media-party parallelism and its effects: a cross-national comparative study

This article was downloaded by: [Stony Brook University]On: 30 October 2014, At: 01:43Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registeredoffice: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Political CommunicationPublication details, including instructions for authors andsubscription information:http://www.tandfonline.com/loi/upcp20

Media-Party Parallelism and Its Effects: ACross-National Comparative StudyHetty van Kempen aa The Netherlands Institute for Social Research/ SCP. The study wasconducted as part of a PhD project at the Amsterdam School ofCommunications Research ASCoR, University of Amsterdam ,Published online: 09 Aug 2007.

To cite this article: Hetty van Kempen (2007) Media-Party Parallelism and Its Effects: A Cross-NationalComparative Study, Political Communication, 24:3, 303-320, DOI: 10.1080/10584600701471674

To link to this article: http://dx.doi.org/10.1080/10584600701471674

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the“Content”) contained in the publications on our platform. However, Taylor & Francis,our agents, and our licensors make no representations or warranties whatsoever as tothe accuracy, completeness, or suitability for any purpose of the Content. Any opinionsand views expressed in this publication are the opinions and views of the authors,and are not the views of or endorsed by Taylor & Francis. The accuracy of the Contentshould not be relied upon and should be independently verified with primary sourcesof information. Taylor and Francis shall not be liable for any losses, actions, claims,proceedings, demands, costs, expenses, damages, and other liabilities whatsoever orhowsoever caused arising directly or indirectly in connection with, in relation to or arisingout of the use of the Content.

This article may be used for research, teaching, and private study purposes. Anysubstantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &Conditions of access and use can be found at http://www.tandfonline.com/page/terms-and-conditions

http://www.tandfonline.com/loi/upcp20

http://www.tandfonline.com/action/showCitFormats?doi=10.1080/10584600701471674

http://dx.doi.org/10.1080/10584600701471674

http://www.tandfonline.com/page/terms-and-conditions

http://www.tandfonline.com/page/terms-and-conditions

Political Communication, 24:303–320, 2007Copyright © Taylor & Francis Group, LLCISSN: 1058-4609 print / 1091-7675 onlineDOI: 10.1080/10584600701471674

303

UPCP1058-46091091-7675Political Communication, Vol. 24, No. 3, Jun 2007: pp. 0–0Political Communication

Media-Party Parallelism and Its Effects: A Cross-National Comparative Study

Media-Party Parallelism and Its EffectsHetty van Kempen HETTY VAN KEMPEN

In 1974, Seymour-Ure introduced the concept of press-party parallelism into massmedia studies. Many agree on the importance of this concept for describing media sys-tems, but thus far little systematic comparative research of press-party parallelism andits possible effects has been conducted. The concept remained mainly theoretical. Toovercome this lacuna, the present article introduces a measure of the extent of media-party parallelism in the 15 countries of the European Union in 1999. The studyemploys survey data from the European Election Study 1999, and shows that media-party parallelism varies considerably between countries and that it structures citizens’political behavior such as electoral participation. The article shows that media-partyparallelism mobilizes citizens to vote, especially those who are not politicallyinterested.

Keywords Cross-national comparative research, electoral participation, EuropeanUnion, Media context, Media system, Press-party parallelism

In the study of political communication and its consequences, scholars have mainlystressed individual-level media exposure effects on political attitudes and behavior. Inmost studies, the media context has not been included, even though scholars seem to agreethat this context is important because it may structure individual behavior (Semetko, 1996).

This article seeks to contribute to the field of political communication by exploringone specific feature of the media context, namely the extent to which media-party parallel-ism is present. This concept was introduced by Seymour-Ure (1974), but it has notappeared in many comparative studies of political communication. The article proposes amethod of measuring media-party parallelism (MPP) and estimates the effect of MPP onindividual political behavior.

According to Seymour-Ure, press-party parallelism is present if ties exist betweennewspapers and political parties. It exists in its strongest form when each newspapersupports a party that is highly visible in the leader columns and in the editorial parts ofnewspapers—but sometimes also in the news items themselves. In the 1970s when Seymour-Ure developed the concept, parallelism existed mainly for the written press and not for TVnews, which aimed at large audiences. Nowadays, with the emergence of more (public andprivate) TV channels, this situation might have changed. For this reason, the article looks

Hetty van Kempen is a research associate at The Netherlands Institute for Social Research/SCP. The study was conducted as part of a PhD project at the Amsterdam School of Communica-tions Research ASCoR, University of Amsterdam.

