other years.
Health care was gender-implicated in 1994 both directly and by its
association with a prominent and highly gendered first lady.
Subgroup Analyses
Gender
These results show that women's and men's opinions differ
little from each other once we take account of gender predispositions and
the other independent variables. As I have argued here, I expect men and
women to react similarly to gendered framing. To explore this expectation,
I ran the gendering analysis separately among men and women; the relevant
results are presented in Table 3. Women may be
slightly more prone to perceive health care through the gender
schema in the years other than 1994 (the average coefficient is 0.075
among women and 0.051 among men in these years). Most importantly, women
and men reacted identically to the gendering rhetoric of 1993–94. In
1994, the effect of gender ideology on health care opinion is 0.196 among
women and 0.202 among men, which confirms the expectation that men and
women would react to the gendering rhetoric similarly.
Gendering of health care 1988–2000, by
respondent gender
Partisanship
We might also expect citizens' partisan attachments to have
conditioned their reactions to the very partisan health care debate. To
assess this possibility, I ran the basic model separately among
identifiers of the two major parties and among independents. Table 4 presents the relevant results.
Gendering of health care 1988–2000, by partisanship
Here, we do find some variation. Democrats and independents followed
the pattern observed so far: They gender health care quite substantially
in 1994 (b = 0.252 and 0.224, respectively), and much less both before and
after (average b = 0.054 and 0.041). In contrast, Republicans seem
entirely unaffected by the gendering discourse of 1993–94: in 1994
they gendered health care a bit less, if anything, than in other
years (b = 0.109 in 1994, compared to 0.157). It is not entirely clear why
this would be the case for Republicans, but it does suggest that the
gendered frames employed by opponents were effective in broadening
opposition to the Clinton plan by appealing in particular to
gender-traditionalist Democrats and independents.
Political Engagement
Finally, we know that citizens vary greatly in the attention they pay
to politics and in their exposure to political discourse (e.g., Converse 1990; Zaller 1992).
If changes in political discourse truly caused the effects I observed
here, then those effects should be strongest among respondents most
exposed to the discourse. Insofar as the gendered discourse was subliminal
and people were not aware of the gender implication, I expect that
gendering should increase with the reception of gendering messages, which
should itself increase with political engagement. To assess this
expectation, I divided each year's sample into thirds, using a
multi-item political knowledge scale (Zaller
1992). Table 5 presents the results of
the health care opinion model, run separately among these groups.
Gendering of health care 1988–2000, by
political engagement
As expected, political engagement sharply conditions the results. The
least-engaged respondents reacted not at all to the gendering rhetoric. It
would seem that the relatively subtle nature of the gender implication
passed them by in 1994. Middle- and high-information respondents, on the
other hand, reacted sharply to the gendering rhetoric of 1993–94.
Before the reform debate, middle-information respondents gendered health
care a bit (b averaged 0.058 in 1988 and 1992, n.s. both years). In 1994
the impact of gender ideology is much stronger (b = 0.285, p < 0.01),
and it drops back essentially to zero in 1996 (average b = 0.017 in 1996
and 2000).
Highly engaged respondents also gendered health care much more in 1994
(b = 0.225, p < 0.01) than in 1988 and 1992 (average b = 0.111). Among
this well-informed group, however, the effect persisted through 1996 (b =
0.197, p < 0.01) before fading by 2000 (b = 0.122, n.s.). Thus, those
who pay at least moderate attention to politics picked up on the gendering
rhetoric, and the best-informed remembered it for some time.15
Egalitarianism also shows an interesting
pattern among the top two-thirds in political information. From 1988 to
1992, these respondents came to frame health care much more in terms of
equality; they then abandoned the egalitarian frame for the implicit
gendered frame in 1994. These findings for the moderating role of
political information are consistent with those of Koch (1998), who also found that those with moderate
information were the most influenced by the reform debate.
The
fact that political engagement conditions gendering so sharply serves as
additional confirmation that this gendering was driven by the political
discourse, insofar as only those who were reasonably engaged in politics
were affected by it.
CONCLUSIONS
Clearly, there are many deep roots of the failure of health care
reform in 1994. The American political system makes major policy
innovation difficult to achieve under the best conditions, and health
reform has failed repeatedly during the twentieth century. Many factors
having nothing to do with public opinion contributed to the failure in
this instance (e.g., Hacker 1997). However, the
administration's choice of a public opinion strategy does raise the
question of what impact, if any, the gender implication of opinion had on
the overall fate of the reform effort and whether there might have been
ways the administration could have countered the gender implication.
Because gender implication implies polarization on gender ideology, it
is theoretically possible that gender implication increased
support for reform by increasing the support of gender egalitarians over
what it otherwise would have been. However, gendering frames came
primarily from reform opponents, so it is reasonable to assume that the
net effect of gender implication was to depress opinion by reducing the
support of gender traditionalists below what it would have otherwise been.
This is consistent with Figure 1, which
indicates that (holding constant the other factors in the model) from 1992
to 1994, support among the most gender-traditional fell by 0.170 on the
zero–one scale (from 0.524 to 0.353); this corresponds to a change
of about one point on the original seven-point scale. In contrast, support
among the most gender-egalitarian fell by 0.058 (from 0.609 to 0.551).
