This paper was presented at the Association for Education in Journalism and
Mass Communication in San Antonio, Texas August 2005.
If you have questions about this paper, please contact the author
directly. If you have questions about the archives, email
rakyat [ at ] eparker.org. For an explanation of the subject line,
send email to
[log in to unmask] with just the four words, "get help info aejmc," in the
body (drop the "").
(Jan 2006)
Thank you.
Elliott Parker
====================================================================
The Impact of Source Types on Perceived Bias of Online News Sources
Li Zeng
College of Communications
Arkansas State University
Mailing Address: 5372 Knollwood Dr, #8, Parma, OH 44129
Telephone: (870) 972-3625
E-mail: [log in to unmask]
and
Walter B. Jaehnig
School of Journalism
Southern Illinois University Carbondale
(Paper submitted to the Communication Theory and Methodology Division
for consideration for presentation at the AEJMC annual convention,
August 2005, San Antonio, TX)
Abstract
This study examines whether different types of sources quoted within
online news stories affect individuals' perception of source bias.
One hundred and five college students participated in an experimental
setting. The findings provide evidence for the applicability of the
Elaboration Likelihood Model in an online news environment. When
exposed to stories arousing a high level of motivation, participants
reported that official sources were more biased than individuals
affiliated with non-government organizations.
The Impact of Source Types on Perceived Bias of Online News Sources
Introduction
Two in three American adults who are online now seek news on the
Internet (Pew Internet & American Life Project, 2005). As a response
to this increasing online news demand, Internet news Web sites have
sprung up like mushrooms. Internet Service Providers (ISP) such as
America Online (AOL) offer news on their Web sites, along with many
other online-only news organizations. Newspapers, television, radio,
and other traditional news media also joined the digital mix, partly
for fear of losing audiences if they fail to have an online presence.
The explosion of online news brings about tremendous diversity in
both the content and form of news. However, it has become apparent
that online news is sometimes highly biased or occasionally even
simply cyber-gossip. Some media scholars and professionals address
the problem by advocating quality journalism on the Internet. The
Online News Association, for example, specifies in its mission that
"[o]nline journalists should uphold traditional high principles" of
journalistic practice (Online News Association, 2001). Traditional
news media have few reservations about implementing strict editorial
rules on their Web sites, as they use in their traditional delivery.
However, many online-only news Web sites do not have vigorous
editorial policies. The multimedia features of the Internet have
invited people skillful in multi-media interaction but lacking
journalism training to become news disseminators. The ease with which
everyone can be transformed into a potential online journalist also
spurred amateur journalists to practice on the Internet. A case in
point is Matt Drudge's gossip site (http://www.drudgereport.com),
where he shows to the online audience what can happen to news absent
the limitation of traditional journalistic rules. There is not yet
broad agreement as to whether traditional journalistic principles and
practices should be applied to the delivery of online news.
In traditional news reporting, for example, journalists have long
been told that an integral component of good journalistic practice is
quoting from credible sources (Gans, 1979; Izard, Culbertson, &
Lambert, 1977; Jones, 1976). Editors frequently require that
reporters contact legitimate sources and incorporate quotations in
their stories so as to increase credibility and vividness of stories.
Sourcing is so essential to news reporting that a news story, however
newsworthy and truthful it is, would rarely be published without
being properly attributed to a credible source (Sundar, 1998).
However, for almost every news story there are numerous potential
sources. These sources contribute their own interpretations and may
influence audiences' assessment of the events being covered.
This study examines whether certain types of sources quoted within
online news reports affect individuals' evaluation of source bias.
Based upon previous content analyses of traditional news reports, the
study intends to answer whether the type of sources quoted within an
online news report affects perceived source bias.
Review of the Literature
Source Type in Media Coverage
For decades researchers have been interested in the selection of news
sources in media coverage, particularly in the framework of the
framing theory. According to Goffman (1974), a news frame is a
"schemata of interpretation" (p. 21). It allows the audience to
understand an event or an issue presented in the media by including
(versus excluding) and emphasizing (versus deemphasizing) certain
aspects. The framing of news sources is mostly concerned with the
selection of or granting more space to certain sources in preference to others.
Sources are important to news for several reasons. As Shoemaker and
Reese (1991) noted, news sources "have a tremendous effect on mass
media content, because journalists can't include in their news
reports what they don't know" (p. 150). More importantly, some types
of sources are so powerful that they "may also influence the news in
subtle ways by providing the context within which all other
information is evaluated…and by monopolizing the journalists' time so
that they don't have an opportunity to seek out sources with
alternative views" (p. 150).
Nonetheless, source type has been a vague variable in the literature.
