Content-Type: text/html

Mediating the media: Frames, attribution of responsibility, and individual media
use
 
 
 
Ben Kilpatrick
1250 Ephesus Church Rd Apt M-8
Chapel Hill, NC 27514
Phone: (919) 929-2655
e-mail: <a href="/cgi-bin/wa?LOGON=A3%3Dind9710B%26L%3DAEJMC%26E%3D7BIT%26P%3D908720%26B%3D--%26T%3DTEXT%252FPLAIN%3B%2520charset%3DUS-ASCII" >[log in to unmask]</a>
University of North Carolina at Chapel Hill
Doctoral Student
 
Robert Wayne Leweke
111 St. Ayers Way
Chapel Hill, NC 27514
Phone: (919) 408-8204
e-mail: <a href="/cgi-bin/wa?LOGON=A3%3Dind9710B%26L%3DAEJMC%26E%3D7BIT%26P%3D908720%26B%3D--%26T%3DTEXT%252FPLAIN%3B%2520charset%3DUS-ASCII" >[log in to unmask]</a>
University of North Carolina
Doctoral Student
 
 
 
 
ABSTRACT
        Following the research of Shanto Iyengar, this study examines how episodic and
thematic framing in television and print media relate to public attribution of
responsibility for poverty, racial inequality and violent crime.  In the
methodological tradition of Agenda Setting, it couples computer-assisted content
analysis with survey data.  It extends the analysis in the uses and
gratifications tradition by including media use style as an independent
variable.  The study finds limited but significant results.
 
 
Mediating the media: Frames, attribution of responsibility, and individual media
use
 
INTRODUCTION
        The question of how media framing of issues influences the public has led
political communication researchers down different paths to measure both the
kind and extent of the frames, and the effects those frames have on their
audience.1  Framing has been defined as the selection, emphasis and exclusion of
particular pieces of information that lead observers to particular conclusions
or narrow possible courses of action.  Communicators, according to Entman,
"select some aspects of a perceived reality and make them more salient in a
communicating text, in such a way as to promote a particular problem definition,
causal interpretation, moral evaluation and/or treatment recommendation for the
item described."2       While the definitions of framing vary from one researcher to
another, this study follows the framing research of Shanto Iyengar;3 it examines
media framing of social problems  and public attribution of responsibility for
those problems.         Like agenda setting or spiral of silence, Iyengar's concept of
framing is both a general metaphor and a specific hypothesis.  Iyengar argued
that differences in the framing of television news stories led to different
attributions of responsibility.  Iyengar defined news media framing in terms of
thematic frames and episodic frames.  He described thematic frames as placing
"political issues and events in some general context"; episodic frames, on the
other hand, focus "on specific events or particular cases."4  In other words,
thematic frames give a relatively broad background of  information on the news
story; an example might be discussing poverty in terms of federal welfare
policy, or structural causes of the problem.  Episodic frames, in contrast,
usually place a specific human face or faces in front of the audience as symbols
of the issue; for example, episodic framing of poverty might tell the story of a
single person or family on welfare, using their situation to frame the issue.
        In Iyengar's studies, participants, who were recruited through a newspaper
advertisement,  viewed edited broadcasts of television news with particular
stories manipulated to be framed either episodically or thematically.  The
experiments were designed to be as natural as possible, with a family room type
setting.5 Iyengar found that first, network news framed certain issues mostly
episodically; thematic frames were less prevalent.6  Second, Iyengar found that
episodic framing tends to lead the audience to place responsibility for the
problem, and for solving it, mainly on those personified in the stories.7  To
continue the above poverty example, subjects viewing poverty stories that were
framed episodically were more likely to place blame on the destitute individuals
in the stories themselves than were subjects who viewed poverty stories framed
thematically.8  Thus, the way news, in this case television network news, framed
issues may indeed influence the opinions the viewing audience forms about a
given issue.    Iyengar's findings have important implications for media
researchers and practitioners.  If episodic framing, an artifact of the
story-telling process, is responsible for powerful media effects, then media
researchers should examine the process in more detail.  For media practitioners,
knowledge of the potential impact of framing is crucial both to generate
intended effects and avoid unintended ones.
        This study further tests the concepts of episodic and thematic framing by
combining a content analysis of ABC World News Tonight transcripts and the front
page of The New York Times with a telephone survey that included questions about
media use and attribution of responsibility for poverty, racial inequality, and
violent crime.  The content analysis covered the one month period preceding the
survey.  While this study does not offer the same level of control as was
enjoyed in Iyengar's experiments, the exposure to the media is more realistic
and demand characteristics are minimized.  The New York Times was included  to
allow for a comparison of framing between two forms of media.  This adds to the
understanding of how media frame certain issues.  In addition we also expand on
Iyengar's concept by examining how audience members interact with media.  One
reason for basing the study on real-world survey data is to test Iyengar's
findings  in field conditions; his original studies were controlled lab
experiments.  The original thesis is extended in two ways.  The first is the
inclusion of print media in the form of the front page of The New York Times.
The second is to go beyond mere exposure to media and, in the uses and
gratifications tradition, to measure the media effect as a product of specific
functions of media use.  In short, the study addresses three questions:
1.  What is the nature of framing across media and issues?  The presence and
proportion of episodic and thematic frames are analyzed between the two media
and among the three issues.
2.  How does framing relate to attribution of responsibility?  Iyengar's
hypothesis that episodic framing causes attribution of responsibility to be in
the direction of individuals rather than government is addressed by comparing
results of the content analysis and survey data.
3.  How do media use and media use style relate to attribution of
responsibility?  With this question we extend the process of the possible impact
of framing to include how audience members use the news media in their daily
lives.  Reasonable arguments can be made to support opposite predictions for
this question.  On the one hand, the impact of framing might be enhanced when
audience members pay more attention to the media.  Attention is, after all, a
crucial step in the communication process.  On the other hand, those who pay
more attention to the media may be influenced less by framing in that they may
analyze the stories more critically or generate more counter arguments that
those for whom the news could be likened to background clutter.
 