Address correspondence to Hetty van Kempen, P.O. Box 16164, 2500 BD The Hague, TheNetherlands. E-mail: [email protected]

Dow

nloa

ded

by [

Ston

y B

rook

Uni

vers

ity]

at 0

1:43

30

Oct

ober

201

4

304 Hetty van Kempen

into parallelism for press and television news, which will be referred to as “media-partyparallelism.”

Moreover, the article investigates the consequences of media-party parallelism forpolitical behavior in a cross-national comparative setting. Exposure to partisan informa-tion may strengthen existing party-political preferences, thereby elevating the probabilityto vote for that party (Brynin & Newton, 2003; Lazarsfeld, Berelson, & Gaudet, 1944;Newton & Brynin, 2001). Consequently, turnout levels are expected to be higher in sys-tems where media-party parallelism is strong. Before moving to the analyses, however, Iwill first elaborate on the concept of media-party parallelism. Subsequently, I will discussits consequences for electoral participation.

Media-Party Parallelism and Its Consequences

In many Western European countries, press-party parallelism is a major feature of themedia system (Hallin & Mancini, 2004), but in some countries much more so than in oth-ers. For example, a high level of parallelism characterizes the Greek press system; Greeknewspapers are rooted in passionate ideological divisions, as reflected in the strong parti-san slant of the newspaper contents (Dimitras, 1997; Hallin & Mancini, 2004). In thesecircumstances, close ties exist between newspapers and political parties. In other situa-tions, rather general bonds exist between papers and parties or ideologies (Voltmer, 2000).This is the case in the Netherlands after the age of pillarization (Brants & Van Kempen,2002). In other countries, such as Ireland, press-party parallelism is virtually absent. InIreland political neutrality is the typical reporting style of newspapers, avoiding clearpolitical stances (Hallin & Mancini, 2004).

In their cross-national comparative study, Hallin and Mancini (2004) distinguish threesorts of media systems on the basis of four variables, one of which is press-party parallelism.They argue that press-party parallelism is strongest in the Mediterranean systems (Greece,Portugal, Spain, Italy, and France) and the Northern/Central European systems (Austria,Denmark, Belgium, Finland, Germany, the Netherlands, Norway, Sweden, and Switzer-land). In these countries, many of the newspapers are associated with particular parties orwith general political tendencies. Contrarily, press-party parallelism is virtually absent in theNorth Atlantic media systems (U.S., Ireland, and Canada, with Britain as an exception, inthat country, press-party parallelism is traditionally very strong). In these countries, catch-allnewspapers predominate, indicating that these papers try to appeal to a wide public acrosssocial divisions. Institutional ties to political parties are avoided, and papers attempt to main-tain balance and neutrality in their contents (Hallin & Mancini, 2004). Within this dichot-omy of countries (the North Atlantic countries on the one hand, and the Mediterranean andCentral/Northern European countries on the other), however, the nuances are missing. It islikely that press-party parallelism also varies between countries that belong to the same cate-gory in the dichotomy. This is what I will examine in the first part of this article.

The development of (partisanship in) a press system is strongly associated with theparty system. Seymour-Ure (1974, p. 159) explains this relationship as follows: “Thesame social forces that find expression in a party or parties of a political system tend tofind expression also through the press.” This clarifies why press-party parallelism is weakin countries such as the U.S. and Canada, where majoritarianism has resulted in catch-allparties with vague ideological identities (Hallin & Mancini, 2004). This “catch-all-ism” isalso reflected in the media contexts of these countries, with low levels of party parallel-ism. In most European countries, however, press-party parallelism is stronger (except for

Dow

nloa

ded

by [

Ston

y B

rook

Uni

vers

ity]

at 0

1:43

30

Oct

ober

201

4

Media-Party Parallelism and Its Effects 305

Ireland, where catch-all parties as well as catch-all media predominate), influenced by thepolitical system: A multiparty system typically coincides with strong partisanship of thepress (Hallin & Mancini, 2004).