Few previous studies, if any, have provided a clear and unambiguous
measure of the variable. The most relevant measure is perhaps the
scale used for affiliation, which has been frequently adopted over
the years in analysis of traditional news content. Initially employed
in the 1979 ABC News-Harris Survey (Harris, 1979), the variable of
affiliation contained eight categories: U.S. government, state
government, local government, foreign government, affiliated U.S.
citizen, unaffiliated U.S. citizen, foreign citizen, and other. Using
this measure, Brown and associates (1987) found that source patterns
in newspaper stories featured diversity, but it was limited diversity
in the sense that sources were mostly in the range of "government."
Berkowitz (1987) reported the same pattern in television newscasts.
However, these eight categories were created not by using a single
standard of classification, thus violating a fundamental rule of
measurement. In the scale, the standard of affiliation was
contaminated by such other standards as the level of government and
locations. For instance, the first three categories were created by
the level of government organizations in the United States, but the
fourth category, foreign government, was far from being another level
of government in the United States. Similarly, the distinction
between affiliated U.S. citizens and unaffiliated U.S. citizens was
their connection or non-connection to non-governmental organizations.
But foreign citizens differed from the previous two categories by
their citizenship. In this study the ABC News-Harris scale were
modified so that its weaknesses were overcome.
A closer look at the ABC News-Harris scale helps determine the source
types in this study. As Brown et al. (1987) pointed out, the first
four categories of sources were all government officials. In studies
of news content and credibility (Armstrong & Nelson, 2003; Paletz,
Fozzard, & Ayanian, 1982; Soloski, 1989; Zoch & Turk, 1998),
government officials have been included in a broader source type
called officials, which also included legal personnel in the United
States or abroad. Therefore, the first type of source in this study
were officials, most of whom were government officials.
The remaining four categories used in the ABC News-Harris study also
can be reclassified. Foreign citizens can be affiliated or
unaffiliated to different non-governmental organizations, just as
their U.S. counterparts are. The news media tend not to consider as
legitimate news sources average citizens absent authority status
unless such individuals have direct experience (e.g., as a victim,
beneficiary, or witness), possess unique knowledge as a result of
their affiliation (e.g., a professor at University of California
Berkeley talking about environment protection), or represent a
special group (e.g., a spokesperson of the union). They can be
divided into two types. One type, called involved parties, includes
individuals who become news sources due to their first-hand
experience in a news issue being covered. The other type, called
affiliated individuals, refers to individuals who become news sources
due to their connections with non-governmental organizations. Rarely
are individuals lacking the credentials of these three types sought
as news sources. Therefore, this study examined three types of
sources: officials, involved parties, and affiliated individuals.
Numerous studies have been designed to examine the pattern of source
types that contribute to the images painted by journalists and to
public understanding of social events. With few exceptions, findings
showed a heavy dominance by officials in news stories, regardless of
the type and level of news media through which they are delivered.
Officials. One of the earliest comprehensive studies of source
diversity was Sigal's (1973) longitudinal content analysis of major
U.S. newspapers, wherein he discovered a limited diversity in source
types, with most sources falling under an official label. Brown and
associates (1987) replicated Sigal's study in a content analysis of
front-page stories published in the New York Times, the Washington
Post, and four local/state newspapers in North Carolina. They found
that among the 5,248 sources identified in the stories, the most
dominant were government sources, of which approximately one-third
(31%) were affiliated with the U.S. federal government, and another
quarter (24.4%) affiliated with other levels of U.S. government or
foreign governments. These data supported what Brown et al. (1987)
proposed as the distinction between diversity and range: diversity
from an elite pluralistic perspective is a constrained concept and
can best be "a great deal of variety within a relatively narrow range
of sources through the presentation of trivial differences among them" (p. 45).
The tendency to rely heavily upon official sources is more pronounced
when political violence, especially that against the United States or
its citizens, is concerned. Paletz, Fozzard, and Ayanian (1982)
reported that in the New York Times and television network news
coverage of three terrorist groups, the news media depended upon
official sources and denied legitimacy of terrorists and justness of
terrorist causes. In a content analysis of magazine news stories and
network newscasts on two U.S. terrorist incidents, Gallimore (1991)
found that authority sources (most of whom were officials) accounted
for more than 40 percent of the total number of sources. In the more
recent case of the "9/11" attack, numerous studies have revealed the
media's excessive dependency on officials (e.g., Zeng, 2002; Zeng,
2003; Zeng & Xiao, 2003).
Involved parties. Despite the many journalist-preferred features
possessed by official sources, different types of non-official
sources are allowed occasional appearance under various situations.
In news coverage of a major accident, for example, primary sources
such as victims, witnesses, and even perpetrators are provided with
space, although usually limited compared with the room granted to
official sources. Atwater (1987) analyzed the 1985 TWA hijacking and
hostage taking and found that 53 percent of the sources were victims
and their families. In a study of political violence coverage in the
New York Times, the Washington Post, and the Los Angeles Times,
Picard and Adams (1991) found that 2.2 percent of the total
characterization words about attacks of political violence and their
perpetrators were from the mouths of witnesses to the events.
