Research in Media Effects
(Revised October 2009)
Mass Media Research: An Introduction, 9 Edition
Roger D. Wimmer and Joseph R. Dominick
While much research is conducted in professional or industry settings, a great deal of mass
media research is conducted at colleges and universities. There are several differences between
research in the academic and the private sectors, including, but not limited to:
• Academic research tends to be more theoretical in nature; private-sector research is
generally more applied.
• The data used in academic research are public, whereas much industry research is based
on proprietary data.
• Top management often determines private-sector research topics; academic researchers
have more freedom in their choice of topics.
• Projects in private-sector research usually cost more to conduct than do academic
The two research settings also have some common features:
• Many research techniques and approaches used in the private sector emerged from
• Industry and academic researchers use the same basic research methodologies and
• The goal of research is often the same in both settings—to explain and predict audience
and consumer behavior.
This chapter describes some of the more popular types of research carried out by academic
investigators and shows how this work relates to private sector research.
Obviously, not every type of scholarly research used in colleges and universities can be
covered in one chapter. What follows is not an exhaustive survey but rather an illustrative
overview of the history, methods, and theoretical development of five research areas: antisocial
and prosocial effects of specific media content, uses and gratifications, agenda setting by the
media, cultivation of perceptions of social reality, and the social impact of the Internet. Readers
who want a more comprehensive treatment of media effects research should consult Bryant and
Wimmer & Dominick: Media Effects – Page 1 Antisocial and Prosocial Effects of Media Content
The antisocial effect of viewing television and motion pictures is one of the most heavily
researched areas in all mass media studies. Comstock, Chaffee, and Katzman (1978) reported
that empirical studies focusing on this topic outnumbered work in all other problem areas by four
to one, and this emphasis is still apparent more than a decade later. Paik and Comstock (1994)
reviewed the results of 217 such studies conducted between 1959 and 1990.
The impact of prosocial content is a newer area and grew out of the recognition that the
same principles underlying the learning of antisocial activities ought to apply to more positive
behavior. Applied and academic researchers share an interest in this area: All the major
networks have sponsored such research, and the effects of antisocial and prosocial content have
been popular topics on college and university campuses for the past 30 years. It is not surprising
that there has been a certain amount of friction between academic researchers and industry
Concern over the social impact of the mass media was evident as far back as the 1920s, when
many critics charged that motion pictures had a negative influence on children. In 1928, the
Motion Picture Research Council, with support from the Payne Fund, a private philanthropic
organization, sponsored a series of 13 studies on movies’ influence on children. After
examination of film content, information gain, attitude change, and influence on behavior, it was
concluded that the movies were potent sources of information, attitudes, and behavior for
children. Furthermore, many of the things that children learned had antisocial overtones. In the
early 1950s, another medium, the comic book, was chastised for its alleged harmful effects
In 1960, Joseph Klapper summarized what was then known about the social impact of mass
communication. In contrast to many researchers, Klapper downplayed the potential harmful
effects of the media. He concluded that the media most often reinforced an individual’s existing
attitudes and predispositions. Klapper’s viewpoint, which became known as the minimal effects
position, was influential in the development of a theory of media effects.
In the late 1950s and early 1960s, concern over the antisocial impact of the media shifted to
television. Experiments on college campuses by Bandura and Berkowitz (summarized in
Comstock & Paik, 1991) showed that aggressive behavior could be learned by viewing violent
media content and that a stimulation effect was more probable than a cathartic (or cleansing)
effect. Senate subcommittees examined possible links between viewing violence on television
and juvenile delinquency, and in 1965, one subcommittee concluded that televised crime and
violence were related to antisocial behaviors among juvenile viewers.
The civil unrest and assassinations in the middle and late 1960s prompted the formation of
the National Commission on the Causes and Prevention of Violence, chaired by Milton
Eisenhower. The staff report of the Eisenhower Commission, which concluded that television
violence taught the viewer how to engage in violence, included a series of recommendations
about reducing the impact of television violence.
Wimmer & Dominick: Media Effects – Page 2 The early 1970s saw extensive research on the social effects of the mass media. Just three
years after the publication of the Eisenhower Commission report came the release of a multi-
volume report sponsored by the Surgeon General’s Scientific Advisory Committee on Television
and Social Behavior (1972, p. 10). In Television and Growing Up, the committee cautiously
summarized its research evidence:
There is a convergence of fairly substantial evidence on short-run causation of
aggression among children by viewing violence . . . and the much less certain
evidence from field studies that . . . violence viewing precedes some long-run
manifestation of aggressive behavior. This convergence . . . constitutes some
preliminary evidence of a causal relationship.
The committee tempered this conclusion by noting that in accordance with the reinforcement
notion, “any sequence by which viewing television violence causes aggressive behavior is most
likely applicable only to some children who are predisposed in that direction” (p. 10).
At about the same time, the three television networks were sponsoring research in this area.
CBS commissioned two studies: a field experiment that found no link between television
viewing and subsequent imitation of antisocial behavior (Milgram & Shotland, 1973), and a
longitudinal study in Great Britain that found an association between viewing violence on
television and committing antisocial acts such as damaging property and hurting others (Belson,
1978). ABC sponsored a series of studies by two mental health consultants who concluded that
television stimulated aggression to only a tiny extent in children (Heller & Polsky, 1976). NBC
began a large-scale panel study, but results were not released until 1983. In addition to television
violence, the potential antisocial impact of pornography was under scrutiny. The Commission
on Obscenity and Pornography (1970), however, reported that such material was not a factor in
determining antisocial behavior. The commission’s conclusions were somewhat controversial in
political circles, but in general they supported the findings of other researchers in human
sexuality (Tan, 1986). Subsequent efforts in this area were directed primarily toward examining
links between pornography and aggression.
Along with violence and pornography, the contrasting prosocial effect of television was
investigated as well. One stimulus for this research was the success of the television series
Sesame Street. A substantial research effort went into the preparation and evaluation of these
children’s programs. It was found that the series was helpful in preparing young children for
school but not very successful in narrowing the information gap between advantaged and
disadvantaged children (Minton, 1975). Other studies by both academic researchers and industry
researchers demonstrated the prosocial impact of other programs. For example, the series Fat
Albert and the Cosby Kids was found to be helpful in teaching prosocial lessons to children (CBS
Broadcast Group, 1974).
Studies of these topics continued between 1975 and 1985, although there were far fewer than
in the early 1970s. An update to the 1972 Surgeon General’s Report, issued in 1982, reflected a
broader research focus than the original document; it incorporated investigations of socialization,
mental health, and perceptions of social reality. Nonetheless, its conclusions were even stronger
than those of its predecessor: “The consensus among most of the research community is that
violence on television does lead to aggressive behavior” (National Institute of Mental Health,
Wimmer & Dominick: Media Effects – Page 3 1982, p. 8). Other researchers, notably Wurtzel and Lometti (1984) and Bear (1984), argued that
the report did not support the conclusion of a causal relationship, whereas Chaffee (1984) and
Murray (1984), among others, contended that the conclusions were valid.
