Content-Type: text/html
The Influence of Structural Forces on the
Reporting of Racially Comparative Risk
Oscar H. Gandy, Jr., Katharina Kopp, Tanya Hands,
Karen Frazer and David Phillips
Annenberg School for Communication
University of Pennsylvania
3620 Walnut Street
Philadelphia PA 19104-6220
215-898-7030
Fax: 215-898-2024
[log in to unmask]
Paper prepared for submission to the Faculty Research Paper
Competition, Minorities and Communication Division, Association for Education in
Journalism and Mass Communication, 1996.
The Influence of Structural Forces on
the Reporting of Racially Comparative Risk
Abstract
Content analysis of 1,245 articles published in 29 newspapers
demonstrates the existence of structural influences which determine the ways in
which newspapers frame stories about differences in the risks that whites and
blacks face. Structural characteristics of a newspaper's market, including the
proportion of African Americans and estimates of their political power, are
revealed as important influences on the selection of headlines and leads for
these stories.
The Influence of Structural Forces on the
Reporting of Racially Comparative Risk
Abstract
Substantial evidence suggests that the way stories are framed
influences exposure, processing and retention of the information contained in
those stories. Framing is therefore an important factor in determining how
citizens understand issues of personal and social concern. There is also
important evidence that indicates that structural factors, such as community
pluralism, influence how newspapers frame issues of public concern.
Content analysis of 1,245 articles published in 29 newspapers
demonstrates the existence of structural influences which determine the ways in
which newspapers frame stories about differences in the risks that whites and
blacks face. Structural characteristics of a newspaper's market, including the
proportion of African Americans and estimates of their political power, are
revealed as important influences on the selection of headlines and leads for
these stories. Structural features reflecting journalistic traditions are seen
to influence the framing of stories about particular disparities in different
ways. Economic measures, representing differences in the relative economic
status of whites and blacks and recent trends in circulation are among the most
important predictors of the use of the discrimination frame.
The influence of structure on the reporting of racially
comparative risk
Introduction
In ways reflecting the trends within political discourse around
the globe (Schiller, 1995), political leaders in the United States have raised
serious questions about the role the government should be expected to play in
guaranteeing the basic necessities of life. While the debate is not always
framed in racial terms, and talk about equality has been subordinated to
discussions about efficiency and global competitiveness, social programs
originally designed to improve the social status of African Americans are being
threatened by the so-called "Republican revolution"( Bowie, 1988; Starobin,
1993; Steinberg, 1995; Verba and Orren, 1985).
The US Congress and numerous state legislatures have begun active
reconsideration of the merits of affirmative action. This examination follows an
apparently widespread commitment to "reinvent" if not eliminate the social
welfare programs that provide assistance to the poor and disadvantaged. Debates
surrounding the development of the National Information Infrastructure have
raised the very real possibility that even the long-standing national commitment
to universal service will be weakened dramatically, if not eliminated altogether
(Browning, 1994).
These political discussions are based in part on an assessment of
the success or failure of past efforts. The speed with which this process of
reexamination has been able to move forward reflects a perception that is
apparently shared by a substantial part of the population,--that programs have
either failed miserably, or have succeeded beyond our expectations. Both
conclusions lend support to a demand for change in social policy.
Of course, because of their scope, these perceptions can not be
based on personal observation, but are dependent upon the analysis and
representation of these matters in the mass media. The media are important
sources of information we use in forming our impressions of the extent and
distribution of hardship in society. They also provide causal analyses that help
to shape these impressions by pointing to the individuals and institutions that
bear primary responsibility for these problems (Baumgartner and Jones, 1993;
Iyengar, 1991). Ultimately, these impressions form the basis of public support
or opposition to specific government policies. When causal analysis suggests
that the victims bear the responsibility for their circumstances, either because
of their actions, or because of their failure to act, people who say they
believe in equality are still likely to express opposition to government
programs designed to support such goals. People identified as victims are no
longer seen to deserve affirmative action or other interventions on their behalf
(Sniderman, Brody, Tetlock, 1991).
The commitment to equality
For the general public to support programs designed to eliminate
racial disparities, there seem to be two fundamental requirements. First, the
public must see this inequality as substantial, and second, this inequality
should not be readily explained in terms of individual responsibility. For
example, press coverage might help the general public see a 20% disparity in of
rates of unemployment among black and white youth as a significant social
problem. But, before they are ready to act, or to support government action,
there would also need to be some evidence of discrimination, or the influence
of structural factors beyond invididual control. Press coverage of the decline
of industrial investment in predominantly black areas would provide the kind of
information that would make it difficult to blame the victims for being
unemployed.
Unfortunately, there are no objective standards which are brought
to bear that determine which disparities will be seen as significant, or which
structural factors will be assessed as external determinants. Indeed,the data
from opinion surveys would suggest that the race of the respondent predicts the
use of entirely different standards and explanatory models when questions of
race are involved in the issue(Morin, 1995).
Among social scientists outside the discipline of communication,
the mass media are not usually seen as important sources of observed differences
in social perception. For example, Lee Sigelman and Susan Welch (1991) explore
the striking differences between whites and blacks in their perception of racial
inequality. While their analysis is detailed and provides solid evidence of
perceptual gaps between the races, these authors rely primarily upon social
structural factors linked to direct experience as the likely origins of these
different perceptions. These theorists seem unaware of the influential role of
the mass media in the cultivation of social perceptions, and only reluctantly
include explanations based in personality or other psychological
domains(Signorielli and Morgan, 1990; Potter, 1994; Tapper, 1995).
Among communication scholars, racial differences are not always
seen to be reliable predictors of perceptions and related policy views. Richard
Allen (1994), for example, has explored the influence of social location and
media use on the commitments of blacks and whites to structural equality. Using
a comprehensive structural equation model approach, Allen analyzed data gathered
for the 1988 National Election Survey. He found in general, that greater
exposure to television news was associated with less willingness to support
policies directed toward improving structural equality.
