international communication gazette 2010 lee 759 76

http://gaz.sagepub.com/Gazette

International Communication

http://gaz.sagepub.com/content/72/8/759The online version of this article can be found at:

DOI: 10.1177/1748048510380814

2010 72: 759International Communication GazetteGunho Lee

effects on primingWho let priming out? Analysis of first- and second-level agenda setting

Published by:

http://www.sagepublications.com

can be found at:International Communication GazetteAdditional services and information for

http://gaz.sagepub.com/cgi/alertsEmail Alerts:

http://gaz.sagepub.com/subscriptionsSubscriptions:

http://www.sagepub.com/journalsReprints.navReprints:

http://www.sagepub.com/journalsPermissions.navPermissions:

http://gaz.sagepub.com/content/72/8/759.refs.htmlCitations:

What is This?

- Dec 9, 2010Version of Record >>

at University of Bucharest on December 4, 2012gaz.sagepub.comDownloaded from

http://gaz.sagepub.com/

http://gaz.sagepub.com/content/72/8/759

http://www.sagepublications.com

http://gaz.sagepub.com/cgi/alerts

http://gaz.sagepub.com/subscriptions

http://www.sagepub.com/journalsReprints.nav

http://www.sagepub.com/journalsPermissions.nav

http://gaz.sagepub.com/content/72/8/759.refs.html

http://gaz.sagepub.com/content/72/8/759.full.pdf

http://online.sagepub.com/site/sphelp/vorhelp.xhtml


Article

Who let priming out?Analysis of first- andsecond-level agendasetting effects onpriming

Gunho LeeEwha Womans University, Korea

AbstractThis study examines the first- and second-level agenda setting process and primingeffects in the same experimental setting. Specifically, it explores the association betweenthe two levels of agenda setting effects and priming effects. The main purpose of thestudy is to determine which of the two agenda setting effects better explains primingeffects by using the global warming issue as the target object. By exposing three groupsof subjects, respectively, to three kinds of pseudo online newspaper (High, Medium andNo Exposure) stimuli, the experiment found that even if there were two levels of agendasetting and priming effects detected, second-level agenda setting effects were moreclosely connected to priming effects. With regard to the relationship between agendasetting and priming effects, it is recommended that various issues beyond the environ-ment issue, which this study took, and their attributes should be tested in order to fur-ther understand media effects.

Keywordsagenda setting, attribute, first level, framing, global warming, issue, media effects, priming,second level

Corresponding author:

Gunho Lee, #612 Ewha-Posco B/D, Division of Media Studies, College of Social Sciences, Ewha Womans

University, 11-1 Daehyun-Dong, Seodaemun-Gu, Seoul, 120-750, Korea

Email: [email protected]

the InternationalCommunication Gazette

72(8) 759–776ª The Author(s) 2010

Reprints and permission:sagepub.co.uk/journalsPermissions.nav

DOI: 10.1177/1748048510380814gaz.sagepub.com

759



Introduction

Since the seminal agenda setting effects research of McCombs and Shaw (1972), the

agenda setting theory has evolved through methodological challenges and theoretical

convergence (Takeshita, 2005). Among such challenges and convergence, the introduc-

tion of priming effects into the agenda setting realm was believed to be a prominent

opportunity to broaden its theoretical boundary (McCombs, 2004). McCombs argued

that priming effects shed light on the hidden values of the agenda setting theory, which

originally focused on the cognitive aspects of the media’s influence on the public

agenda. With priming effects, he insisted that the agenda setting theory can explain how

people form their opinions about certain issues.

Initially, priming effects were considered to be linked to first-level agenda setting

effects, which dealt with the salience transfer of issues or objects from the media to the

public (Iyengar and Kinder, 1987). Iyengar and Kinder’s work, which was believed to

illuminate the relationship between agenda setting and priming, highlighted the role

of selected information in influencing people’s judgement standards. While the agenda

setting theory evolved, another element caught scholars’ attention — salience transfer of

attributes. Attributes refer to the particular characteristics of issues or objects (Ghanem,

1997; Kiousis et al., 1999; McCombs et al., 2000). The salience transfer of attributes

from the media to the public was conceptualized as second-level agenda setting

effects. While it has been studied for its theoretical and conceptual validation (Price and

Tewksbury, 1997; Scheufele, 2000), its relationship with priming effects has also been

explored (Kim et al., 2002). The role of priming in association with the two levels of

agenda setting effects was synthesized as the function of shaping the strength and the

direction of public opinion (McCombs, 2004).

McCombs’ typology for agenda setting and priming associates the first level with

opinion strength, and the second level with opinion direction. However, the explanation

of the connection seems to be mixed in with some complicated ideas. According to his

typology, although opinion formation is a synthetic work of strength and direction, these

seem to work independently. When the typology explains the media’s role in shaping

opinion strength, it seems to disregard the direction of opinion and vice versa. It may

be right ‘abstractly’, but it is dubious if we can distinguish strength from direction with

regard to priming theory. In the discipline of psychology, the strength and direction of

opinion are not usually distinguished (Goethals and Nelson, 1973; Hilmert et al.,

2006; Jones and Gerard, 1967; Suls et al., 2000). Opinion is supposed to have direction,

and the strength of an opinion can be measured by how far the opinion goes in a certain

direction. In other words, the strength of an opinion is innately involved with its direc-

tion, or simply put, its strength cannot be thought of without considering its direction

(Johnson, 2003; Zimmerman and Kitsantas, 2002).

As explained in McCombs’ typology, opinion formation can consist of strength and

direction. Strength and direction, however, may not be independent of each other as

McCombs argued, but are instead interdependent with regard to priming effects. The fol-

lowing question then arises: if strength and direction work interdependently, which of

the first- and second-level agenda setting effects has a closer relationship with priming

effects indicating the media’s influence on people’s opinion? To answer this question,

760 the International Communication Gazette 72(8)

760



this article explores the links between first-level agenda and priming effects and between

second-level agenda and priming effects together in an experimental setting in order to

determine which of the two levels induce priming effects. For the experimental approach

that this article undertook, pseudo online newspapers were used. The author hopes that

the simultaneous examination of the three kinds of media effects in an experiment, which

is a rare case, will help understand the media function.