Not long after the Surgeon General’s report was updated, the results of the NBC panel study
begun in the early 1970s were published (Milavsky, Kessler, Stipp, & Rubens, 1983). This panel
study, which used state-of-the-art statistical analyses, found a nonsignificant relationship
between viewing television violence during the early phases of the study and subsequent
aggression. The NBC data have been reexamined by others, and at least one article suggests that
the data from this survey do show a slight relationship between violence viewing and aggression
among at least one demographic subgroup—middle-class girls (Cook, Kendzierski, & Thomas,
From 1985 to 2001, the controversy subsided, but this topic remained popular among
academic researchers. Williams (1986) conducted an elaborate field experiment in three
Canadian communities. One town was about to receive television for the first time, another
received Canadian TV, and the third received both Canadian and U.S. programs. Two years
later, Williams and her colleagues found that when compared to children in the other two
communities, children in the town that had just received TV scored higher on measures of
physical and verbal aggression.
Additional evidence on the topic of television and violence comes from a series of panel
studies conducted by an international team of researchers (Huesmann & Eron, 1986). Data were
gathered from young people in the United States, Finland, Australia, Israel, and Poland. Findings
from the U.S. and Polish studies reached a similar conclusion: Early TV viewing was related to
later aggression. The Finnish study found this relationship for boys but not for girls. The Israeli
study found that TV viewing seemed to be related to aggression for children living in urban areas
but not for those in rural areas. The Australian study failed to find a relationship. In all countries
where a relationship between TV viewing and violence was found, the relationship was relatively
weak. Rosenthal (1986), who concluded that even a weak relationship could have substantial
social consequences, examined the practical implications of this weak relationship.
More recently, Congress passed the Telecommunications Act of 1996. Part of the act
specified that newly manufactured TV sets had to contain a V-chip, a computer chip that allows
parents to block out violent and other objectionable programming from their TV sets. The chip
would work in concert with a ratings system developed by the industry. (Recent research
suggests that consumers have largely ignored the V-chip. One study found that 53% of
consumers who had recently purchased a new TV set were not even aware they had a V-chip. A
Kaiser Family Foundation study discovered that only 17% of families were using the V-chip to
Another recent research area examined mediating effects on the viewing of TV violence.
Nathanson (1999), for example, confirmed that parental mediation of TV viewing helped curtail
the antisocial inclinations of their children. The same researcher (Nathanson, 2001) also
examined the influence of peer mediation on antisocial TV viewing. She found that peer
influence was more frequent and more potent than parental mediation and that it tended to
promote a positive attitude toward antisocial TV.
Wimmer & Dominick: Media Effects – Page 4 The violence at Columbine High School in Littleton, Colorado, and in other high schools at
the end of the century, sparked renewed interest in media violence among parents and policy
makers. Media leaders were called before a congressional committee investigating this topic. In
2001, the Surgeon General issued a report entitled Youth Violence, a document that included a
study of the factors that contributed most to antisocial behavior among young people. The report
concluded that media violence was less of a risk factor than family influences, peer group
attitudes, socioeconomic status, and substance abuse (U.S. Department of Health and Human
The increasing popularity of video games during the early years of this decade opened up
another avenue of inquiry for researchers. Since more than 90% of young people report that they
sometimes play these games, and since some of the more popular games feature graphic and
explicit violence (Doom, Grand Theft Auto), social concern over their impact was widespread.
Results of some of the early studies in this area (for example, Silvern & Williamson, 1987)
suggest that playing video games can lead to increased aggression levels in young children and is
related to their self-concepts (Funk & Buchman, 1996). More recent research, however, has
Results from both surveys and experiments have been mixed with some studies finding a
relationship between exposure to violent games and antisocial behavior while others found no
relationship. Meta-analyses have also reached different conclusions. For example, Anderson
and Bushman (2001) and Anderson (2004) found a small but significant correlation between
violent game-playing and aggression while Sherry (2001, 2007) concluded that no relationship
existed. A meta-analysis by Ferguson (2007) suggested that publication bias, the tendency of
journals to publish only those studies with significant effects, was a factor in those meta-analyses
that found a significant link. When publication bias was controlled, Ferguson found no evidence
that violent games were associated with aggressive behavior.
Research about the antisocial effects of pornography increased in the late 1980s but has
recently declined. One controversial research area examined if prolonged exposure to nonviolent
pornography had any antisocial effects (Donnerstein, Linz, & Penrod, 1987; Zillmann & Bryant,
1989; Allen, D’Alessio, & Brezgel, 1995). The most recent studies have focused on exposure to
pornographic Internet sites. For example, Peter and Valkenburg (2008) found a link between
exposure to pornographic Internet sites and adolescents’ positive attitudes toward casual sex.
Research interest in the prosocial effects of media exposure decreased in the 1980s and has
remained at low level into the end of the 2000s. Sprafkin and Rubinstein (1979) reported on a
correlational study in which the viewing of prosocial television programs accounted for only 1%
of the variance in an index of prosocial behavior exhibited in school. The apparent lack of a
strong relationship between these two variables, coupled with the absence of general agreement
on a definition of prosocial content, might have discouraged researchers from selecting this area.
In any case, few studies of the media impact on prosocial behavior have appeared in the
scholarly literature in the last five years. The meta-analysis of Anderson and Bushman (2001)
found only a handful of prosocial studies to analyze but concluded that playing violent video
games is linked to a decline in prosocial behavior.
Wimmer & Dominick: Media Effects – Page 5 Methods
Researchers who study the effects of mass media have used most of the techniques discussed in
this book: content analysis, laboratory experiments, surveys, field experiments, observations, and
panels. In addition, they have used some advanced techniques, such as meta-analysis, that have
not been discussed. Given the variety of methods used, it is not possible to describe a typical
approach. Instead, this section focuses on five different methods as illustrations of some
The Experimental Method. A common design used to study the antisocial impact of the
media is to show one group of subjects violent media content while a control group sees
nonviolent content. This was the approach used by Berkowitz and Bandura in their early work.
The dependent variable, aggression, is measured immediately after exposure—either by a pencil-
and-paper test or by a mechanical device like the one described next. For example, Liebert and
Baron (1972) divided children into two groups. The first group saw a 3.5-minute segment from a
television show depicting a chase, two fistfights, two shootings, and a knifing. Children in the
control group saw a segment of similar length in which athletes competed in track and field
events. After viewing, the children were taken one at a time into another room that contained an
apparatus with two buttons, one labeled “Help” and the other labeled “Hurt.” An experimenter
explained to the children that wires from the device were connected to a game in an adjacent
room. The subjects were told that in the adjacent room, another child was starting to play a game.