An explanation for this outcome may be found in work by
Armstrong, Neuendorf and Brentar (1992), who suggest that "beliefs about actual
socioeconomic outcomes must provide a foundation upon which racially related
political and social disputes are argued and racial conflicts over political and
economic resources are played out" (p.173). They suggest that assessments
regarding the fairness of the distribution of socioeconomic resources are formed
"at least in part from patterns of exposure to relevant mass media content"
(p.154).
Depending upon the ways in which the problem of inequality is
framed, then, press coverage may lead citizens toward, or away from support of
particular public policies. Although the response of individuals to media
constructions seems also to vary in response to similarities and differences in
their background and social circumstance, the influence of media is substantial.
The perceived fairness regarding socioeconomic outcomes is likely to be based in
part on the assessments that people make regarding the extent of meaningful
differences between racial and ethnic groups, as well as the causal explanations
for those differences that are made available in the mass media.
Inequality perceived
Unfortunately, the predominant orientation, even within
communication research is toward explaining common, rather than diverging
perceptions of social reality. Mainstreaming, a modification of the cultivation
hypothesis associated with Gerbner, Gross, et al,(Signorielli, 1990) argues that
the differences in social perception that are likely to be produced by
differences in concrete social experience, are likely to be moderated, or even
erased through extensive exposure to prime time television fare. Even though it
is clear that race-linked differences in the perception of social reality exist,
and may in fact contribute to racial hostility (Williams, 1992), the reality of
differences between groups is often either ignored, or is treated as something
to be explained away as an artifact of study design.
Communication scholars in general have been interested more in
differences between media forms, than in differences between media consumers.
For example, on the basis of research indicating an increase in the portrayal of
African-Americans in integrated settings (Gandy and Matabane, 1989), Armstrong,
Neuendorf and Brentat (1992) hypothesized that the greater one's exposure to
fictional television, the higher black socioeconomic outcomes would be perceived
relative to those of whites. The opposite relationship was hypothesized for
exposure to television news. As predicted, exposure to TV entertainment was
associated with a more sanguine view of black socioeconomic outcomes, with
exposure to TV news contributing to less positive views. The strength of these
relationships was greater among white students who had only limited opportunity
for direct interaction with African-Americans. Beyond noting this, there was no
effort to explain the differences in the perceptions among blacks and whites.
The authors' primary concern was with the media's construction of a broadly
held, negative view of blacks that might lead toward increased opposition to
affirmative action.
Robert Entman's efforts to explore the concept of modern racism
through an examination of television news (1990; 1994) move in a similar
direction. His analysis of network news suggested that nearly 60% of network
news stories centered on negative news about blacks. "The third most common
topic was blacks as victims of social misfortunes other than crime, such as
fires, poverty, bad schools, and racial discrimination" (1994, p. 511). But
Entman also notes that the ways in which the news media cover blacks may have an
impact on whites different from the way it may affect blacks. The fact that
there are relatively few blacks in the news, and the fact that whites have
relatively few contacts with blacks, may lead whites to treat those few blacks
as representative of the entire population. This is thought to operate in part
as a mechanism of outgroup classification (McAdams, 1995). Entman suggests that
the "essence of racial prejudice is homogenizing and generalizing about the
disliked outgroup: a tendency to lump most individual members of the outgroup
together as sharing similar undesirable traits, while seeing one's own group as
a diverse colection of clearly differentiated individuals" [Entman, 1994, 517).
From our perspective, we think it important to recognize that
important differences in perception do occur, and that they are influenced by
the ways in which stories are framed. The literature in cognitive science that
relates to the processing of information argues that mental structures, referred
to as schema, are generated, strengthened and linked to other images and
impressions from past experiences (Graber, 1984; Gunter, 1987). This literature
informs our thinking about the ways in which differences in social experience
may interact with media framing to produce racial and ethnic differences in the
perception of inequality.
Framing and social perceptions
A number of social psychologists, led in part by the pioneering
work of Amos Tversky and Daniel Kahneman (1977; 1990) have challenged
assumptions about economic rationality by pointing out the ways in which subtle
differences in framing can influence the choices people make between essentially
equivalent options. Generally this work focuses on shifts in preferences that
are produced through the use of evaluative terms, such as calling a cost a tax
rather than a charge, or by suggesting cues that lead respondents to associate a
single case with a particular distribution in ways that bias their assessment of
probability (Kahneman, Slovic and Tversky, 1982). These framing cues might,
however also involve racial identifications that would trigger implicit
evaluative responses in certain subgroups of the general population.
For example, when people make a choice between two policy options
based on the ways in which the equivalent risks to some mysterious disease are
framed, it is clear that rules regarding invariance are violated. That is,
respondents are risk-avoiding under one condition, and risk-seeking under
another, simply because the options are presented in terms of lives saved,
rather than lives lost (Tversky and Kahneman, 1990).
Although we have found no specific test of our assumptions about
the influence of racial framing in the press, we believe the literature on
racial and ethnic group identification provides a basis for assuming that
identifying the potential victims by race would produce significantly different
shifts among respondent populations that classified by race. Several studies
reported by social psychologists indicate quite subtle and persistent negative
stereotypes of blacks are readily evoked, even in respondents who claim to
harbor little racial prejudice (Devine, 1989; Gaertner and McLaughlin, 1983).
We note that the most consistent differences between white and
black opinions over the years seem to be with regard to policies that apply to
matters of racial inequality (Page and Shapiro, 1992). The direction in which
these opinions differ clearly reflects an identification by blacks with the
circumstances of other blacks.
To the extent that framing involves statistical representations
involving social groups, there is also a strong basis for concern about people's
understanding of the facts as presented (Mauro, 1992). Evaluations, including
feelings of threat, appear to be linked to misperceptions of the relative
proportions of particular racial groups within the population(Nadeau, Niemi and
Levine, 1993). The data seem to suggest that people who feel threatened by
blacks or Hispanics will tend to overestimate their numbers. This tendency is
even greater in areas where minority populations are actually relatively high.
These misperceptions may interact with the ways in which stories about minority
groups are framed.
