Sunteți pe pagina 1din 13

F.

Arendt:
Journalof Cultivation
Media Psychology Ef fects
©2010;
2010 of22(4):147–159
Vol. a Newspaper
Hogrefe Publishing

Original Article

Cultivation Effects of a Newspaper


on Reality Estimates and
Explicit and Implicit Attitudes
Florian Arendt
Department of Communication, University of Vienna, Austria

Abstract. This paper explores the cultivation effect of a newspaper on its readers’ reality estimates and attitudes. Additionally, the study
tries to advance cultivation research by examining implicit attitudes (i.e., automatic affective reactions toward an object). A content
analysis of four months of news coverage in one particular newspaper showed that foreigners were overrepresented as offenders and that
the newspaper had a negative view of the EU. According to cultivation theory, it is assumed that the more people read a newspaper, the
more their reality estimates and attitudes correspond to the most recurrent, stable, and overarching patterns of the newspaper’s content.
To test this hypothesis, a total of 453 students participated in a study that used a cross-lagged panel design with two waves and a time-lag
of two months. Consistent with the cultivation hypothesis, those who spent more time reading the newspaper were more likely to
overestimate the frequency of foreigners as offenders (i.e., first-order cultivation) and had more negative self-reported attitudes toward
the EU (i.e., second-order cultivation). Additionally, those who read more of the newspaper showed more negative implicit attitudes
toward the EU (i.e., implicit cultivation). The data show evidence of a significant causal influence of newspaper exposure on implicit
attitudes, and a marginally significant causal effect on the overestimation of foreigners as offenders and on explicit attitudes toward the
EU. The consideration of implicit attitudes as an additional dependent variable could advance cultivation theory and research.

Keywords: cultivation, newspaper, media effects, implicit attitudes

Introduction reality. For example, content analyses have shown that par-
ticular occupations such as doctor, judge, and police officer
occur with much greater frequency on television than in
We live in a mass mediated society. The mass media play reality (Weimann, 2000). The central hypothesis of this the-
important roles in our daily lives. For that reason, the mass ory, called cultivation hypothesis, states that people who
media are considered important social agents, transmitting watch more television are more likely to perceive the real
social perceptions and attitudes (Weimann, 2000). In fact, world in ways that reflect the most common and recurrent
the content of the mass media and their possible effects on messages from television, as compared to people who
recipients have been major concerns of social scientists for watch less, but are otherwise comparable in terms of im-
decades (Bryant & Oliver, 2009). portant demographic characteristics (Morgan et al., 2009).
The empirical work reported in this article investigated For example, if there is an overrepresentation of doctors on
the possible cultivation effects that reading a newspaper television, heavy viewers should overestimate the real-
had on readers’ social perceptions and attitudes. The study world incidence of doctors.
also attempted to advance existing cultivation research by Although the original formulation of the cultivation hy-
using implicit attitudes as an additional dependent variable. pothesis and much of the subsequent research focused on
television, several studies have investigated the cultivation
effect of newspapers (Arendt, 2008, 2009; Bauer, 2005;
Previous Cultivation Research Grabe & Drew, 2007; Guo, Zhu, & Chen, 2001; Hicks &
Lee, 2004; Nisbet & Wang, 2004; Reber & Chang, 2000;
Many studies have claimed to show the effect of exposure Reimer & Rosengren, 1990; Stroman & Seltzer, 1985; Ver-
to a mass medium on recipients’ social perceptions and at- geer, Lubbers, & Scheepers, 2000). The focus on television
titudes. One prominent theory in this area is cultivation the- is due to the assertion that television is the “central cultural
ory (Gerbner & Gross, 1976; Morgan, Shanahan, & Signo- arm of American society” (Gerbner & Gross, 1976, p. 175).
rielli, 2009). Cultivation theory is based on the findings that No claim has been made regarding the validity of this state-
television content often presents a systematic distortion of ment outside of the American context (see Reimer & Ro-

© 2010 Hogrefe Publishing Journal of Media Psychology 2010; Vol. 22(4):147–159


DOI: 10.1027/1864-1105/a000020
148 F. Arendt: Cultivation Effects of a Newspaper

sengren, 1990). It must be noted that the specific media concerning the cultivation effect of specific newspaper
structure of the country and context that is being studied is products (Vergeer et al., 2000).
of major importance. In Austria, where more than 75% of Shanahan and Morgan (1999) conducted a meta-analy-
the population reads a newspaper daily, but only 63% of sis based on a database consisting of 5,799 separate find-
the population watches television daily (Verein Arbeitsge- ings, derived from 97 cultivation studies conducted since
meinschaft Media-Analysen, 2008), the omission of news- 1976. Average Pearson r’s were about .10. Partial r’s were
papers from a cultivation discussion would be absurd. slightly lower at .09. The authors concluded that the culti-
Whether there are different cultivation effects for different vation literature has demonstrated that there is a small pos-
media channels due to different content (e.g., television is itive relationship between watching television and various
mostly fictitious; newspapers are mostly factual) is an em- dependent variables. Another meta-analysis conducted by
pirical question (see Grabe & Drew, 2007). Rossmann (2008) verified Shanahan and Morgan’s results.
The media market in Austria (Steinmaurer, 2009) is a There is no meta-analysis regarding newspaper cultivation.
typical example of a small state market with a giant next- Based on the findings from the small number of available
door neighbor sharing the same language. The print media studies mentioned above, we can assume that newspapers
market is highly concentrated. Kronen Zeitung (Krone for are also able to produce a cultivation effect.
short) is the dominant player. Because Krone has approxi- In Austria, two cultivation studies have been conducted
mately 3 million readers a day in a country of 8 million to investigate the effects of television. Both studies found
inhabitants (Verein Arbeitsgemeinschaft Media-Analysen, support for the cultivation hypothesis (Appel, 2008; Barth,
2008), it can be seen as the mainstream of the Austrian 1988). Additionally, two cultivation studies have tested the
print-based symbolic environment. This newspaper has cultivation hypothesis with regard to the effects of a partic-
been criticized by many observers for its biased news cov- ular newspaper product and found supporting empirical ev-
erage. For example, Wise (2002) wrote: idence (Arendt, 2008, 2009).
“With three million readers a day, Austria’s Kronen Zeitung
has perhaps the highest per capita circulation of any newspaper
in the world. It gained this preeminence with an almost dadaist Implicit Cultivation
collage of stories fanning fears that Vienna is being swamped
by undesirable refugees, editorials tinged with anti-Semitic in- In cultivation research, a distinction is made between two
nuendo, and articles trivializing the Holocaust” (p. 12). kinds of dependent variables. First-order cultivation refers
to effects on social perceptions (set-size and probability
In addition, many people in the Austrian public think that judgments; see Shrum, 2009), while second-order cultiva-
Krone has a strong impact on its readers (Plasser & Leng- tion refers to effects on attitudes and values (Shanahan &
auer, 2010). In the words of one former Austrian Chancel- Morgan, 1999). This article tries to advance cultivation the-
lor, “it is impossible to govern without the support of the ory and research by using implicit attitudes as an additional
Krone” (Art, 2007, p. 336). dependent variable, and thus investigate the phenomenon
From a theoretical point of view, there is a difference of implicit cultivation. Shrum (2007) referred to this new
between assuming that newspapers in general (e.g., Aus- dimension as a “promising area of research” (p. 265).
trian newspapers) produce a cultivation effect and claim- In Gawronski and Bodenhausen’s (2006) associative-
ing a cultivation effect from one particular newspaper propositional evaluation (APE) model, implicit attitudes
(e.g., Krone). However, in the case of Austria, it seems are interpreted as “automatic affective reactions resulting
appropriate to study the effect of Krone for at least two from the particular associations that are activated automat-
reasons. First, the constructed realities in different news- ically when a person encounters a relevant stimulus”
papers are not the same. Like many other scholars who (p. 697). This process does not require much cognitive ca-
promote content-specific cultivation research, Bilandzic pacity or an intention to evaluate an object. Associations
and Rössler (2004) criticized Gerbner and Gross’ (1976) can be activated irrespective of whether a person considers
assumption regarding the homogeneity of the media con- these evaluations as accurate or inaccurate. Processes of
tent. There are too many differences among, for example, propositional reasoning assess the validity of automatic, af-
Der Standard (quality newspaper, leftist), Die Presse fective reactions. This assessment provides the basis for
(quality newspaper, conservative), and Krone (tabloid, evaluative judgments, which are reflected in explicit (i.e.,
opportunistic/conservative) with regard to constructed self reported) attitudes that are termed second-order judg-
realities (Berkel, 2006; Stark & Kraus, 2008). Second, ments in cultivation research. Research on explicit and im-
according to cultivation theory, different media realities plicit attitudes has shown that implicit attitudes predict
lead to different cultivation hypotheses and thus, should spontaneous behavior better while explicit attitudes predict
have different cultivation effects. Consistent with this no- controlled behavior better (Gawronski & Conrey, 2004).
tion, research has found content-specific cultivation ef- Arendt (2009) included implicit attitudes for the first
fects (Bilandzic & Rössler, 2004). Although the majority time in cultivation research. That research, which involved
of newspaper-cultivation studies have investigated the ef- a sample of 92 participants, demonstrated a small cross-
fects of general newspaper use, there is empirical work sectional correlation between newspaper exposure and im-