Theoretical background

Agenda setting effects

In the years after the Chapel Hill study (McCombs and Shaw, 1972), the core agenda

setting literature focused on the fundamental maxim, ‘Elements prominent in the mass

media’s picture of the world influence the prominence of those elements in the audi-

ence’s picture’ (McCombs et al., 2000: 77). This was simply called ‘salience transfer’.

Throughout the development of the theory, the elements have been mainly summarized

into two categories – the first level (focusing on issues or objects) and the second level

(focusing on issue attributes).

First-level agenda setting proposes that if the media give importance to certain

issues in their reporting, then the audience will perceive these issues as important too.

First-level agenda setting effects are originally focused on political issues. In relation

to this, McCombs and Shaw’s (1972) seminal study examined the correlation between

the 1968 presidential election news coverage and the voters’ thinking, and several sub-

sequent studies followed a similar track (Shaw and McCombs, 1977; Weaver et al.,

1981). Even adopting more complicated concepts and methods, many other agenda set-

ting studies dealt with political events (Ansolabehre et al., 1993; Harrison et al., 1991;

King, 1997). The theory, however, made a huge leap in explaining media effects when

the research topic was diversified into various issues. Examining non-political issues like

living costs, unemployment and so on, Zucker (1978) argued that agenda setting effects

could not take place everywhere, but that the effects would depend on the degree of the

issue’s obtrusiveness. Yagade and Dozier (1990) insisted that the abstractness influenced

the size of agenda setting effects. A piece of research dealing with the civil rights issue

boosted the studies on the time lag of agenda setting effects (Winter and Eyal, 1981),

while a crime issue study pointed out the difference between reality and media presenta-

tion (Ghanem, 1996).

With the first-level study being the initial point of agenda setting research, the second-

level study on attributes had been there as the seed for its theoretical development (Shaw

and McCombs, 1977). Second-level agenda setting came to light when scholars began to

ask how the various attributes of an issue can be transferred from the media to the public

(Golan and Wanta, 2001; Kim et al., 2002; Kiousis et al., 1999; McCombs et al., 2000).

In agenda setting theory, attributes are defined as ‘characteristics and traits that fill out

the picture of each object’ (McCombs et al., 2000: 78). Regarding media messages, attri-

butes are also referred as ‘the set of perspectives or frames that journalists and the public

employ to think about each object’ (Ghanem, 1997: 5). In short, attributes are certain fea-

tures of objects or issues. Second-level agenda setting suggests that certain attributes

Lee 761

761



depicted in the media message are accentuated over other elements, and in turn, the

attributes depicted in the media influence the public’s perception of issues.

Taking attributes into its realm of theory, agenda setting was able to explain media

effects as not only telling people what to think but also telling them how to think. In par-

ticular, the introduction of the substantive and affective dimensions of attributes to

second-level agenda setting studies helped theories explain in detail how people perceive

issues (McCombs et al., 2000). Although there are other arguments urging the further

development of the dimensions of attributes (Ghanem, 1997), these two main dimen-

sions were discussed as the fundamental aspects of attributes (Kiousis et al., 1999). The

substantive dimension is defined as ‘the characteristics of news that help us cognitively

structure news and discern among various topics’ (Kiousis et al., 1999: 417). Meanwhile,

affective attributes are the factors drawing emotional responses from the audience

(Kiousis et al., 1999; McCombs et al., 2000). These two dimensions are believed to

involve people’s opinion direction (McCombs, 2004). Additionally, the idea to tell how

to think about widened the range of theoretical explanation from agenda setting into the

territory of priming, which proposes the media’s influence on the audience’s evaluative

dimension in connection with the world outside (Kim et al., 2002).

Priming in conjunction with agenda setting effects

Priming refers to ‘the process in which the media attend to some issues and not others

and thereby alter the standards by which people evaluate’ objects in the real world

(Severin and Tankard, 2001: 226). This concept shares some aspects with those of

agenda setting theory. Priming focuses on some issues, not others, selected by the media,

as does agenda setting theory. However, priming goes one step further. Not staying at the

level of salience transfer, it scrutinizes the media’s role in influencing the audience’s

evaluative dimension (Kim et al., 2002). Iyengar et al. (1982) found that the media set

the criteria by which people could assess presidential candidates’ ability, and Iyengar

and Kinder (1987) formalized this process as priming effects, which set the guidelines

that people use for shaping their opinions. Linked together with the human information

processing structure, while the agenda setting effects explain salience transfer, the

priming function seems to elucidate the media’s role as opinion organizers.

Iyengar and Kinder (1987) specifically supported the linkage between issue agenda

setting effects and the priming function. They pointed out that salience transfer from the

media to the public (agenda setting function) became the creative basis of the evaluative

dimension (priming function). However, from the initial stages of priming studies and as

such seeking the opinion formation function from the issues presented in the media, it

has been continuously questioned as to how people form their opinions on such issues

(Kim et al., 2002). This kind of question seems to ask more about the fundamental

reasons for the formation of opinions. In the last 15 years or so, a few studies on

second-level agenda setting and priming effects opened the way to answer such ques-

tions (Iyengar and Simon, 1993; Kim et al., 2002). Iyengar and Simon’s (1993) study

revealed that TV audiences who were heavily exposed to news with a military approach

as a solution to the Gulf crisis favoured a military solution over diplomatic options. Kim

et al.’s (2002) study on the city of Ithaca, New York, found a significant role played by


762



the attributes of an issue in setting audiences’ opinions about an issue. With these

scholars’ efforts based on considerable first- and second-level agenda setting studies,

McCombs (2004) referred the priming function to a consequence of agenda setting

effects and put the priming function under the wide umbrella of agenda setting theory.

He argued that opinion formation is not only a matter of building opinion strength, which

can be associated with first-level agenda setting effects, but also a matter of building

opinion direction, which can be associated with second-level agenda setting effects.

McCombs’ typology, however, can be refined by stressing second-level agenda

setting effects further than the first level in influencing priming effects. In psychology,

opinions are believed to involve beliefs and values (Goethals and Nelson, 1973; Jones

and Gerard, 1967). Beliefs are related to the degree of truthfulness or credibility of facts,

and values are related to the degree of preference (Hilmert et al., 2006). Questions

regarding beliefs are generally about whether some facts are true and correct or not, and

questions regarding values are about the degree of liking something (Suls et al., 2000).