Additionally, we have been concerned that efforts to use media
for social purposes have ignored the fact that messages prepared specifically
for one group are also likely to be read by members of other groups. We believe
that the impact of those messages designed to mobilize one group may be quite
different and problematic for members of the groups that are identified as
victims.
Given what is known about the ways in which the level of
political involvement varies by race and class (Milburn, 1991), it is likely
that investigative journalists who prepare stories about some form of outrageous
action, such as discriminatory lending practices by banks, will write their
stories so as to attract the attention of upscale, politically active white
readers. The fact that such stories generally frame blacks as victims, and that
such framing may influence how black readers assess their own circumstances is
not likely to be taken into account by investigative journalists concerned with
producing social change (Protess, et al, 1991).
The Influence of Structure
The press is frequently criticized for reflecting ideological or
racial bias (van Dijk, 1988; 1991; Small, 1994). We sought to explore the kinds
of structural forces that might influence the ways in which the news media frame
stories about race that reflected more complex and more subtle influences. Those
influences are structural and operate on several levels of the news organization
at the same time.
One structural influence that may be seen to influence the
framing of race is the racial composition of the newsroom. Stocking and Gross
(1989) provide evidence to support the view that journalists make use of a
causal hypothesis to guide their construction of stories (Stocking and Gross,
1989). White journalists are likely to differ from their African American
colleagues in the extent to which they believe individual or institutional
racism is the primary influence over the outcome of some conflict. Depending
upon the number and influence of minority journalists within the organization,
the mere presence of black journalists on staff may influence the coverage and
framing of stories with a racial component.
John McManus (1994) among numerous others who have commented upon
the challenge to journalistic values represented in the turn toward markets and
a marketplace logic (Baker, 1994b; Hirsch and Thompson, 1994), expresses concern
about the consequences that flow from this trend. This is a concern about
injustice. It seems unfair that the economic subsidy that advertising represents
actually goes to support the tastes and preferences of a readership that could
actually afford to pay the full costs of a high quality information product. As
a result, it seems likely that the specialized information needs of the poorer,
minority audiences will not be served, because the information they require will
not be the subject of upmarket media features.
The literature on segmentation and targeting that reflects a
growing concern by editors and publishers about the composition of newspaper
readership (Bogart, 1991) has yet to explore the ways in which news content
changes as the racial character of the readership changes. The concept of
audience as a product produced by means of a technology that involves the
careful combination of different kinds of editorial material is just beginning
to be articulated theoretically (Webster and Phalen, 1994). Audience production
functions have been estimated, but there has been little effort to observe how
these functions change as the nature of the "preferred" audience changes.
It is also not clear whether editors respond to changes in the
composition of their potential, rather than their actual readership. To the
extent that the racial composition of a newspapers' audience matters, and
references to comments made by Otis Chandler at the Los Angeles Times suggest
that it does (Baker, 1994a, p. 68), then we should assume that changes may
become more salient as they reach that critical status sociologists refer to as
the "tipping point."
In housing, education and other social systems in which racial
composition is especially salient, there is an commonly observed phenomenon in
which "white flight" begins to accelerate and schools, or communities that were
predominantly white, almost overnight become predominantly black. We assume that
managing editors are sensitive to these structural changes, and may adjust their
racial content in ways that may appeal to the audience they are about to lose,
or the audience they are about to depend on for their survival.
Donohue, Tichenor and Olien (1972; 1985) suggest that newspapers
serving smaller communities tend to avoid conflict, while those serving larger
and more pluralistic communities seemed more willing to cover struggles within
the community. An important aspect of the community pluralism model is the
assumption of relative power. Editors are thought to respond to powerful
interests within the community. These interests might be economic entities, such
as large corporations which employ a large number of the newspaper's readers, or
they might be a racial or ethnic group within the community that wields
political or economic power. Whatever the case, editors are likely to treat
these interests with respect, and they will be careful about risking offense.
However, the notion of pluralism is one which suggests that power is rarely
concentrated, nor is it associated with a particular group from moment to
moment. As a result, as community pluralism increases, the power of any
identifiable group is thought to decline. As a result, newspapers and other
media are seen as relatively more autonomous, and therefore more able to risk
covering controversial issues (Griffin and Dunwoody, 1995).
Alternative explanations for this observation exist, however. In
finding that major market television stations "devote more attention to stories
on controversial issues of unpleasant aspects of community life," Carroll (1989,
p. 56) suggests that the larger the community, the greater the variety of such
stories that are actually available to be reported. Life in the big city is just
different! Although this hypothesis is not tested by comparisons of objective
indicators of status, we suspect that traditional "quality of life" measures
would indeed support such a view. By the same logic, communities with
proportunately more African Americans would have more cases of discrimination
from which editors might choose stories to publish.
Inequality and Risk
Although the literature within communications has begun to pay
considerable attention to the representation of risk in media (Singer and
Endreny, 1993), it has missed the opportunity that this subject presents for the
study of inequality and racial comparisons. The first oversight is no doubt due
to the emphasis within the area on technological rather than social risk
(hazards associated with the release of chemicals, radiation, or system
failures, as with airplane crashes). While there is just beginning to be some
discussion of racial differences in this area under the rubric of environmental
racism, the number of these studies is still quite slim (Krimsky and Plough,
1988). It is here, and in the general area of acceptable risk and the assignment
of risk that racial comparisons might be especially informative about the need
for a public policy response.
Although the literature has examined those risks that are
associated with individual decision making, especially health-related risks that
involve the decision to smoke, engage in unprotected sex, drive carelessly, and
indeed, racial group characteristics have been used to explain some of the
disparities in life expectancy, the extent to which these risks are also
reflections of structural or institutionalized racism has apparently been
subordinated to other concerns (Singer and Endreny, 1993).
Yet, it is clear that many risks are the result of decisions that
are made by individuals in ways that assign personal risks to others. Because
these decisions reflect either personal, or institutional preferences linked to
race, they produce significant racial disparities in access to goods, services,
or social opportunity. Many of these risks are covered in the press, and the way
these stories are framed is bound to influence both the perception of how
serious the disparity is, as well as the extent to which a policy-related
conclusion about discriminatory intent is warranted (Goshorn and Gandy, 1995).
