can anyone do a 10 pages paper of the communication theory class?
ryoaga
3969718.pdf
Exposure to Newspapers and Attitudes toward Ethnic Minorities : A Longitudinal
Analysis
Maurice Vergeer Department of Methodology
University of Nijmegen The Netherlands
Marcel Lubbers and Peer Scheepers Department of Sociology University of Nijmegen
The Netherlands
This study aims to test whether exposure to certain newspapers leads people to perceive ethnic minorities as more threatening. Building on culti- vation theory as developed by Gerbner, this study extends his theory by looking at the possible eþects of exposure to newspapers.
Results indicate that exposure to a newspaper characterized by nega- tive reporting about ethnic crime leads people to perceive ethnic minor- ities as more of a threat than exposure to other newspapers does.
KEYWORDS ethnic minorities, mass media, cultivation theory
T here has been much debate in recent years about the presence of ethnic minor-
ities in West-European countries. Much of this debate can be attributed to the
increase in ethnic immigration in West-European countries, including the
Netherlands. After a sharp drop in 1981, the inýux of ethnic minorities consistently
increased from 10,000 per year in the period 1982–1985 to over 65,000 in 1994
(Centraal Bureau Voor de Statistiek [ CBS ], 1995, 1996). As in many other countries, there has been much social unrest in the Nether-
lands, especially about the temporary housing of ethnic minorities. More generally,
many indicators show that during this period the traditional norm of ethnic tolerance
has declined and, in turn, ethnic intolerance has grown. One of the indicators of this
growth in ethnic intolerance is the support the Dutch population expresses for ethnic
discrimination : This has increased since 1984. In both 1985 and 1989, about 31% of the Dutch population expressed support for ethnic discrimination. By 1992 this
support had increased to almost 35%, whereas by 1993 it exceeded 40% (Coenders
Address correspondence to Maurice Vergeer, Department of Methodology, Uni- versity of Nijmegen, P.O. Box 9104, 6500 HE Nijmegen, The Netherlands. E-mail : [email protected]
T he Howard J ournal of Communications, 11 :127–143, 2000
Copyright 2000 Taylor & FrancisÓ 1064-6175/00 $12.00 1 .00 127
128 M. Vergeer et al.
& Scheepers, 1998 ; Scheepers, 1996). In 1994 almost 40% of the voters agreed to restrict ethnic immigration, and almost 40% opposed the equal treatment of ethnic minorities in the labor market (Scheepers, Schmeets, & Felling, 1997).
Newspapers have given wide coverage to this recent inýux of ethnic minorities
and the increased intolerance. Quantitative analysis (Lubbers, Scheepers, & Wester,
1998) of all articles about ethnic issues published in three major Dutch newspapers
between 1990 and 1995 has shown that over time the number of these news reports of
ethnic minority issues has increased (see Figure 1). This trend accompanied the
increase in ethnic immigration. Our study focuses on readers’ perception of these
ethnic issues in three important daily newspapers in the Netherlands, namely, the
V olkskrant, the T elegraaf and the Gelderlander. The Volkskrant, a rather progressive
newspaper, has the country’ s third largest market share. The T elegraaf, a somewhat
conservative newspaper, has the largest market share. The Gelderlander is the largest
regional newspaper in the Netherlands.
Bovenkerk (1978) and Van Dijk (1991, 1993) used content analysis to show that
the news media tend to portray ethnic minorities in a somewhat unfavorable light.
News reports link ethnic minorities predominantly to problematic topics such as
crime, drugs, violence, and evictions (Van Dijk, 1983, 1987, 1991). Figure 1 conürms
these previous studies. It shows that in 1992 the number of newspaper articles about
ethnic crime increased sharply. The period from 1992 through 1994 is relatively
stable in terms of the number of articles published about ethnic crime. Then in 1995
there is a slight decrease.
Van Dijk has argued that, to maintain their power and that of the White major-
ity, White elites tend to favor a positive representation of their own group and a
negative representation of minorities (1993, p. 248). This means that they select
topics that characterize ethnic minorities negatively—news that deünes minorities as
Figure 1 Number of articles about ethnic crime per year (total and per newspaper). (Derived from Lubbers et al., Ethnic minorities in dutch newspapers 1990–1995 : Patterns of criminalization and prob- lematization, 1998).
Exposure to Newspapers 129
diþerent, problematic, deviant, or threatening (1993, p. 249). The news also tends to
present the topic of immigration as problematic. It draws attention to the large
numbers, immigration rackets, and ‘‘economic’’ refugees who are reported to be
‘‘living oþ our pockets’’ (p. 278). Also, ethnic minorities themselves hardly have a
voice in the news : Competent and critical minority experts, especially those special-
ized in ethnic aþairs, tend to be ignored, marginalized, discredited, or even attacked
in the White press (p. 254).
Van Dijk also detects diþerences between conservative and liberal newspapers.
The conservative newspapers take a less favorable approach toward ethnic minorities
than the liberal newspapers do. These ündings have led Van Dijk to conclude that
newspapers are responsible for a continued process of polarization between ‘‘us’’ and
‘‘them.’’ This, in turn, may lead readers to believe that ethnic immigration, and the
presence of ethnic minorities as such, pose a threat to society as a whole and to
certain social groups in particular.
Research Problem
Few studies on the nature of news reports about ethnic minorities researched the
consequences of exposure to these reports. In this article the authors continue and
build on this line of research. Bovenkerk (1978) and Van Dijk (1983, 1991) have
assumed a relationship between media messages and readers’ attitudes. However, it
has yet to be established whether exposure to this kind of ‘‘ethnic news’’ does indeed
have a detrimental eþect on people’s attitudes toward ethnic minorities. This study
sets out to test this crucial assumption of whether being exposed to newspaper articles
about ethnic minorities is related to unfavorable attitudes toward ethnic minorities.
In other words, to what extent exposure to newspapers is associated with people’s
perceptions of ethnic threat will be investigated. Also tested will be whether this eþect
is constant over time, or whether it increases or decreases over the years as the media
coverage of ethnic issues increases or decreases. The following research questions will
be answered :
1. Do people exposed to diþerent newspapers diþer in the extent to which
they perceive ethnic minorities as a threat ?