These characteristics of opinion seem to be related more to second-level agenda setting

effects, which deal with the substantive and affective perceptions influenced by media

presentation. Truthfulness is related to the substantive dimension of attributes in

second-level agenda setting effects (Lee and Yoo, 2004). Furthermore, the degree of

perceiving truthfulness somewhat decides the degree of preference. The better the cred-

ibility is, the greater the positive preference in opinion is (Johnson, 2003; Zimmerman

and Kitsantas, 2002). The degree of preference is related to the attributes’ affective

dimension of second-level agenda setting effects, which indicates emotional tone (i.e.

positive, neutral and negative). Then as truthfulness is connected with preference, affec-

tive attributes hinge on substantive attributes (Lee and Yoo, 2004). Therefore, the ques-

tions regarding opinion studies are closely related to attribute salience transfer. In short,

priming effects are more likely related to second-level agenda setting effects, which can

answer why opinion formation occurs, than first-level agenda setting effects, which

mainly show the facts of importance transfer.

To empirically test the argument mentioned so far, the current study first investigates

first- and second-level agenda setting and priming effects together. Then the study scru-

tinizes whether first-level and/or second-level agenda setting effects have a relationship

with priming effects. The global warming issue was chosen as the object for the study.

Issue and attributes

The environment is described in some earlier studies as an issue that the media can make

salient in the public’s thinking through their news reporting relatively easier than other

issues (Soroka, 2002a, 2002b; Xu and Bengston, 1997; Yagage and Dozier, 1990;

Zucker, 1978). Global warming has been one of the highly debated environmental issues

in recent years. Specifically, the issue gradually became prominent with several interna-

tional events, including the 1992 Earth Summit and the 1997 Kyoto Protocol. In partic-

ular, the Kyoto Protocol linked economic aspects clearly to this issue when it tried to

regulate individual countries’ industrial activities, which are alleged to cause emissions

and thus creating great quantities of greenhouse gases. In addition to the attributes of

economics, those of natural disasters, energy policy and regional conflict have been

Lee 763

763



largely reported in news articles in relation to global warming effects. The change in

ecosystems is another factor that is believed to be an effect of global warming. In view

of this, this study employs the global warming issue as a target object and its attributes as

target attributes in order to explore first- and second-level agenda setting effects

together. The major purpose of this study is to investigate the association between the

two kinds of agenda setting effects and priming effects, so determining such effects

is a necessary condition. The environment issue, which was believed to be easily trans-

ferable from the media to the public, would be a feasible selection to meet the study’s

final goal. In short, with an examination of the agenda setting effects of the issue and

attributes, the study measures whether such effects create an impact on shaping the

evaluative dimension relative to the global warming issue. Specifically, as noted earlier,

it is expected that first-level agenda setting effects do not have a clear association with

priming effects, but second-level agenda setting effects do.

For this study, a professional journalist who has been covering environmental issues

for more than two years was employed to select the attributes of the global warming

issue. Monitoring for two months an online news search engine, Google news, he chose

five attributes that were frequently reported in connection with the global warming issue

on the website. These are the ‘Crisis of Species (Ecosystem)’, ‘Increased Potential for

Flooding (Flood)’, ‘Impact on Industrial Business (Business)’, ‘Effects on the Energy

System (Energy)’ and ‘International and/or Provincial Conflict (Conflict)’. These attri-

butes were described in news presentations as traits reflecting the nature of global warm-

ing effects and/or the potential aftermaths of global warming effects. Specifically, the

ecosystem, flood and conflict were the negative consequences of global warming effects,

while business and energy were the neutral outcomes of the effects. That is, the attributes

of global warming effects in the experiment were chosen as somewhat skewed to nega-

tive aspects in order to explore second-level agenda setting and priming effects by

manipulating the tones of the attributes as such.

Hypotheses

This study’s explorative approach was made in line with the thinking that priming effects

are an extended function of agenda setting effects (McCombs, 2004). The general

assumption of agenda setting theory is that the more the exposure, the greater the effects.

That is, if the media present an issue and/or issue attributes more frequently than other

issues and/or issue attributes, the audience will perceive it (or them) as more important.

The strength of importance in audiences’ minds made by such exposure is conceptua-

lized as perceived importance or perceived salience (McCombs, 2004). In the current

study, following the basic assumption, the degree of exposure is the main stimulus for

examining first- and second-level agenda setting effects, and in turn, the priming effects

and the association between the two. For the study, the following hypotheses were

suggested:

H1: The global warming news presented by the media influences the audience’s

perceived salience of the global warming issue.


perceived salience of global warming issue attributes.


764




evaluative dimension relevant to the global warming issue.

H4: The issue on the salience transfer of global warming news presented by the

media is not associated with the audience’s evaluative dimension relevant to the global

warming issue.

H5: The attribute on the salience transfer of global warming news presented by the

media is associated with the audience’s evaluative dimension relevant to the global

warming issue.

Method

An experiment with several stages was conducted to test the hypotheses. For the

three-week experiment period, the subjects were divided into three different groups,

namely, High Exposure, Medium Exposure and No Exposure to the global warming

issue and its attributes. This study’s experimental design featured a controlled indepen-

dent variable (High Exposure vs Medium Exposure vs No Exposure condition) with pre-

and post-experimental measures for detecting issue salience transfer, attribute salience

transfer and priming effects.

Stimulus materials

This study employed a simple form of online newspaper. These online papers were

delivered to the subjects, who were divided into three groups (High vs Medium vs No

Exposure), via email every week during the three-week experiment period. Each subject

had three online newspapers, and in each online paper, 10 news links were attached

through which the subjects in the study were guided in reading the actual news stories.

A critical feature of the experiment was the selection of news articles reflecting the global

warming issue and its five different attributes. The same professional journalist chose the

stories. He retrieved 60 news stories from an online news search engine, Google news.

Half of the stories were about the global warming issue and its attributes, while the other

half were stories about random issues other than the global warming issue. They were

selected from news stories in Google news a week prior to each experimental session.