Although comprising only 2.2 percent, this figure is not negligible
because Picard and Adams included in the total a category called
media source, which is generally not considered a type of news source
but accounted for over 90 percent of the total in their 1991 study. A
content analysis (Gallimore, 1991) of magazine news stories and
network newscasts on two U.S. terrorist incidents identified 6
percent of the sources in the coverage as perpetrators. In another
study of the coverage of the "9/11" attack in the three U.S. elite
newspapers, Zeng (2003) noted that witnesses contributed to 1.6
percent of the total characterizations of the "9/11" attack and
people who perpetrated it.[1]
Empirical data also revealed that quotations from primary sources,
despite their non-official status and their consequential lack of
suitability as news sources for journalists, are influential on the
audience's perception of news. Personal exemplifications from
individuals involved in an issue under coverage, such as victims and
witnesses, for example, are well documented to affect individuals'
understanding of the issue.
Gibson and Zillman (1993) found that exemplification from victims and
witnesses considerably affected people's understanding of issues
reported in news stories. They asked college students to read a print
news report about the problem of inadequate safety in amusement
parks. Some students read a version featuring personal testimony from
individuals who had been involved in or witnessed amusement park
accidents, some read a version with the testimony paraphrased by the
journalist, and others read a version containing merely factual
information about the injuries suffered by several victims, with a
summary of the arguments from both sides of the safety issue. The
results showed that those who read the version with personal
testimony believed that safety in amusement parks was less adequate
than those who were not exposed to personal testimony by primary sources.
Affiliated individuals. Janowitz (1991) proposed that one of the
three-fold roles of the mass media was to prevent one side of an
issue from monopolizing or excessively overshadowing other sides.
This is mostly achieved through granting news space to affiliated
individuals, whom journalists also consider legitimate and credible
(Thurwanger, 2003). Among this type of sources are individuals
affiliated with non-governmental institutions, such as business
organizations and special interest groups. In a content analysis of
community newspaper coverage of prison site selection, affiliated
individuals were cited in nearly 20 percent of the coverage, with
business sources representing 11.8 percent and organizations with
opposing opinions about the issue accounting for 7.1 percent
(Thurwanger, 2003).
However, the way affiliated individuals are covered in the news media
renders them more vulnerable to accusation of bias than officials.
Although viewed as legitimate by journalists, affiliated individuals
possess a social position that is never equivalent in power and
prominence to that occupied by officials. As revealed in previous
studies, affiliated individuals are merely supplementary to the
already established official power structure as reflected in media
frames. Their chances of being quoted are even smaller when they
represent viewpoints conflicting with those of the U.S. government,
for instance, when an individual is affiliated with an "unfriendly"
organization. When indeed cited, these sources face a greater threat
of being viewed as biased.
News Sources and Perception of News Stories
Media researchers have examined at least three types of sources,
story bylines, news media, and sources quoted in news stories. This
study is interested in sources for quotation in a news story, which
are what journalists usually refer to as news sources. Essentially,
this line of research overlaps with the research on quotations, where
primary attention was paid to whether sources are directly quoted,
paraphrased, or not mentioned at all in a news story. The findings
about the effects of quotations on audiences' perceptions are
ambiguous, some even conflicting with each other.
Largely overlooked is the influence that source selection, especially
the type of sources selected, might have upon individuals'
perceptions of news credibility and understanding of the issue
concerned. Over the years of credibility research, only a few
scholars have noted that journalists' reliance on limited types of
sources might prevent them from obtaining a complete view of the
event they cover (Ericson, Baranek, & Chan, 1987), which in turn
affects their audience's assessment of news credibility. For example,
Rouner, Slater and Buddenbaum (1999) proposed that the pattern of
sources typically used may constrain and bias news coverage, although
empirical data suggested that journalists and the general public have
different understanding of bias.
Armstrong and Nelson (2003) used official and non-official source
cues to test people's perception of credibility. College students
were asked to read identical stories about a missing hiker quoted
from a gender-neutral name but with either an official or
non-official title. It was found that the subjects considered the
official source (a park ranger) more credible than the non-official
source (a fellow hiker), despite the same name and identical
information they conveyed. The authors further argued that the source
might be used by individuals as a heuristic cue, such that the
credibility of the source will be transferred to the information
content. When people are exposed to a news report that quotes an
authority source, they believe the source is less biased and more
credible, and tend to process the information based on such
heuristics. As a consequence, they will assign more credibility to
the message as a whole, even though the report contains the same
information as the one quoting a non-official source. More
importantly, the non-official source was manipulated as "a friend who
had been camping with the missing hiker" (p. 13). As a witness he/she
knew more about the condition under which the incident occurred and
was able to provide more background information than the official
park ranger. Despite his/her knowledge as an eyewitness,
unfortunately, his/her lack of authority status disqualified him/her
from being considered as an unbiased source.