Journal of Media Psychology 2010; Vol. 22(4):147–159 © 2010 Hogrefe Publishing


F. Arendt: Cultivation Effects of a Newspaper 149

plicit attitudes toward the EU. This finding must be repli- Variables
cated with a larger sample. Because of the cross-sectional
nature of both Arendt studies, and their lack of first-order Each EU-related article was coded in terms of whether it
measures, an empirical investigation that addressed these exhibited a positive, negative, ambiguous, or no evaluation
drawbacks was needed. of the EU. The coding was made according to a coding
schema from Früh (2007, pp. 249–260), which was used in
previous studies (Arendt, 2008, 2009). An article was cod-
ed “negative” if the EU was associated with negative con-
cepts such as “corruption” or “democratic deficit.” An ar-
Overview of the Present Empirical ticle was coded “positive” if the EU was associated with
positive connotations. The code “ambiguous” was used
Work when both positive and negative associations were present.
The code “no evaluation” was used when there were no
Cultivation research has two steps. First, researchers must evaluative associations. Because the same coding schema
conduct (or refer to) a content analysis to reliably outline was used in two previous studies in which articles about
selected features and trends in the world that newspapers the EU from the same newspaper were coded by the same
present to their readers. Based on this empirical evidence, person, and Cohen’s Kappa-values indicated a highly reli-
researchers formulate cultivation hypotheses. In a second able measurement (Arendt, 2008, 2009; κ > .90), no sepa-
step, researchers aim to determine whether those who rate reliability analysis was conducted for the coding of the
spend more time reading the newspaper are more likely to EU evaluation variable.
perceive social reality in ways that reflect the world of the With regard to the crime topic, units were coded accord-
mass medium. Following this research paradigm, Study 1 ing to whether an offender was labeled as an “Austrian” or
attempted to outline the Krone world with regard to specific as a “foreigner” or “no relevant information available.” A
aspects of social reality. Based on these findings, cultiva- total of 1,912 cases were analyzed. For the reliability anal-
tion hypotheses were formulated. Study 2 attempted to test ysis, a second trained coder coded 100 cases. The Cohen’s
these hypotheses. Kappa-value for the coded variable “nationality of the of-
fender” was .91, indicating a highly reliable measurement.

Results
Study 1
With regard to the EU, 48.5% articles were coded as neg-
As previously mentioned, Krone has been criticized by ative. Only 1.5% were coded as positive. Most of the re-
many observers. Two topics in particular were often men- maining articles had no evaluation, and only a few were
tioned: the European Union and crime coverage. Krone is considered ambiguous. Regarding the crime topic, the na-
accused of having news coverage that (a) is very negative tionality of the offender (Austrian or foreigner) was coded
toward the EU and (b) overrepresents foreigners as crimi- in 1,004 cases. For the rest of the articles, no information
nal offenders (Horvath, 2006). To produce empirical evi- pertaining to the nationality of the offender was available.
dence to deduce evidence-based cultivation hypotheses Based on these 1,004 cases, 65.3% of all those suspected
with regard to these two different topics, a content analysis of having committed an offense were labeled as foreigners.
that focused on these two aspects of social reality was con- Offenders from Serbia, 10.8%, and Hungary, 9.4%, were
ducted. the most frequently represented in the foreigner category.

Discussion
Method
First, the results of the content analysis indicate that Krone
Sample and Procedure has a very negative view of the EU. Second, data from the
Federal Criminal Police Office (Bundeskriminalamt, per-
The Krone news coverage between February and May 2009 sonal communication, January 28, 2009) show that in 2008,
was analyzed. All articles were selected and coded accord- 27.2% of all those suspected of having committed a crime
ing to the two topics. The unit of analysis for the EU topic were foreigners. This leads to the conclusion that there was
was an EU-related article. The unit of analysis for the crime a massive overrepresentation of foreigners in Krone’s
topic was a person who was suspected of having committed crime reporting (reality: 27.2%, Krone: 65.3%). Offenders
an offense. The coding was done by the author. 1,140 arti- from Serbia (reality: 3.26%, Krone: 10.8%) and Hungary
cles were coded as relating to the EU topic and 1,920 per- (reality: 0.91%, Krone: 9.4%) were the most overrepresent-
sons as pertaining to the crime topic. ed.