Based on the selected news articles, three different kinds of online newspaper (High

vs Medium vs No Exposure) were prepared, and the groups, who received these papers

respectively, were labelled accordingly. All online newspapers had the same format with

10 hyperlinks, but had different mixtures of stories. The High Exposure newspaper con-

tained only stories about the global warming issue and its five attributes. Half of the stor-

ies in the Medium Exposure newspaper were about the global warming issue and its five

attributes, and the other half were about random issues other than the global warming

issue. The No Exposure newspaper had only random stories other than global warming

stories. Each of the three kinds of online paper (High, Medium and No Exposure) was

delivered once a week for the three-week period to the subjects in the respective groups.

Thus, an individual subject received three online newspapers of the same kind, but the

contents of the first, second and third papers were different from one another. The

subjects received the paper every Wednesday for three weeks.

Lee 765

765



Additionally, to explore the hypotheses, the number of stories related to the individual

attributes was manipulated. For instance, over the three exposure times, the subjects in

the High Exposure group had a total of 10 stories about the Ecosystem attribute, 8 stories

about the Flood attribute, 6 stories about the Business attribute, 4 stories about the

Energy attribute and 2 stories about the Conflict attribute. This was executed in such

a way that the author could rank-order the attributes in terms of exposure frequency

to the subjects. Thus, over the three-week period, the respondents received 30 stories,

all of which were about the global warming issue and its attributes, but the number of

stories for each attribute was different from the others. For the same time period, the sub-

jects in the No Exposure group had 30 random stories, which were not relevant to the

global warming issue. The subjects in the Medium Exposure group had half the number

of stories with regard to each attribute of the global warming issue, compared to the High

Exposure group. That is, they had 5 stories about the Ecosystem attribute, 4 stories about

the Flood attribute, 3 stories about the Business attribute, 2 stories about the Energy attri-

bute and 1 story about the Conflict attribute. All stories on the global warming issue and

its attributes for the Medium Exposure group were included in those for the High Expo-

sure group. Fifteen other stories, which were the other half of the stories for the Medium

Exposure group, came from the random stories of the No Exposure group. The stories

containing the target issue and attributes were dispersed so that those stories for the same

attribute were not clustered together. The No Exposure group was not intended to have

the target issue and its attributes, so the dispersion of the stories was not considered.

Experimental procedures

The pre-experimental questionnaire was administered a week before the first experiment

as a form of online survey, and the post-experimental questionnaire was completed a

week after the third and final exposure. For the pre-experimental survey, 166 undergrad-

uates from a large US southwestern university were recruited by email to participate in

the study in order to earn extra course credit for the fall semester of 2004. They com-

pleted the questionnaire attached to the email, which asked about their perceived issue

and attribute salience of the global warming issue. For issue salience, they were asked

to mark the degree of their perceived importance of the issue. For attribute salience, they

were asked to rank-order the perceived importance of the attributes. The answers were

used to set the baseline figures to later test first- and second-level agenda setting effects.

In particular, the change in perceived importance of issue between the pre- and post-

experimental survey was used to test the relationship between first-level agenda setting

and priming effects. The questionnaire also asked the respondents’ opinions on the sup-

port for the reduction of global warming effects (i.e. judgement whether to support the

efforts to reduce global warming effects or not) in order to set the baseline figures to later

test priming effects. Finally, the questionnaire asked the respondents’ beliefs about the

attributes of the global warming issue (i.e. the likelihood of the five attributes’ being

affected by global warming effects) in order to set the baseline figures to later test the

relation between second-level agenda setting and priming effects.

For the three-week experimental procedure after the pre-experimental survey, the

subjects were asked to read every news story and write a one-sentence summary for each


766



story. The goal of the task was to get the subjects to read the stories carefully. However,

during the three-week experimental procedure, 59 subjects were unable to complete

all the tasks.

A week after the final exposure, the 107 subjects who had completed the pre-survey

and the three experimental tasks were asked to complete the post-experimental survey.

Among the 107 participants, 12 provided incomplete data, which left 95 subjects in the

data pool. While the original participants had been assigned almost equally in terms of

number but randomly to each of the three conditions (High vs Medium vs No Exposure),

each group had a different number of participants left due to the dropouts and incomplete

data — High (30), Medium (34) and No Exposure (31). The ages of the subjects left in

the final data pool ranged from 17 to 27, with a mean age of 19. Eventually, 18 males and

77 females completed the post-survey, which consisted of the same questionnaire as in

the pre-survey.

Measure

Issue salience transfer. The current study measured subjects’ beliefs about the importance

of the global warming issue in pre- and post-experimental questionnaires. Four items

were borrowed from Iyengar and Kinder’s (1987) study. On both occasions, the subjects

assessed the following questions on nine-point Likert-type scales: ‘how important is the

global warming issue today?’ ‘how many news reports about the global warming issue

do you regularly pay attention to?’ ‘to what extent do you think the global warming issue

is deserving of additional government action?’ and ‘how often do you talk about the glo-

bal warming issue in your everyday conversation?’ The internal consistency reliability

for these four measures was satisfactory (Cronbach’s a ¼ .81), and the scales were

summed up to form an index score for issue importance. The difference in the subjects’

perceived importance between the pre- and post-experiment was then measured to test

first-level agenda setting effects.

Attribute salience transfer. The subjects were asked to rank-order the five different attri-

butes of the global warming issue according to their relative importance both before and

after the experiments. The question was, ‘Regarding the global warming issue, which do

you think is most important among the ecosystem, flood, industrial business, energy pol-

icy and regional conflict?’ Importance here refers to the significance or degree of chance

for each attribute’s reflection of the traits of global warming effects. The subjects were

asked to rank the attributes from 1 (most important) to 5 (least important). Then the rank-

order correlation between the frequency of attribute exposure and the subjects’ perceived

salience of the attributes, as expressed in their pre- and post-exposure survey response,

was determined.

Priming effects. For the subjects’ evaluative dimension on the global warming issue, the

current study measured the levels of support for the efforts to reduce global warming

effects. The author assessed possible support on a nine-point Likert-type scale anchored

by ‘extremely oppose’ and ‘extremely support’ by asking the subjects whether they per-

sonally support or oppose these efforts. The question was, ‘Do you support or oppose the

Lee 767

767



efforts to reduce the global warming effects?’ The difference in the level of the subjects’

support between the pre- and post-experiment was measured to test priming effects.