In general, one might expect a decrease of partisanship in the press, as in most casesthe institutional bonds with political parties have vanished. Nevertheless, many bigger andsmaller papers still apply an ideological slant, which is most visible in newspaper editorials.As bias seems to be the rule in editorial pages, partisanship as such will not easily disap-pear, despite the fact that voters are becoming more and more party-politically dealigned.

As opposed to newspapers, television in Western Europe has typically developedautonomously from political parties (Semetko, 1996). This is partly due to political rulesand regulations about impartiality in broadcasting, and for another part to technologicaland economic constraints. With only one or two channels per country, television newsprograms mainly applied a catch-all format, aiming at a large audience that was not con-fined by distinct party-political preferences. The appearance of more (commercial) chan-nels in all European countries does not seem to have changed that tendency: Theavailability of more channels seems in most countries not to have resulted in diversificationon party-political grounds.

However, there are a few exceptions. In Italy, for example, the three public televisionnetworks have developed into three partisan channels since the end of the 1970s(Roncarolo, 2002). Such may be the case in other countries as well, and therefore I con-sider television news–party parallelism alongside press-party parallelism. This results in abroadening of the original concept by Seymour-Ure (1974) into media-party parallelism.Television news and papers do not capture the whole media spectrum; however, they areby far the most important news sources for most citizens.

What are the possible consequences of media-party parallelism? The partisan press, ormore generally, partisan media, can be instruments of political mobilization, which isexpressed in stronger party-political attitudes and higher participation levels. Rokkan andTorsvik (1970) argued on the basis of a Norwegian study that exposure to the partisan pressinvolves a process of two-way enforcement. On the one hand, a pronounced orientation towarda specific party results in a strong motivation to keep papers in the household that promote thatparty. On the other hand, regular exposure to such papers will provide arguments for maintain-ing the orientation and strengthen the commitment to the party. Hallin and Mancini (2004)similarly argue that exposure to the partisan press strengthens the bonds between citizens andparties. In other words, regular exposure to partisan news may reinforce party attachment andreduce ambiguity of political opinion building (Miller, 1991; Newton & Brynin, 2001; Norris,Curtice, Sanders, Scammel, & Semetko, 1999; Patterson, 1998; Voltmer, 2000).

This is in line with findings from early American studies, which found that people areinclined to read and listen to information that corresponds with their own political views,and that this information helps to crystallize and reinforce the vote choice (Berelson,Lazarsfeld, & McPhee, 1954; Lazarsfeld et al., 1944). As a result, citizens exposed toinformation that reinforces previously held party-political beliefs may be more likely toturn out. Newton and Brynin (2001) observed a strong relationship between party votingand reading a (specific type of) newspaper. They found that exposure to a paper of one’sown party outlook reinforces existing partisan attitudes and increases the likelihood tovote for that specific party. They found the same logic applies to voting turnout: Thosewho read a newspaper that matches their party-political preferences turn out in larger num-bers (Brynin & Newton, 2003).

I do not only expect this phenomenon of reinforcement to take place at the individuallevel, but at the aggregate (contextual) level as well. A strong partisan media context may

Dow

nloa

ded

by [

Ston

y B

rook

Uni

vers

ity]

at 0

1:43

30

Oct

ober

201

4


well also affect those who do not read a newspaper with links to their party, or those whodo not read a paper (or watch television news) at all. Most people have a social network offriends and relatives, and through interpersonal communication newspapers do not onlyaffect their own readers, but nonreaders as well, through two-step flow processes (Katz &Lazarsfeld, 1955). By these two-step flow phenomena, an effect that may appear at firstsight at the individual level may turn into a contextual effect. In that case, everybody ben-efits from the party-political information in the news media, independent of their exposureto it. This notion is supported by findings of a follow-up study which reported in VanKempen (2006), which shows that the effect of parallelism is the same for everybody, irre-spective of the amount of attention one pays to political news1. Hence, I expect higher par-ticipation levels if media-party parallelism is strong, irrespective of media exposure. Thisresults in the following hypothesis.

H1: Media-party parallelism has a positive effect on the level of voter turnout.