2. To what extent can people’s perception of ethnic threat be attributed
solely to their exposure to newspapers, and to what extent can it be
attributed to other social forces such as their own sociostructural charac-
teristics and political orientation ?
3. Does the relationship between exposure to newspapers and the perception
of ethnic threat vary over the years ?
Theories about Media Exposure and the Perception of Ethnic Threat
Our ürst question is a speciüc formulation of an old question : Does being
exposed to media messages aþect people’s attitudes ? One important approach to the
study of media inýuences is Gerbner’s theory of cultural indicators (1969, 1973). This
theory consists of a three-part analysis. First is the analysis of institutional processes,
which focuses on the manner in which media institutions operate and conduct their
130 M. Vergeer et al.
business. Second is the analysis of the message system that focuses on the media
institutions’ products, namely, their messages. And third comes the cultivation
analysis. This is perhaps the most important analysis because it focuses on the impli-
cations of the previous two. The cultivation analysis investigates the inýuence of
media messages on people and society as a whole. The most general hypothesis that
can be derived from cultivation theory is that exposure to television in general aþects
viewers’ attitudes in such a way that these conform to the opinions and messages
expressed on television. In other words, cultivation theory states that exposure inýu-
ences people to perceive the world in a certain manner.
Cultivation theory has been criticized heavily (Potter, 1993, 1994 ; Stilling, 1997 ;
Van Den Bulck, 1995). Potter (1993, 1994) has pointed out several methodological
ýaws generally ignored in previous research. For example, in earlier studies the mea-
surement of exposure to television often was very crude : It was categorized as either
light or heavy. Furthermore, some authors (Hirsch, 1980 ; Potter, 1993) questioned
whether exposure to television was the cause or the eþect or both of some societal
phenomenon. Also, the empirical tests did not include other relevant concepts that
might establish the extent to which the relationship between television exposure and
measures of cultivation was in fact spurious.
Recent research (Vergeer, Rutten, & Scheepers, 1996 ; Vergeer & Scheepers,
1998) implemented the recommendations of these critics. It has shown that being
exposed to television leads people to perceive ethnic minorities as a threat. In con-
trast to earlier research, however, these studies have found that it is not general
exposure to television that makes people perceive ethnic minorities as more of a
threat. Exposure to üctional and entertainment programs (Vergeer et al., 1996), and
to commercial television stations (Vergeer & Scheepers, 1998) in particular, has this
eþect. Even after controlling for relevant variables, those that test for spuriousness,
these relationships persist. An additional test performed to gauge the opposite
relationship, namely, the inýuence of people’s perception of ethnic threat on their
exposure to television, rendered a negligible eþect. In other words, when controlling
for this competing opposite relationship, we found even stronger support for cultiva-
tion theory.
Like television, newspapers are an important source of information. Our contri-
bution in this article will be to build on the cultivation theory of media exposure.
Previous content analyses (Bovenkerk, 1978 ; Van Dijk, 1991, 1993) show that news-
papers depict ethnic minorities as a general threat to society. While not intended as
such, these earlier analyses function as the message system analyses that Gerbner and
his associates promoted. The next step is to establish whether the ündings that expo-
sure to certain television programs and television stations aþects people’s attitudes
toward ethnic minorities is also true for newspapers. Applying Gerbner’s most crucial
hypothesis to newspapers, one would expect that exposure to newspapers aþects
people’s perception of ethnic threat. But such a general proposition may be too
crude. Like television stations and television programs, newspapers diþer greatly
(Van Dijk, 1993). In particular, newspapers diþer in the amount of articles they
publish about ethnic minorities. Just like exposure to diþerent television stations and
television programs aþects people’s perception of ethnic threat diþerently, one would
expect exposure to diþerent newspapers to have diþerent eþects on people’s percep-
tion of ethnic threat.
Exposure to Newspapers 131
Some newspapers may consistently pay more attention to ethnic minorities.
Some may emphasize the unfavorable characteristics of these ethnic minorities more
strongly than other newspapers do. Figure 1 shows that the T elegraaf publishes arti-
cles about ethnic crime more frequently than the Volkskrant and the Gelderlander do.
Previous analyses of three important daily newspapers in the Netherlands (Lubbers
et al., 1998) also have shown that particularly the T elegraaf often associates ethnic
minorities with threatening features such as crime. The Volkskrant, by contrast, associ-
ates ethnic minorities less often with threatening phenomena.
Gerbner (1969 ; Gerbner, Gross, Morgan, & Signorielli, 1986) posed that,
assuming a homogeneous media content, a general measure of exposure would
suffice. However, having ascertained that newspapers diþer, it would be ill-advised
not to take this into account : To suffice with a general measure of exposure could
obscure existing eþects, ones that could surface when using diþerentiated measures of
exposure (Potter & Chang, 1990 ; Rubin, Perse, & Taylor, 1988). Hence, we expect
readers of the T elegraaf to perceive ethnic minorities as more threatening than readers
of the V olkskrant do. This leads us to formulate the following hypothesis :
People who are exposed to the news coverage of the Volkskrant or the Gelderlander
perceive ethnic minorities as less of a threat than people exposed to news coverage of
the T elegraaf do.
Alternative Explanations for the Perception of Ethnic Threat
To answer our second question we turn to alternative explanations for the per-
ception of ethnic threat. Potter (1993, 1994) voiced one of the main criticisms of
cultivation theory, namely, that the relationship between media messages and atti-
tudes, in this case attitudes toward ethnic minorities, might be spurious. Potter has
suggested including other social indicators that might be related to these ethnic atti-
tudes in the analysis. Fortunately, existing theories about ethnic attitudes predict
which other factors might be related to the perception of ethnic threat.
Earlier sociological research has suggested that certain groups are more likely
than others to perceive ethnic minorities as threatening. Two theories shed light on
this process, namely, social identity theory and realistic conýict theory (Coenders &
Scheepers, 1998 ; Eisinga & Scheepers, 1989). According to social identity theory, as
developed by Tajfel (1981), every individual has an urge to belong to a social group.