Research Questions
On the basis of the literature that we have reviewed several
related questions present themselves for exploration.
1. What patterns are evident in the ways in which newspapers
frame stories of racial disparity?
2. To what extent do these patterns vary in relation to the
structural character of newspaper markets and newspaper organizations?
Design
In order to answer our research questions we developed a search
strategy that we believed would capture a large number of examples of stories
about the kinds of risks that affect African Americans that might also be
presented in racially or ethnically comparative terms. We made use of the
"papers" file within the DIALOG database. Although many of the 53 papers in this
file have been indexes since 1988, several were not included until 1990. The
analyses reported in this paper are limited to 29 papers which were included in
the database since 1989.
The initial sample was generated by a request to produce all
stories that included the word "blacks" occurring within ten words of "more
likely" or "less likely." This strategy generated more than 6000 individual
stories. We were cognizant of the difficulties associated with the use of
computerized databases, including the different outcomes that are produced by
varying the size of the "within __ words" parameter (Kaufman, Dykers and
Caldwell, 1993; Tankard, Hendrickson and Lee, 1994). We note, for example, that
the database excluded the Gannett newspapers (other than USA Today). This is
particularly troublesome because of the special editorial policies that Gannett
has developed regarding coverage of race and minority concerns (Hale, 1988).
Two coders made the determination of which stories to retain. On
the basis of a reliability test involving 56 stories, a 91% rate of agreement
was achieved regarding inclusion. Average agreement on 17 primarily dichotomous
variables was 89.4%, for which agreement was never below 72% for any item. A
third coder was added after approximately one third of the stories had been
processed and the level of agreement between coders on a 10 story comparison was
98.4%.
Risks were defined broadly to include negative outcomes that were
probabilistic or uncertain, and would therefore include stories about the
probability of being denied a mortgage, or a job, or an agressive medical
treatment. Many articles were rejected because they were about opinions, or
practices, such as might be generated by a story including the phrase, "Blacks
were more likely to vote democratic in the last election." This sampling
strategy produced a sample of 1,245 articles. The average number of articles per
paper was approximately 32, with 8 as the smallest, and 70 as the greatest
number of articles published between 1989 and 1993.
Seven dichotomous variables [coded 0 for absence, and 1 for
presence] were used to characterize the headlines. We sought to classify these
headlines in ways that might capture the editorial orientation of the newspaper
toward the framing of race and risk. Because the stories were frequently about
comparisons between whites and blacks, we noted whether the headline spoke about
disparity [DISPARIT], and/or made use of the term race, or its root as in racism
or racist [RACE]. Because disparity might be interpreted as the product of bias
or discrimination, or merely as an objective fact, we noted whether an
evaluative assessment was included in the headline [BIAS].
The fact of disparity need not always mean that blacks lost and
whites won. Therefore we noted whether the headlines indicated that blacks had,
or were likely to experience a bad or negative outcome [BLACKNOT], or if whites
had or were likely to experience a positive outcome [WHITES].
It is frequently the case that statistical evidence from the past
is framed in terms implying a future possibility, as in "blacks are more likely
to be denied." This is clearly an assumption, an inference drawn from
observations, and interpreted as representative of a recurring pattern, or a
situation likely to obtain in the future. We were therefore, interested to note
whether or not the headline made use of an "actuarial assumption", and used
terms like probability or chance to characterize the level of certainty about
some risk [ACTUARY]. Similarly, because claims or explanations might be
presented as controversial, we noted if headlines indicated that there was
uncertainty about the interpretation, cause, or process involved with the
particular risk [QUESTION].
The subject matter of the article was classified into 8
categories on the basis of the headline and lead: Health; Finance; Education;
Employment; Government Service; Criminal Justice; Income; and Other Social
Risks. We also determined the centrality of racial comparisons in an ordinal
scale ranging from 0-2: no focus (0), substantial, if racial comparisons
featured in the headline or in the lead (1), and primary, if racial comparisons
were in both the headline and the lead (2).
Seven variables (six of them dichotomous) were used to
characterize the lead paragraphs. We noted the level of sophistication in
statistics, ranging from none at all to ratios or coefficients in three levels
(0-2). Comparisons in terms of the likelihood of loss or gain, or negative or
positive outcomes were made for blacks and whites: 1) a high probability of
black loss [BLOSE], 2) a low probability of black success [BGAIN], 3) a high
probability of white success [WGAIN], and 4) a low probability of white loss
[WLOSS]. Finally, we assessed the level of statistical precision used in
probabilistic speech where "more likely" was considered to be less precise than
"much more likely," or some quantitative measure, such as a risk ratio.
In order to capture the structural influences we thought might be
reflected in the selection and framing of these articles, we made use of several
different sources. Statistical data were gathered at the level of the
Metropolitan Statistical Area (where it was available) for each of the markets
in which the papers in our sample were based.[1] On the basis of these data,
estimates of general population size (1990), average household income (1989),
and change in household income (1989-1979), black presence (computed as blacks
as proportion of general population in 1990), and change in black presence
between 1980 and 1990 were gathered or computed. Newspaper circulation in 1990,
and change in circulation, 1980-1990 (as computed, declines in circulation would
be positive), as well as a 1993 estimate of the proportion of minorities on the
newspaper's professional newsroom staff, were used as structural features of the
newspaper.[2]
We included several critical structural variables that we believe
reflected the political and economic power of African Americans in the newspaper
markets. Several were comparisons between blacks and whites expressed as
black/white ratios: household income, home ownership, educational attainment (%
with some college), percent unemployed, in addition to estimates of black infant
mortality, and black elected officials per 1000 black population as a index of
proportional representation (Karnig, 1976, 1979; Laveist, 1989, 1993; Taebel,
1978).
Findings
Searching on basis of a "more likely/less likely" frame produced
a sample in which nearly three quarters of the articles were about explicit
racial comparisons. Table One describes some of the more important
characteristics of the articles. Because the variables have been measured as
dichotomous dummy variables variables where the value of 1 is assigned to the
target characteristic, the mean of the distribution is the proportion of ones.