Association between agenda setting and priming effects. To investigate first-level agenda set-

ting effects’ influence on priming effects, the change in perceived importance of issue

salience and the change in support level for reducing global warming effects between

the pre- and post-experimental survey were calculated. Regression analysis was then

conducted by using the change in perceived issue salience as the independent variable

and the change in support level as the dependent variable. To explore the influence of

attribute salience transfer on priming effects, the subjects answered additional questions

asking about their beliefs on each of the five attributes of the global warming issue. The

beliefs were measured on nine-point Likert-type scales anchored by ‘strongly disagree’

and ‘strongly agree’ as to ‘how likely it was for each attribute to happen as a conse-

quence of global warming effects’. The basic format of the questions followed Kim

et al.’s (2002) study. For determining the association between second-level agenda set-

ting and priming effects, the change in support level for reducing global warming effects

was used as a dependent variable and the changes in beliefs on each of the five attributes

were used as independent variables for regression analysis.

Results

Issue agenda setting

First-level agenda setting effects were investigated by a set of paired-samples t-tests.

Prior to conducting the t-tests, the author analysed the primary dependent variables to

descriptively examine the changes of means before and after the exposure. The direc-

tions of the changes in the first-level agenda setting effects for all three groups were pos-

itive. The mean of the High Exposure group moved from 20.77 to 25.10, the mean of the

Medium Exposure group moved from 19.47 to 25.35 and the mean of the No Exposure

group moved from 18.77 to 19.65. Despite the same directions of change, however, there

seemed to be differences among the groups. The amount of mean change of the Medium

Exposure (M ¼ 5.88, SD¼ 3.46) and High Exposure (M¼ 4.33, SD ¼ 3.97) groups was

larger than that of the No Exposure (M ¼ 0.87, SD ¼ 2.93) group. In addition, while the

standard deviations of both the High (from 5.92 to 5.40) and Medium (from 5.12 to 4.83)

groups decreased, that of the No Exposure group increased from 4.45 to 4.68. The results

seem to indicate that the subjects exposed to the global warming issue were more likely

to change their issue importance than those who were not. To determine whether such

indications are to be statistically supported, paired-samples t-tests were conducted for

each of the three groups, respectively, and the results are reported in Table 1.

The results showed the first-level agenda setting effects for the subjects who were

exposed to the global warming issue, while no such effects were detected for the subjects

who were not. Table 1 shows that there was a statistically significant change in the sub-

jects’ issue importance after high exposure, t(29)¼ 5.98, p < .01, and medium exposure,

t(33)¼ 9.91, p < .01, to the target issue in the online newspaper. However, the No Expo-

sure group did not demonstrate such effects, t(30) ¼ 1.66, p ¼ .11. The results showed


768



that the subjects who were exposed to the global warming issue became more likely to

believe that the issue is important.

Attribute agenda setting

Table 2 shows two sets of results: (1) the subjects’ opinions on each of the five

attributes, which were measured for both pre- and post-surveys by rank-order from 1

(most important) to 5 (least important); (2) and the rank-order correlation between the

frequency of attributes exposed to the subjects and their perceived importance of the

attributes. The rankings for each of the attributes were simply added to reveal the over-

all opinions of the subjects in the individual groups. Thus, the smaller the number in the

table, the more important the subjects think the attribute is. The far left column in

the upper part of Table 2 contains the attributes in the order of their exposure frequency

to the subjects of the High and Medium groups.

As seen in the upper part of Table 2, the rank-orders of the five attributes in the pre-

survey are exactly the same among the three groups. In all groups, the ecosystem was

thought to be the most important attribute of the global warming issue; energy, the

Table 2. Ranks of attributes and rank-order correlation between the media presentation andsubjects’ perception for second-level agenda setting effects

High Exposure Medium Exposure No Exposure

Pre Post Pre Post Pre Post

Nc Rd N R N R N R N R N R

Aa Ecosystem 45 1 40 1 55 1 54 1 42 1 40 1Flood 95 3 78 2 111 3 91 2 91 3 94 3Business 100 4 116 4 119 4 127 4 118 4 122 5Energy 81 2 86 3 91 2 95 3 88 2 90 2Conflict 129 5 130 5 134 5 143 5 126 5 119 4

Bb rse .700 .900* .700 .900* .700 .500

a Ranks of subjects’ perceived importance of the attributes.b Rank-order correlation between the frequencies of the attributes exposed to the subjects and their per-ceived importance.c N ¼ Accumulated number of each attribute’s rank given by the subjects.d R ¼ Rank of importance based on N.e rs ¼ Spearman’s rho.* p < .05.

Table 1. Results of the paired t-tests on first-level agenda setting effects

M SD d.f. t

High Exposure Issue salience transfer 4.33 3.97 29 5.981**Medium Exposure 5.88 3.46 33 9.907**No Exposure 0.87 2.93 30 1.655

** p < .01.

Lee 769

769



second most important; flood, the third; business, the fourth; and conflict, the fifth.

Additionally, before exposure to the stimuli, the rank-orders of the attributes in all three

groups were not significantly correlated to the rank-order of what would be the fre-

quency of the attributes’ exposure to the subjects. The lower part of Table 2 shows that

the rank-order correlation between the frequency of the attributes’ exposure to the sub-

jects through the experiments and the perceived importance of such attributes among the

subjects before the experiments was statistically insignificant, rs ¼ .70, p ¼ .19, N ¼ 5.

However, after exposure to the differentiated news contents, the subjects in all

groups changed their perceived importance of the attributes. Notably, the rank-order

of the High and Medium Exposure groups went in the same direction, but that of the

No Exposure group went a different way. More importantly, the rank-order correlation

between the frequency of attributes’ exposure and the perceived importance of the

attributes among the subjects in the High and Medium groups became statistically

significant, rs ¼ .90, p < .05, N ¼ 5, while that of the No Exposure group worsened,

rs¼ .50, p¼ .39, N¼ 5. The results of the rank-order correlation showed that the expo-

sure of the attributes of the global warming issue to the subjects led to attribute salience

transfer from the media to the audience.