Belonging to a group that provides values and norms allows the individual to com-
prehend his social reality. Groups that do not conform to the values and norms to
which the in-group adheres are evaluated negatively, while groups that hold similar
values and norms are evaluated positively. This process involves social identiücation,
the selective perception of favorable characteristics of one’s own group, and social
contraidentiücation, the selective perception of unfavorable characteristics of other
groups called out-groups. As a result of these mental processes, individuals are favor-
ably inclined toward their in-group and unfavorably inclined toward out-groups.
These two related attitudes, known as ethnocentrism, have been shown to correlate
positively (Billiet, Eisinga, & Scheepers, 1996 ; Pieterse, Scheepers, & Van Der Ven,
1991 ; Scheepers, Felling, & Peters, 1990 ; Smooha, 1987).
132 M. Vergeer et al.
Although these ündings have not been refuted empirically, they do not apply to
all people equally : People belonging to certain social categories may subscribe more
strongly to ethnocentrism than others. Realistic conýict theory (Blalock, 1967 ;
Levine & Campbell, 1972) suggests that, in acquiring scarce goods, social groups
have to compete with each other, which results in conýicting interests. Especially
people who believe they must compete with ethnic minorities tend to look for a
scapegoat whom they can hold accountable for their own threatened social position.
People who occupy underprivileged positions, similar to the ones occupied by many
ethnic minorities, feel more threatened by them than do people in privileged social
positions. These underprivileged groups are particularly likely to perceive the pres-
ence of ethnic minorities as threatening.
Research has shown that people who belong to less privileged social groups (for
example, less educated people, people with low-status occupations, and the
unemployed) tend to be the most ethnocentric. Between 1985 and 1990, ethnocen-
trism has increased among the less educated, as it did among the better educated and
those with high-status occupations (Scheepers, Eisinga, & Linssen, 1994). According
to realistic conýict theory, young people are more likely than older generations to feel
threatened by ethnic minorities because they are new to the labor market and thus
must compete with ethnic minorities. Research (Scheepers & Coenders, 1996) has
shown that during the nineties younger cohorts have indeed been more inclined to
support ethnic discrimination than have older cohorts. Dependents, particularly the
unemployed, constitute another underprivileged social group. Like young people,
they seek access to the labor market. However, to gain access to that labor market
they must compete with ethnic minorities. Given these ündings, we expect that a
person’s social class, education, and age will aþect his or her perception of ethnic
threat.
Political orientation also is considered to be important regarding attitudes
toward ethnic minorities in general and ethnic threat in particular. A conservative
orientation is characterized by an apprehension about political and social change
(Sniderman & Carmines, 1997). As a consequence, people who favor political conser-
vatism defend the ‘‘status quo’’ and consequently oppose societal changes, in particu-
lar those related to ethnic issues. Therefore, people who are politically conservative
may be more likely to reject and feel threatened by ethnic minorities.
Taking these factors into consideration allows one to ascertain the extent to
which the relationship between being exposed to certain newspapers and people’s
perception of ethnic threat is spurious. This helps to test our hypotheses more rigor-
ously, which is why a second hypothesis has been formulated :
Even after controlling for social class, education, age, and political orientation, the
eþects of being exposed to certain newspapers on people’s perception of ethnic threat
remain signiücant.
Longitudinal Trend
The frequency with which ethnic crime is reported diþers not only by newspaper,
but also by year. Particularly 1992 saw a rise in the news reporting ethnic crime.
This was the case for all three newspapers (see Figure 1). In total, the number of
Exposure to Newspapers 133
Figure 2 Diþerence in number of articles about ethnic minorities in news- papers (1990 = 100). (Derived from Lubbers et al., Ethnic minorities in dutch newspapers 1990–1995 : Patterns of criminalization and prob- lematization, 1998).
articles published between 1992 and 1994 was nearly 100% higher than in 1990. However, this increase in articles was not equal for all newspapers (see Figure 2). 1
In particular, the indexed diþerences (1990 5 100) between the T elegraaf and the Gelderlander are considerable : For all years the indexed diþerence is larger than
200, achieving its maximum of 548 in 1992. In the case of the T elegraaf versus the
Volkskrant, the indexed diþerences are most substantial for the years 1992, 1994, and
1995. We hypothesized earlier that exposure to a larger number of articles about
ethnic crime would be accompanied by a larger diþerence in perception of ethnic
threat. Therefore, larger diþerences between newspapers over the years, in terms of
number of articles, should be accompanied by larger diþerences in perception of
ethnic threat by people exposed to those newspapers over the years. Based on the
ündings presented in Figure 2, the third hypothesis can be formulated in the follow-
ing manner :
People exposed to the Volkskrant in 1992, 1994, and 1995 and people exposed to the
Gelderlander in 1992 through 1995 perceive less ethnic threat than people exposed to
the T elegraaf in those respective years.
Data and Measurements
We have used data from üve surveys conducted in the Netherlands between 1990
and 1995. The ürst survey involved a random sample of the Dutch population in
1990 (Eisinga, Felling, Peters, Scheepers, Schreuder, Konig, & Jacobs, 1992). The
other surveys, those conducted in 1992 through 1995, involved random samples of
the population of Nijmegen. Nijmegen is a Dutch city of approximately 150,000
134 M. Vergeer et al.
residents. Intensively trained interviewers collected the data using prestructured
questionnaires.
To examine the representativeness of each sample, we compared the samples
with the populations from which they were drawn. Although we found minor dis-
crepancies, the samples can be considered representative of their populations. All
dataüles have been well documented and submitted for secondary analyses to the
Dutch Steinmetz Archive (De Graaf, 1994 ; Scheepers, 1995 ; Te Grotenhuis &
Scheepers, 1995).
In spite of minor methodological problems, the data serve our purpose of estab-
lishing the changes over time in the relationship between exposure to newspapers and
people’s perception of ethnic threat. In fact, the data contain similar measurements
of the key variables included in our hypotheses.