© 2010 Hogrefe Publishing Journal of Media Psychology 2010; Vol. 22(4):147–159


150 F. Arendt: Cultivation Effects of a Newspaper

papers on relatively stable and deeply rooted variables such


Study 2 as symbolic racism should be small. A cultivation effect
would only be observable over a long period of time (e.g.,
Development of the Cultivation Hypotheses several years). From that argumentation, an explorative re-
Based on the theoretical background discussed above and search question could be formulated:
the findings of the content analysis, the following cultiva-
tion hypotheses could be derived: RQ1: Is there a cross-sectional correlation between expo-
sure to a newspaper (Krone) and symbolic racism; does
Hypothesis 1: The more someone reads Krone, the more exposure have a causal influence on symbolic racism; and
this person overestimates the frequency of foreigners, par- does symbolic racism have a causal influence on the fre-
ticularly Serbs and Hungarians, as criminal offenders (i.e., quency-estimates?
first-order cultivation).
To broaden the perspective regarding the EU-topic, two ad-
Hypothesis 2: The more someone reads Krone, the more ditional variables were considered. First, the valence of the
negative the explicit attitudes this person has toward the free associations toward the EU was chosen to give partic-
European Union (i.e., second-order cultivation). ipants the opportunity to freely recall some of their associ-
ations with the EU. The rationale behind this lies in the
Hypothesis 3: The more someone reads Krone, the more heuristic processing model for first-order cultivation ef-
negative the implicit attitudes this person has toward the fects proposed by Shrum (2009). Based on the model, it
European Union (i.e., implicit cultivation). can be assumed that exposure to Krone, which reports a lot
of negative attributes regarding the EU, enhances the ac-
Because cultivation theory predicts that at least a part of cessibility of those attributes. Thus, readers should more
the correlation between newspaper exposure and a depend- easily recall negative attributes than non-readers (while
ent variable can be attributed to the causal influence of other variables are held constant). Because Shrum’s model
newspaper exposure (Gerbner & Gross, 1976), an addition- deals with first-order cultivation, and the present study
al hypothesis can be formulated: measures free associations in the context of second-order
measures with regard to newspaper cultivation, the theoret-
Hypothesis 4: There is a causal influence of newspaper ex- ical knowledge is not sufficient to develop a hypothesis.
posure on first-order, second-order, and implicit measures. For that reason, another explorative research question was
formulated.

Additional Theoretical Considerations RQ2: Is there a cross-sectional correlation between expo-


sure to a newspaper (Krone) and the valence of the free
Because it is likely that personality-related factors also associations with the EU, and does exposure have a causal
have a causal effect on the first-order measures, one addi- influence?
tional variable was considered: symbolic racism (Henry &
Sears, 2002). This variable is conceptualized as a relatively
stable personality trait. “Old-fashioned” racism incorpo- A second variable, which should broaden the perspective
rates social distance between the races, beliefs in the bio- pertaining to the EU-topic, is behavioral intention (e.g.,
logical inferiority of others, and support for formal discrim- intentions to vote in favor of Austria exiting the EU). Nabi
ination and segregation. In contrast, symbolic racism rep- and Sullivan (2001) attempted to determine if heavy tele-
resents another kind of “new” racism. The word vision viewing influences not only first- and second-order
“symbolic” was chosen to highlight the idea that the beliefs measures (violence in society), but also behavioral inten-
that are part of this type of racism are rooted in an abstract tions (intentions to engage in protective behavior). Their
system of moral values and ideals. Henry and Sears con- findings indicated that amount of television viewing direct-
ceptualized symbolic racism on four dimensions focusing ly influenced prevalence estimates of violence in society as
on more contemporary racism issues: (a) work ethic and well as intentions to take protective measures. Because the
responsibility for outcomes (i.e., the sense that foreigners’ present study has a different main focus and because there
failure to progress comes from their unwillingness to work is a lack of theoretical knowledge concerning this variable
hard enough), (b) excessive demands (i.e., the sense that in the context of newspapers, another explorative research
foreigners are demanding too much), (c) denial of continu- question was formulated.
ing discrimination (i.e., the sense that foreigners no longer
face much prejudice in society today), and (d) undeserved RQ3: Is there a cross-sectional correlation between expo-
advantage (i.e., the sense that foreigners have received sure to a newspaper (Krone) and behavioral intentions (to
more than they deserve). Following the argumentation of vote in favor of Austria exiting the EU), and does exposure
Reimer and Rosengren (1990, p. 199), the effects of news- have a causal influence?

Journal of Media Psychology 2010; Vol. 22(4):147–159 © 2010 Hogrefe Publishing


F. Arendt: Cultivation Effects of a Newspaper 151

Method Frequency of Foreigners

Overview Participants were asked: “What percentage of all people


who are suspected of having committed an offense are for-
Most cultivation studies are based on a cross-sectional de- eigners?” They had the following choices for answering
sign. By definition, correlations cannot be interpreted caus- this question: “27, 29, 31, 33, 35, 37, 39, 41, 43.” The first
ally without additional theoretical assumptions. The cross- number, 27, reflected the reality answer (as indicated by
lagged panel design (Burkholder & Harlow, 2003; Kenny, official statistics), and the last number, 43, reflected the
1975; Reinecke, 2005) with two waves, which is used in Krone answer (as indicated by the content analysis of Study
the present study, allows more confident causal inferences. 1). When the survey questions were formulated, it was un-
clear whether the proportion of foreigners was stable above
60 percent during the whole period of two months. Because
Participants of that, the Krone answer was chosen to be lower than it
actually was. It should be noted that there is an ongoing
A total of 453 students enrolled in communication courses debate about the most accurate question format (Hetsroni,
at the University of Vienna, Austria, participated in this 2007; Rossmann, 2008, p. 36).
study. A time lag of two months for the two-wave panel
design was chosen for economic reasons (because of the Frequency of Serbs
length of the semester). 73.8% were female. The partici-
Participants were asked: “What percentage of all people
pants ranged in age from 18 to 36 years (M = 21.15, SD =
suspected of having committed an offense are Serbs?”
2.34). 262 students participated in the study both at Time
They had the following choices for answering the question:
1 and Time 2. The panel mortality was due to some students
“3, 4, 5, 6, 7, 8, 9, 10, 11.” The first number, 3, reflected
not attending courses at both times. 42.9% of participants
the reality answer, and the last number, 11, reflected the
at Time 1 and 43.8% of participants at Time 2 read Krone.
Krone answer.

Frequency of Hungarians
Materials
Participants were asked: “What percentage of all people
The materials consisted of a questionnaire, a 13-page taped suspected of having committed an offense are Hungari-
booklet that contained the paper-format Implicit Associa- ans?” They had the following choices for answering the
tion Test (IAT, Greenwald, McGhee, & Schwartz, 1998; question: “1, 2, 3, 4, 5, 6, 7, 8, 9.” The first number, 1,
Lemm, Lane, Sattler, Khan, & Nosek, 2008) on the first six reflected the reality answer, and the last number, 9, reflect-
pages, and a survey on the following seven pages. The same ed the Krone answer.
questionnaire was used at both times.
Symbolic Racism
Because of space limitations in the questionnaire, each di-
Measures
mension of symbolic racism was assessed with one item,
which could be rated on a seven-point scale from totally
Krone Exposure. Exposure was conceptualized as a latent
agree to totally disagree: (a) “Foreigners work as hard as
variable. Data for four manifest variables were collected:
Austrians,” (b) “Discrimination against foreigners is no
1. Bipolar measure (question: “How often do you read Kro-
longer a problem in Austria,” (c) “Foreigners are demand-
nen Zeitung?”; answer: seven-point bipolar scale from nev-
ing too much from the rest of society” and (d) “Foreigners
er to every day); 2. Days-per-week measure (question:
have received too many rights, but must not attend to a
“How many days a week do you read Kronen Zeitung?”);
comparable amount of duties.” The outcome of two com-
3. Minutes-per-week measure (question: “How many min-
mon factor analyses (one for Time 1 and one for Time 2)
utes a day do you read Kronen Zeitung?”; calculation: min-
was one factor at both times, which explained 57.83% of
utes-per-day multiplied by days-per-week); 4. Last-14-
the variance at Time 1 and 58.90% of the variance at Time
days measure (question: “On how many of the last 14 days
2. Correlation analyses were performed with the factor
did you read Kronen Zeitung?”). The outcome of two com-
scores of these latent variables. The composite reliability
mon factor analyses (one for Time 1 and one for Time 2)
indicated a reliable measurement at both times (Time 1:
was one factor at both Time 1 and Time 2, which explains
.87, Time 2: .86).
most of the variance of the indicator variables (Time 1:
85.77%, Time 2: 81.80%). Correlation analyses were per-
Explicit attitudes
formed with the factor scores of these latent variables. The
composite reliability (i.e., construct reliability; see Dia- The generalized attitude measure from McCroskey and
mantopoulos & Siguaw, 2007, p. 90) indicates a highly re- Richmond (1996) was used. This seven-point bipolar scale
liable measurement at both times (Time 1: .96, Time 2: .96). has six items (good-bad, positive-negative, beneficial-