Elaboration Likelihood Model
A way to avoid being misled by over-generalization of the impact of
low-credible and high-credible sources is to consider the Elaboration
Likelihood Model (ELM). The model was developed by Petty and Cacioppo
(1986), who noted that research on cognitive responses to persuasion
in the previous decade failed to consider an important audience
variable. According to them, elaboration was "the extent to which a
person thinks about the issue-relevant arguments contained in a
message" (p. 128). There are various levels of elaboration and it is
best viewed as falling along a continuum (Perloff, 1993). The amount
of elaboration individuals engage in depends on how active they are
as information processors. According to Petty and Cacioppo (1986),
active information processors process the message they receive
through the central route, which involves heavy cognitive
elaboration; passive receivers take the peripheral route for
information processing and hardly engage in any elaboration work.
Under the Elaboration Likelihood Model, which pathway an individual
chooses for message processing is determined by his/her motivation
and ability to process. Motivation is more extensively studied in the
literature, of which involvement has been identified as the key
element. People are high in involvement when they perceive an issue
as highly relevant to their personal lives. They are low in
involvement when an issue bears little personal relevance to them.
Under high involvement situations individuals are motivated to select
the central route and engage in issue-relevant thinking. As a result
they pay more attention to fundamental features such as the quality
of an argument. Under low involvement conditions, however,
individuals are more likely to take the peripheral route, which
usually requires nothing but superficial thinking about the issue.
One of the possibilities is that under low involvement message
recipients make simple decisions by such peripheral cues as source
expertise (Petty, Cacioppo, & Goldman, 1981) or even the number of
arguments in a message (Eagly & Chaiken, 1984; Perloff, 1993). As a
consequence, they may agree with the message simply because it
includes an expert source or numerous, albeit weak, arguments. When
exposed to news, it is possible that low-involvement individuals
focus on the authoritativeness of sources, rather than attending to
the bias with which such sources present their account of a certain issue.
Hypotheses
Some unique features of the Internet require that online news readers
be active news seekers. For example, unlike traditional news reports,
online news stories are usually presented in a non-linear format with
the use of hyperlinks. Very often online news readers seek news they
are interested in by selecting certain hyperlinks and click on them.
As predicted by the Elaboration Likelihood Model (ELM), individuals
with high motivation tend to choose the central route for information
processing. As a consequence, they are less likely to make simple
decisions based on such source cues as authority and expertise.
Rather, they would pay more attention to the quality of the
arguments, including inherent bias on the part of the sources
providing these arguments. Therefore, officials, despite their
authority status, may be perceived as no less biased than the other
two types of sources.
H1a: Individuals with high motivation will perceive officials more
biased than involved parties;
H1b: Individuals with high motivation will perceive officials equally
biased as involved parties;
H2a: Individuals with high motivation will perceive officials more
biased than affiliated individuals;
H2b: Individuals with high motivation will perceive officials equally
biased as affiliated individuals.
When surfing a news Web site, however, some online news readers may
accidentally click on a link and come across news stories that they
are usually not interested in, or they might try any link on a Web
page and read a story to pass time. As a consequence, they are not
highly motivated to read those news stories and may simply judge the
stories by such peripheral cues as the authoritativeness of news
sources. However, the biases of different source types are not
equally apparent to the general audience. Because officials occupy
the center of the power structure in a society and have been the
dominant sources in news coverage, their biases are rarely noticeable
under the cover of authority and credibility the society assumes them
to possess. However, there is no conclusive evidence as to whether
involved parties or affiliated public members are considered biased.
In some studies it was found that traditional news audience tended to
ignore the bias of such involved parties as victims and witnesses
(Aust & Zillman, 1996; Gibson & Zillman, 1998; Zillman, Gibson,
Sundar, & Perkins, 1996), but were concerned about the bias of a
spokesperson of a non-governmental organization (Fedler & Counts,
1981). However, one study (Armstrong & Nelson, 2003) reported that a
witness (a fellow hiker in a story about a missing hiker) was
considered more biased than a non-governmental official (a park ranger).
H3: Individuals with low motivation will perceive officials less
biased than: a) involved parties, and b) affiliated individuals.
Methods
Overall Design
This study employed a 2_7 experimental design that combined a
within-subject and a between-subject design. A within-subject design
was used for the manipulation of the independent variable
"motivation." Each participant was asked to read four news stories on
different topics, two stories high in "motivation," and two low in
"motivation" (meaning that the stories aroused high/low levels of
motivation among the respondents). A between-subject design was
chosen for the manipulation of the other independent variable "story
type." As a consequence, there were seven treatment groups in this study.
One hundred and five college students at a Southwestern university
participated in this study in September 2004. Although a random
sample was not used in this study, efforts were made to randomly
assign treatments to the participants. This randomization helped
control for extraneous variances that might exist in certain groups
of participants.