Three items measure perceived ethnic threat. One item dealt with concern about
the level of education, the second item concerned the distribution of housing, 2 while
the third item dealt with concern about cutbacks on social security. 3 These items
were scored on a üve-point Likert continuum, ranging from ‘‘strongly disagree’’ to
‘‘strongly agree.’’ We applied a factor analysis to establish whether the items were
one dimensional. The internal consistency is quite acceptable according Cronbach’s
alpha (.76). To determine whether the factor structure is identical in all samples, we
used LISREL 8.30 (JoÈ reskog & SoÈ rbom, 1996). Using the multisample option, we found that this was indeed the case.
The goodness-of-ü t, measured by the root mean square error of approximation
(RMSEA), is .031. This allows us to conclude that the factor structure is identical in
every year. 4
We measured exposure to newspapers by asking respondents whether they read a
national newspaper regularly, and, if so, which newspaper(s). They could list a
maximum of three newspapers. We also asked respondents whether they read
regional newspapers. This resulted in a nominal variable that indicated the
newspaper(s) to which people had been exposed : the V olkskrant, the T elegraaf, the
Gelderlander, and ‘‘other’’ national newspapers.
Nearly 43% of the 2,661 respondents appear to read more than one newspaper. A particularly common combination is that of the Gelderlander and one of the national
newspapers ; 32.1% of the sample read this combination of papers. By including addi- tional categories of people exposed to more than one newspaper we are able to show
which eþects are unique to exposure to the T elegraaf, to exposure to the Volkskrant,
and to exposure to the Gelderlander. We combine a few categories of those exposed to
multiple newspapers into single categories to make sure each of the nominal cate-
gories contained enough respondents.
We certainly are aware of the fact that our measurement of exposure to news-
papers is both rather crude and rather meager. Including other features of exposure
to newspapers, like the time spent reading the newspaper or the speciüc content
elements of the newspaper, would certainly have increased the validity of these mea-
surements. Unfortunately, we know of no such extensive measurement included in
Dutch secondary longitudinal data. However, if we are able to detect a statistical
relationship between our crude measurement of exposure to newspapers and the per-
ception of ethnic threat, then this suggests that developing a more extensive measure-
ment of this relationship is a worthwhile endeavor for future research.
Exposure to Newspapers 135
To construct the variable social class, we used the EGP-classiücation devised by
Erikson, Goldthorpe, and Portocarero (1983). This nominal classiücation originally
consisted of 10 categories. We reduced these to üve categories to ensure enough cases
in each category. Furthermore, we added three categories of nonworking people so
that we ended up with an eight-category classiücation. We measured education by
asking respondents about the highest level of education they completed. Age was
determined by asking the respondents for their date of birth. We constructed the
variable year using the year in which the respondents participated in the survey. A
proxy for political orientation is the political party people would vote for in the event
of national elections for parliament. Because of the great number of political parties
in the Netherlands, we constructed four categories : left-wing political parties (PvdA,
CPN, D66, PSP, SP, ‘‘Groen Links’’), right-wing parties (VVD, CD), religious
parties (CDA, SGP, GPV, RPF), and other special interest political parties (e.g.,
elderly and environmental).
Method
To determine the strength of the relationship between exposure to newspapers
and attitudes toward ethnic minorities, we used multiple regression analysis. By ürst
demonstrating that there is a bivariate relationship between exposure and the culti-
vation concept and consequently adding variables belonging to the competing
theory, one can explicitly test and demonstrate whether the original relationship is
spurious.
Prior to the regression analysis, we performed various tests to ensure that impor-
tant technical assumptions (linearity, homoscedasticity, and interval measurement)
required for regression analysis were not violated. Because exposure to newspapers,
social class, education, and age did not satisfy one or more of these technical assump-
tions, we treated them as nominal dummy variables.
To ascertain whether the causal order of exposure to newspapers and perceived
ethnic threat should be altered, we would have liked to perform a structural equa-
tions analysis to model a reciprocal relation between exposure to newspapers and
perceived ethnic threat. However, because of the nominal measurement level of expo-
sure to newspapers, it is not possible to model this reciprocal relationship.
To determine the relative eþects of the predictors, we usually look at the stan-
dardized regression coefficients (b’s). To interpret the relationship between the
nominal predictor and the dependent variable, one must look at the unstandardized
regression coefficient (b’s) of the original dummy variables. However, the relative
eþect of a nominal variable cannot be determined in a straightforward manner. The
relative eþect of nominal variables is usually determined by constructing so-called
compound variables, based on the unstandardized regression coefficients of the
dummy variables (Eisinga, Scheepers, & van Snippenburg, 1991). One cannot inter-
pret the slope of this standardized regression coefficient because it is by deünition
positive and therefore not appropriate in the case of a nominal variable.
In summary, we will conduct the regression analysis in two phases. First, we test
whether people’s exposure to news coverage from certain newspapers is associated
with people’s perception of ethnic minorities as more threatening. This test answers
136 M. Vergeer et al.
our ürst question and the ürst hypothesis. Second, we test whether alternative expla-
nations provided by social identity theory and realistic conýict theory prove that the
original relationship is spurious. This test corresponds to our second question and
hypothesis. Next, we test whether the relationship we ünd between exposure to news-
papers and the perception of ethnic threat varies over time according to variations in
news coverage. This step answers our third question and is related to our third
hypothesis. To establish whether these eþects of exposure to newspapers vary over
time, we will perform a LISREL 8.30 analysis, using the multisample option. The
multisample option in LISREL allows us to analyze several mutually exclusive
groups simultaneously. Any degree of invariance in the model can be tested, ranging
from no constraints at all to the speciücation that all parameters are equal for all
groups. In our case, to test whether eþects of exposure to newspapers diþer over the
years, we will specify so-called equality constraints on all eþects of people’s exposure
to newspapers and their education, age, and social class (JoÈ reskog & SoÈ rbom, 1996). The criteria for the evaluation of the eþects are (1) modiücation indices larger than
üve for exposure to newspapers, (2) signiücant standardized residuals (larger than
1.96 or smaller than 2 1.96), and, if these criteria are met, (3) a change in the eþects as predicted by our third hypothesis. Furthermore, we use a number of model üt
indices to evaluate the model as a whole.