METHOD
        The practice of following content analysis with survey data is well-established
in the Agenda Setting tradition.  This study adds a factor analysis of media use
style which allows for discussion of the importance of audience interaction with
media in the uses and gratifications tradition.  Particulars of the content
analysis and survey are detailed below.
 
Content Analysis
        The first step in this study was a computer-assisted content analysis of frames
of three different issues in two types of media.  The three issues were poverty,
racial inequality, and violent crime.9  The media analyzed for content were
transcripts of ABC World News Tonight broadcasts and front-page stories in The
New York Times, both found in a keyword search of the NEXIS on-line database.
The New York Times  was chosen because as the nation's leading newspaper, it
often influences the news agendas of other media, both print and broadcast.10
ABC's World News Tonight provided the only transcript available on-line of the
three major network evening newscasts, but at the time of the analysis ranked
first among them.11  News stories were located using keywords12 for both media
during the month leading up to the Carolina Poll (Sept. 15-Oct. 14, 1995).      The
KWIK keyword search function was used to find the phrases surrounding the
keywords.  This method found 305 entries which were downloaded and coded.  Word
processing software was then used to locate the keywords within the downloaded
files, and the context of each phrase was then coded as episodic, thematic, or
neither (if neither of the frames was clearly dominant).  The unit of analysis
was the sentence surrounding each keyword; the sentence was used to code the
entry as specified above.13  The date of the entry and the keyword used were
also recorded.  These data were analyzed to look for differences in framing both
among the media and among the issues.  The sentence in which each keyword was
found was used as the unit of analysis for both methodological and theoretical
reasons.  Methodologically, the KWIK function in NEXIS uses keywords to extract
one or more sentences surrounding the keyword chosen, so using the keyword's
sentence as the unit of analysis enabled the coder both to take advantage of the
speed of the KWIK function to build a data set, and to use the common components
of the data set to increase reliability in coding.  From a standpoint of theory,
the unit of analysis demonstrates that episodic and thematic frames can be found
within units of the stories themselves, and that the same story could contain
more than one frame.  By using the sentence instead of the entire story as the
unit of analysis, we therefore avoided a weakness in the Iyengar analysis, which
coded entire stories as episodic or thematic regardless of the variable
proportion of frames within them.14  Survey
        The survey was conducted in the week of October 14, 1995.  A sample of 623
residents of North Carolina were surveyed using a form of random digit dialing.
The median age of the respondents was 42.  There were 270 males and 353 females.
Median level of education was 13 years, while the median household income was
between $30,000 and $40,000 per year.  Variables were operationalized as
described below.
 