© 2010 Hogrefe Publishing Journal of Media Psychology 2010; Vol. 22(4):147–159


152 F. Arendt: Cultivation Effects of a Newspaper

harmful, fair-unfair, wise-foolish, right-wrong). The out- explicit attitude measure (good-bad, wrong-right, harmful-
come of two common factor analyses (one for Time 1 and beneficial, fair-unfair). Words related to Austria were “Vi-
one for Time 2) was one factor at both times, which ex- enna,” “Falco” (a celebrated Austrian musician), “Ste-
plained 69.25% of the variance at Time 1 and 69.29% of phansdom” (a famous Austrian cathedral), and “Red-
the variance at Time 2. Correlation analyses were per- White-Red” (the colors of the Austrian flag). Words related
formed with the factor scores of these latent variables. The to EU were “monetary union,” “Brussels,” “member
composite reliability indicated a highly reliable measure- states,” and “entry talks.” Arendt’s (2009) study showed
ment at both times (Time 1: .92, Time 2: .94). that the participants were able to categorize the word stim-
uli correctly. The design of the IAT was constructed accord-
Implicit attitudes ing to the procedure described by Greenwald et al. (1998).
Each participant completed the test for the incongruent
One method used to measure implicit attitudes is the IAT condition before the test for the congruent condition. The
(Greenwald et al., 1998). The implicit attitudes variable IAT score was calculated using the following formula: (A –
was measured with a paper-format IAT, which is a validat- B)/(A + B), where A represents the performance in the con-
ed method for measuring implicit attitudes (Lemm et al., gruent condition and B represents performance in the in-
2008). The same paper-format IAT was used by Arendt congruent condition.1 The higher the score, the more com-
(2009). Figure 1 illustrates an example. Participants had to patible are the combinations “EU” with “negative” and
classify words (e.g., good, bad, Vienna, Brussels) into four “Austria” with “positive” (than the combinations “EU”
categories (EU, Austria, positive, negative). Two of those with “positive” and “Austria” with “negative”). Thus, the
categories were concepts (EU, Austria) and two were eval- higher the score, the more negative is the automatic, affec-
uative attributes (positive, negative). There were two im- tive reaction toward the EU.

Free Associations
In a first step, participants were asked to write down up to
five spontaneous associations to the EU. In a second step,
they were asked to evaluate the valence of these associa-
tions on a seven-point scale ranging from very positive to
very negative (participant-as-coder method, see Quick &
Stephenson, 2008). The mean of the evaluations was used
as the free associations measure.

Figure 1. Illustration of a paper-format Implicit Associa- Behavioral intentions


tion Test.
Participants were asked whether they would vote in favor
of Austria exiting the EU if there would be an election held
portant sequences. In the first combined task (congruent on the following Sunday (possible answers: yes or no).
condition), EU was paired with negative and Austria was
paired with positive. For example, if the word “Vienna”
was presented in the middle, participants had to classify it Procedure
as belonging to the combination “Austria or positive.” In
the second combined task (incongruent condition), Austria Due to the cross-lagged panel design of this study, data
was paired with negative and EU was paired with positive. from the participants were collected on two occasions with
If the word “fair” was presented, participants had to clas- a time-lag of two months (March and May 2009). This is
sify it as belonging to the combination “EU or positive.” a relatively short time-lag, and was chosen for economic
The essential assumption is that categorization should be reasons. Nevertheless, the results could be considered ac-
faster when the pairing of a concept with an attribute re- curate because it is known that a cultivation effect can oc-
flects a stronger memory association. Thus, a person who cur in such a short time (Rössler & Brosius, 2001). Partic-
has a negative automatic affective reaction toward the EU ipants completed the IAT in groups of approximately 20
(compared to Austria), should more quickly categorize people. The male experimenter informed them that they
word stimuli when EU and negative are paired together as would have to classify words into categories. The newspa-
a category as opposed to when EU and positive are paired. per Krone was not mentioned in order to avoid a systematic
Performance was defined as the number of correct discrim- influence (Shanahan & Morgan, 1999; Shrum, 2009). The
inations in each condition. The words were taken from the participants were then given the questionnaire with the IAT

1 This formula was chosen because it was assumed that this algorithm performs better than the algorithm recommended by Lemm et al. (2008).
It must be noted that the two algorithms produce similar values, which correlate almost perfectly (see Arendt, 2009).

Journal of Media Psychology 2010; Vol. 22(4):147–159 © 2010 Hogrefe Publishing


F. Arendt: Cultivation Effects of a Newspaper 153

Table 1. Intercorrelations (Pearson) between Krone exposure and crime-related variables


Variable 1 2 3 4 5 Range M t1 SD t1
Time 1 (n = 339–392, above the diagonal), Time 2 (n = 271–392, under the diagonal)
1. Krone Exposure – .18** .18** .13** .21** standard-
ized
2. Frequency Foreigners .17** – .50** .30** .42** 1–9 4.16 2.33
3. Frequency Serbs .14* .53** – .52** .30** 1–9 4.19 2.21
4. Frequency Hungarians .10* .44** .55** – .19** 1–9 4.11 2.00
5. Symbolic Racism .27** .39** .32** .15** – standard-
ized
M t2 3.75 3.94 3.89
SD t2 2.24 2.12 2.05
Correlations between Time 1 (rows) and Time 2 (columns) (n = 250–264)
1. Krone Exposure .83** .24** .19* .11* .27**
2. Frequency Foreigners .18** .59** .44** .30** .42**
3. Frequency Serbs .19** .37** .62** .37** .32**
4. Frequency Hungarians .21** .26** .38** .50** .15**
5. Symbolic Racism .28** .45** .30** .14** .80**
Note. p < .05, ** p < .01, one-tailed.