Results
In Table 1 we present the results of the multiple regression analyses.
According to our ürst hypothesis, exposure to the V olkskrant and the Gelderlander
leads people to perceive ethnic minorities as less threatening than exposure to the
T elegraaf does. Exposure to newspapers explains 16.4% of the variance (b 5 .405, signiücant). When we take a look at the unstandardized eþects, representing mean
diþerences as deviations from the reference category (Hardy, 1993), we see that
people exposed to the V olkskrant perceive ethnic minorities as less threatening than
those exposed to the T elegraaf do (b 5 2 2.911, signiücant). However, people exposed to the Gelderlander do not diþer from those exposed to the T elegraaf. Notice
that people exposed to other national newspapers also perceive ethnic minorities as
less threatening than people exposed to the T elegraaf do (b 5 2 .892, signiücant). Furthermore, people who are exposed to more than one newspaper perceive ethnic
minorities as less of a threat than those exposed only to the T elegraaf do. This is not
true for people exposed to both the Gelderlander and the T elegraaf. This last ünding is
not surprising since those only exposed to the Gelderlander do not diþer from those
exposed to the T elegraaf. The results from this ürst step of our test support our ürst
hypothesis with respect to the Volkskrant, however, not with respect to the Gelderlander.
To test whether this ürst hypothesis for the diþerence between the Volkskrant and
the T elegraaf must nevertheless be refuted, we introduce other relevant variables con-
cerning people’s sociostructural characteristics and political orientation into the
regression equation. This addition increases the percentage of variance explained
from 16.4 to 28.4, a highly signiücant increase. It turns out that education is the most
important of the three sociostructural characteristics (b 5 .235). Less-educated people (those who only completed elementary school, lower vocational school, or
Exposure to Newspapers 137
Table 1 Regression of Perception of Ethnic Threat on Exposure to Newspapers, Socio-Structural Characteristics, and Year (n = 2,298)
Hypothesis 1 2
Newspaper .405 .208 Telegraaf (ref) (ref) Gelderlander ( 2 .087) (.029) Volkskrant (- 2.911) (- 1.330) No newspaper (.151) (.249) Other newspaper(s) ( - .892) (- .510) Gelderlander 1 Volkskrant (- 2.539) (- 1.176) Gelderlander 1 Telegraaf (.212) (.238) Gelderlander 1 other (- 1.722) (- 1.057) Volkskrant and/or Telegraaf and/or other ( - 1.781) (- .788) Telegraaf 1 other (- 1.121) (- 1.098) Gelderlander and/or Volkskrant and/or Telegraaf and/or other (- 1.734) (- .804)
Education .235 Elementary school (1.558) Lower vocational school (1.895) Lower secondary school (1.073) Secondary vocational school (.788) O- and a-levels (.663) College (.427) University completed (ref)
Age .088 18–24 (ref) 25–29 (.040) 30–39 (.119) 40–49 ( 2 .051) 50–59 (.320) 60 1 (.638)
Social Class .108 Managers (ref) Routine nonmanual workers (.259) Self-employed (.818) Skilled manual workers (.417) Unskilled manual workers (.661) Dependents ( 2 .532) Students (.194) Housewives (.743)
Political Orientation .207 Left ( 2 .502) Right (1.043) Religious (.047) Other (ref)
R Square .164 .284 Adjusted R Square .160 .274 Increment of R Square .120
Coefficients printed bold indicates p , .05 (one-tailed).
138 M. Vergeer et al.
lower secondary school) perceive ethnic minorities as more of a threat than those
with a university degree (the reference category) do. People with an average level of
education (secondary vocational school and o- and a-levels) 5 perceive ethnic minor-
ities as more threatening than the university-educated people do but as less threat-
ening than the less-educated do. The eþects of age (b 5 .088) and social class (b 5 .108) are somewhat smaller than education. People older than 60 perceive ethnic minorities as more threatening than young people between the ages of 18 and
24 do (b 5 .638, signiücant). People who belong to üve diþerent categories of social class diþer signiücantly from managers (the reference category) : the routine non-
manual workers, self-employed, skilled manual workers, unskilled manual workers,
and housewives all perceive ethnic minorities as more of a threat than managers do.
Contrary to our expectation, dependents do not ünd ethnic minorities to be more
threatening than managers do : They are equal in the extent to which they perceive
ethnic minorities as threatening. Political orientation is of considerable importance
(b 5 .207). In line with the expectation, people who voted for right-wing political parties perceive ethnic minorities as more threatening than people who voted for
other parties (b 5 1.043). People who voted for left-wing or religious political parties did not perceive ethnic minorities as being more threatening than people who voted
for other parties do. This analysis provides evidence that will allow us to evaluate our
hypotheses. We ünd that, even after statistically controlling for alternative explana-
tions, the eþect of exposure to newspaper(s) remains statistically signiücant
(b 5 .208). Although the original eþect (b 5 .405) has been strongly reduced, we see that people exposed to the Volkskrant (b 5 2 1.330) still perceive ethnic minorities as less threatening than people exposed to the T elegraaf do. Therefore, neither the ürst
crucial hypothesis nor the second hypothesis can be refuted. 6
Table 2 shows the relative importance of exposure to newspapers and individual
sociostructural characteristics, as expressed in the percentage uniquely explained
variance. 7 The unique contribution of exposure to newspapers to the variance
explained in people’s perception of ethnic threat (11.62%) conürms the results of previous research (Vergeer et al., 1996 ; Vergeer & Scheepers, 1998). People’s indi-
vidual sociostructural characteristics and political orientation are responsible for
nearly 43% of the variance explained. Again, we conclude that these sociostructural characteristics and political orientation are more important than exposure to news-
papers but that the latter does contribute signiücantly to the total variance that our
model explains.