(There was, in fact, no other child.) At various times, by pressing the appropriate buttons, each
child was given a chance either to help the unseen child win the game or to hurt the child. The
results showed that children who had seen the violent segment were significantly more likely
than the control group to press the “Hurt” button. Of course, there are many variations on this
basic design. For example, the type of violent content shown to the subjects can be manipulated
(cartoon versus live violence, entertainment versus newscast violence, justified versus unjustified
violence). Also, some subjects may be frustrated before exposure. The degree of association
between the media violence and the subsequent testing situation may be high or low. Subjects
can watch alone or with others who praise or condemn the media violence. Media exposure can
be a one-time event or it can be manipulated over time. For a thorough summary of this research,
see Comstock and Paik (1991) and Liebert and Sprafkin (1992).
Experimental studies to examine the impact of media exposure on prosocial behavior have
used essentially the same approach. Subjects see a televised segment that is either prosocial or
neutral, and the dependent variable is then assessed. For example, Forge and Phemister (1987)
randomly assigned preschoolers to one of four conditions: prosocial animated program (The Get-
along Gang), neutral animated (Alvin and the Chipmunks), prosocial nonanimated (Mr. Rogers’
Neighborhood), and neutral nonanimated (Animal Express). The children watched the program
and were then placed in a free-play situation where their prosocial behaviors were observed and
recorded. The results demonstrated an effect for the program variable (prosocial programs
prompted more prosocial behaviors than did neutral programs) but no effect for the animated
versus nonanimated variable.
The operational definitions of prosocial behavior have varied widely: Studies have examined
cooperative behaviors, sharing, kindness, altruism, friendliness, creativity, and absence of
stereotyping. Almost any behavior with a positive social value seems to be a candidate for study,
Wimmer & Dominick: Media Effects – Page 6 as exemplified by the experiment by Baran, Chase, and Courtright (1979): Third-graders were
assigned to one of three treatment conditions. One group saw a condensed version of a segment
of The Waltons demonstrating cooperative behavior; the second group saw a program portraying
noncooperative behavior; and the third group saw no program. After answering a few written
questions dealing with the program, each subject left the viewing room only to encounter a
confederate of the experimenter who passed the doorway and dropped an armload of books.
There were two dependent measures: whether the subject attempted to retrieve the books and
how much time elapsed until the subject began to help. The group that saw the cooperative
content was more likely to help, and their responses were quicker than those of the control group.
It is interesting that there was no difference in helping behavior or in time elapsed between the
group that saw The Waltons and the group that saw the noncooperative content.
The Survey Approach. Most survey studies have used questionnaires that incorporate
measures of media exposure (such as viewing television violence or exposure to pornography)
and a pencil-and-paper measure of antisocial behavior or attitudes. In addition, many recent
studies have included measures of demographic and sociographic variables that mediate the
exposure–antisocial behavior relationship. Results are usually expressed as a series of
A survey by McLeod, Atkin, and Chaffee (1972) illustrates this approach. Their
questionnaire contained measures of violence viewing, aggression, and family environment.
They tabulated viewing by giving respondents a list of 65 prime-time television programs with a
scale measuring how often each was viewed. An index of overall violence viewing was obtained
by using an independent rating of the violence level of each show and multiplying it by the
frequency of viewing. Aggression was measured by seven scales. One measured respondents’
approval of manifest physical aggression (sample item: “Whoever insults me or my family is
looking for a fight”). Another examined approval of aggression (“It’s all right to hurt an enemy
if you are mad at him”). Respondents indicated their degree of agreement with each of the items
on the separate scales. Family environment was measured by asking about parental control over
television, parental emphasis on nonaggressive punishment (such as withdrawal of privileges),
and other variables. The researchers found a moderate positive relationship between the
respondents’ level of violence viewing and their self-reports of aggression. Family environment
showed no consistent association with either of the two variables.
Sprafkin and Rubinstein (1979) used the survey method to examine the relationship between
television viewing and prosocial behavior. They used basically the same approach as McLeod,
Atkin, and Chaffee (1972), except their viewing measure was designed to assess exposure to
television programs established as prosocial by prior content analysis. Their measure of
prosocial behaviors was based on peer nominations of people who reflected 12 prosocial
behaviors, including helping, sharing, following rules, staying out of fights, and being nice. The
researchers found that when the influence of the child’s gender, the parents’ educational level,
and the child’s academic level were statistically controlled, exposure to prosocial television
explained only 1% of the variance in prosocial behaviors.
Field Experiments. Parke, Berkowitz, and Leyens (1977) conducted a field experiment in a
minimum-security penal institution for juveniles. The researchers exposed groups to unedited
feature-length films that were either aggressive or nonaggressive. On the day after the last film
Wimmer & Dominick: Media Effects – Page 7 was shown, in the context of a bogus learning experiment, the boys were told they had a chance
to hurt a confederate of the experimenters who had insulted one group of boys and had been
neutral to the other. The results on an electric shock measure similar to the one used in the
Liebert and Baron (1972) study, described previously, revealed that the most aggressive of all the
experimental groups were the boys who had seen the aggressive films and had been insulted. In
addition to this laboratory measure, the investigators collected observational data on the boys’
aggressive interpersonal behavior in their everyday environment. These data showed that boys
who saw the violent movies were more interpersonally aggressive. However, there was no
apparent cumulative effect of movies on aggression. The boys who watched the diet of
aggressive films were just as aggressive after the first film as after the last.
Figure 9.11 illustrates the design of the Canadian field experiment (Williams, 1986)
discussed earlier. The dependent variable of aggression was measured in three ways:
observations of behavior on school playgrounds, peer ratings, and teacher ratings. On the
observational measure, the aggressive acts of children in the town labeled A (the town that just
received TV) increased from an average of 0.43 per minute in Phase 1 to 1.1 per minute in Phase
2. Children in the other towns showed only a slight and statistically insignificant increase in the
same period. Peer and teacher ratings tended to support the behavioral data. As yet, there have
been no large-scale field experiments examining prosocial behavior.
Panel Studies. Primarily because of the time and expense involved in panel studies, this
method is seldom used to examine the antisocial effects of the media. Five studies relevant to
this topic are briefly reviewed here. Lefkowitz, Eron, Waldner, and Huesmann (1972), using a
catch-up panel design, reinterviewed 427 of 875 youthful subjects 10 years after they had
participated in a study of mental health. Measures of television viewing and aggression had been
administered to these subjects when they were in the third grade, and data on the two variables
were gathered again a decade later. Slightly different methods were used to measure television
viewing on the two occasions. Viewing in the third grade was established based on mothers’
reports of their children’s three favorite television shows. Ten years later, respondents rated their
own frequency of viewing. The data were subjected to cross-lagged correlations and path
analysis. The results supported the hypothesis that aggression in later life was caused in part by
television viewing during early years. However, the panel study by Milavsky and colleagues
(1983), sponsored by NBC, found no evidence of a relationship.
The difference between the results of these studies might be due to several factors. The
Milavsky study did not vary its measure of “violent television viewing” throughout its duration.