and the survey. The experimenter had a stopwatch, which cremental index of fit). Models with RMSEA values of less
was visible to the participants, to measure the time partic- than .05 are indicative of good fit, between .05 and under
ipants had for each condition. For each combined task, par- .08 of reasonable fit, between .08 and .10 of mediocre fit,
ticipants had 20 seconds. First, they took the test, then they and > .10 of poor fit (Diamantopoulos & Siguaw, 2007,
filled out the survey questions in the following order: ex- p. 85). CFI-values above .90 are desired (Holbert & Ste-
plicit attitudes, behavioral intentions, free associations, fre- phenson, 2008). Additionally, a power analysis was per-
quency of foreigners, frequency of Serbs, frequency of formed (MacCallum, Browne, & Sugawara, 1996). Be-
Hungarians, symbolic racism, Krone exposure. At Time 1, cause of the high statistical power of the models, a model
the experimenter told the participants that he would see cannot be considered bad with regard to the model-fit only
them again in two months. At Time 2, the experimenter because of a significant p-value. The probability of a Type
gave a debriefing after the data were collected. The data I error was maintained at .05 for all subsequent analyses.2
collection took approximately 15 minutes each time. Be-
cause Krone was not emphasized at Time 1, it is assumed
that data collection regarding Krone exposure at Time 1 had Results
no systematic influence on the dependent variables two
months later. Hypothesis 1 predicted a positive cross-sectional correla-
tion between the amount of exposure to Krone and first-or-
der measures. As Table 1 shows, this hypothesis was con-
Statistical Analysis firmed. Those who spent more time reading Krone were
more likely to overestimate the real-world incidence of for-
The correlational analyses were performed with SPSS for eigners, particularly Serbs and Hungarians, as people who
Windows (version 15; SPSS Inc, Chicago, IL, USA). To were suspected of having committed crimes. The values of
test the causal influence of Krone exposure, two structural the Pearson-coefficients did not change substantially if
equation models (SEM) were calculated with Amos (ver- third variables (age, gender, nationality, rural/urban) were
sion 7.0; Amos Development Corporation, Spring House, controlled for using the partial correlation technique (not
PA, USA). According to Holbert and Stephenson (2008), shown in the table, all p’s < .01).
three indices are reported: 1. χ²-distributed test statistic ac- Hypothesis 2 assumed a positive cross-sectional corre-
companied by degrees of freedom, the sample size, and the lation between Krone exposure and the second-order mea-
p-value; 2. RMSEA (absolute index of fit); and 3. CFI (in- sure. As Table 2 shows, this hypothesis was confirmed. The

2 Some of the greatest misuses of structural equation modeling occur within the model respecification phase. This process should not be
approached with blind adherence to the data (Holbert & Stephenson, 2008). If there is only a data-driven model modification, the modified
model, while demonstrating an excellent fit in one particular sample, may fail when applied to a different sample. According to Diamanto-
poulos and Siguaw (2007, p. 102), model modifications should be firmly resisted unless a clear interpretation can be offered. Because of
that reason, no model respecification will be made until the a priori specified model is absolutely inadequate.

© 2010 Hogrefe Publishing Journal of Media Psychology 2010; Vol. 22(4):147–159


154 F. Arendt: Cultivation Effects of a Newspaper

Table 2. Intercorrelations (Pearson) between Krone exposure and EU-related variables


Variable 1 2 3 4 5 Range M t1 SD t1
Time 1 (n = 340–386, above the diagonal), Time 2 (n = 274–315, under the diagonal)
1. Krone Exposure – .23** .11* .14** .22** standard-
ized
2. Explicit Attitudes .27** – .18** .58** .47** standard-
ized
3. Implicit Attitudes .16* .28** – .10* .05 (–.5)–.5 .06 .14
4. Free Associations .17** .58** .18* – .28** 1–7 3.24 1.19
5. Behavioral Intentions .19** .48** .11* .32** – 0–1 .14 .34
M t2 .03 3.24 .10
SD t2 .13 1.17 .30
Correlations between Time 1 (rows) and Time 2 (columns) (n = 250–264)
1. Krone Exposure .83** .29** .24* .17* .17**
2. Explicit Attitudes .30** .77** .26** .41** .41**
3. Implicit Attitudes .15* .16** .35** .07 –.03
4. Free Associations .17** .50** .10 .55** .21**
5. Behavioral Intentions .29** .41** .13* .24** .50**
Note. * p < .05, ** p < .01, one-tailed

more someone read Krone, the more negative were that sure on explicit attitudes. This is illustrated in Figure 3 (in-
person’s explicit attitudes. After controlling for the same dicated by the cross-lag between Krone exposure at Time
third variables as above, the coefficients still showed al- 1 and explicit attitudes at Time 2).
most the same values (not shown in the table, all p’s < .01). The third and last part of hypothesis 4 assumed a causal
Hypothesis 3 predicted a positive cross-sectional corre- influence of exposure on implicit attitudes. The data fully
lation between Krone exposure and implicit attitudes. As supported this assumption, as shown in Figure 3 (indicated
can be seen in Table 2, there is a positive correlation be- by the cross-lag between Krone exposure at Time 1 and im-
tween exposure and implicit attitudes. Controlling for the plicit attitudes at Time 2). Surprisingly, Krone exposure
same third variables as above reduced the coefficients, in- showed the strongest causal effect on implicit attitudes. In
dicating that at least a part of the relationship is due to a summary, hypothesis four was only partially confirmed.
third variable. Despite this fact, the coefficients had the Because the results of all of the cultivation hypotheses
right sign and had borderline significance at Time 1, have been reported, I will now present the results of the ex-
r(353) = .07, p < .10, and Time 2, r(280) = .09, p < .10. planatory data analyses for the research questions. Research
This leads to the conclusion that hypothesis 3 was only question one asked if there would be a cross-sectional rela-
partially supported. tionship between Krone exposure and symbolic racism. Table
The first part of hypothesis 4 assumed a causal effect of 1 shows that there was a significant correlation between
Krone exposure on first-order measures. A SEM was calcu- amount of exposure and symbolic racism. The second part of
lated to test this assumption. The model shows a mediocre fit Research Question 1 dealt with the question of whether Kro-
to the data (N = 247; χ² = 613.93; p < .001; power > .96; ne exposure had a causal effect on symbolic racism. This was
RMSEA = .099; 90% CI RMSEA [.090, .108]; CFI = .89). As not supported by the data. Moreover, symbolic racism had a
indicated in Figure 2, Krone only had a borderline significant small but significant effect on Krone exposure. Besides this
causal effect on the frequency estimate with regard to for- finding, the data in Figure 2 show that symbolic racism had
eigners (indicated by the cross-lag between Krone exposure a causal influence on the frequency estimate with regard to
at Time 1 and the foreigner variable at Time 2). The data did foreigners, and that the foreigner variable had a significant
not show a causal influence of Krone exposure on the fre- effect on the Serbs variable and the Hungarians variable.
quency estimate with regard to Serbs and Hungarians. Research Question 2 asked if there would be a cross-sec-
The second part of hypothesis 4 referred to a causal in- tional correlation between Krone exposure and the valence
fluence of exposure on the second-order measure. Because of the free EU associations, and if exposure would have a
the EU topic is not directly related to the crime topic, an- causal effect. As indicated by the data presented in Table
other SEM was calculated to test the second part of the 2, exposure correlated with the valence of the free associ-
fourth hypothesis. The model (N = 234; χ² = 624.68; p < ations. Those who read more Krone were more likely to
.001; power > .96; RMSEA = .075, 90% CI RMSEA [.067, remember negative EU associations. Figure 3 shows that
.083]; CFI = .92) showed a reasonable fit to the data. The there was only a borderline significant causal effect of ex-
data showed a borderline significant causal effect of expo- posure on free associations.