------------------------------------------
Table One about here
----------------------
A majority of headlines indicated that blacks lose, or experience
negative outcomes (Blacknot=52%). As more than 75 percent of the stories involve
racial comparisons to some degree, it is not surprising that 44% of the
headlines also framed the stories in terms of disparity. However, only a
relatively small proportion (11%) of the headlines framed the story in terms of
bias or discrimination and just slightly more identify them as being
specifically about race (14%).
Although, given the nature of the facts presented in the
stories, we might expect to find any one of four different risk framings within
the lead, the overwhelming editorial preference was for statements pointing to
the high probability of blacks experiencing a negative outcome (71.3%).
Of course, as journalists and their observers are quick to note,
there are traditional ways of framing stories. Stories about disparity involving
comparisons between whites and blacks are stories about black
loss...journalistic tradition makes it so! Yet, we suspect that even within the
general tendency to privilege black victimization, there are other generally
held views that construct crime stories differently from stories about economic
disparity, or about differences in health outcomes. Indeed, we suggest that just
as it was difficult for journalists to win acceptance of a frame that positioned
the president of the United States as a criminal following Watergate(Lang and
Lang, 1981), it is more difficult for journalists to frame doctors and judges as
engaging in discrimination than it is for them to construct bankers in that
fashion. Table Two presents the results of an analysis of the differences in
framing we observed when stories were identified by type or theme.
--------------------------------------------------------
Table Two about here
----------------------
Analysis of variance suggests that story type matters in nearly
all the comparisons, and as we suspected, it matters the most with regard to
whether or not the story is framed as being about discrimination, rather than
merely a story about disparity. Stories about economics, which includes stories
about the denial of mortgages, or the disparity in insurance coverage, indicates
a willingless to characterise these financial institutions as possibly engaging
in discriminatory behavior. Such a framing is incredibly rare in health stories,
and occurs only about one-third as often in stories about criminal justice.
-------------------------------------
Table Three about here
--------------------------
We used a similar analytical strategy to determine if authorship
was a determinant of framing. It is reasonable to expect that national stories
originating from wire services might be framed differently from local stories.
Although we also recognize that local editors frequently make up their own
headlines to accompany wire stories. As we see in Table Three, the framing of
stories does vary as a function of story source, but the differences are
generally not as striking as in the case of story type. We note in particular,
that framing with regard to discrimination does not vary significantly as a
function of story source. Local stories are least likely to use a headline that
emphasizes black loss.
----------------------------------
Table Four about here
--------------------------
Table Four presents an attempt to examine patterns in framing
along lines suggested by van Dijk (1991). We would expect a high correlation
between a story's lead and its headline unless for some reason an editor decided
to "promote" or "demote" some aspect of a story. While stories that are framed
as being about disparity are frequently matched by leads that talk about the low
probability of black loss, and the high probability of white gain (r=.14), such
headlines are rarely matched with leads indicating the high probability of black
loss. When placed in the context of knowledge that more than 70% of the leads
make such a statment, this particular co-occurrence is relatively rare. More
striking, perhaps, is the tendency for headlines that emphasize the racial
character of the story to be associated with leads emphasizing the high
probability of white success (r=.23).
--------------------------------------
Table Five about here
-----------------------------------
Table Five uses simple correlations to explore the relationships
between framing in the headline and lead and the structural characteristics of
the newspapers and their markets. The table includes only those variables that
had at least two significant correlations with framing measures. On that basis,
the most important structural influence appears to be a measure of economic
inequality, the ratio of white to black unemployment. White unemployment is
always smaller than black unemployment. Thus, the larger the number, the better
off blacks are relative to whites. The signs of the correlations indicate that
as the relative employment status of blacks improves, we find fewer stories
framed in terms of disparity, even though stories do tend to be framed in racial
terms. We also note, that in those markets where the comparative status of
blacks is better than in other markets, editors seem less driven to emphasize
the uncertainty which might surrount stories about disparity or discrimination.
however that under conditions of The papers in these towns are partial to leads
which talk about the high probability of black loss, or the low probability of
white gain, but were less likely to publish leads noting a low probability of
black gain.
Although significant, the magnitude of these correlations is
really quite small.
-----------------------------------
Table Six about here
-----------------
Table six represents an effort to examine the influence of
several predictors combined. A hierarchical design is used to enter predictor
variables into a regression equation. The blocks enter static measures of the
market structure, newspaper characteristics, changing characteristics of markets
and newspapers, and a final block which introduces attributes of the stories.
While each of the equations is highly significant indicating a genuine
relationship between stucture and content, the amount of explained variance
(R-squared) is quite small. Only the equation predicting the use of the bias of
discrimination frame explains more than 10% of the variance, and virtually all
of that explanatory power is linked to journalistic standards about how to tell
particular kinds of stories. Our measure of the tipping point, or change in the
black population, emerges as a significant influence in only one of the
equations, the use of a lead emphasizing the high probability of black loss.
Overall, we might interpret these Beta coefficients to suggest that in bigger
cities, especially those which have seen a relatively substantial increase in
the presence of African Americans, we find more stories with leads emphasizing
the high probability of black loss, especially those stories having to do with
the criminal justice system.
--------------------------------------
Table Seven about here
----------------------------
A final analysis (Table Seven) involves use of discriminant
analysis as an approach to classifying articles. In this case, we use the same
set of independent or predictor variables as used in the regression model, but
we are interested in estimating a discriminant function which assigns cases
correctly upon the basis of information contained in the variables defining that
function. In assessing the sign and magnitude of the correlations with the
discriminant function,[3] we note that the discrimination tag was most likely to
be used with stories about finance (the denial of mortgages), somewhat less
likely to be associated with stories about disparities in government service
delivery (eligibility for disability payments was a key story), and rarely
associated with stories about differences in health care delivery (here,
differences in coronary bypass surgery was a major story).
Discussion
We believe this preliminary analysis has made an important
beginning in our efforts to determine the nature of framing by the press in the
area of racially comparative risks. It is clear that there is considerable
variation in the ways that newspapers frame stories about the risks faced by
blacks, especially when blacks are compared with whites.