In addition, the NBC researchers used LISREL (linear structural equations), a more powerful
statistical technique, which was not available at the time of the Lefkowitz study. Finally, the
Lefkowitz measures were taken 10 years apart; the maximum time lag in the NBC study was 3
Another panel study of the media and possible antisocial effects was conducted by
Huesmann and Eron (1986). The investigators followed 758 children who were in the first and
third grades in 1977 and reinterviewed them in 1978 and 1979. Aggression was measured by
both peer nominations and self-ratings. Multiple regression analyses disclosed that, for both boys
and girls, watching TV violence was a significant predictor of the aggression they would later
demonstrate. Other significant variables were the degree to which children identified with
Wimmer & Dominick: Media Effects – Page 8 violent TV characters, the perceived reality of the violence, and the amount of a child’s
More recently, two longitudinal panel studies have found long-term effects of viewing TV
violence. Huesmann, Moise-Titus, Podolski and Eron (2003) did a 15-year follow-up study with
more than 300 respondents from surveys originally conducted in the 1970s. They found that
respondents who watched violent shows at age 8 were more likely to be more aggressive in their
20s. The results remained significant even when such factors as IQ, social class, and parenting
differences were statistically controlled. A second study (Johnson, Cohen, Smailes, Kasen, &
Brook, 2002) found a significant association between the amount of time spent watching TV
during their respondents’ teenage years and aggressive behavior as young adults. The results of
this study, however, were criticized because the researchers measured general TV viewing rather
than viewing of violent programs.
Meta-analysis. A complete description of the techniques of meta-analysis is beyond the
scope of this book. For our purposes, meta-analysis is defined as the quantitative aggregation of
many research findings and their interpretations. It allows researchers to draw general
conclusions from an analysis of many studies that have been conducted concerning a definable
research topic. Its goal is to provide a synthesis of an existing body of research. Given the large
number of research studies that have been conducted concerning antisocial and prosocial
behavior, it is not surprising that the mid- to late-1990s saw the growth in popularity of meta-
analytic research in this area. Five examples of meta-analysis are discussed here.
Paik and Comstock (1994) performed a meta-analysis on 217 studies from 1959 to 1990 that
tested 1,142 hypotheses. They concluded that the magnitude of the impact of exposure to media
violence varied with the method used to study it. Experiments produced the strongest effects,
and time-series studies the weakest. Nonetheless, there was overall a highly significant positive
association between exposure to portrayals of violence and antisocial behavior. In addition, they
found that males were affected by exposure to media violence only slightly more than females
and that violent cartoons and fantasy programs produced the greatest magnitude of effects. The
latter finding is at odds with the conventional argument that cartoon violence does not affect
viewers because it is unrealistic.
A second meta-analysis on the impact of exposure to pornography and subsequent aggressive
behavior was done by Allen, D’Alessio, and Brezgel (1995). They analyzed the results of 30
studies and found that there was indeed a connection between exposure to pornography and
subsequent antisocial behavior. More specifically, they noted that exposure to nudity actually
decreased aggressive behavior. In contrast, consumption of material depicting nonviolent sexual
activity increased aggressive behavior, while exposure to violent sexual activity generated the
highest levels of aggression. These findings are in accord with those discussed by Paik and
Comstock (1994). A meta-analysis of studies examining exposure to pornography and
acceptance of rape myths (Allen, Emmers, Gebhardt, & Geiry, 1995) revealed that experimental
studies showed a positive relationship between pornography and rape myth acceptance but
nonexperimental studies displayed no such effects.
Friedlander (1993) reported the results of a meta-analysis that compared the magnitude of
effects reported by studies that looked at antisocial behavior with those that examined prosocial
Wimmer & Dominick: Media Effects – Page 9 behavior. He found that, with few exceptions, the effects found for prosocial media messages
were larger than the effect found for antisocial messages. Finally, Hogben (1998) looked at the
results of 56 analyses from 30 studies and concluded that viewing televised violence was
associated with a small increase in viewer aggression. In addition, there was a correlation
between the year a study was done and the effect size; the later the study, the greater the effect
size, suggesting that prolonged exposure has a greater effect on viewers. Last, justified violence
and violence that did not accurately portray the consequence of violence generated greater effect
Summary. Experiments and surveys have been the most popular research strategies used to
study the impact of media on antisocial and prosocial behavior. The more elaborate techniques
of field experiments and panel studies have been used infrequently. Laboratory experiments
have shown a stronger positive relationship between viewing media violence and aggression than
have the other techniques. Meta-analyses have offered general conclusions about the scope and
magnitude of these effects.
One of the earliest theoretical considerations in the debate on the impact of media violence was
the controversy of catharsis versus stimulation. The catharsis approach suggests that viewing
fantasy expressions of hostility reduces aggression because a person who watches filmed or
televised violence is purged of his or her aggressive urges. This theory has some obvious
attraction for industry executives because it implies that presenting violent television shows is a
prosocial action. The stimulation theory argues the opposite: Viewing violence prompts more
aggression on the part of the viewer. Research findings in this area have indicated little support
for the catharsis position. A few studies did find a lessening of aggressive behavior after
viewing violent content, but these results apparently were an artifact of the research design. The
overwhelming majority of studies found evidence of a stimulation effect.
Since these early studies, many experiments and surveys have used social learning as their
conceptual basis. As spelled out by Bandura (1977), the theory explains how people learn from
direct experience or from observation (or modeling). Some key elements in this theory are
attention, retention, motor reproduction, and motivations. According to Bandura, attention to an
event is influenced by characteristics of the event and by characteristics of the observer. For
example, repeated observation of an event by a person who has been paying close attention
should increase learning. Retention refers to how well an individual remembers behaviors that
have been observed. Motor reproduction is the actual behavioral enactment of the observed
event. For example, some people can accurately imitate a behavior after merely observing it, but
others need to experiment. The motivational component of the theory depends on the
reinforcement or punishment that accompanies performance of the observed behavior.
Applied to the effects area, social learning theory predicts that people can learn antisocial or
prosocial acts by watching films or television. The model further suggests that viewing repeated
antisocial acts makes people more likely to perform these acts in real life. Another suggestion is
that desensitization accounts for people who are heavily exposed to violence and antisocial acts
becoming less anxious about the consequences.
Bandura (1977) summarized much of the research on social learning theory. In brief, some
Wimmer & Dominick: Media Effects – Page 10 key findings in laboratory and field experiments suggest that children can easily perform new
acts of aggression after a single exposure to them on television or in films. The similarity
between the circumstances of the observed antisocial acts and the post-observation
circumstances is important in determining whether the act is performed. If a model is positively
reinforced for performing antisocial acts, the observed acts are performed more frequently in real
life. Likewise, when children are promised rewards for performing antisocial acts, they exhibit
more antisocial behavior. Other factors that facilitate the performance of antisocial acts include
the degree to which the media behavior is perceived to be real, the emotional arousal of the
subjects, and the presence of cues in the post-observation environment that elicit antisocial
behavior. Finally, as predicted by the theory, desensitization to violence can occur through
repeated exposure to violent acts.