Dependent Variables
        As with the content analysis, poverty, racial inequality and violent crime were
analyzed.  Measures of responsibility for the three issues were assessed for use
as dependent variables in regression analyses.  Respondents were asked: "Who is
responsible for dealing with poverty?"; "Who is responsible for dealing with
racial inequality?"; and "Who is responsible for dealing with violence?".
Possible answers for each question were "Government," "People themselves," or
"It depends".  For regression purposes, categories were combined to create a
bivariate variable for each of the issues. The responses "Government" and "It
depends" were combined and represent the idea that the government bears at least
some responsibility for the issue at hand.
 
Independent Variables
        Media use, style of media use, political leaning, and items measuring locus of
control and political efficacy were assessed.  To account for potential
demographic differences age, gender, education and income were used as control
variables.
        Media use was assessed by asking how many days in the last seven respondents
watched television and read newspapers, and whether or not they read news
magazines.  While Iyengar's hypotheses focus on the effects of mere exposure to
media, we would hypothesize that the relationship between media use and
attribution of responsibility is a more complex one that includes how
respondents interact or use media on a daily basis.  For this reason, five
questions designed to tap individual media use style were included.  These were
"I keep up with the news (1) to make sure nothing affecting me has happened; (2)
because it's important; (3) because it's exciting; (4) because it's a habit; and
(5) to pass the time."  Factor analysis indicated two underlying constructs
which were labeled "active" and "passive" in terms of media use style. These
results should be interpreted with caution, however, because there were not a
large number of values above .3 in the correlation matrix.
 
Factor Loadings "Passive" "Active"  1. to make sure nothing affecting me has
happened .21 .68  2. because it's important -.01 .68  3. because it's
exciting .55 .29  4. because it's a habit .80 -.17  5. to pass the
time .57 .20    Items 1 and 2 clearly indicate active use of media, while 4 and
5 are clearly passive.  Item 3, "I keep up with the news because it's exciting,"
loads more on the passive factor, but also has a moderate loading on the active
factor.  This may indicate that item 3 relates to both factors, or that it does
not differentiate the factors as well as the other items.       Political leaning was
operationalized through a seven point Liberal/Conservative scale with higher
values representing the more conservative response.  Locus of control was
assessed by having respondents agree or disagree with the statement "Life is
basically a matter of luck, so it's better not to plan."  Those who disagree
with the statement are likely to have a more internal locus of control.
Political efficacy, meanwhile, was operationalized through a similar item which
stated "More often, government does the right thing."  Agreement would be
expected to relate to higher feelings of political efficacy.
        Stepwise regression was used and variables were entered in three blocks.  To
control for demographics, age, gender, education and income were entered first.
Then, political leaning, locus of control and political efficacy were entered.
Finally, television viewing, newspaper reading, magazine reading and the
"active" and "passive" media use style variables were entered.  Findings are
discussed in the results section.
 
RESULTS
Content Analysis
        The purpose of the content analysis was to determine what frames were commonly
used for the selected issues and in the two media.
        As Table 1 shows, the dominant frames differed depending on the medium, taking
all three issues together as a whole{X2 =(N= 305) 10.45, p<.01}.15  The ABC news
broadcasts were more likely to frame the three issues episodically than were the
entries found in the New York Times.  Out of 192 entries in the ABC transcripts,
63.0% were framed episodically.  Only 46.0% of the Times' 113 entries were
framed episodically.  Only 16 of the entries were coded as being neither clearly
episodic or thematic.
        However, when the frames are analyzed by issue, the relationship between the
medium and the issue being framed is specified.  The race and violent crime
issues tended to be framed episodically by both media, but the poverty issue was
framed thematically by both media, as Tables 2 and 3 show.  This relationship
between the medium and the type of frame for poverty entries is significant{X2
=(N= 103) 9.06, p<.01}.
 