Regarding the third hypothesis whether people exposed to the Volkskrant and the
Gelderlander in speciüc years perceive ethnic minorities as less threatening as compared
Table 2 Uniquely Explained Variance
Uniquely explained % Uniquely variance explained variance
Total .284 Exposure to newspaper .033 11.62 Sociostructural characteristics
and political orientation .120 42.54 Not unique 45.84
Exposure to Newspapers 139
Table 3 Model Fit Indices
Df 108 Chi-square 178.54 Goodness of Fit Index (GFI) .99 Standardized Root Mean Square Residual (standardized RMR) .011 Root Mean Square Error of Approximation (RMSEA) .035 Comparative Fit Index (CFI) 1.00
with those who are exposed to the T elegraaf, we test whether the eþects of exposure to
newspapers varied over time. Therefore, we specify equality constraints on all eþects
for all years to test whether these eþects are invariant over time. The actual hypothe-
sis then states that the regression coefficient for a particular predictor in the year
1990 (t 5 0) is equal to the regression coefficients of that same predictor in 1992, 1993, 1994, and 1995 (i.e., bt/ 0 5 bt/ 1 5 bt/ 2 5 bt/ 3 5 bt/ 4).
To assess the quality of the model, Jaccard and Wan (1996) have recommended
using several indices of model üt. Table 3 shows that according to these indices we
have an excellent model.
Yet there are some modiücation indices larger than üve and some signiücant
standardized residuals. According to these modiücation indices of exposure to news-
papers, the regression model should be adjusted at one point. The modiücation index
of people exposed to the Gelderlander in 1990 indicates that releasing this constraint
improves the üt of the model signiücantly. The eþect would decrease slightly
( D b 5 2 .98), indicating that people exposed to the Gelderlander in 1990 perceive less ethnic threat than those exposed to the T elegraaf in 1990. This ünding so far supports
the hypothesis. However, the other predictions concerning exposure to the Volkskrant
do not support the third hypothesis : Larger diþerences in the number of published
articles between these newspapers do not lead to larger diþerences in perceived ethnic
threat. Eventually, we conclude that this longitudinal hypothesis is refuted.
Conclusions
The goal of our study has been to elaborate cultivation theory by applying it to
attitudes toward ethnic minorities. Although originally devised for television mes-
sages, we applied cultivation theory to the message system of newspapers. We found
that three diþerent Dutch newspapers diþer in the extent to which they reported
ethnic crime. The T elegraaf reported extensively about ethnic crime, particularly
when compared with the Volkskrant and the Gelderlander. Consequently, our ürst ques-
tion has been whether readers exposed to these diþerent newspapers diþer in the
extent to which they perceive ethnic minorities as threatening. This question resulted
in the crucial hypothesis that people exposed to the news coverage of the T elegraaf in
particular perceive ethnic minorities as more of a threat than people exposed to the
news coverage of other newspapers do. This hypothesis should in part be refuted.
People exposed to the T elegraaf indeed perceive ethnic minorities as more threatening
than people not exposed to the T elegraaf do, with the exception of people who are
exposed to the Gelderlander. They do not diþer from people exposed to the T elegraaf.
140 M. Vergeer et al.
The second test for this crucial hypothesis involves a second question, namely,
whether exposure to news coverage in newspapers is solely responsible for the extent
to which people perceive ethnic minorities as threatening. Drawing on two theories
that explain unfavorable attitudes toward ethnic minorities (social identity theory
and realistic conýict theory), we formulated a second hypothesis, according to which
the initial bivariate eþect of exposure to the news coverage by newspapers persists
even after controlling for social class, education, age, and political orientation. This
hypothesis also could not be refuted : Even after controlling for people’s socio-
structural characteristics, the eþect of their exposure to news coverage remains signiü-
cant. Finally, because both ethnic immigration and news coverage of ethnic crime
increased during the period from 1990 to 1995, we formulated a third question. We
asked whether the eþect of exposure to newspapers varied over the years.
Our hypothesis was that the diþerences between people who were exposed to, on
the one hand, the T elegraaf and on the other hand the Volkskrant or the Gelderlander in
the extent they perceive ethnic minorities as a threat changed over the years. This
hypothesis was refuted.
The ündings regarding the eþects of social characteristics are somewhat mixed.
As predicted by realistic conýict theory, less-educated people and people with low-
status occupations perceive ethnic minorities as more threatening than other people
do. However, some predictions derived from realistic conýict theory were refuted.
According to this theory, younger people and dependents perceive ethnic minorities
as more of a threat than others do. Our ündings, however, show that this is not the
case. People over 60, not younger people, perceive ethnic minorities as more threat-
ening. The composition of the population of the city Nijmegen might explain the
ünding that people who depend on some kind of outside ünancial support do not
diþer from managers in terms of their perception of ethnic threat. Because Nijmegen
is a university city, the number of well-educated dependents is relatively high. 8
Eventually, we have to conclude that cultivation theory is supported to some
degree. However, there are some ündings that cast some doubt upon aspects of the
theory. The results show that exposure to more newspaper messages does not always
lead to more perceived threat regarding ethnic minorities.
A remarkable result is that people exposed to more than one newspaper perceive
ethnic minorities as less threatening than others do. According to cultivation theory,
exposure to more media messages should lead to increased perception of ethnic
threat. However, this is not the case here : Exposure to more media messages from
diþerent newspapers, as compared with exposure to media messages from only one
newspaper, leads people to perceive ethnic minorities as less threatening. This means
a falsiücation of one particular aspect of cultivation theory. A possible explanation
for this ünding is that exposure to a larger number of diþerent opinions about ethnic
minorities, as articulated in diþerent newspapers, leads to a more sophisticated atti-
tude than does just being exposed to the messages of a single newspaper.
Furthermore, larger diþerences between newspapers over the years, in terms of
articles about ethnic issues, is not necessarily accompanied by larger diþerences in the
extent people perceive ethnic minorities as a threat.
A ünal remarkable ünding is that people without exposure to newspapers per-
ceive ethnic minorities as equally threatening as people exposed to the T elegraaf do.