We have no basis for determining whether or not the press is
accurate in the ways it represents the risks in the social environment. Indeed,
risk assessment is a most uncertain science. There is already evidence that the
representation of risk in the media often departs from the best available
scientific measures. We should not be surprised that this departure occurs in
news in very much the same way as it occurs in prime time television fiction.
What we were concerned to discover is the way in which these risks are presented
in order to discern the ways in which those representations reflected the
influence of professional and structural forces. The data seem surprisingly
clear if we consider the ways in which these representations actually vary.
Our analyses demonstate quite clearly that the subject of the
story is an important determinant of the ways in which the story will be framed
in the headline and in the lead. This indicates that journalistic practice is an
institutionalized constraint on the framing of such stories. At the same time,
we would suggest that the links between story type and the selection of leads
may also reflect a constraint within the popular conception of risk (Beck, 1992;
Douglas and Wildavsky, 1982). This suggests that reporters, assume that their
readers will find it more difficult to accept discrimination as the explanation
for particular racial disparities. The regression and discriminant analyses
clearly support this conclusion in that the most important predictors in the
model are the indicators of whether the story is about health, or finance, or
criminal justice.
Yet, if we examine the published stories, and indeed, if we
examine the reports upon which the stories are based, there is considerable
evidence to suggest that there is in fact, racial discrimination in the delivery
of health care, just as there is in the administration of justice. The fact that
journalists, or their editors tend to avoid the assignment of that frame to
these stories has important consequences for social perceptions that inform
policy preferences. We suggest further, that because the selection of the frame
is so important in determining whether a story will be read, and if read, how it
will be understood (Iyengar, 1991), a tendency to avoid particular kinds of
frames may help to widen and reinforce disparate views of the nature of social
justice in society. The social consequences that this divergence may produce
ought not be underestimated.
Structural forces did not emerge as very powerful predictors of
story framing. The most influential structural measure is the ratio of black to
white unemployment. It was significant in six of ten correlations. The strongest
relationship was a negative correlation [r=-.16] suggesting that papers in
markets where the gap between blacks and whites was the smallest, were unlikely
to emphasize uncertainty about the facts or the analysis of published stories
about risk. Contrary to expectation, change variables were not especially useful
as predictors of framing. We believe some, but not all of this reflects problems
of measurement.
Although measures of black presence, change in black presence,
and relative change in black presence are all representations of the racial
composition of the market, the analysis is actually static, comparing five year
totals cross-sectionally. The critical measure needed is a longitudinal
assessment of changes in a newspaper's framing that can be linked to changes in
market structure. Few newspapers have published enough stories across the five
years to support such an analysis. Our hopes for attracting other scholars to
this question must also be evaluated in light of the historical difficulty that
communication scholars have faced in trying to link ownership and competitive
structure to the quality of editorial content (Hale, 1988). Yet,as Sig Gissler,
former editor of the Milwaukee Journal suggests, race is "America's rawest nerve
and most enduring dilemma" and we ignore it at great peril [Gissler, 1994).
References
Allen, Richard L. 1994. "Structural equality in black and
white." Howard Journal of Communications, 5(1):69-91.
Armstrong, G. Blake, Neuendorf, Kimberly A. with Brentar,
James.1992. "Tv entertainment, news, and racial perceptions of
college
students." Journal of Communication 42(3):153-176.
Baker, C. Edwin. 1994a. "Ownership of newspapers: The view
from the positivist social sciences." Research Paper R-12. The
Joan
Shorenstein Center. Harvard University, September, 1994.
-------1994b. Advertising and a Democratic Press. Princeton
NJ: Princeton University Press.
Baumgartner, F., and Jones, B. 1993. Agendas and Instability
in American Politics. Chicago:University of Chicago Press.
Beck, Ulrich. 1992. Risk Society. Towards a New Modernity.
London: Sage.
Bowie, Norman E. 1988. Equal Opportunity. Boulder
CO:Westview.
Browning, John. 1994. "Universal service. An idea whose time
is past." Wired. (September):102-105;152-154.
Caroll, Raymond, L. 1989."Market size and tv news values."
Journalism Quarterly,69(1):49-56.
Devine, Patricia. 1989. "Stereotypes and prejudice: Their
automatic and controlled components." Journal of Personality and
Social
Psychology, 56(1):5-18.
Donohue, George, Tichenor, Philip, and Olien, Clarice. 1972.
"Gatekeeping: Mass media systems and information control." pp.
41-69 in
F. Kline and P. Tichenor (eds.) Current Perspectives in Mass
Communication Research, V. 1. Beverly Hills CA: Sage.
---------.985. "Reporting conflict by pluralism, newspaper
type and ownership." Journalism Quarterly,
62(3)(Autumn):489-499,507.
Douglas, Mary, and Wildavsky, Aaron. 1982. Risk and Culture.
Berkeley CA: University of California Press.
Entman, Robert M. 1990. "Modern racism and the images of
blacks in local television news." Critical Studies in Mass
Communication, 7(4):332-345.
Entman, Robert M. 1994. "Representation and reality in the
portrayal of blacks on network television news." Journalism
Quarterly,
71(3):509-520.
Gandy, Oscar H. and Matabane, Paula W. 1989. "Television and
social perceptions among African-Americans and Hispanics." pp.
318-348
in M. K. Asante and W. B. Gudykunst [eds.] Handbook of
International and
Intercultural Communication. Newbury Park CA: Sage.
Gaertner, Samuel, and McLaughlin, John. 1983. "Racial
stereotypes and ascriptions of positive and negative
characteristics."
Social Psychology Quarterly, 46(1): 23-30.
Gissler, Sig. 1994."Newspapers' quest for racial candor."
Media Studies Journal (Summer):123.
Goshorn, Kent and Gandy, Oscar. 1995. "Race, risk and
responsibility: Editorial constraint in the framing of
inequality."
Journal of Communication, 45(2)(Spring):133-151.
Graber, Doris.1984. Processing the News. New York: Longman.