Survey results
        Table 4 shows the frequencies of responses to the questions about attribution
of responsibility.  In general, respondents attributed responsibility for social
problems to "people themselves" rather than to government.  Respondents were
least likely to attribute responsibility to government for dealing with racial
inequality and most likely to attribute responsibility for dealing with violence
to the government.
        Regression analysis with government responsibility as the dependent variable
and demographics, self-reported media use and the factors described above
provided small, but significant results.  They are discussed in more detail
below.  For each issue, age, gender, education and income were entered as a
block.  Political leaning, locus of control and political efficacy were entered
as a second block.  Finally, television viewing, newspaper reading, magazine
reading and the "active" and "passive" media use style variables were entered.
Stepwise regression was used with an inclusion level of .05 and exclusion level
of .10.
 
Poverty
        As Table 5 shows, responsibility for dealing with poverty showed a
statistically significant relationship with income, political leaning, and
"active" use of media.  Together these account for about 7 percent of the
variance in responsibility for dealing with poverty.  Those with higher incomes
and those who reported being more conservative were more likely to attribute
responsibility for poverty to individuals.  However, more active users of the
media were more likely to attribute responsibility for dealing with poverty to
the government.  Each block represents a different equation generated by
stepwise regression.
 
Racial Inequality
        In Table 6, responsibility for dealing with racial inequality showed a
statistically significant relationship with income, "active" use of media, and
television viewing.  These three variables account for about 3 percent of the
variance in responsibility for dealing with racial inequality.  Those with
higher incomes and those who watch more television were more likely to attribute
responsibility for dealing with racial inequality to individuals, while, as with
poverty, more active media users were more likely to attribute responsibility
for dealing with racial inequality to government.  Each block represents a
different equation generated by stepwise regression.  No variables were entered
in the second block.
 
Violent Crime
        Responsibility for dealing with violent crime showed a statistically
significant relationship only with "active" use of media.  Table 7 shows that
this variable accounted for a mere 1 percent of the variance in responsibility
for dealing with violent crime.  As with responsibility for dealing with poverty
and responsibility for dealing with racial inequality, the relationship was such
that more active users of media were more likely to attribute responsibility for
dealing with violent crime to government rather than individuals.  Since only
"active media use" was found to have a significant relationship with attribution
of responsibility for dealing with violent crime, it is the only variable shown.
No variables were entered in blocks 1 and 2.
 