The effect of MPP is not necessarily the same for different groups. The cognitive hypoth-esis predicts that those who are not so much involved or interested in politics, and are nothighly educated, are most susceptible to media effects (Ha & McComles, 2004; Iyengar &Kinder, 1987; Iyengar, Peters, & Kinder, 1982; Krosnick & Kinder, 1990; McLeod,Becker, & Byrnes, 1973). Their argument repertoire (Cappella, Price, & Nir, 2002) issmaller, which makes them more susceptible to new information. Contrarily, people whoare much more involved into politics develop more resistance mechanisms through theinformation they have already obtained in the past. This indicates that a person who ishighly informed and has strong predispositions is better able to decide whether or not toaccept new information. The studies mentioned above focus on agenda-setting and prim-ing effects, but the cognitive hypothesis may also apply to the effect of partisan news onparticipation. Exposure to partisan news may increase identification and the likelihood toturn out especially for those who are not so much interested or involved in politics. Iexpect those who have high levels of political interest to be least susceptible to partisannews, whereas those with low levels of interest may be more prone to accept partisannews2. This leads me to expect the following.

H2: MPP has a larger effect on participation in elections for persons with lowlevels of political interest, than for persons with high levels of interest.

This hypothesis indicates a negative interaction effect of MPP and interest on turnout. Asargued before, MPP consists of press-party parallelism (PPP) and television news–partyparallelism (TPP). Nevertheless, I expect differences in the level of PPP and TPP. Due tothe partisan tradition of newspapers, and the catch-all tradition of television news (e.g.,Semetko, 1996), I expect PPP in general to be higher than TPP.

H3a: The level of press-party parallelism is higher than the level of televisionnews–party parallelism.

What are the implications of H3a for the effect of PPP and TPP on participation? The rein-forcement effect is likely to be stronger for newspapers than television, as newspapers ingeneral will apply a stronger partisan slant. This is supported by a study by Schönbach(1983), who found evidence that in Germany the press has greater potential than televisionto stimulate political participation. In addition, it is easier for people to stick to their

Dow

nloa

ded

by [

Ston

y B

rook

Uni

vers

ity]

at 0

1:43

30

Oct

ober

201

4


newspaper than to a TV news program (Kleinnijenhuis, 1990), especially in the case ofsubscriptions. Repeated exposure to partisan contents—which is more likely to occur inthe case of newspapers—will result in a larger reinforcement effect, leading to higherelectoral participation. Therefore, I expect the effect of PPP on participation to be largerthan the effect of TPP, which results in the following hypothesis.

H3b: Press-party parallelism has a stronger effect on voter turnout thantelevision news–party parallelism.

Measuring Media-Party Parallelism

There are four manifestations of media-party parallelism (Hallin & Mancini, 2004;Seymour-Ure, 1974). It is discernible in media contents; in the ownership of the newsmedia; in the affiliations of journalists, owners, and managers; and in readership patterns.Studying media contents may be the most obvious thing to do when one is interested in theextent to which media-party parallelism is present. However, to do so in a cross-nationalcomparative setting is very time consuming and costly, and large-scale comparative dataon political affiliations of owners and journalists are lacking. Nevertheless, these dataare available for media publics. Partisan bias in media contents is also expressed bynewspaper readership. This is supported by British, Dutch, and Swedish studies show-ing a strong relationship between political preferences of readers and those of theirnewspapers (Kleinnijenhuis, 1990; Seymour-Ure, 1974; Weibull, 1995). Therefore, thisarticle looks at readership patterns in order to indicate the extent to which media-partyparallelism is present. It investigates the relationship between media exposure and party-political preferences. By doing that, I left the three other indicators of media-party paral-lelism out of consideration. One has to bear in mind that my operationalization is just oneof four possibilities. By studying the other three indicators, one might come to (slightly)different conclusions. To simplify matters, I speak in this article about MPP, but one hasto reckon with the fact that it is one of four possible operationalizations of the concept.