Griffin, Robert J. and Dunwoody, Sharon. 1995. "Impacts of
information subsidies and community structure on local press
coverage of
environmental contamination." Journalism Quarterly, 72
(2)(Summer):271-284.
Gunter, Barrie. 1987. Poor Reception. Misunderstanding and
Forgetting Broadcast News. Hillsdale NJ: Lawrence Erlbaum.
Hale, F. Dennis. 1988. "Editorial diversity and
concentration." pp 161-176 in R. Picard, M. McCombs, J. Winter
and
S.Lacy [Eds. Press Concentration and Monopoly. Norwood, NJ:
Ablex.
Hirsch, Paul and Thompson, Tracey. 1994. "The stock market
as audience: The impact of public ownership on newspapers." pp
142-158
in J.Ettema and D.Whitney [eds.] Audiencemaking: How the Media
Create
the Audience.Thousand Oaks CA: Sage.
Iyengar, Shanto.1991. Is Anybody Responsible? How Television
Frames Political Issues. Chicago: University of Chicago Press.
Kahneman, Daniel, Slovik, Paul, and Tversky, Amos
[eds].1982. Judgement Under Uncertainty: Heuristics and Biases.
NY:
Cambridge University Press.
Karnig, Albert. 1976. "Black representation on city
councils." Urban Affairs Quarterly, 12(2)(December):223-242.
_________1979. "Black resources and city council
representation." The Journal of Politics, 41:134-149.
Kaufman, Philip A., Dykers, Carol Reese, and Caldwell,
Carole. 1993."Why going online for content analysis can reduce
research
reliability." Journalism Quarterly, 70(4):824-832.
Krimsky, S. and Plough, A. 1988. Environmental Hazards.
Communicating Risks as a Social Process. Dover MA:Auburn House.
Lang, Gladys, and Lang, Kurt. 1981. "Watergate: an
exploration of the agenda-building process." pp. 447-468 in G.
Wilhoit
and H. deBock (Eds). Mass Communication Review Yearbook, Vol 2.
Beverly
Hills, CA: Sage.
LaVeist, Thomas. 1989. "Linking residential segregation to
the infant-mortality race disparity in US cities." SSR,
73(2)(January):90-94.
____________1993. "Segregation, poverty, and empowerment:
Health consequences for African Americans." The Milbank
Quarterly,
71(1):41-64.
Mauro, John. 1992. Statistical Deception at Work. Hillsdale
NJ: Lawrence Erlbaum.
McAdams, Richard. 1995. "Cooperation and conflict: The
economics of group status production and race discrimination."
Harvard
Law Review, 108, 1003-1084.
McManus, John M. 1994.Market-Driven Journalism: Let the
Citizen Beware? Thousand Oaks CA: Sage.
Milburn, Michael.1991. Persuasion and Politics. The Social
Psychology of Public Opinion. Pacific Grove CA:Brooks/Cole.
Morin, Richard. 1995 (October 8). "A distorted image of
minorities; Poll suggests that what whites think they see may
affect
their beliefs." The Washington Post, A1.
Nadeau, Richard, Niemi, Richard, and Levine, Jeffrey. 1993.
"Innumeracy about minority populations." Public Opinion
Quarterly, 57
(3):332-347.
Page, Benjamim and Shapiro, Robert. 1992. The Rational
Public. Fifty Years of Trends in Americans' Policy Preferences.
Chicago:
University of Chicago Press.
Potter, James. 1994. Cultivation Theory and Research. A
Methodological Critique. Journalism Monographs,147 (October).
Protess, D., Cook, F., Doppelt, J., Ettema, J., Gordon, M.,
Leff, D., and Miller, P. 1991. The Journalism of Outrage:
Investigative
Reporting in America. New York: Guilford Press.
Schiller, Herbert I. 1996. Information Inequality. New York:
Routledge.
Sigelman, Lee and Welch, Susan.1991. Black Americans' Views
of Racial Inequality. New York:Cambridge University Press.
Signorielli, Nancy. 1990. "Television's mean and dangerous
world. A continuation of the Cultural Indicators Perspective." pp
85-106
in N. Signorielli and M. Morgan [eds]. Cultivation Analysis.
Newbury
Park CA: Sage.
Signorielli, Nancy and Morgan, Michael. 1990. Cultivation
Analysis. New Directions in Media Effects Research. Newbury Park
CA:Sage.
Singer, Eleanor, and Endreny, Phyllis.1993. Reporting on
Risk. New York:Russell Sage Foundation.
Small, Melvin. 1994. Covering Dissent. The Media and the
Anti-Vietnam War Movement. New Brunswick,NJ: Rutgers University
Press.
Sniderman, Paul, Brody, Richard, and Tetlock, Richard. 1991.
Reasoning and Choice. Explorations in Political Psychology. NY:
Cambridge University Press.
Starobin, Paul. 1993. "Unequal shares." National Journal.
(September, 11 ):2176-2179.
Steinberg, Stephen.1995. Turning Back. The Retreat from
Racial Justice in American Thought and Policy. Boston: Beacon
Press.
Stocking, S. Holly and Gross, Paget. 1989. How do
Journalists Think? A Proposal for the Study of Cognitive Bias in
Newsmaking.Bloomington, IN: ERIC Clearinghouse.
Taebel, Delbert.1978. "Minority representation on city
councils: The impact of structure on Blacks and Hispanics."
Social
Science Quarterly, 59(1):142-152.
Tankard, James W., Hendrickson, Laura J. and Lee, Dong-Geun.
1994. "Using Lexis/Nexis and other databases for content
analysis:
Opportunities and risks." Conference paper. AEJMC, Atlanta,
GA.August.
Tapper, John. 1995. "The ecology of cultivation: A
conceptual model for Cultivation Research." Communication Theory,
5
(1):36-57.
Tversky, Amos and Kahneman, Daniel. 1977. "Judgement under
uncertainty: Heuristics and biases."pp. 309-327 in W.B. Fairly
and F.
Mosteller [eds]. Statistics and Public Policy. Reading Mass:
Addison-Wesley.