Journal of Media Psychology 2010; Vol. 22(4):147–159 © 2010 Hogrefe Publishing


F. Arendt: Cultivation Effects of a Newspaper 155

Figure 2. Cross-lagged panel analysis: first-order cultiva- Figure 3. Cross-lagged panel analysis: cultivation effects
tion. The figure shows the standardized coefficients. To on explicit attitudes, implicit attitudes, free associations,
keep the figure simple, we show only latent variables. All and behavioral intentions. The figure shows the standard-
links shown are significant, p < .05, except the dashed ized coefficients. To keep the figure simple, only latent
lines, which are borderline significant, .05 < p < .10. Mea- variables are shown. All links shown are significant, p <
surement model Krone exposure: all λ’s > .79 at Time 1 .05, except the dashed lines, which are borderline signifi-
and Time 2. The bipolar measure was used as the reference cant, .05 < p < .10. Measurement model Krone exposure:
variable. Measurement model Symbolic Racism: all λ’s > all λ’s > .79 at Time 1 and Time 2. The bipolar measure
.38 at Time 1 and Time 2. The comparatively small value was used as the reference variable. Measurement model
is because of only one item (denial of discrimination). All Explicit Attitudes: all λ’s > .53 at Time 1 and all λ’s > .67
other indicators show λ’s > .67. The work ethic measure at Time 2. The negative measure was used as the reference
was used as the reference variable. CR = composite reli- variable. CR = composite reliability; AVE = average vari-
ability; AVE = average variance extracted. ance extracted.

The third and last research question asked if there ingly, the data in Figure 3 show that behavioral intentions
would be a cross-sectional relationship between Krone had a small but significant causal effect on Krone expo-
exposure and behavioral intentions, and if exposure sure. Additionally, as indicated by the SEM in Figure 3,
would have a causal effect on that variable. The coeffi- within the dependent measures, explicit attitudes showed
cients in Table 3 show that those who read more Krone a causal influence on the rest of the EU-related dependent
were more likely to vote in favor of Austria exiting the variables and correlated significantly with implicit atti-
EU. There was no causal influence of exposure. Interest- tudes (also shown in Table 2).

© 2010 Hogrefe Publishing Journal of Media Psychology 2010; Vol. 22(4):147–159


156 F. Arendt: Cultivation Effects of a Newspaper

Discussion distinct ways in which people process social information


(Hogg & Vaughan, 2005, p. 68). On the one hand, people
This study aimed to gain insights into the cultivation effect can carefully and deliberately consider a lot of informa-
of newspapers and to advance cultivation theory and re- tion to build a judgment (i.e., systematic processing). On
search by using implicit attitudes as an additional dimen- the other hand, people can rely on very few pieces of
sion of the dependent variable. Data from a content analysis information (i.e., heuristic processing). In cultivation re-
search, this phenomenon is best addressed by Shrum
of one particular newspaper found that (a) foreigners, es-
(2009) in his heuristic processing model for first-order
pecially Serbs and Hungarians, were overrepresented as
effects. Two basic assumptions of this model are that (a)
people suspected of having committed crimes, and that (b)
people build their first-order judgments heuristically, and
the newspaper (Krone) had a very negative view toward the
that (b) the accessibility of a construct operates as a me-
EU. Based on these findings, it was assumed that those who
diating variable between amount of exposure and first-
spent more time reading this particular newspaper (a) were
order measures (i.e., individuals infer prevalence, such as
more likely to overestimate the frequency of foreigners,
the real-world incidence of doctors, from the ease of re-
particularly Serbs and Hungarians, as people suspected of
trieval of relevant examples). If people process informa-
having committed crimes (i.e., first-order cultivation), (b)
tion systematically, the cultivation effect is reduced be-
had more negative explicit attitudes toward the EU (i.e.,
cause people use other information than that which is
second-order cultivation), and (c) had more negative auto-
most accessible.
matic, affective reactions toward the EU (i.e., implicit cul-
There are some promising thoughts on a model for sec-
tivation). The data of the cross-lagged panel study with two
ond-order effects (Shrum, 2009). However, the amount of
waves supported these hypotheses with regard to the cross-
empirical evidence and theoretical knowledge is not com-
sectional relationships. Only the cross-sectional correlation
parable to that of first-order cultivation effects. Shrum con-
between Krone exposure and implicit attitudes failed to
ceptualized second-order cultivation as a form of persua-
achieve significance after controlling for third variables.
sion in the tradition of dual-process models such as the
However, the coefficients still showed the right sign and
Elaboration Likelihood Model (Petty, Brinol, & Priester,
achieved borderline significance (p < .10). It must be noted
2009). One additional hypothesis could be that automatic
that the magnitude of the coefficients was fully consistent affective reactions play a similar role in the second-order
with the magnitude observed in most of the other cultiva- cultivation process as construct accessibility does in the
tion studies (Shanahan & Morgan, 1999). first-order cultivation process. It is hypothesized that im-
Krone exposure had a small and only marginally sig- plicit attitudes operate as a mediating variable between
nificant causal influence on explicit attitudes and on the amount of exposure and second-order measures (i.e., indi-
overestimation of foreigners as people suspected of hav- viduals build their attitudes to some extent on their “feel-
ing committed crimes. Because Krone exposure showed ings” toward the attitude object). As in first-order effects,
no causal effect on the overestimation of Serbs and Hun- whether people construct their attitudes through heuristic
garians as suspected offenders, these empirical findings or systematic processing is important. That is, in forming
should be interpreted with caution. It seems that reading attitudes, people should rely more on their automatic affec-
Krone only affects the overestimation of a more general tive reactions when they process heuristically because it is
social category (i.e., foreigners) and does not affect spe- assumed that cognitive elaboration (i.e., during systematic
cific social categories (i.e., Serbs and Hungarians). This processing) affects the complexity of propositional think-
should be taken into consideration in future research. In ing. It does so by influencing the number of judgment-rel-
addition, the relatively stable personality trait symbolic evant propositions that people consider in addition to auto-
racism had a small causal influence on Krone exposure. matic affective reactions. To the extent that any of these
This finding supports the assumption that newspaper’s additional propositions are inconsistent with the automatic
contribution to conceptions of social reality is not a one- affective reactions, systematic processing is likely to re-
way, monolithic “push” process. Rather, the influence of duce the correlation between implicit and explicit attitudes.
a mass medium is subtle, complex, and intermingled with For example, if systematic processing identifies an addi-
other influences (Morgan et al., 2009). tional proposition (e.g., “There is no war in Europe because
The most interesting finding of the present study is that of the EU”) that is inconsistent with the propositional im-
newspaper exposure showed the strongest causal influ- plication of the automatic reaction (e.g., “I dislike the EU”),
ence on implicit attitudes. This is of major importance to expressed explicit attitudes should be more positive than
cultivation research. Thus, implicit attitudes should be without this additional proposition. It should be noted that
used as a dependent variable in future research. In addi- systematic processing could also increase the correlation
tion, implicit attitudes can also be conceptualized as a between implicit and explicit attitudes. This should be ob-
mediating variable: Research on the inferential processes servable if the additional proposition (e.g., “There is a lot
that people use to identify, sample, and combine infor- of corruption in the EU”) is consistent with the proposition-
mation to form impressions, and to make judgments and al implication of the automatic affective reaction (e.g., “I
guide behavior, has shown that there appear to be two dislike the EU”).