The dependent variable, source bias, is a perceived attribute of a
news source whereby the individual news source, or the group the news
source represents, has an apparent vested interest in a cause or
action relative to maintaining or changing the status quo (Rouner,
Slater, & Buddenbaum, 1999). In a broader sense, it refers to whether
a news source presents only one side of a story. Researchers
(Andsager, 2003; Rouner et al., 1999) frequently used a single-item
scale "bias" to measure source bias. To increase the validity of the
measure, three additional items were used to measure the variable,
which are one-sided, neutral, and fair. Six-point Likert-type
statements were presented about the three sources quoted in each
story. Respondents were asked to indicate their level of agreement on
those statements (strongly agree, agree, slightly agree, slightly
disagree, disagree, and strongly disagree).
A Web site was designed specifically for this study. To rule out
story-specific effect (Jackson, 1992), two stories were used in each
treatment conditions. In the high motivation condition were a story
about prisoner abuse in Iraq and one about a proposed tuition
increase in the program where the respondents were enrolled in. The
low motivation condition included a story about protection of eagle
habitats in Ohio and one about budget changes for summer schools in
New York City.
Seven versions were created for each of the four stories. The basic
information in all seven versions was identical, with variation in
only the type of sources quoted within each story. The first three
versions each cited only one type of source, with version one citing
an official, version two citing an involved party, and version three
citing an affiliated public member. Versions four to six each cited
two types of sources, with version four citing both an official and
an involved party, version five citing both an official and an
affiliated public member, and version six citing an involved party
and an affiliated public member. The last version, version seven,
cited all three types of sources. In all versions except the last
one, quotations from the sources who were not included in each
version were converted into statements by the reporter. Slight
changes in language were made in the first six versions when needed
for grammatical and logical purposes. To help the respondent identify
the sources and refer back when necessary, all source names appeared
in boldface on the Web site.
A different version of a paper and pencil questionnaire was prepared
for each treatment group. All seven versions of the questionnaire
were identical in content. The major difference between the versions
was that they directed participants to different links on the Web
site, and the links took the participants to Web pages with matching
versions of the four treatment stories. For example, version one of
the questionnaire directed the participant to the link "A1," and this
hyperlink took him/her to a page with version one of story A, which
cited only an official source. After reading the first story and
completing the questions about the official source and the story
itself in the questionnaire, the participant read directions that
told him/her to click on "B1" on the current Web page, and that took
him/her to version one of story B, which also cited only an official
source. After reading the second story, he/she answered the questions
about the source and story B1. Similarly, the respondent completed
stories C1 and D1. Subjects assigned to other treatment groups read
similar instructions and went to the links that matched the
treatments they were assigned to.
Procedures for Data Collection
The experiment was carried out in computer labs with Internet access
at a mid-sized Southwestern university. To rule out the possibility
that students in one session were systematically different from those
in another session, individual subjects were randomly assigned to the
seven conditions.
Two graduate students administered the study in a computer laboratory
setting. All seven versions of the questionnaire were collated and
handed out in such a way that a randomly selected first participant
in the first session received version one, the second version two,
until the eighth receiving version one again. This systematic way of
distributing the questionnaire was employed during the whole process
of this study to ensure that each respondent was randomly assigned to
a treatment condition.
After completing the questions on demographics and general media use,
the respondents were instructed to visit the Web site of this study.
The first page of the Web site displayed the text "Welcome to
participate in this study" in a large font at the center of the
screen. A "Start" button was located below the welcoming message. The
respondents were told to click the "Start" button and open another
Web page, which listed seven buttons. Following the directions at the
bottom of page 2 of the questionnaire, the respondent clicked on the
button corresponding to the version he/she was assigned to (button A1
for version one; A2 for version two…A7 for version seven) and read
the first story.
At the end of the first story on the computer screen, there was an
instruction "End of Story A1(A2…A7). Now please turn to p. 3 of the
questionnaire and answer the questions about the story you just
read." After completing the questions about the first story,
respondents read another instruction on the questionnaire saying,
"Now please click on the button B1 (B2, …B7) at the bottom of your
computer screen and read story B1 (B2, …B7)." Similar to the first
story, at the end of the second story respondents read an
instruction, "Now please turn to p.4 of the questionnaire and answer
the questions about the story you just read." Similar instructions
were provided for stories C and D. At the end of the questionnaire,
respondents read, "That's all. Thank you!" After everybody in a
session completed, the questionnaires were collected. The respondents
were thanked and dismissed.
Findings
This study analyzed data from 105 students enrolled in classes at a
mid-sized public university. Most of the respondents were
Communications majors (84.8%). The youngest respondent was 18 years
old, and the oldest 43, with an average age of 20.7 years old. There
were slightly more females (51.4%) than males.
Repeated-measure ANOVA tests established that the effect of source
type on perception of source bias was uniform across stories.
Therefore, a composite measure of the dependent variable, source
bias, was created by averaging scores across the two stories under
the same treatment condition.