Tversky, Amos and Kahneman, Daniel. 1990. "Rational choice
and the framing of decisions." pp 60-89 in K. Cook and M. Levi
[eds].
The Limits of Rationality. Chicago: University of Chicago Press.
van Dijk, Teun a. 1988. News Analysis. Hillsdale, NJ:
Lawrence Erlbaum.
van Dijk. 1991. Racism and the Press. London: Routledge.
Verba, Sydney and Orren, Gary. 1985. Equality in America. The
View from the Top. Cambridge: Havard University Press.
Webster, James and Phalen, Patricia. 1994. "Victim, consumer, or
commodity? Audience models in communication policy." pp. 19-37 in J. Ettema and
D. Whitney [eds] Audiencemaking. How the Media Create the Audience. Thousand
Oaks CA: Sage.
Williams, Celeste.1992. "Distrust and hope. Many blacks see a
white conspiracy." The Milwaukee Journal.(June 7):1.
Table One
Framing Patterns [n=1245]
Headline and Lead Mean [Percent=1]
Disparity
42.8
Race
14.9
Bias
11.2
Question
21.0
Blacknot
51.2
Whites
3.9
Blacks Gain
17.3
Blacks Lose
71.3
Whites Gain
10.3
Whites Lose
2.1
Table Two
Headline and Lead by Subject(Means)
Headline and Lead
Anovar
Health
(n=489)
Economics
(n=103)
Cr Justice
(n=250)
Disparity
8.12***
.49
.43
.32
Race
1.65
.16
.20
.15
Bias
24.65***
.03
.39
.11
Question
3.56***
.23
.23
.18
Blacknot
1.79
.53
.41
.52
Whites
3.46***
.03
.03
.02
Blacks Gain
7.00***
.20
.14
.04
Blacks Lose
4.16***
.71
.76
.82
Whites Gain
2.67**
.10
.09
.06
Whites Lose
0.89
.02
.02
.03
Table Three
Headline and Lead by Credit (Means)
Headline and Lead
Anovar
Named Staff
(n=410)
Wire Svs
(N=427)
Staff/Wire (N=123)
Disparity
8.19***
.41
.46
.52
Race
4.77***
.19
.11
.16
Bias
1.70
.14
.09
.09
Question
5.61***
.25
.24
.18
Blacknot
7.84***
.48
.57
.58
Whites
3.65**
.03
.04
.08
Blacks Gain
2.56*
.15
.21
.20
Blacks Lose
2.53*
.75
.71
.64
Whites Gain
3.06*
.09
.10
.16
Whites Lose
0.73
.02
.02
.01
Table Four
Structure of Emphasis
Correlation of Headlines with Actuarial Leads
Blacks
Gain
Blacks
Lose
Whites
Gain
Whites
Lose
Disparity
.14**
-.14**
.14**
-.00
Race
.00
-.05
.23**
.03
Bias
.02
-.02
.05
-.02
Question
.03
-.03
.00
-.03
BlackNot
.10**
.12**
-.12**
.01
Whites
-.02
-.17**
.41**
.03
Table Five
Correlates of Framing
1
2
3
4
5
6
7
8
A
-.10
**
.01
.01
.05
-.09
**
-.03
.12
**
-.01
B
.04
-.06
*
-.03
-.06
*
.06
*
.09
**
.07
*
.07
**
C
-.04
.05
.06
*
-.04
-.02
-.03
-.02
-.00
D
-.04
-.03
-.06
*
.03
-.16
**
.02
.05
.01
E
.01
.04
-.00
.06
*
-.01
-.04
.03
-.06
*
F
.04
-.07
**
-.07
*
.02
.03
-.03
-.03
.01
G
-.10
**
.02
.00
.04
-.09
**
-.02
.03
.00
H
.13
**
.01
.02
.01
.07
*
.08
**
-.00
-.03
I
.00
-.02
-.02
.02
.03
-.08
**
.02
.00
J
.15
**
-.01
-.01
.07
*
.07
*
.10
**
-.08
**
-.04
1. General population size
6. 1990 Circulation
2. Change in Bk Pop share
7. Ciculation Change, 1980-90
3. Magnitude of racial change
8. Black Proportional Representation
4. Black/White Education Ratio
5. White/Black Unemployment
A=Disparity
B=Race
C=Bias
D=Question
E=Blacks
F-=Whites
G=Blacks Win
H=Blacks Lose
I=Whites Win
J=Whites Lose
Table Six
The Nature of Influence
Hierarchical Regression
(R-Square Change, Final Beta)
Disparity
Race
Black Loss
White Loss
Bias
Block 1
W/B Unem
-.07*
.05
.03
.03
-.01
B/W Edu
.09**
-.06*
.01
.07*
-.04
Size
-.06
.03
.13***
.09**
-.01
Block 2
Circ'90
.01
.07*
.06
.10**
-.03
Block 3
ChCirc
.12***
.07*
.03
-.06
-.00
ChBlk
.03
-.01
.06*
.03
.03
Block 4
Gserv
.05
.02
.03
-.01
.12***
Finance
.02
.07*
.09**
.01
.24***
CrimJu
-.07*
.04
.18***
.04
-.02
Health
.08*
.06
.10**
.03
-.14***
R-Squared
.05***
.02***
.05***
.04***
.11***
Table Seven
Pooled within-groups correlations
between discriminating variables
and canonical discriminant function.
Finance Theme
.812
Health Theme
-.589
Gov't Service Theme
.365
Circulation Change
-.036
W/B Unemployment Ratio
.033
Change in Black Pop %
.033
Criminal Justice Theme
.018
B/W Education Ratio
.011
Circulation in 1990
-.009
Population
.000
Percent of cases correctly classified: 86.7%
[1] County and City Data Book; State and Metropolitan Area Data
Book; US Cities and Counties CD-ROM; 1980 and 1990 Census data.
[2] Estimates were gathered from the American Society of
Newspaper Editors 1993 Survey of daily newspapers. Defined as full-time
newsroom professionals who are minorities.
[3] We interpret these coefficients cautiously because the high
degree of multicollinearity makes them