Journal of Media Psychology 2010; Vol. 22(4):147–159 © 2010 Hogrefe Publishing


F. Arendt: Cultivation Effects of a Newspaper 157

Although a lot of questions remain unanswered, this References


study presents empirical evidence that the concept of im-
plicit attitudes could advance cultivation theory and
Appel, M. (2008). Fictional narratives cultivate just-world be-
therefore should be considered in future research. Be- liefs. Journal of Communication, 58, 62–83.
cause the present investigation only shows a relationship Arendt, F. (2008). Zur kultivierenden Wirkung der Kronen Zei-
between newspaper exposure and implicit attitudes, fu- tung [Cultivation effect of Kronen Zeitung]. Medien Jour-
ture research should test the assumption that automatic nal, 32(4), 3–21.
affective reactions play a similar role in second-order cul- Arendt, F. (2009). Explizite und implizite kultivierende Wir-
tivation to the role construct-accessibility plays in first- kung der Kronen Zeitung. Eine empirische Untersuchung
order cultivation. von Kultivierungseffekten auf explizite und implizite politi-
There are several limitations to the present study. First, sche Einstellungen [The newspaper Kronen Zeitung’s long-
a cross-lagged panel design allows more confident causal term cultivation effects on explicit and implicit attitudes].
Medien & Kommunikationswissenschaft, 57, 217–237.
inferences. However, this does not mean that the design
Art, D. (2007). Reacting to the radical right. Lessons from Ger-
can prove a causal effect. For example, the higher cross-
many and Austria. Party Politics, 13, 331–349.
lag between newspaper exposure and implicit attitudes,
Barth, B. (1988). Fernsehnutzung und Realitätswahrnehmung:
which indicates a causal effect of newspaper exposure on Zur Überprüfung der Kultivierungshypothese [Exposure to
implicit attitudes, could partly be due to the different au- television and reality estimates: Testing the cultivation hy-
tocorrelations of newspaper exposure (.84) and implicit pothesis]. Rundfunk und Fernsehen, 36, 67–79.
attitudes (.31). Additionally, third variables could influ- Bauer, M. W. (2005). Distinguishing red and green biotechnol-
ence the findings. Unfortunately, the best method for ogy: Cultivation effects of the elite press. International
demonstrating causality, the experiment, has some limi- Journal of Public Opinion Research, 17, 64–89.
tations with regard to cultivation research, which focuses Berkel, B (2006). Political parallelism in news and commentar-
on long-term effects (Rossmann & Brosius, 2004). Sec- ies on the Haider-conflict. A comparative analysis of Aus-
ond, the sample in the present study was not representa- trian, British, German, and French quality newspapers.
tive. Because the cultivation effect has been repeatedly Communications, 31, 85–104.
demonstrated with representative samples (Shanahan & Bilandzic, H., & Rössler, P. (2004). Life according to televi-
sion. Implications of genre-specific cultivation effects: The
Morgan, 1999), including the cultivation effect of Krone
Gratification/Cultivation model. Communications, 29,
(Arendt, 2009), a sample of students was chosen for eco- 295–326.
nomical reasons. Third, the chosen time-lag of two Bryant, J., & Oliver, M. B. (Eds.). (2009). Media effects. Ad-
months is very short for the purpose of cultivation. The vances in theory and research (3rd ed.). New York, NY:
findings indicated that Krone exposure had an effect Routledge.
within the time-lag of two months. Future research Burkholder, G. J., & Harlow, L. L. (2003). An illustration of a
should investigate whether a causal effect of newspaper longitudinal cross-lagged design for larger structural equa-
exposure also has a cultivation effect with a different tion models. Structural Equation Modeling, 10, 465–486.
time-lag (e.g., several years). Fourth, future research Diamantopoulos, A., & Siguaw, J. A. (2007). Introducing LIS-
should use other topics when investigating implicit cul- REL. London, UK: Sage Publications.
tivation in order to enhance the generalizability of the Früh, W. (2007). Inhaltsanalyse. Theorie und Praxis. [Content
results. Fifth, the results cannot be generalized to televi- analysis. Theory and practice] (6th ed.). Konstanz, Germa-
ny: UVK Verlagsgesellschaft mbH.
sion. Cultivation researchers should also use implicit at-
Gawronski, B., & Bodenhausen, G. V. (2006). Associative and
titudes as a new dimension of the dependent variable
propositional processes in evaluation: An integrative review
when doing studies that focus on television. Sixth, data of implicit and explicit attitude change. Psychological Bul-
collection regarding Krone exposure at Time 1 could letin, 132, 692–731.
have had a systematic influence on the dependent vari- Gawronski, B., & Conrey, F. (2004). Der Implizite Assozia-
ables two months later. tionstest als Maß automatisch aktivierter Assoziationen:
Despite the limitations, one far-reaching consequence Reichweite und Grenzen. [The Implicit Association Test as
for cultivation research arises. The data indicate that an- a tool to measure automatic associations: Scope and limits].
other dependent variable should be considered: Cumula- Psychologische Rundschau, 55, 118–126.
tive exposure to a mass medium seems to influence the Gerbner, G., & Gross, L. (1976). Living with television: The
automatic affective reactions toward social objects. violence profile. Journal of Communication, 26, 173–199.
Grabe, M. E., & Drew, D. G. (2007). Crime cultivation: Com-
parisons across media genres and channels. Journal of
Broadcasting & Electronic Media, 51, 147–171.
Acknowledgment Greenwald, A. G., McGhee, D. E., & Schwartz, J. L. K. (1998).
Measuring individual differences in implicit cognition: The
implicit association test. Journal of Personality and Social
The author would like to thank Peter Vitouch, University Psychology, 74, 1464–1480.
of Vienna, for his helpful comments. Guo, Z., Zhu, J. J., & Chen, H. (2001). Mediated reality bites:

© 2010 Hogrefe Publishing Journal of Media Psychology 2010; Vol. 22(4):147–159


158 F. Arendt: Cultivation Effects of a Newspaper

Comparing direct and indirect experience as sources of per- of trait reactance and sensation seeking on perceived threat,
ceptions across two communities in China. International state reactance, and reactance restoration. Human Commu-
Journal of Public Opinion Research, 13, 398–418. nication Research, 34, 448–476.
Henry, P. J., & Sears, D. (2002). The symbolic racism 2000 Reber, B. H., & Chang, Y. (2000). Assessing cultivation theory
scale. Political Psychology, 23, 253–283. and public health model for crime reporting. Newspaper Re-
Hetsroni, A. (2007). Open or closed – this is the question: The search Journal, 21, 99–112.
influence of question format on the cultivation effect. Com- Reimer, B., & Rosengren, K. E. (1990). Cultivated viewers and
munication Methods and Measures, 1, 1–12. readers: A life-style perspective. In N. Signorielli & M. Mor-
Hicks, G., & Lee, T. T. (2004, May). Learning to be prejudiced? gan (Eds.), Cultivation analysis: New directions in media
Media usage and anti-gay attitudes. Paper presented at the effects research (pp. 181–206). Newbury Park, CA: Sage.
annual meeting of the International Communication Associ- Reinecke, J. (2005). Strukturgleichungsmodelle in den Sozial-
ation, New Orleans, LA. wissenschaften [Structural equation models in the social sci-
Hogg, M. A., & Vaughan, G. M. (2005). Social psychology. Ed- ences]. München, Germany: Oldenburg.
inburgh Gate, UK: Pearson Education Limited.
Rössler, P., & Brosius, H. B. (2001). Do talk shows cultivate
Holbert, R. L., & Stephenson, M. T. (2008). Commentary on the
adolescents’ views of the world? A prolonged-exposure ex-
uses and misuses of structural equation modeling in commu-
periment. Journal of Communication, 51, 143–163.
nication research. In A. F. Hayes, M. D. Slater, & L. B. Sny-
der (Eds.), The sage handbook of advanced data analysis Rossmann, C. (2008). Fiktion Wirklichkeit. Ein Modell der In-
methods for communication research (pp. 185–218). Thou- formationsverarbeitung im Kultivierungsprozess [Fiction re-
sand Oaks, CA: Sage. ality. An information processing model for the cultivation
Horvath, P. (2006). Die Feindbilder der Kronen Zeitung [The process]. Wiesbaden, Germany: VS Verlag.
“enemies” of Kronen Zeitung]. Medienimpulse, 14, 30–32. Rossmann, C., & Brosius, H. B. (2004). The problem of cau-
Kenny, D. A. (1975). Cross-lagged panel correlation: A test for sality in cultivation research. Communications, 29,
spuriousness. Psychological Bulletin, 82, 887–903. 379–397.
Lemm, K. M., Lane, K. A., Sattler, D. N., Khan, S. R., & Nosek, Shanahan, J., & Morgan, M. (1999). Television and its viewers.
B. A. (2008). Assessing implicit cognitions with a paper-for- Cultivation theory and research. Cambridge, UK: Cam-
mat implicit association test. In M. A. Morrison & T. G. bridge University Press.
Morrison (Eds.), The psychology of modern prejudice Shrum, L. J. (2007). Social cognition and cultivation. In D. R.
(pp. 123–146). Hauppauge, NY: Nova Science Publishers. Roskos-Ewoldson & J. L. Monahan (Eds.), Communication
MacCallum, R. C., Browne, M. W., & Sugawara, H. (1996). and social cognition: Theories and methods. (pp. 245–272).
Power analysis and determination of sample size for covari- Mahwah, NJ: Erlbaum.
ance structure modeling. Psychological Methods, 1, Shrum, L. J. (2009). Media consumption and perceptions of so-
130–149. cial reality: Effects and underlying processes. In J. Bryant
McCroskey, J. C., & Richmond, V. P. (1996). Fundamentals of & M. B. Oliver (Eds.), Media effects. Advances in theory and
human communication: An interpersonal perspective. Pros- research (pp. 50–73). New York, NY: Taylor & Francis.
pect Heights, IL: Waveland Press.
Stark, B., & Kraus, D. (2008). Crossmediale Strategien überre-
Morgan, M., Shanahan, J., & Signorielli, N. (2009). Growing gionaler Tageszeitungen. Empirische Studie am Beispiel des
up with television: Cultivation processes. In J. Bryant & Pressemarkts in Österreich [Cross-media strategies of news-
M. B. Oliver (Eds.), Media effects. Advances in theory and
papers. An empirical investigation in Austria]. Media Per-
research (pp. 34–49). New York, NY: Taylor & Francis.
spektiven, 38, 307–317.
Nabi, R. L., & Sullivan, J. L. (2001). Does television viewing
Steinmaurer, T. (2009). Diversity through delay? The Austrian
relate to engagement in protective action against crime? A
cultivation analysis from a theory of reasoned action per- case. International Communication Gazette, 71, 77–87.
spective. Communication Research, 28, 802–825. Strohman, C. A., & Seltzer, R. (1985). Media use and percep-
Nisbet, E., & Wang, Z. (2004, May). Polarization versus the tions of crime. Journalism Quarterly, 62, 340–345.
mainstream: Differentiating the effects of the news media. Verein Arbeitsgemeinschaft Media-Analysen. (2008). Media
Paper presented at the annual meeting of the International Analyse 2008 [Media Analysis 2008]. Retrieved from
Communication Association, New Orleans, LA. http://www.media-analyse.at
Petty, R. E., Brinol, P., & Priester, J. R. (2009). Mass media at- Vergeer, M., Lubbers, M., & Scheepers, P. (2000). Exposure to
titude change. Implications of the elaboration likelihood newspapers and attitudes toward ethnic minorities: A longi-
model of persuasion. In J. Bryant & M. B. Oliver (Eds.), Me- tudinal analysis. The Howard Journal of Communications,
dia effects. Advances in theory and research (pp. 125–164). 11, 127–143.
New York, NY: Taylor & Francis. Weimann, G. (2000). Communicating unreality: Modern media
Plasser, F., & Lengauer, G. (2010). Politik vor Redaktions- and the reconstruction of reality. London, UK: Sage.
schluss: Kommunikationsorientierungen von Macht- und
Wise, M. Z. (2002, January). Austria’s troubling tabloid. Co-
Medieneliten in Österreich [Orientation of media elites in
lumbia Journalism Review, 40(5), 12. Retrieved from:
Austria]. In F. Plasser (Ed.), Politik in der Medienarena.
http://www.michaelzwise.com/articleDisplay.php?article_
Praxis politischer Kommunikation in Österreich [Political
id=6
communication in Austria] (pp. 53–100). Vienna, Austria:
Facultas.
Quick, B. L., & Stephenson, M. T. (2008). Examining the role Date of acceptance: May 17, 2010

Journal of Media Psychology 2010; Vol. 22(4):147–159 © 2010 Hogrefe Publishing


F. Arendt: Cultivation Effects of a Newspaper 159

Florian Arendt is a scientific assistant at the


Department of Communication, University
of Vienna, Austria. His research interests
include media effects (especially cultiva-
tion research), media psychology, and re-
search methods.

Florian Arendt
Department of Communication
University of Vienna
Lammgasse 8/8
1080 Vienna
Austria
Tel. +43 1 42-774-9342
Fax +43 1 42-774-9366
E-mail florian.arendt@univie.ac.at

© 2010 Hogrefe Publishing Journal of Media Psychology 2010; Vol. 22(4):147–159

S-ar putea să vă placă și