Other research has continued to refine and reformulate some of the elements in social
learning theory. For example, the arousal hypothesis (Tannenbaum & Zillmann, 1975) suggests
that, for a portrayal to have a demonstrable effect, increased arousal may be necessary.
According to this model, if an angered person is exposed to an arousing stimulus, such as a
pornographic film, and is placed in a situation to which aggression is a possible response, the
person will become more aggressive. (Excitation transfer is the term used by the researchers.)
Zillmann, Hoyt, and Day (1979) offer some support for this model. It appears that subjects
in a high state of arousal after seeing a violent film will perform more prosocial acts than
nonaroused subjects. Like aggressive behavior, prosocial behavior seems to be facilitated by
media-induced arousal (Mueller, Donnerstein, & Hallam, 1983).
Other research has shown that social learning theory can be applied to the study of the effects
of viewing pornography. Zillmann and Bryant (1982) showed that heavy exposure to
pornographic films apparently desensitized subjects to the seriousness of rape and led to
decreased compassion for women as rape victims. A similar finding was obtained by Linz,
Donnerstein, and Penrod (1984). Men who viewed five movies depicting erotic situations
involving violence toward women perceived the films as less violent and less degrading to
women than did a control group not exposed to the films. In sum, social learning theory is a
promising framework for integrating many findings in this area.
Another promising theory, outlined by Berkowitz and Rogers (1986), is based on priming
effects analysis. Drawing upon the concepts of cognitive neo-associationism, priming effects
analysis posits that elements of thought, feeling, or memories are parts of a network connected
by associative pathways. When a thought element is activated, the activation spreads along the
pathways to other parts of the network. Thus, for some time after a concept is activated, there is
an increased probability that it and other associated parts of the network will come to mind
again, thus creating the priming effect. As a result, aggressive ideas prompted by viewing media
violence trigger other semantically related thoughts, thereby increasing the probability that
associated aggressive thoughts will come to mind. Berkowitz and Rogers note that priming
analysis can explain why much exposure to media violence results in short-term, transient
effects. They point out that the priming effect attenuates over time to lower the probability of
subsequent violent effects.
Van Evra (1990) suggests that “script theory” might also be useful in explaining the impact
Wimmer & Dominick: Media Effects – Page 11 of viewing TV violence. Since most viewers, particularly younger ones, have little real-life
experience with violence but see a lot of it on TV, their behavior patterns or scripts might be
influenced by the TV exposure. Those who watch a large amount of violent TV might store
these scripts in their memory and display violence when an appropriate stimulus triggers the
acting out of their scripts. Moreover, Huesmann and Eron (1986) argue that if a young child
learns early in his or her developmental cycle that aggression is a potent problem-solving
technique; that behavior will be hard to change because the script has been well rehearsed by the
Drawing upon the this information, Comstock and Paik (1991) proposed a three-factor
explanation of the influence of media violence on antisocial and aggressive behavior:
1. Violent portrayals that are unique, compelling, and unusual are likely to prompt viewer
aggression because of their high attention and arousal.
2. Social cognition theory suggests that repetitive and redundant portrayals of violence prompt
viewers to develop expectations and perceptions of violence.
3. Violent media content encourages the early acquisition of stable and enduring traits.
Children who are only 3 or 4 years old may learn some violent scripts.
Sander (1997) proposed a new theoretical approach, the dynamic transaction model, to
explain how viewers perceive violence. The model posits that a person’s reaction to media
violence is a function of the precise form of the media stimulus and the interpretive ability of the
receiver. A quasi-experimental study of viewers revealed that audience members and
researchers perceive violence differently and that specific content variables (physical vs.
psychological violence, serious vs. comic violence, real vs. fantasy violence, and so on) have the
greatest influence on perceptions, followed by the emotional state of the receiver while watching
violence. Krcmar’s (1998) study suggested that family communication patterns are also
important in determining how children perceive violence. These last two studies support the idea
that perceptions of violence may be a key concept in formulating theories about the impact of
this kind of material.
Comstock (2007) argued for a sociological approach to theory. He maintained that the
research on TV violence should move beyond focusing on the individual and examine how
violence has an impact on various social groups. Using the results of meta-analyses, Comstock
identified five social groupings that were related to vulnerability for negative influence: those
with a predisposition to aggression, indifferent parenting, unsatisfactory social relationships, low
psychological well-being, and those who exhibited disruptive behaviors.
Uses and Gratifications
The uses and gratifications perspective takes the view of the media consumer. It examines how
people use the media and the gratifications they seek and receive from their media behaviors.
Uses and gratifications researchers assume that audience members are aware of and can
articulate their reasons for consuming various media content.
Wimmer & Dominick: Media Effects – Page 12 History
The uses and gratifications approach has its roots in the 1940s, when researchers became
interested in why people engaged in various forms of media behavior, such as radio listening or
newspaper reading. These early studies were primarily descriptive, seeking to classify the
responses of audience members into meaningful categories. For example, Herzog (1944)
identified three types of gratification associated with listening to radio soap operas: emotional
release, wishful thinking, and obtaining advice. Berelson (1949) took advantage of a New York
newspaper strike to ask people why they read the paper. The responses fell into five major
categories: reading for information, reading for social prestige, reading for escape, reading as a
tool for daily living, and reading for a social context. These early studies had little theoretical
coherence; in fact, many were inspired by the practical needs of newspaper publishers and radio
broadcasters to know the motivations of their audience in order to serve them more efficiently.
The next step in the development of this research began during the late 1950s and continued
into the 1960s. In this phase, the emphasis was on identifying and operationalizing the many
social and psychological variables that were presumed to be the antecedents of different patterns
of consumption and gratification. For example, Schramm, Lyle, and Parker (1961), in their
extensive study, found that children’s use of television was influenced by individual mental
ability and relationships with parents and peers, among other things. Gerson (1966) concluded
that race was important in predicting how adolescents used the media. These studies and many
more conducted during this period reflected a shift from the traditional effects model of mass
media research to the functional perspective.
According to Windahl (1981), a primary difference between the traditional effects approach
and the uses and gratifications approach is that a media effects researcher usually examines mass
communication from the perspective of the communicator, whereas the uses and gratifications
researcher uses the audience member as a point of departure. Windahl argues for a synthesis of
the two approaches, believing that it is more beneficial to emphasize their similarities than to
stress their differences. He has coined the term conseffects of media content and use to
categorize observations that are partly results of content use in itself (a viewpoint commonly
adopted by effects researchers) and partly results of content mediated by use (a viewpoint
adopted by many uses and gratifications researchers).