When reading stories arousing high motivation, the student
respondents rated affiliated individuals the least biased (mean =
3.51) among all three types of sources. For a six-point scale with a
mean of 3.50, the bias rating of affiliated individuals at 3.51
suggested that affiliated individuals were not actually considered
biased. Involved parties, such as the mistreated prisoner in Baghdad
and the local high school student who might be affected by a proposed
tuition increase, were considered to be the most biased (mean =
4.14). Officials were seen as slightly less biased (mean = 4.05) than
involved parties, but somewhat more biased than individuals
affiliated with non-governmental organizations. Different source
types were rated significantly different in terms of bias, as
indicated by the significant ANOVA result of the effect of source
type on the composite measure of source bias after controlling
demographic variables (F(2, 179) = 7.707, p. < .05).
Involved parties were also considered the most biased source (mean =
4.74) for the students reading stories that stimulated low
motivation. Officials (mean = 3.92) and affiliated individuals (mean
= 3.91) received identical bias ratings, both considerably lower than
involved parties. The source type had a significant effect on
perceived source bias under the low motivation condition (F. =
20.255, p. < .05) after controlling the variance of demographic
variables, as the ANOVA analysis suggested. The bias ratings also
showed that both involved parties and affiliated individuals were
rated as far more biased when respondents were not motivated than
when they were highly motivated. As the level of motivation
increased, however, respondents tended to believe that officials were
less biased.
H1a: Individuals with high motivation will perceive officials more
biased than involved parties;
H2b: Individuals with high motivation will perceive officials equally
biased as involved parties.
When the student participants were exposed to stories that evoked
high motivation, they considered officials slightly less biased than
involved parties. However, as shown in Table 1, the difference was
not statistically significant (t (1, 118) = -.554, p. > .05).
Therefore, H1a, which proposed that respondents with a high level of
motivation would consider officials more biased than involved
parties, was rejected and H1b, which predicted that respondents with
highly motivation would consider officials equally biased as involved
parties, was supported.
H2a: Individuals with high motivation will perceive officials more
biased than affiliated individuals;
H2b: Individuals with high motivation will perceive officials equally
biased as affiliated individuals.
For students reading the stories on prisoner abuse and tuition
increase (both arousing high motivation), officials were considerably
more biased (mean = 4.05) than individuals affiliated with
non-government organizations (mean = 3.51). T-test results in Table 1
suggested that the difference in bias scores was significant (t (1,
118) = 3.398, p. < .05), leading to the retention of H2a, which
stated that respondents with high motivation would rate officials as
more biased than affiliated individuals, and the rejection of H2b,
which proposed that the two source types would be equally biased.
Table 1
Perceived Source Bias as a Function of Source Type
Source type
t (1, 118)
p.
Official
Involved
Affiliated
High motivation
4.05
4.14
-.554
.580
4.05
3.51
*3.398
.001
Low motivation
3.92
4.74
*-5.689
.000
3.92
3.91
.089
.930
(Note: * indicated a significant effect at alpha = .05)
H3a: Individuals with low motivation will perceive officials less
biased than involved parties;
H3b: Individuals with low motivation will perceive officials less
biased than affiliated individuals.
When reading the stories on eagle protection and summer schools,
which aroused a low level of motivation, the students believed that
officials (mean = 3.92) were considerably less biased than involved
parties (mean = 4.74). As displayed in Table 1, officials were rated
as significantly less biased than involved parties (t (1, 118) =
-5.689, p. < .05), providing support for H3a, which predicted that
respondents with low motivation would perceive officials as less
biased than involved parties. However, for respondents under the same
low motivation condition, officials and affiliated individuals were
nearly even in bias (t (1, 118) = .089, p. > .05), leading to the
rejection of H3b, a proposition that officials would be seen as less
biased than affiliated individuals.
Discussion and Conclusions
This study examined how different types of online news sources were
perceived in terms of bias. College students from a mid-sized
Southwestern public university participated in the study in an
experimental setting. Participants in each of the seven treatment
groups were asked to read a version of four news stories posted on a
Web site designed specifically for this study. Two of the stories
were about topics that were considered highly relevant to the student
participants and hence would arouse a high level of motivation from
the participants, and the other two stories were about topics that
were seen to bear little relevance to the students, thus arousing a
low level of motivation. Each version of the stories quoted one of
the three types of sources (officials, involved parties, and
affiliated individuals) or a combination of the three source types.
The findings provide evidence that the Elaboration Likelihood Model,
a theoretical model in understanding how people process information
delivered through traditional mass communication and interpersonal
communication, bears strong explanatory power in how individuals
process news delivered on the Internet. The results from this study
also support the proposition that source type plays an important role
in people's perception of source bias and online news credibility.
Among the three types of sources quoted in different versions of
stories manipulated in this study, some sources were considered
biased, while others were not, despite the presumption that any
source could present only one perspective of a certain issue or event.