Support for the plan fell across the board, but the drop was about three
times as large among gender traditionalists. An instructive exercise is to
imagine that support among gender traditionalists had not declined any
more precipitously than among egalitarians—that support declined by
0.058 across the board. In this scenario, the left-hand end of the line in
the middle pane of Figure 1 would be rotated
upward until the 1994 line had the same slope as the 1992 line, albeit at
a lower level. In this case, the overall average opinion in 1994 would
have been 0.524 on the zero–one scale—about equal to its level
in 1988.16
Of course, this is entirely
hypothetical and heuristic. If we imagine instead that gender egalitarians
dropped as much as traditionalists between 1992 and 1994 (which would
imply equalizing the slopes in Figure 1 by
rotating the right-hand end of the 1994 line downward), average
opinion in 1994 would have been 0.412. Without experimental evidence, we
cannot be sure which scenario, or what intermediate one, in fact would
have obtained without gender-implicating framing.
This is not
overwhelmingly higher, but still represents a potentially significant
difference politically. It is implausible that this difference by itself
would have turned the tide of opposition to the administration plan, but
it does suggest that gender implication added an additional nail to the
coffin.
17 Lawrence Jacobs and Robert Shapiro
discuss the ultimately limited influence of public opinion on
congressional action, although they do acknowledge the ways that fading
public support contributed to the loss of an important group of moderate
Republican legislators (2000,
125–48).
If we grant for a moment that the gendering frames depressed net
opinion and thereby hurt reform, how might the administration have
countered or avoided the gender implication? Unless the plan itself had
been radically different, the administration could probably not have
avoided antigovernment frames from opponents. However, if the Clinton team
had managed to keep that aspect of the debate focused on who should be
responsible for paying, rather than who would be responsible for
making health care decisions, then the specifically gendered impact of
opposition frames might have been muted. Mark Schlesinger finds that
Americans are much more supportive of government financing of health care
compared with government influence on the content of health care provision
(Schlesinger 2004); perhaps if the
administration had not opened the door by focusing on personal benefits,
then opponents would have been prevented from deploying the potent
combination of limited government and private-sphere interference frames.
Or perhaps not. In any case, it seems likely that a traditional debate
over the relative efficiency of government versus private-sector provision
and over the need for a systemic approach to universal coverage would have
been better for reform advocates.
More broadly, the results presented in Table
4 indicate that the impact of gendering was strongest among
Democrats and independents, which suggests that the gendering frames were
particularly effective at decreasing support among these groups. Thus, the
gendered frames may have been particularly effective for opponents of
reform insofar as they separated gender-traditionalist Democrats (and
independents) from the rest of the Democratic coalition. In this sense,
gender implication may have served an analogous role to the implicitly
racialized rhetoric deployed by Republicans to attract support from
racially conservative Reagan Democrats in the 1980s (Edsall and Edsall 1992; Kinder and
Sanders 1996).
Gender issues have come on and off the political agenda over the
years, but in contrast with matters of race, gender issues have not served
as a fundamental basis of partisan alignment.18
Although Christina Wolbrecht (2000)
demonstrates that parties' elites polarized on gender issues
beginning in 1980, she does not explore the relationship between that
polarization and mass opinion, nor the relationship among the public
between gender attitudes and partisanship.
This likely means that
elite debate does not invoke gender frames as frequently as racial frames,
and that the public is therefore less well trained to view political
issues through the “lens of gender.”
19 This is consistent with Burns and Kinder's (2003) findings that people's explanations for
gender inequality predicted opinion much less pervasively than did their
racial explanations, and also with my finding (Winter
2001) that racialization is much easier than gendering to induce
experimentally.
Nevertheless, the current analysis shows that
gender
can serve as an organizing principle for a political issue
under the right circumstances. We can imagine this happening for other
issues, when political elites choose frames that trade on gender schemas
among the mass public and convey those frames loudly enough. Because
gender is not central to the mainstream partisan alignment, we would not
expect this to happen often. However, precisely for this reason, gendering
can be a useful strategy for fracturing an existing coalition (e.g.,
Riker 1986). Just as the gendered frames moved
gender-traditional Democrats and independents against health care reform,
we might expect there to be other issues where Republicans can use
gendered frames in this way. In fact, Republicans' ability to attract
gender-traditionalist Democrats and independents with implicitly gendered
political rhetoric is a pattern we may be seeing continued today with the
explicit emphasis on so-called cultural issues, many of which involve
matters of gender ideology at their heart.
More broadly, this article demonstrates that gender can matter for
public opinion in ways that go beyond our current approaches. Virginia
Sapiro lays out a typology of three ways that public policies may be
gendered: because they are “manifestly about gender,” because
men and women have “different experiences, needs, or problems”
relating to the policy, or because policies inadvertently affect men and
women differently. She points out, though, that there is no necessary
correspondence between the gendered content of policies and the
public's perception of those policies in gendered terms, and
suggests that more research is needed for “investigating the
conditions under which culturally derived stereotypes and frames are
activated” (2003, 619–20). This
article is an example of this kind of research for a policy that the
public does not consciously associate with gender.
The theoretical approach presented here, as well as the results for
this case, suggest that one important route to the gendered
perception of issues—what I call gender
implication—is a correspondence in structure between elite frames
and mass schemas. It further implies that gendered issue
perceptions can be largely or entirely symbolic and metaphorical:
The gender implication of health care opinion in 1993–94 turned not
on the fact that women and men have different health problems. Rather,
gender implication occurred because the frames that were deployed
structured reform as interfering metaphorically with intimate power
relations within the private sphere of health care provision.