Windahl’s perspective links the earlier uses and gratifications approach to the third phase in
its development. Recently, uses and gratifications research has become more conceptual and
theoretical as investigators have offered data to explain the connections between audience
motives, media gratifications, and outcomes. As Rubin (1985, p. 210) notes: “Several typologies
of mass media motives and functions have been formulated to conceptualize the seeking of
gratifications as variables that intervene before media effects.” For example, Rubin (1979)
found a significant positive correlation between the viewing of television to learn something and
the perceived reality of television content: Those who used television as a learning device
thought television content was more true to life. DeBock (1980) notes that people who
experienced the most frustration at being deprived of a newspaper during a strike were those who
used the newspaper for information and those who viewed newspaper reading as a ritual. These
and many other recent studies have revealed that a variety of audience gratifications are related
to a wide range of media effects. These “uses and effects” studies (Rubin, 1985) have bridged
Wimmer & Dominick: Media Effects – Page 13 the gap between the traditional effects approach and the uses and gratifications perspective.
In the last several years, the uses and gratifications approach has been used to explore the
impact of new technologies on the audience. For example, Lin (1993) posited that audience
activity (planning viewing, discussing content, remembering the program) would be an important
intervening variable in the gratification-seeking process because of the viewing options opened
up by cable, VCRs, and remote controls. Her results supported her hypothesis. Viewers who
were most active had a greater expectation of gratification and also reported obtaining greater
Albarran and Dimmick (1993) combined the uses and gratifications approach with niche
theory in their study of the utility of the video entertainment industries. They found that
broadcast TV was the most diverse in serving the cognitive gratifications of the audience,
whereas cable TV and the VCR were the most effective in meeting needs related to feeling and
The advent of the Internet has spurred a renaissance in uses and gratifications research as
investigators describe Internet motivations and compare and contrast their results with the uses
and gratifications from traditional media. To illustrate, Valkenburg and Soeters (2001) found
that Internet use among their sample of 8- to 13-year-olds was most related to an enjoyment of
using computers and finding information. Ferguson and Perse (2000) examined the World Wide
Web as a functional alternative to TV and discovered that many of the motivations for using the
web were similar to those for viewing television. Finally, Papacharissi and Rubin (2000) came
up with a set of five motivations for using the Internet: utility, passing time, seeking information,
convenience, and entertainment.
The uses and gratifications approach continued to be popular throughout the first decade of
the new century as investigators applied the technique to study emerging media. For example,
researchers used the approach to study:
• Motives for viewing YouTube (Haridakis and Hanson, 2009).
• Gratifications from user-generated media (Guosong, 2009).
• Uses and gratifications of social media (Raacke & Bonds-Raacke, 2008)
• Gratifications associated with e-mail, cell phones and instant messages (Ramirez,
Dimmick, Feaster, & Lin, 2008).
Uses and gratifications researchers have relied heavily on the survey method to collect their
data. As a first step, researchers have conducted focus groups or have asked respondents to write
essays about their reasons for media consumption. Closed-ended Likert-type scales are then
constructed based on what was said in the focus group or written in the essays. The closed-ended
measures are typically subjected to multivariate statistical techniques such as factor analysis,
which identifies various dimensions of gratifications.
For example, in their study of the uses and gratifications of VCRs, Rubin and Bantz (1989)
first asked selected groups of respondents to list 10 ways in which they used their VCRs and to
Wimmer & Dominick: Media Effects – Page 14 provide reasons for those uses. This procedure resulted in a list of categories and statements
describing VCR usage. A questionnaire was then developed from this master list and
administered to respondents, who were asked to indicate how frequently they used their VCRs
for these purposes and to rate how much importance they placed on the statements detailing the
reasons for usage. After revisions, a final questionnaire was developed; it contained 95
motivational statements. This questionnaire was administered to a sample of 424 VCR owners.
Through factor analysis, the 95 statements were then reduced to eight main motivational
categories. These are some examples of the factors and statements that went with them: “I want
to keep a permanent copy of the program” (library storage); “I use music video for parties”
(music videos); “I don’t have to join an exercise class” (exercise tapes). Rubin and Bantz then
correlated these factors with demographic and media exposure variables. Note that this technique
assumes that the audience is aware of its reasons and can report them when asked. The method
also assumes that the pencil-and-paper test is a valid and reliable measurement scale. Other
assumptions include an active audience with goal-directed media behavior; expectations for
media use that are produced from individual predispositions, social interaction, and
environmental factors; and media selection initiated by the individual.
The experimental method has not been used widely in uses and gratifications research.
When it has been chosen, investigators typically manipulated the subjects’ motivations and
measured differences in their media consumption. To illustrate, Bryant and Zillmann (1984)
placed their subjects in either a state of boredom or a state of stress and then gave them a choice
of watching a relaxing or a stimulating television program. Stressed subjects watched more
tranquil programs, and bored subjects opted for the exciting fare. McLeod and Becker (1981) had
their subjects sit in a lounge that contained public affairs magazines. One group of subjects was
told that they would soon be tested about the current situation in Pakistan; a second group was
told they would be required to write an essay on U.S. military aid to Pakistan; while a control
group was given no specific instructions. As expected, subjects in the test and essay conditions
made greater use of the magazines than did the control group. The two test groups also differed
in the type of information they remembered from the periodicals. Experiments such as these two
indicate that different cognitive or affective states facilitate the use of media for various reasons,
as predicted by the uses and gratifications rationale.
AN INSIDE LOOK
Media Effects Research: Whether the Weather Makes a Difference
Uses and gratifications research has shed a good deal of light on viewer motivations for watching
TV, but the approach has not been particularly successful in predicting the actual amount of
television use. Roe and Vandebosch (1996) suggest that one reason for the inability to predict is
that researchers sometimes overlook the obvious—such as the weather.
Seasonal variations in TV viewing are well documented: People watch more in the winter
and less in the summer. Roe and Vandebosch, however, suggest that specific weather effects
occur with each season. The researchers gathered detailed meteorological data in Belgium for a
year, including temperature, precipitation amount, wind speed, cloud cover, barometric pressure,
and hours of sunlight. They also collected television-viewing statistics encompassing the
Wimmer & Dominick: Media Effects – Page 15 percentage viewing and the daily average amount of time spent watching.
Their results showed strong correlations between all their weather-related measures, except
for barometric pressure, and viewing with some correlations reaching as high as .75. In addition,
there was consistency within each individual season. People watched more TV when there were
fewer hours of daylight, when the temperature was low, when wind speed was high, and when
there was some precipitation.
The implication in this finding for broadcasters was clear. The single most important
determiner of TV audience size was wholly beyond their control.
As mentioned earlier, researchers in the academic sector are interested in developing theory
concerning the topics they investigate. This tendency is well illustrated in the history of uses and
gratifications research. Whereas early studies tended to be descriptive, later scholars have
attempted to integrate research findings into a more theoretical context.