In addition to the strong support this study lends to the ELM in
online news, an interesting finding of this study was that for the
college student subjects in this study as a group, affiliated
individuals were considered as not biased or low in bias, but
involved parties were always seen as biased. This is consistent with
an earlier study (Armstrong & Nelson, 2003), which found that a
witness was believed to be more biased than an individual affiliated
with a non-governmental organization (a park ranger). That affiliated
individuals were considered to be low in bias can be explained by
saying that affiliated individuals are usually perceived as
disinterested and less likely to exercise authority over the public,
at least those individuals who hold distrust of politicians might
think so. This explanation is relevant to the ELM, but also differs
from the ELM in that it carries across different levels of
motivation. It is not yet clear, however, why involved parties were
considered high in bias, despite their apparent involvement in the
issue and hence first-hand experience about what has happened. A
possible explanation might be that involved parties are usually
perceived to be highly emotional and lacking sufficient detachment to
provide neutral and rational elaboration of an occurrence.
For the respondents with low motivation, officials were no longer
considered biased, although the same type of source was believed to
be highly biased among individuals reading highly motivating stories.
This is not surprising, however, considering the authority officials
possess and the ELM prediction that individuals under the low
motivation condition base their judgment on the authority of a
source. What deserves attention is that although both officials and
affiliated individuals received low bias ratings among respondents
reading stories arousing low motivation, only the low bias of
officials was carried over to participants' perception of story
credibility. In other words, while the perceived low bias of
officials helped foster the credibility of stories quoting officials,
the similarly low bias of affiliated individuals did not seem to have
affected the credibility of the stories quoting affiliated
individuals. This disparity in how perceived bias of officials and
that of affiliated individuals affected the perceived credibility of
the stories quoting these sources may be related to how valid the
perception of the low bias of each source type is among the student
participants when they read stories arousing little motivation. A
possibility is that for individuals exposed to low motivation
stimulus, officials and affiliated individuals were both considered
to possess low levels of bias but according to differing standards.
This study is among the first to use the Elaboration Likelihood Model
to explain people's perception of online news messages. The strong
support for the ELM found in this study indicates that the ELM should
not be limited to traditional media research. Although online news is
an extensive category and may still lack a clear definition among the
public, online versions of traditional news media are an important
part of online news. This study provides evidence that, in addition
to its strong explanatory power established in traditional media
research, the ELM is also helpful in our understanding of online news
perception.
Suggestions for Future Research
This study is an exploratory examination of the impact of source type
on people's perception of source bias and news credibility in the
online environment. Much research is desirable to test the
explanatory power of source bias as the predictor variable and to
address the limitations mentioned in the previous section.
Future research should study other samples from the online news
population. First, it is important to know whether the findings from
this study are representative of the perceptions of college students
in other parts of the country, e.g., students in a metropolitan
institution, students at a private research university, or those at a
small community college. Second, because it is currently impossible
to study a random sample of all online news users in the United
States, further studies should be conducted to examine other segments
of this population, especially those with highly different
demographics than the somewhat homogenous student sample employed in
this study. For example, because college students represent a narrow
range of age, it is important to study individuals of other age
groups, such as the middle-aged or the elderly. It can also be
revealing if similar studies can be carried out with individuals with
less education, or with people who are more dependent upon
traditional news media than most college students are.
References
Andsager, J. (2003). Consider the source: How perceived bias affects
message credibility. In Paper presented to the Midwest Association
for Public Opinion Research annual conference, November 2003,
Chicago, Illinois.
Armstrong, C. L., & Nelson, M. R. (2003, November). Conditional
characterizations: The influence of cognition and personality on
gender assessments of official and non-official newspaper sources. In
Paper presented at the Midwest Association of Public Opinion Research
annual conference, Chicago, Illinois.
Atwater, T. (1987). Network evening news coverage of the TWA hostage
crisis. Journalism Quarterly, 64, 520-525.
Aust, C. F., & Zillman, D. (1996). Effects of victim exemplification
in television news on viewer perception of social issues. Journalism
and Mass Communication Quarterly, 73, 787-803.
Berkowitz, D. (1987). TV news sources and news channels: A study in
agenda-building. Journalism Quarterly, 64, 513.
Brown, J. D., Bybee, C. R., Weardon, S. T., & Straughan, D. M.
(1987). Invisible power: Newspaper news sources and the limits of
diversity. Journalism Quarterly, 64, 45-54.
Eagly, A. H., & Chaiken, S. (1984). Cognitive theories of persuasion.
In L. Berkowitz (Ed.), Advances in Experimental Social Psychology
(pp. 267-359). Hillsdale, NJ: Lawrence Erlbaum Associates.
Ericson, R. V., Baranek, P. M., & Chan, J. B. L. (1987). Visualizing
Deviance: A Study of News Organization. Toronto: University of Toronto Press.
Fedler, F., & Counts, T. (1981). Variations in attribution affect
readers' evaluations of stories. Newspaper Research Journal, 2, 25-34.
Gallimore, T. (1991). Media compliance with voluntary press
guidelines for covering terrorism. In Y. Alexander & F. G. Picard
(Eds.), In the camera's eye: News coverage of terrorist events (pp.