This ünding also challenges cultivation theory (Hirsch, 1980). However, it is possible
Exposure to Newspapers 141
that the category of people who are not exposed to newspapers are more frequently
exposed to television, as suggested by a recent study (Sociaal en Cutureel Planbureau
[ SCP ], 1996, p. 385). Consequently, they derive their information from other sources such as üctional and entertainment programs. Earlier research (Vergeer et al., 1996)
has shown that the more people are exposed to üctional and entertainment television
programs, the more they perceive ethnic minorities as threatening. In general, future
research should focus on the interplay between exposure to newspapers on the one
hand and to television on the other. It is possible that people who depend exclusively
on television or on one newspaper are more ‘‘cultivated’’ (that is, inýuenced to per-
ceive ethnic minorities as more threatening). People who read more than one news-
paper and watch television are exposed to a larger variety of opinions than people
who rely on only one medium are.
Future research should not only focus on the eþects of exposure to one mass
medium. It should take into account also the cumulative eþects of being exposed to
several mass media simultaneously. Furthermore, the explanation of unfavorable atti-
tudes toward ethnic minorities should incorporate another important, often ignored,
source of information, namely, interpersonal contact with ethnic groups in the neigh-
borhood. Information based on these personal experiences might intensify or weaken
the eþects of exposure to media messages. Also, exposure to newspapers should be
measured on an interval level. This enables more precise analyses and it also allows
for testing whether a reciprocal relationship between exposure to newspapers and
perceived ethnic threat is plausible.
Notes
1 The larger the indexed diþerence, the larger the diþerences between on the one hand the T elegraaf and on the other hand the Volkskrant or the Gelderlander .
2 The Netherlands, being a densely populated country, has trouble providing enough and adequate housing for its population. Therefore, certain institutions regulate the distribution of rental houses for its population. The literal English translation of the item reads as follows : ‘‘Regarding the distribution of houses, people from ethnic minorities get their turn before Dutch people do.’’
3 The Netherlands has an elaborate system of social welfare beneüts. Social security in this context includes a wide variety of beneüts extended to the unemployed, the disabled, the aged, and so on.
4 To determine whether this model should be improved, we looked at the RMSEA. According to Browne and Cudeck (1993), a RMSEA of about .05 or less indicates a close üt. In our case, the small RMSEA of .031 indicates that our model üts well.
5 Obtaining one’s a- and o-levels indicates a certain level of education, roughly comparable to com- pleting high school.
6 Because people’s perception of ethnic threat might be caused by a more general sense of threat, we performed an additional test not reported in Table 1. We checked whether fear of and experience with crime had an additional eþect on the perception of ethnic threat. The results indicate that these factors have only slight signiücant eþects. Thus the results pertaining to exposure to newspapers still hold true.
7 The uniquely explained variance of a predictor variable equals the decrease in variance explained if the given predictor variable is deleted from the regression model.
8 After obtaining a university degree, former students may not directly ünd a job. In the mean time they still live in Nijmegen, which results in highly educated dependents.
142 M. Vergeer et al.
References
Billiet, J., Eisinga, R., & Scheepers, P. (1996). Ethnocentrism in the low countries : a comparative perspective. N ew Community, 22, 401–416.
Blalock, H. M. (1967). T oward a theory of minority g roup relations. New York : Wiley. Bovenkerk, F. (1978). Rassendiscriminatie bij misdaadverslaggeving in de pers [Racial discrimi-
nation in crime reports by the press ]. In F. Bovenkerk (Ed.), Omdat zij anders zijn : Patronen van rassendiscri- minatie in N ederland (pp. 119–135). Amsterdam, The Netherlands: Boom.
Browne, M. W., & Cudeck, R. (1993). Alternative ways of assessing model üt. In K. A. Bollen & J. Scott Long, T esting structural equation models (pp. 136–162). London : Sage.
Centraal Bureau voor de Statistiek [CBS ] (1995). Statistical yearbook 1994. Voorburg, The Nether- lands : CBS.
Centraal Bureau voor de Statistiek [CBS ] (1996). Statistical yearbook 1995. Voorburg, The Nether- lands : CBS.
Coenders, M., & Scheepers, P. (1998). Support for ethnic discrimination in the Netherlands 1979– 1993 : Eþects of period, cohort, and individual characteristics. European Sociological Review, 14, 405–422.
De Graaf, N. D. (1994). De sociale en culturele situatie van N ijmegen [The social and cultural situation of Nijmegen]. Amsterdam, The Netherlands: Steinmetz Archive.
Eisinga, R., Felling, A., Peters, J., Scheepers, P., Schreuder, O., Konig, R., & Jacobs, E. (1992). Religion in Dutch society. D ocumentation of a national survey on religious and secular attitudes in 1990. Amsterdam, The Netherlands: Steinmetz Archive.
Eisinga, R., & Scheepers, P. (1989). Etnocentrisme in N ederland : T heoretische en empirische modellen [Ethnocentrism in the Netherlands: Theoretical and empirical models ]. Nijmegen, The Netherlands: ITS.
Eisinga, R., Scheepers, P., & van Snippenburg, L. (1991). The standardized eþect of a compound of dummy variables or polynomial terms. Quality & Quantity, 25, 103–114.
Erikson, R., Goldthorpe, J. H., & Portocarero, L. (1983). Intergenerational class mobility and the convergence thesis : England, France and Sweden. British J ournal of Sociology, 34, 303–343.
Gerbner, G. (1969). Toward ‘‘Cultural Indicators’’ : The analysis of mass mediated public message systems. In G. Gerbner, O. Holsti, K. Krippendorf, W. J. Paisley, & P. Stone (Eds), T he analysis of communication content, developments in scientiüc theories and computer techniques (pp. 123–132). London : Wiley.
Gerbner, G. (1973). Cultural indicators : The third voice. In G. Gerbner, L. P. Gross, & W. H. Melody (Eds.), Communications technology and social policy (pp. 555–573). London : Wiley.
Gerbner, G. (1979). Television’s inýuence on values and behaviour. M assacommunicatie, 7, 215–222. Gerbner, G., Gross, L., Morgan, M., & Signorielli, N. (1986). Living with television : The dynamics
of the cultivation process. In J. Bryant & D. Zillmann (Eds.), Perspectives on media eþ ects (pp. 17–40). Hillsdale, NJ : Erlbaum.