DISCUSSION
        The three questions listed earlier and limitations of the study are discussed
below.  A summary and implications follow.
1.  What is the nature of framing across media and issues?  The content analysis
told us that, on the whole, the issues we examined were framed episodically. The
three issues were framed episodically most of the time (57 percent) with 63
percent of ABC News and 46 percent of The New York Times entries being coded as
episodic.  In addition, consonant with the idea that television is more
episodic, ABC News was less thematic and more episodic than The New York Times.
However, in our sample, the particular issue did make a difference.  There was a
strong tendency for both the ABC nightly news broadcasts and the front-page
entries in The New York Times to frame poverty in clearly thematic terms.  In
contrast, both media covered race and crime in expectedly episodic ways for most
entries.
        One reason for this large difference is the national news environment during
the month of analyzed coverage.  The main stories in the race category were the
criminal trial of O.J. Simpson and the "Million Man March" led by Louis
Farrakhan.  Such major, ongoing stories provided the national news media with
issues that were almost always framed in episodic terms (e.g., Simpson and
Farrakhan as continuing symbols of race relations).  Although there were no
major stories in the crime category that dominated the month's coverage the way
those two stories dominated race, the crime entries also presented the issue
mostly as one involving individuals rather than society.  In contrast, nearly
all the stories about poverty during the period were framed in terms of the
budget battle between Congress and the president, especially regarding the
effect of changes in the Medicaid program.  This type of coverage matched
Iyengar's definition of thematic coverage as characterized in part by public
policy debates or outcomes.16   This suggests that the national print and
broadcast media may influence each other both in agenda-setting terms17 and in
the way they frame particular issues.  The results related to poverty also seem
to indicate that the frame used to cover an issue may be determined by more than
just the medium involved.  Particular events, such as budget battles in
Congress, may lead to more thematic or episodic coverage, regardless of the
medium. 2.  How does framing relate to attribution of responsibility?  Based on
the greater proportion of episodic framing and given Iyengar's hypothesis, the
attributions of responsibility should be in the direction of individuals rather
than government.  In support of this hypothesis, univariate findings indicate
that respondents did tend to attribute responsibility to individuals rather than
government.  However, these findings only mean that it is possible that media
are influencing the respondents, but should not be taken as strong evidence
since the correlation could be a response to any number of other variables.
Particularly, the journalists' decisions to frame issues in episodic ways may be
a reflection of generally held public views rather than a cause of those views.
In addition, the attribution of responsibility and the type of framing do not
indicate a clean pattern.  While poverty was framed more thematically than the
other issues, government was not seen as more responsible for dealing with the
problem of poverty.  Instead, with 36 percent of respondents placing some
responsibility on government, poverty fell between racial inequality, which had
the least amount of government responsibility (21 percent), and violent crime,
which had the most government responsibility of the issues examined (41
percent).
3.  How do media use and media use style relate to attribution of
responsibility?  The regression analysis also provides limited support for the
impact of framing on attribution of responsibility.  It should first be noted
that the variance explained (R2), while statistically significant, is quite
limited: 7 percent for poverty, 3 percent for racial inequality, and 1 percent
for violent crime.  Second, direct media use, in the form of television viewing,
appears in only one issue (racial inequality).  This indicates that more than
mere exposure to media is at work in influencing how attributions of
responsibility for social problems are made.
        The most interesting finding is the tendency of "active" media users--those who
keep up with the news "because it's important" and "to make sure nothing
affecting me has happened" but not "because it's a habit or "to pass the
time"--to be more likely to attribute responsibility to the government for
dealing with all three of the social problems.  This may indicate that frames
are more likely to influence members of the audience who use the media more as
resource to fill time than those who interact with the media on a more
deliberate basis.  It would stand to reason that more active users might produce
counter arguments to the persuasive impact of episodic frames, while passive
users would simply take the message of the frames at face value.
 
Limitations
        This study does not provide definitive answers for those who want to know what
the impact of media frames may be.  Through use of a survey, we gained breadth
but sacrificed depth and control.  A study of additional indicators of media use
and media use style may have shown a clearer relationship between framing and
attribution of responsibility.  Nevertheless, the data do have the advantage of
being drawn from a diverse population, making it more generalizable than the
traditional experiment with college students as participants.  In addition, the
impact of media frames may be a fait accompli.  That is, the process of forming
public opinion towards the issues studied may have already taken place.  Indeed,
a majority of people already believe that individuals rather than government are
responsible for dealing with these problems.  If, in fact, a ceiling has already
been reached, no amount of correlational research will show a great effect.
 
SUMMARY AND IMPLICATIONS
        This study supports the original Iyengar framing study showing that media do
tend to frame social issues in episodic rather than thematic terms.  As a whole,
in both media, issues were framed episodically.  However, it also indicates that
particular events have an impact on the way an issue is covered at a given time.
Television news is not always episodic.  Poverty, for example, was framed
thematically by both media in the time period analyzed.  These differences among
the issues suggest that researchers interested in this type of framing must
actually conduct analyses of media content and cannot simply assume that the
coverage will be consistently framed in particular ways across issues and time.
        A direct relationship between the framing pattern in the content analysis and
attribution of responsibility in the survey was not apparent.  One possibility
is that this indicates the absence of a real world media effect.  This may also
reflect differences among the issues examined that mask a real media effect.
Further studies that include more issues, additional indicators both of media
use and attribution of responsibility could help to resolve this question.
        In addition, if frames do indeed operate as Iyengar hypothesized, we also see
evidence that how a person uses the media is important in determining what
impact framing may have on attribution of responsibility.  Active use of media
seems to moderate the impact of framing.  From a theoretical standpoint,
examining the psychological basis of media effects and the interaction between
audience members and the media becomes a priority.  For media consumers, the
implication is that possible manipulation by the media is reduced by more active
use of media.
        Taken together, these findings show that the relationship between media framing
and attribution of responsibility is a complex one that may be mediated both by
events outside the media and by the individual media use patterns of individual
audience members.  The answer to Iyengar's question "Is anyone responsible?"
seems to be that both the audience and the media may bear some responsibility
for the impact of media frames.
 