Priming

The proposed hypothesis was examined by a set of paired-samples t-tests. Prior to

conducting the t-tests, the author analysed the primary dependent variables to descrip-

tively examine the changes of means before and after the exposure. The directions of

the changes in priming effects for all three groups were positive. The mean of the

High Exposure group moved from 7.70 to 8.33, the mean of the Medium Exposure

group moved from 7.56 to 8.41, and the mean of the No Exposure group moved from

7.19 to 7.45.

Although the groups moved in the same direction, there seemed to be differences

among them. While the amount of mean change of the Medium Exposure (M ¼ 0.85,

SD ¼ 1.16) group was greater than that of the High Exposure (M ¼ 0.63, SD ¼ 1.03)

group, both changes were greater than that of the No Exposure (M ¼ 0.26, SD ¼ 1.18)

group. In addition, while the standard deviations of both High (from 1.21 to 0.96) and

Medium (from 1.50 to 0.86) groups decreased, that of the No Exposure group increased

slightly from 1.35 to 1.36. The results seem to indicate that the subjects exposed to the tar-

get issue and its attributes were more likely to change their evaluative dimensions on the

issue than those who were not. To determine whether such indications are to be statistically

supported, paired-samples t-tests were conducted for the High, Medium and No Exposure

groups, respectively, and the results are reported in Table 3.

The results showed priming effects for the subjects who were exposed to the target

issue and its attributes, while no such effects were detected for the subjects who were

not. Table 3 shows that there was a statistically significant change in the subjects’

evaluative dimension after high exposure, t(29) ¼ 3.36, p < .01, and medium exposure,

t(33) ¼ 4.29, p < .01, to the target issue and its attributes in the online newspaper. How-

ever, the No Exposure group did not demonstrate such statistically significant effects,

t(30) ¼ 1.22, p ¼ .23. The results showed that the subjects who were exposed to the


770



global warming issue and its attributes became more likely to judge that they should

support the efforts to reduce global warming effects.

Association between agenda setting and priming effects

Although the paired t-tests above showed priming effects for the High and Medium

Exposure groups, it was not clear whether these effects were from the influence of issue

salience transfer or attribute salience transfer. While the High and Medium Exposure

groups’ subjects were exposed to the five different attributes, the issue of global warming

itself was exposed simultaneously. Thus, the priming effects detected above may be the

results of issue salience transfer and/or attribute salience transfer. To determine if the

issue or attributes salient in the media play significant roles as determinants of issue eva-

luations, the author used two regression models. These models were applied to the two

groups – High and Medium – once combined and once separately. The results of the

regression tests are presented in Table 4. The No Exposure group did not show any

agenda setting and priming effects, so it was excluded from the regression analyses.

Issue agenda setting and priming. The regression model for the association between first-

level agenda setting and priming effects includes the subjects’ perceived issue impor-

tance change of the global warming issue before and after the experiments, in predicting

the degree of change for supporting the efforts to reduce global warming effects before

Table 4. Regression analysis of agenda setting effects predicting priming effects

High þ Medium (N ¼ 64) High (N ¼ 30) Medium (N ¼ 34)

Aa Issue importance .067 (.036) .078 (.047) .047 (.059)R2 .052 .090 .019

Bb Ecosystem .247* (.101) .076 (.141) .473** (.167)Flood .147y (.087) .235y (.122) .065 (.140)Business .132 (.094) .109 (.140) .054 (.163)Energy –.063 (.100) .055 (.131) –.091 (.182)Conflict –.103 (.069) –.157 (.123) –.067 (.087)R2 .220* .251 .328*

Note: The entries are unstandardized regression coefficients with standard errors in parentheses.a Issue salience predicting support for the efforts to reduce global warming effects.b Attributes salience predicting support for the efforts to reduce global warming effects.y p < .1; * p < .05; ** p < .01.

Table 3. Results of the paired t-tests on priming effects

M SD d.f. t

High Exposure Priming effects .63 1.03 29 3.357**Medium Exposure .85 1.16 33 4.294**No Exposure .26 1.18 30 1.215

** p < .01.

Lee 771

771



and after the experiments. As seen in the upper part of Table 4, the model does not

explain priming effects. None of the three cells show the statistically significant relation-

ship between first-level agenda setting and priming effects: for the combined group of

High and Medium Exposure, R2 ¼ .05, p ¼ .07; High Exposure, R2 ¼ .09, p ¼ .11; and

Medium Exposure, R2¼ .02, p¼ .43. Strictly speaking, first-level agenda setting effects

did not have an influence on priming effects.

Attribute agenda setting and priming. The regression model for the association between

second-level agenda setting and priming effects includes the subjects’ belief changes

on the five attributes of the global warming issue before and after the experiments, in

predicting the degree of change for supporting the efforts to reduce global warming

effects before and after the experiments. As seen in the lower part of Table 4, the model

efficiently explains priming effects for the combined group of High and Medium expo-

sures with R2 ¼ .22, p < .05. That is, the set of five attributes can predict the priming

effects for the two groups, and specifically among the subjects who were exposed to the

attributes, the Ecosystem attribute worked as a significant predictor in predicting support

for the efforts to reduce the effects of global warming, t¼ 2.45, p < .05. Additionally, the

Flood attribute worked as a marginally significant predictor for priming effects, t¼ 1.68,

p < .10. Thus, it can be concluded that attribute salience transfer is highly associated with

priming effects.

However, when the model was applied to the exposed groups separately, some com-

plicated results were found. As seen in the table, only the Medium group showed that the

model could explain priming effects efficiently, R2 ¼ .33, p < .05, and specifically, the

Ecosystem attribute worked as a significant predictor in predicting support for the efforts

to reduce global warming effects, t ¼ 2.83, p < .01. However, the model did not produce

clear support for the High group in the attributes’ predicting priming effects, R2 ¼ .25,

p ¼ .20. Even though the Flood attribute worked as a marginally significant predictor,

t ¼ 1.93, p < .10, it was not strongly meaningful because the model does not support the

overall relationship.