Hardy, M. A. (1993). Reg ression with dummy variables. London : Sage. Hirsch, P. M. (1980). The ‘‘Scary World’’ of the nonviewer and other anomalies. A reanalysis of
Gerbner et al.’s ündings on cultivation analysis. Part I. Communication Research, 7, 403–456. Jaccard, J., & Wan, C. K. (1996). LISREL approaches to interaction eþ ects in multiple reg ression. London :
Sage. JoÈ reskog, K. G., & SoÈ rbom, D. (1996). LISREL 8 user’s reference g uide. Chicago : Scientiüc Software
International. Levine, R. A., & Campbell, D. T. (1972). Ethnocentrism , theories of conýict, ethnic attitudes and g roup
behavior. New York : Wiley. Lubbers, M., Scheepers, P., & Wester, F. (1998). Ethnic minorities in dutch newspapers 1990–1995 :
Patterns of criminalization and problematization. Gazette, 60(5), 415–431. Pieterse, H., Scheepers, P., & Van Der Ven, J. (1991). Religious beliefs and ethnocentrism, a com-
parison between the Dutch and the white church-going South-Africans. J ournal of Empirical T heology, 4, 64–85.
Potter, W. J. (1993). Cultivation theory and research. A conceptual critique. Human Communication Research, 19, 564–601.
Potter, W. J. (1994). Cultivation theory and research, a methodological critique. J ournalism M ono- g raphs (Serial No. 147).
Potter, W. J., & Chang, I. C. (1990). Television exposure measures and the cultivation hypothesis. J ournal of Broadcast and Electronic M edia, 34(3), 313–333.
Rubin, A. M., Perse, E. M., & Taylor, D. S. (1988). A methodological investigation of cultivation. Communication Research, 15, 107–134.
Scheepers, P. L. H. (1995). Sociaal-culturele ontwikkeling en in N ijmegen 1995. Amsterdam, The Nether- lands : Steinmetz Archive.
Scheepers, P. (1996). Etnocentrisme en rechts-extremisme. Reacties op de komst en aanwezigheid van etnische minderheden [Ethnocentrism and right wing extremism. Reactions to the arrival and pres- ence of ethnic minorities ]. In H. B. G. Ganzeboom & W. C. Ultee (Eds.), De sociale segmentatie van N eder-
Exposure to Newspapers 143
land in 2015. Voorstudies en achtergronden. Wetenschappelijke raad voor het regeringsbeleid (pp. 247–299). Den Haag, The Netherlands: SDU.
Scheepers, P., & Coenders, M. (1996). Trends in etnische discriminatie in Nederland 1980–1993 [Trends in ethnic discrimination in the Netherlands 1980–1993 ]. Verkenningen, 22(3), 8–25.J ustitieÈ le
Scheepers, P., Eisinga, R., & Linssen, E. (1994). Etnocentrisme in Nederland. Verandering bij kans- arme en/of gepriviligeerde Groepen? [Ethnocentrism in the Netherlands. Changes among underprivileged and privileged groups ? ] Sociologische Gids, 41, 185–201.
Scheepers, P., Felling, A., & Peters, J. (1990). Social conditions, authoritarianism, and ethnocen- trism : A theoretical model of the early Frankfurt School updated and tested. European Sociological Review, 6(1), 15–29.
Scheepers, P., Schmeets, H., & Felling, A. (1997). Fortress Holland ? Support for ethnocentric poli- cies among the 1994-electorate of the Netherlands. Ethnic and Racial Studies, 20, 145–159.
Smooha, S. (1987). Jewish and Arab ethnocentrism in Israel. Ethnic and Racial Studies, 10, 1–26. Sniderman, P. M., & Carmines, E. G. (1997). Reaching beyond race. Cambridge, MA : Harvard Uni-
versity Press. Sociaal en Cultureel Planbureau [SCP ]. (1996). Sociaal en cultureel rapport 1996 [Social and cultural
report 1996 ]. Rijswijk, The Netherlands: SCP. Stilling, E. A. (1997). The electronic melting pot hypothesis : The cultivation of acculturation among
Hispanics through television viewing. T he Howard J ournal of Communications, 8, 77–100. Tajfel, H. (1981). Human groups and social categ ories. Cambridge : Cambridge University Press. Te Grotenhuis, M., & Scheepers, P. (1995). D e sociaal-culturele situatie van de N ijmeegse Bevolking 1994
[The social-cultural situation of the population of Nijmegen 1994 ]. Amsterdam, The Netherlands: Stein- metz Archive.
Van Den Bulck, J. (1995). The selective viewer. Deüning (Flemish) viewer types. European J ournal of Communication , 10, 147–177.
Van Dijk, T. (1983). M inderheden in de media. Een analyse van de berichtg eving over etnische minderheden in de dag bladpers [Minorities in the media : An analysis of reports about ethnic minorities in the daily press ]. Amsterdam, The Netherlands: SUA.
Van Dijk, T. (1987). Communicating racism : Ethnic prejudice in thoug ht and talk. London : Sage. Van Dijk, T. (1991). Racism and the press. London : Routledge. Van Dijk, T. (1993). Elite discourse and racism. London : Routledge. Vergeer, M., Rutten, P., & Scheepers, P. (1996). Cultivatietheorie in een veranderd media-
landschap. Overzicht van eerdere studies en een toetsing voor een middelgrote stad [Cultivation theory in a changed media landscape. An overview of earlier studies and a test for a medium-sized city ]. T ijdschrift voor Communicatiewetenschap, 24, 120–151.
Vergeer, M., & Scheepers, P. (1998). Publieke versus commercieÈ le programmering : Eþecten van blootstelling aan televisie op subjectief ervaren etnische dreiging [Public and commercial programming : Eþects of exposure to television on perceived ethnic threat ]. In H. Verstraeten & El Schelfhout (Eds.), D e rol van de media in de multiculturele samenleving (pp. 271–292). Bruxelles, Belgium : VUBpress.