 
WORKS CITED
 
Blalock, Hubert M., Jr. Social Statistics. New York: McGraw-Hill, 1979.
 
Entman, R.M.  "Framing: Toward Clarification of a Fractured Paradigm."  Journal
of Communication, 43:4 (1993), 51-58.
 
Entman, R.M.  "Framing U.S. Coverage of International News: Contrasts in
Narratives of the KAL and Iran Air Incidents."  Journal of Communication, 41:4
(1991), 6-27.
 
Gandy, Oscar J., Jr. "From Bad to Worse-- The Media's Framing of Race and Risk"
Media Studies Journal, 8:3 (1994), 39-48.
 
Gitlin, Todd.  The Whole World Is Watching: Mass Media in the Making and
Unmaking of the New Left.  Berkeley: University of California Press, 1980.
 
Iyengar, Shanto.  Is Anyone Responsible?  How Television Frames         Political
Issues.  Chicago: University of Chicago Press, 1991.
 
Pan, Zhongdang and Gerald M. Kosicki.  "Framing Analysis: An Approach to News
Discourse."  Political Communication, 10 (1993), 55-73.
 
Parenti, M.  Inventing Reality: The Politics of News Media.  2nd ed.  New York:
St. Martin's Press, 1993.
 
Shoemaker, Pamela J., ed.  Communication Campaigns about Drugs: Government,
Media, and the Public.  Hillsdale, N.J.: Lawrence Erlbaum Associates, 1989.
 
Tuchman, Gaye.  Making News: a Study in the Construction of Reality.  New York:
Free Press, 1978.
 
 
 
Table 1 Frame by Medium
 
 
Count     Row Pct ABC News The New York Times   Column Pct     Total
Pct      121 52 173  Episodic 69.9%   30.1%    63.0%   46.0%    39.7%
 17.0% 56.7%   65   51   116  Thematic 56.0%   44.0%    33.9%   45.1%    21.3%
 16.7% 38.0%   6   10
 16  Neither 37.5% 62.5%    3.1% 8.8%    2.0% 3.3% 5.2%  Column
Total 192 113 305   63.0% 37.0% 100.0%    Table 2  Issue by Frame controlling
for Medium   ABC News   Episodic Thematic Neither  Poverty     Count 8 39 1  Row
% 16.7% 81.3% 2.1%  Col. % 6.6% 60.0% 16.7%  Race     Count 57 14 1  Row
% 79.2% 19.4% 1.4%  Col. % 47.1% 21.5% 16.7%  Violence     Count 56 12 4  Row
% 77.8% 16.7% 5.6%  Col. % 46.3% 18.5% 66.7%
Table 3 Issue by frame controlling for medium   The New York
Times   Episodic Thematic Neither  Poverty     Count 2 46 8  Row
% 3.6% 82.1% 14.3%  Col. % 3.8% 90.2% 80.0%  Race     Count 26 4 0  Row
% 86.7% 13.3% 0.0%  Col. % 50.0% 7.8% 0.0%  Violence     Count 24 1 2  Row
% 89.9% 3.7% 7.4%  Col. % 46.2% 2.0% 20.0%    Table 4  Attribution of
responsibility   Frequency Percent  Who is responsible for
dealing with Poverty?    Government 93 14.9  People themselves 374 60.0  It
depends 134 21.5  Don't know/No Answer 22 3.5  Who is responsible for  dealing
with Racial Inequality?    Government 49 7.9  People themselves 478 76.7  It
depends 80 12.8  Don't know/No Answer 16 2.6  Who is responsible for
dealing with Violence?    Government 124 19.9  People themselves 349 56.0  It
depends 134 21.5  Don't know/No Answer 16 2.6
Table 5 Regression of responsibility for poverty  Variable R R2 Adjusted
R2 F t values b Beta values  Block 1 .15 .022 .020 8.65     Family
Income    (p<.01) 3.01
(p<.01) .037 .150  Block 2 .21 .043 .038 8.38     Family Income    (p<.01) 2.77
(p<.01) .035 .14  Political leaning     2.82
(p<.01) .035 .14  Block 3 .27 .074 .067 9.97     Family Income    (p<.01) 3.01
(p<.01) .15 .150  Political leaning     3.05
(p<.01) .15 .15  "Active" media use     -3.56
(p<.01) -.18 -.17    Table 6 Regression of responsibility for racial
inequality  Variable R R2 Adjusted
R2 F t values b Beta values  Block 1 .12 .015 .012 5.69     Family
Income    (p<.05) 2.39
(p<.05) .026 .122  Block 2 .17 .028 .024 5.54     Family Income    (p<.01) 2.55
(p<.01) .027 .130  "Active" media use     -2.31
(p<.05) -.047 -.118  Block 3 .20 .039 .031 5.02     Family
Income    (p<.01) 2.63
(p<.01) .028 .134  "Active" media use     -2.41
(p<.01) -.048 -.122  Television viewing     1.98
(p<.05) .018 .100
Table 7 Regression of responsibility for violent crime  Variable R R2 Adjusted
R2 F t values b Beta values  Block 3 .12 .014 .011 5.26     "Active" media
use    (p<.05) -2.29
(p<.05) -.058 -.118
 