Table 4 seems to indicate that although attribute salience transfer influences priming

effects, there may be a difference as to the degree or range of the impact based on the

amount of exposure or the varied mixture of the news content.

Conclusions and discussions

The overall results of this study suggest that even if we can find first- and second-level

agenda setting effects and priming effects together, the priming effects were more likely

caused by second-level agenda setting effects than first-level agenda setting effects. That

is, the research findings support the idea that not the issue salience transfer but the attri-

bute salience transfer becomes a significant factor in determining issue evaluations

among the audience. However, beyond the argument of attribute salience transfer and

its influence on priming effects in the experiment, there are some aspects that need to

be discussed further.

First, the study began with a conceptualization of the opinion factor in priming effects

in relation with agenda setting effects. The current study tackled McCombs’ argument


772



about the relationship between agenda setting and priming effects. In his typology, the

formation of opinion, which is the core part of priming effects (Iyengar and Kinder,

1987), resulted from the combination of first- and second-level agenda setting effects

(McCombs, 2004). Specifically, he insisted that the strength of an opinion has something

to do with first-level agenda setting effects, and its direction has something to do with

second-level agenda setting effects. He seemed to argue that strength and direction are

two different things. The current study, however, suggests another way of understanding

the opinion factor. That is, opinion direction innately has opinion strength. Although the

strength and direction of an opinion can be divided in an abstract way as McCombs

argued, it would not be meaningful to gauge the opinion’s strength without considering

its direction, especially when thinking of priming effects, which generally measure how

much to support or oppose something. Conclusively, opinion formation is rather a matter

of setting opinion direction, and the strength of an opinion can be measured by how far

the opinion goes to a certain direction.

Subsequently, another question arises. According to some psychologists, opinion

consists of beliefs and values (Goethals and Nelson, 1973; Jones and Gerard, 1967).

Applying their reasoning to the agenda setting propositions, both elements of opinion

seem to be closely related to the two dimensions of attributes — substantive and affec-

tive. What remains to be determined is which of the two attribute dimensions better

explains priming effects. Because the affective dimension is about the tone — positive,

neutral and negative — we would think it would be more closely related to priming

effects. However, as we saw in the experiment and some previous studies, the affective

dimension seemed to hinge on the substantive dimension (Lee and Yoo, 2004). Even

some psychologists believe that belief and value, or the degree of credibility and level

of preference, are very closely related to each other (Johnson, 2003; Zimmerman and

Kitsantas, 2002). Although this study is one of the few attempts to trace the real influen-

cing factor between first- and second-level agenda setting on priming effects, future

studies that explore the role of the substantive and affective dimensions of attributes

affecting priming effects are certainly recommended.

Second, although the study found second-level agenda setting effects’ influence on

priming effects, it did not explain all the complicated details of the impact. We would

think that the High Exposure group might exhibit greater priming effects and greater

influence of second-level agenda setting on priming effects based on common sense.

However, the study results showed the opposite, and unfortunately, these did not reveal

the concrete and specific reasons why such phenomena occurred; the results only showed

the observable facts.

As a matter of fact, this study’s experiment is not the only case to show such complex

results. McCombs et al.’s Spanish local election study also revealed that the participants

exposed to ‘some’ political information showed more affective attribute agenda setting

effects than those exposed to ‘all’ political information (McCombs, 2004). The results of

their study and this present experiment clearly distinguished the difference of agenda set-

ting effects between exposure and no exposure but showed complicated results between

High Exposure and Medium Exposure, so they seem to make an interesting suggestion:

agenda setting effects and related priming effects, especially in terms of the attribute

agenda, which is supposed to influence the audience’s substantive and affective

Lee 773

773



perceptions, are not monotonously grounded in the simple frequency of exposure but

also on some other factors like the audience’s psychological status (Ha, 2004). However,

because just one or two studies can hardly explain all the aspects of complicated psycho-

logical or social phenomena, the factors discussed previously are worth being studied

further in the future. Specifically, such observations could direct some further studies

on this topic, including the need for orientation, which is believed to explain the psycho-

logical foundations of the agenda setting theory (Weaver, 1977, 1980). The convergence

of the second-level agenda setting study with the two traditional dimensions of the need

for orientation – relevance and uncertainty – may be a good research direction in order to

uncover the reasons for the differing influence of the attributes on priming effects.

Furthermore, the effort to discover such detailed aspects of agenda setting and prim-

ing effects can be incorporated with some other theoretical frameworks in communica-

tion studies including uses and gratification (Blumler and McQuail, 1969; Katz, 1996;

Katz et al., 1973; Zillmann and Bryant, 1985). The uses and gratification theory posits

people’s active learning from media content by seeking meaning for themselves, so the

convergence of agenda setting and the uses and gratification theory could expedite the

findings of the specific details of media effects, including the phenomena observed but

not explained clearly in the current study. Specifically, it would help scholars find some

intervening variables which were not identified in this study’s experiment showing some

complicated results. And such an approach would help further our understandings about

the relationship between the media exposure to a certain issue and/or its attributes and

audiences’ original perception about the issue and/or its attributes, which this experiment

might have missed. In sum, the results and discussions of this study would hopefully

direct future research to explore media effects further.

Funding

This research received no specific grant from any funding agency in the public, commercial, or

not-for-profit sectors.

References

Ansolabehre S, Behr R and Iyengar S (1993) The Media Game, American Politics in the Television

Age. Toronto: Maxwell Macmillan Canada.

Blumler JG and McQuail D (1969) Television in Politics: Its Uses and Influence. Chicago, IL: Uni-

versity of Chicago Press.

Ghanem S (1996) Media coverage of crime and public opinion. Unpublished doctoral dissertation,

School of Journalism, University of Texas at Austin.

Ghanem S (1997) Filling the tapestry: The second level of agenda-setting. In: McCombs M,

Shaw D and Weaver D (eds) Communication and Democracy. Mahwah, NJ: Lawrence

Erlbaum, 3–14.

Goethals GR and Nelson RE (1973) Similarity in the influence process: The belief-value distinc-

tion. Journal of Personality and Social Psychology 25(1): 117–122.