In an early explanation of the uses and gratifications process, Rosengren (1974) suggested
that certain basic needs interact with personal characteristics and the social environment of the
individual to produce perceived problems and perceived solutions. The problems and solutions
constitute different motives for gratification behavior that can come from using the media or
from other activities. Together the media use or other behaviors produce gratification (or
nongratification) that has an impact on the individual or society, thereby starting the process
anew. After reviewing the results of approximately 100 uses and gratifications studies,
Palmgreen (1984) stated that “a rather complex theoretical structure . . . has begun to emerge.”
He proposed an integrative gratifications model that suggested a multivariate approach.
The gratifications sought by the audience form the central concept in the model. There are,
however, many antecedent variables such as media structure, media technology, social
circumstances, psychological variables, needs, values, and beliefs that all relate to the particular
gratification pattern used by the audience. Additionally, the consequences of the gratifications
relate directly to media and nonmedia consumption behaviors and the perceived gratifications
that are obtained. As Palmgreen admits, this model suffers from lack of parsimony and needs
strengthening in several areas, but it does represent an increase in our understanding of the mass
media process. Further refinements in the model will come from surveys and experiments
designed to test specific hypotheses derived from well-articulated theoretical rationales and from
carefully designed descriptive studies. For example, Levy and Windahl (1984) examined the
assumption of an active audience in the uses and gratifications approach. They derived a
typology of audience activity and prepared a model that linked activity to various uses and
gratifications, thus further clarifying one important postulate in the uses and gratifications
Swanson (1987) called for more research to encourage the theoretical grounding of the uses
and gratifications approach. Specifically, Swanson urged that research focus on (1) the role of
gratification seeking in exposure to mass media, (2) the relationship between gratification and
the interpretive frames through which audiences understand media content, and (3) the link
between gratifications and media content. Van Evra (1990) presents an integrated theoretical
Wimmer & Dominick: Media Effects – Page 16 model of television’s impact in which the use of the medium is considered along with the amount
of viewing, presence of information alternatives, and perceived reality of the medium. Her
description highlights the complex interactions that need to be examined in order to understand
the viewing process. Additionally, uses and gratifications researchers have incorporated a theory
from social psychology, expectancy-value theory, into their formulations (Babrow, 1989). This
theory suggests that audience attitude toward media behavior is an important factor in media use.
Rubin (1994) summarized the growth of theory in the area and concludes that single-variable
explanations of media effects are inadequate. He suggests that more attention be given to
antecedent, mediating, and consequent exposure conditions. Finn (1997) investigated a five-
factor personality model as a correlate of mass media use. He found that people who scored high
on the extroversion and agreeableness dimensions of a personality measure were more likely to
choose nonmedia activities (such as conversation) to meet their communication needs. In a
comprehensive review of the theoretical developments relevant to uses and gratifications theory,
Ruggiero (2000) argues that researchers must expand the uses and gratifications model to
accommodate the unique features of the Internet such as interactivity and demassification. He
also contends that the growing popularity of the Internet will make the uses and gratifications
approach even more valuable in the future.
The uses and gratifications approach also illustrates the difference in emphasis between
academic and applied research objectives. Newspaper publishers and broadcasting executives,
who want guidance in attracting readers, viewers, and listeners, seem to be particularly interested
in determining what specific content is best suited to meeting the needs of the audience. College
and university researchers are interested not only in understanding content characteristics but
also in developing theories that explain and predict the public’s media consumption based on
sociological, psychological, and structural variables.
Agenda Setting by the Media
Agenda setting theory proposes that “the public agenda—or what kinds of things people discuss,
think, and worry about (and sometimes ultimately press for legislation about)—is powerfully
shaped and directed by what the news media choose to publicize” (Larson, 1994). This means
that if the news media decide to give the most time and space to covering the budget deficit, this
issue will become the most important item on the audience’s agenda. If the news media devote
the second most coverage to unemployment, audiences will also rate unemployment as the
second most important issue to them, and so on. Agenda setting research examines the
relationship between media priorities and audience priorities in the relative importance of news
The notion of agenda setting by the media can be traced back to Walter Lippmann (1922), who
suggested that the media were responsible for the “pictures in our heads.” Forty years later,
Cohen (1963) further articulated the idea when he argued that the media may not always be
successful in telling people what to think, but they are usually successful in telling them what to
think about. Lang and Lang (1966, p. 468) reinforced this notion by observing, “The mass media
force attention to certain issues. . . . They are constantly presenting objects, suggesting what
individuals in the mass should think about, know about, have feelings about.”
Wimmer & Dominick: Media Effects – Page 17 The first empirical test of agenda setting came in 1972 when McCombs and Shaw (1972)
reported the results of a study done during the 1968 presidential election. They found strong
support for the agenda-setting hypothesis. There were strong relationships between the emphasis
placed on different campaign issues by the media and the judgments of voters regarding the
importance of various campaign topics. This study inspired a host of others, many of them
concerned with agenda setting as it occurred during political campaigns. For example, Tipton,
Haney, and Baseheart (1975) used cross-lagged correlation to analyze the impact of the media on
agenda setting during statewide elections. Patterson and McClure (1976) studied the impact of
television news and television commercials on agenda setting in the 1972 election. They
concluded that television news had minimal impact on public awareness of issues but that
television campaign advertising accounted for increased audience awareness of candidates’
positions on issues.
Agenda setting continued to be a popular research topic through the 1980s and 1990s. Its
focus has expanded from looking at political campaigns to examining other topics. The agenda-
setting technique is now being used in a variety of areas: history, advertising, foreign news, and
medical news. McCombs (1994) and Wanta (1997) present useful summaries of this topic.
In recent years the most popular subjects in agenda-setting research are (1) how the media
agenda is set (this research is also called agenda building), and (2) how the media choose to
portray the issues they cover (called framing analysis). With regard to agenda building, Wanta,
Stephenson, Turk, and McCombs (1989) noted some correlation between issues raised in the
president’s State of the Union address and the media coverage of those issues. Similarly, Wanta
(1991) discovered that the president can have an impact on the media agenda, particularly when
presidential approval ratings are high. Boyle (2001) found that major party candidate political
ads can have an influence on media coverage of a campaign. Reese (1990) presents a review of
the agenda-building research.
Framing analysis recognizes that media can impart a certain perspective, or “spin,” to the
events they cover and that this, in turn, might influence public attitudes on an issue. Framing
analysis has been called the second level of agenda setting. As Ghanem (1997, p. 3) put it:
Agenda setting is now detailing a second level of effects that examines how
media coverage affects both what the public thinks about and how the public
thinks about it. This second level of agenda setting deals with the specific
attributes of a topic and how this agenda of attributes also influences public
For example, Iyengar and Simon (1993) found a framing effect in their study of news
coverage of the Gulf War. Respondents who relied the most on television news, where military
developments were emphasized, expressed greater support for a military rather than a diplomatic
solution to the crisis. In their study of the way media framed breast cancer coverage in the 1990s,
Andsager and Powers (1999) discovered that women’s magazines offered more personal stories
and more comprehensive information, while news magazines focused more on the economic
angle, stressing research funding and insurance. Finally, Andsager (2000) analyzed the attempts
by interest groups to frame the abortion debate of the late 1990s and the impact their efforts had
with news media. She found that the pro-life group was more successful in getting their
Wimmer & Dominick: Media Effects – Page 18 interpretation into press coverage.