 
1See Tuchman, 1978; Gitlin, 1980; Gamson and Modigliani, 1989; Entman, 1991,
1993; Pan and Kosicki, 1993; Parenti, 1993; Gandy, 1994.  2 Entman, 1993, p.
52.  3Iyengar, 1991.  4Ibid., p. 2.  5Ibid., p. 23.  6Iyengar,
1991.  7Especially for poverty and racial inequality issues, two of the issues
studied here.  Ibid., p. 128.  8Ibid., pp. 46 61.  9These issues were chosen to
coincide with three of the issues that Iyengar's study dealt with.  They also
match issues addressed in questions in the Fall 1995 Carolina Poll.
 
10 See, for example, Shoemaker, et al, 1989. 11In Fall 1995 World News Tonight
maintained a rating of over 10 points (as a percentage of all households with
televisions) and a share approaching 20 percent (of sets in use).  NBC's Nightly
News held second place and CBS's Evening News was third.  Also, World News
Tonight had led the other newscasts in the ratings for about three years prior
to our analysis.  Hollywood Reporter, 22 November 1995.
 
12The keywords used were: for "poverty," welfare, AFDC (Aid For Dependent
Children), hunger, homeless, dependent children, Medicaid, poverty, and poor;
for  "racial inequality," racism, affirmative action, African American,
black(s), underclass, and inner city; for "violence," violence, killing, rape
and murder.  13An example of an excerpt coded as thematic: in an ABC story about
the debate over welfare reform in Congress, the sentence was: "Congress put off
a final vote on welfare reform until next week." ABC World News Tonight, 15
September 1995.
 
14Iyengar, 1991, pp. 18-19.  15The values of the chi-square were statistically
significant, allowing us to conclude that the observed frequencies differed
enough from the expected frequencies (had the values been distributed randomly)
to justify rejection of the null hypothesis that there was no relationship
between the medium, the frame, and the issue.  For a discussion of the use of
chi-square with nominal data and the assumptions involved, see Blalock, 1979,
pp. 279-282.
 
16Iyengar, p. 18.  17For example, on the illegal drug issue, see Shoemaker, et
al, 1989.
Mediating the media: Frames, attribution of responsibility and individual media
use
 
 
 
 
 
 
 
 
 
 
 
 
Mediating the media: Frames, attribution of responsibility and individual media
use       20     2