103-118). Brassey's (US), Inc.
Gans, H. J. (1979). Deciding what's news. New York: Vintage Books.
Gibson, R., & Zillman, D. (1993). The impact of quotation in news
reports on issue perception. Journalism Quarterly, 70, 793-800.
Gibson, R., & Zillman, D. (1998). Effects of citation in exemplifying
testimony on issue perception. Journalism Quarterly, 75, 167-176.
Goffman, E. (1974). Frame analysis: An essay on the organization of
experience. NY: Harper & Row.
Harris, L. (1979). ABC news-Harris survey. Chicago: Chicago Tribune.
Izard, R. S., Culbertson, H. M., & Lambert, D. A. (1977).
Fundamentals of news reporting. Dubuque, IA: Kendall/Hunt Publishing Co.
Jackson, S. (1992). Message effects research: Principles of design
and analysis. New York: The Guilford Publications, Inc.
Janowitz, M. (1991). On social organization and social control.
Chicago: University of Chicago Press.
Jones, J. P. (1976). Gathering and writing the news. Chicago: Nelson-Hall.
Online News Association (2001, July 21). "Digital journalism
credibility study." Retrieved April 4, 2004, from
http://www.journalist.org/Programs/StudyText.htm
Paletz, D. L., Fozzard, P. A., & Ayanian, J. Z. (1982). The I. R.A.,
the Read Brigades, and the F.A.L.N. in the "New York Times." Journal
of Communication, 32, 162-171.
Perloff, R. M. (1993). The Dynamics of Persuasion (1st ed.).
Hillsdale, NJ: Laurence Erlbaum Associates.
Petty, R. E., & Cacioppo, J. T. (1986). The elaboration likelihood
model of persuasion. In L. Berkowitz (Ed.), Advances in experimental
social psychology (pp. 123-205). New York: Academic Press.
Petty, R. E., Cacioppo, J. T., & Goldman, R. (1981). Personal
involvement as a determinant of argument-based persuasion. Journal of
Personality and Social Psychology, 41, 847-855.
Pew Internet & American Life Project (2005). Internet: The
mainstreaming of online life. Retrieved March 9, 2005, from
http://www.pewinternet.org/PPF/r/148/report_display.asp
Picard, R. G., & Adams, P. D. (1991). Characterizations of acts and
perpetrators of political violence in three elite U.S. daily
newspapers. In A. O. Alali & K. K. Eke (Eds.), Media Coverage of
Terrorism: Methods of Diffusion (pp. 12-21). Newbury Park, Calif.:
Sage Publications.
Rouner, D., Slater, M. D., & Buddenbaum, J. M. (1999). How
perceptions of news bias in news sources relate to beliefs about
media bias. Newspaper Research Journal, 20, 41-51.
Shoemaker, P. J., & Reese, S. D. (1991). Mediating the message:
Theories of influences on mass media content. New York: Longman.
Sigal, L. V. (1973). Reporters and officials: The organization and
politics of newsmaking. Lexington, Mass.: D.C. Heath and Co.
Soloski, J. (1989). Sources and channels of local news. Journalism
Quarterly, 66, 864-870.
Sundar, S. S. (1998). Effect of source attribution on perception of
online news stories. Journalism and Mass Communication Quarterly, 75, 55-68.
Thurwanger, M. L. (2003). Community Newspapers as Instruments of
Social Control: Framing Prison Site Selection in Rural Illinois
(Doctoral dissertation, Southern Illinois University Carbondale, 2003).
Zeng, L. (2002, November). Coverage of the "9/11" attack in elite
newspapers: A comparative study of the U.S. and U.K. In paper
presented to the Midwest Association of Public Opinion Research
annual conference, Chicago: Illinois.
Zeng, L. (2003, July-August). Characterizations of the "9/11" attack
and perpetrators in three U.S. elite newspapers. In paper presented
to the Association for Education in Journalism and Mass Communication
annual conference, Kansas City, Missouri.
Zeng, L., & Xiao, Z. (2003, November). Who and how? Investigating
sources and characterizations in elite British newspaper coverage of
the "9/11" attack. In paper presented to the Midwest Association for
Public Opinion Research annual conference, Chicago, Illinois.
Zillman, D., Gibson, R., Sundar, S. S., & Perkins, J. W. Jr. (1996).
Effects of exemplification in news reports on the perception of
social issues. Journalism and Mass Communication Quarterly, 73, 427-444.
Zoch, L. M., & Turk, J. V. (1998). Women making news: Gender as a
variable in source selection and use. Journalism and Mass
Communication Quarterly, 75, 762-775.
[1] Although only 1.6 percent of the total, witnesses played a role
that was more than trivial because Zeng (2003) employed the same
scales that Picard and Adams (1991) used. The media, which is not
strictly a news source, contributed to more than 70% of the
characterization terms in Zeng's study, therefore leading to a
smaller percentage for witnesses than they actually accounted for.
|