Agenda setting continued to be an important topic to mass communication researchers well
into the new century. Tai (2009) found that 56 studies of agenda setting appeared in major
communication journals from 1996 to 2005. Not surprisingly, many were conducted in the
context of political campaigns using the methods established by earlier studies. For example,
Dunn (2009) looked at agenda setting in the 2005 Virginia gubernatorial election and found that
the agenda of the major candidates and the media agenda were related while Kiousis and Shields
(2008) examined the influence of public relations efforts in the 2004 presidential election.
In addition, the agenda setting influence of emerging media attracted the attention of several
researchers. Sweetser, Golan and Wanta (2008) found evidence that blog content influenced the
media agenda and Wallstein (2007) discovered a reciprocal relationship between mainstream
media coverage and blog discussions during a presidential election campaign.
Recent research using framing analysis has looked at a variety of topics. Yun, Nah and
McLeod (2008) investigated how news media framed the controversy over stem cell research.
D’Angelo and Lombard (2008) conducted an experiment that revealed that different frames
prompted subjects to rate certain topics more important than others. Finally, Lipshultz (2007)
examined how the media framed the “war on terror.”
The typical agenda-setting study involves several of the approaches discussed in earlier chapters.
Content analysis is used to define the media agenda, and surveys are used to collect data on the
audience agenda. In addition, since determining the media agenda and surveying the audience
are not done simultaneously, a longitudinal dimension is present. More recently, some studies
have used the experimental approach.
Measuring the Media Agenda. Several techniques have been used to establish the media
agenda. The most common method involves grouping coverage topics into broad categories and
measuring the amount of time or space devoted to each category. The operational definitions of
these categories are important because the more broadly a topic area is defined, the easier it is to
demonstrate an agenda-setting effect. Ideally, the content analysis should include all media:
television, radio, newspaper, and magazines. Unfortunately, this is too large a task for most
researchers to handle comfortably, and most studies have been confined to one or two media,
usually television and the daily newspaper. For example, Williams and Semlak (1978) tabulated
the total air time for each topic mentioned in the three television network newscasts over a 19-
day period. The topics were rank-ordered according to their total time. At the same time, the
newspaper agenda was constructed by measuring the total column inches devoted to each topic
on the front and editorial pages of the local newspaper. McLeod, Becker, and Byrnes (1974)
content-analyzed local newspapers for a 6-week period, totaling the number of inches devoted to
each topic, including headlines and pertinent pictures on the front and editorial pages. Among
other things, they found that the front and editorial pages adequately represented the entire
newspaper in their topical areas.
The development of new technologies has created problems for researchers when it comes to
measuring the media agenda. Cable TV, fax machines, email, blogs, online computer services,
Wimmer & Dominick: Media Effects – Page 19 and the Internet have greatly expanded the information outlets available to the public. The role
of these new channels of communication in agenda setting is still unclear.
Measuring Public Agendas. The public agenda has been measured in at least four ways.
First, respondents are asked an open-ended question such as “What do you feel is the most
important political issue to you personally?” or “What is the most important political issue in
your community?” The phrasing of this question can elicit either the respondent’s intrapersonal
agenda (as in the first example) or interpersonal agenda (the second example). A second method
asks respondents to rate in importance the issues in a list compiled by the researcher. The third
technique is a variation of this approach. Respondents are given a list of topics selected by the
researcher and asked to rank-order them according to perceived importance. The fourth
technique uses the paired-comparisons method. Each issue on a preselected list is paired with
every other issue, and the respondent is asked to consider each pair and to identify the more
important issue. When all the responses have been tabulated, the issues are ordered from the
most important to the least important.
As with all measurement, each technique has its own advantages and disadvantages. The
open-ended method gives respondents great freedom in nominating issues, but it favors those
people who are better able to verbalize their thoughts. The closed-ended ranking and rating
techniques make sure that all respondents have a common vocabulary, but they assume that each
respondent is aware of all the public issues listed and restrict the respondent from expressing a
personal point of view. The paired-comparisons method provides interval data, which allows for
more sophisticated statistical techniques, but it takes longer to complete than the other methods,
and this might be a problem in some forms of survey research.
Three important periods used in collecting the data for agenda-setting research are (1) the
duration of the media agenda measurement period, (2) the time lag between measuring the media
agenda and measuring the personal agenda, and (3) the duration of the audience agenda
measurement. Unfortunately, there is little in the way of research or theory to guide the
investigator in this area. To illustrate, Mullins (1977) studied media content for a week to
determine the media agenda, but Gormley (1975) gathered media data for 4.5 months. Similarly,
the time lag between media agenda measurement and audience agenda measurement has varied
from no time at all (McLeod et al., 1974) to a lag of 5 months (Gormley, 1975). Wanta and Hu
(1994a) discovered that different media have different optimum time lags. Television, for
example, has a more immediate impact, whereas newspapers are more effective in the long term.
It is not surprising that the duration of the measurement period for audience agendas has also
varied widely. Hilker (1976) collected a public agenda measure in a single day, whereas
McLeod and colleagues (1974) took 4 weeks. Eyal, Winter, and DeGeorge (1981) suggested
that methodological studies should be carried out to determine the optimal effect span or peak
association period between the media emphasis and public emphasis. Winter and Eyal (1981), in
an example of one of these methodological studies, found an optimal effect span of 6 weeks for
agenda setting on the civil rights issue. Similarly, Salwen (1988) found that it took from 5 to 7
weeks of news media coverage of environmental issues before they became salient on the
In a large-scale agenda-setting study of German television, Brosius and Kepplinger (1990)
Wimmer & Dominick: Media Effects – Page 20 found that the nature of the issue had an impact on the time lag necessary to demonstrate an
effect. For general issues such as environmental protection, a lag of a year or two might be
appropriate. For issues raised in political campaigns, 4 to 6 weeks might be the appropriate lag.
For a breaking event within an issue, such as the Chernobyl disaster, a lag of a week might be
Agenda-setting researchers are now incorporating more complicated longitudinal analysis
measures into their designs. Gonzenbach and McGavin (1997) for example, present descriptions
of time series analysis and time series modeling and a discussion of nonlinear analysis
Several researchers have used the experimental technique to study the causal direction in
agenda setting. For example, Heeter, Brown, Soffin, Stanley, & Salwen (1989) examined the
agenda-setting effect of teletext. One group of subjects was instructed to abstain from all
traditional news media for five consecutive days and instead spend 30 minutes each day with a
teletext news service. The results indicated that a week’s worth of exposure did l