Golan G and Wanta W (2001) Second-level of agenda setting in the New Hampshire Primary:

A comparison of coverage in three newspapers and perception of candidates. Journalism and

Mass Communication Quarterly 78(2): 247–259.


774



Ha S (2004) Who accepts the news? News coverage of presidential campaigns, voters’ information

processing ability, and media effects susceptibility. Unpublished doctoral dissertation, School

of Journalism, University of Texas at Austin.

Harrison TM, Stephen TD and Fehr BJ (1991) Image versus issues in the 1984 Presidential

election. Human Communication Research 18(2): 209–227.

Hilmert CJ, Kulik JA and Christenfeld NJS (2006) Positive and negative opinion modeling:

The influence of another’s similarity and dissimilarity. Journal of Personality and Social

Psychology 90(3): 440–452.

Iyengar S and Kinder DR (1987) News That Matters: Television and American Opinion. Chicago,

IL: University of Chicago Press.

Iyengar S and Simon A (1993) News coverage of the Gulf Crisis and public opinion: A study of

agenda setting, priming and framing. Communication Research 20(3): 365–383.

Iyengar S, Peters MD and Kinder DR (1982) Experimental demonstrations of the ‘not-so-minimal’

consequences of television news programs. American Political Science Review 76(4):

848–858.

Johnson A (2003) Procedural memory and skill acquisition. In: Healy AF and Proctor RW (eds)

Handbook of Psychology: Experimental Psychology, Vol. 4. New York: Wiley, 499–523.

Jones EE and Gerard H (1967) Foundations of Social Psychology. New York: Wiley.

Katz E (1996) Viewers work. In: Hay J, Grossberg L and Wartella E (eds) The Audience and its

Landscape. Boulder, CO: Westview Press, 9–21.

Katz E, Gurevitch M and Haas H (1973) On the use of the mass media for important things.

American Sociological Review 38(2): 164–181.

Kim S, Scheufele DA and Shanahan J (2002) Think about it this way: Attribute agenda-setting

function of the press and the public’s evaluation of a local issue. Journalism and Mass

Communication Quarterly 79(1): 7–25.

King P (1997) The press, candidate images and voter perceptions. In: McCombs M, Shaw D and

Weaver D (eds) Communication and Democracy: Exploring the Intellectual Frontiers in

Agenda Setting Theory. Mahwah, NJ: Lawrence Erlbaum, 29–40.

Kiousis S, Bantimaroudis P and Ban H (1999) Candidate image attributes: Experiment on the sub-

stantive dimension of second level agenda setting. Communication Research 26(4): 414–428.

Lee G and Yoo C (2004) Attribute salience transfer of global warming issue from online papers to

the public: Attribute of environment vs. attribute of economy. Paper presented at the meeting of

the Association for Education in Journalism and Mass Communication Convention, Toronto,

Ontario, August.

McCombs M (2004) Setting the Agenda: The Mass Media and Public Opinion. Cambridge: Polity

Press.

McCombs M and Shaw D (1972) The agenda-setting function of mass media. Public Opinion

Quarterly 36: 176–187.

McCombs M, Lopez-Escobar E and Llamas J (2000) Setting the agenda of attributes in the 1996

Spanish general election. Journal of Communication 50(2): 77–92.

Price V and Tewksbury D (1997) News values and public opinion: A theoretical account of media

priming and framing. In: Barett GA and Boster FJ (eds) Progress in Communication Sciences:

Advances in Persuasion, Vol. 3. Greenwich, CT: Ablex, 173–212.

Scheufele DA (2000) Agenda-setting, priming and framing revisited: Another look at cognitive

effects of political communication. Mass Communication and Society 3(2 and 3): 297–316.

Lee 775

775



Severin WJ and Tankard JW (2001) Communication Theories: Origins, Methods and Uses in Mass

Communication, 3rd edn. New York: Addison-Wesley.

Shaw DL and McCombs M (1977) The Emergence of American Political Issues. St Paul, MN:

West Publishing.

Soroka SN (2002a) Agenda-Setting Dynamics in Canada. Vancouver and Toronto: UBC Press.

Soroka SN (2002b) Issue attributes and agenda setting by media, the public and policy makers in

Canada. International Journal of Public Opinions Research 14(3): 264–285 (electronic

version).

Suls J, Martin R and Wheeler L (2000) Three kinds of opinion comparison: The triadic model.

Personality and Social Psychology Review 4(3): 219–237.

Takeshita T (2005) Current critical problems in agenda setting research. International Journal of

Public Opinions Research 18(3): 275–296.

Weaver DH (1977) Political issues and voter need for orientation. In: McCombs M and Shaw D

(eds) The Emergence of American Political Issues. St Paul, MN: West Publishing, 107–109.

Weaver DH (1980) Audience need for orientation and media effects. Communication Research

7(3): 361–376.

Weaver DH, Graber DA, McCombs ME and Eyal CH (1981) Media Agenda Setting in a Presiden-

tial Election: Issues, Images and Interest. New York: Praeger.

Winter JP and Eyal CH (1981) Agenda-setting for the civil rights issue. Public Opinion Quarterly

45(3): 376–383.

Xu Z and Bengston DN (1997) Trends in national forest values among forestry professionals,

environmentalists, and the news media, 1982–1993. Society and Natural Resources 10: 43–59.

Yagade A and Dozier DM (1990) The media agenda setting effects of concrete versus abstract

issues. Journalism Quarterly 67(1): 3–10.

Zillmann D and Bryant J (1985) Affect, mood, and emotion as determinants of selective exposure.

In: Zillmann D and Bryant J (eds) Selective Exposure to Communication. Hillsdale, NJ:

Lawrence Erlbaum, 157–190.

Zimmerman BJ and Kitsantas A (2002) Acquiring writing revision and self-regulatory skill

through observation and emulation. Journal of Educational Psychology 94: 660–667.

Zucker HG (1978) The variable nature of news media influence. In: Ruben BD (ed.) Communica-

tion Yearbook, 2. New Brunswick, NJ: Transaction, 225–240.


776



international communication gazette 2010 lee 759 76

Documents

unpublished doctoral dissertation

global warming issue

global warming effects

marginally significant predictor

audiences evaluative dimension

audiences perceived salience

issue salience transfer

primary dependent variables