In order to calculate the MPP system variable, I regress party preference scores onnewspaper and television news exposure.3 Party preference scores are measured by askingrespondents to indicate for each party the probability of their ever voting for it on a scalefrom 1 (not at all probable) to 10 (very probable). For exposure to newspapers and TVnews—the independent variables—dummies are included for regular exposure4 to each ofthe national newspapers and TV news programs.5 A score of one indicates regular expo-sure; zero indicates no (regular) exposure. Only those newspapers and TV news programsthat have a substantial audience6 in the sample are included in the analysis.7 The calcula-tion of the MPP variable is inspired by the measure for the strength of sociopolitical cleav-ages used by Franklin (1992). He calculated a weighted mean of explained variances forparty preference scales on variables of social structure.

Party preference scores are regressed on media exposure variables for each party inevery political system. After doing this for all parties, I calculate a weighted mean ofadjusted explained variances in proportion to each party’s strength (measured by electoralsuccess).8 I weighted the mean in order to avoid a small, communist party, for example(for which the preference may be quite strongly correlated with media use), beingweighted as much in the MPP measure as a broad, liberal party. Not weighting wouldresult in bias caused by small parties and news media that are not very broadly used.Averaging the MPP scores of all parties ensures that the values of MPP are not dependentupon the number of parties in each country. Since the values now range between zero9

Dow

nloa

ded

by [

Ston

y B

rook

Uni

vers

ity]

at 0

1:43

30

Oct

ober

201

4


(no media-party parallelism) and 100 (maximum media-party parallelism), in each coun-try, the values can be compared. In practice, a value of 100 is not possible to obtain, sincemedia audiences are overlapping. (Appendix C describes the entire procedure using Austriaas an example.)

Research Method and Data

To investigate the hypotheses presented above, this study compares relatively similarpolitical systems, namely the 15 states that were members of the EU in 1999. Participationin the European Parliament elections of 1999 indicates the level of voter turnout, thedependent variable in the study. These are elections to the same institution, held in thesame week in all member states. This makes the data comparable. Nevertheless, turnoutlevels vary from a low of 24% in the UK to 90% in Belgium. In addition, it is known thatother factors play an important role in turnout to the European elections. The most impor-tant one is the perception that the EU is a “good thing” (e.g., Niedermayer, 1990). More-over, at the aggregate level, the coincidence of local, regional, or general elections, as wellas the presence of a compulsory voting law, increases participation levels in Europeanelections (Niedermayer, 1990; Oppenhuis, 1995; Van der Eijk & Franklin, 1996). Toavoid misspecifying the model, I included these variables in my study as controls.

All analyses are based on data from the European Election Study 1999.10 This large-scale telephone survey includes citizens age 18 and over in all EU member states withinthree and a half weeks after election to the European Parliament. The data set containsinformation on political attitudes and preferences, electoral behavior, media use, andsociodemographic characteristics. The data set yields 13,549 respondents11 in the 15 EUmember states.12 (General information on the data set, including nonresponse, can befound in Appendix A. Question wordings of the variables used in this study can be foundin Appendix B.)

To examine the three hypotheses stated above, the study applies multiple regressionanalyses explaining turnout at two levels.13 First, the study connects aggregate levels ofparticipation to the MPP variable for the EU elections in 1999. This analysis (focusing onthe 15 EU member states, with Belgium divided into two separate political/media-systems14) explains turnout levels. Second, the study explains self-reported participationin the EU elections in 1999 by the MPP variable (n=13,549). By adding variables knownto affect voting participation, I control for possible misattributions of explanatory powerto the MPP variable. Moreover, at the individual level of analysis, I can test the interactioneffect of H2.

In order to test the validity of the MPP variable and the model, I conduct several tests(described below). I conduct these tests to avoid misinterpretation of the results and toexclude rival interpretations.

First, I conduct a jack-knife test in order to conclude whether the results are depen-dent on one specific outlying country. In such a test, the model is estimated 16 times, eachtime leaving one country out. If the findings are robust, the same results should be foundeach time a single country is left out. If this is the case, the findings are robust to a signifi-cant extent.

Second, I control for habitual voting. The MPP effect on participation might be spuri-ous, caused by habitual voting. In other words: Reading a specific newspaper is a habit,associated with the habit of voting for a specific party, but it has nothing to do with anymedia effect. In order to test for this, I calculated the relationship between the strength ofparty identification and the party sympathy score for those who voted for this party, in a

Dow

nloa

ded

by [

Ston

y B

rook

Uni

vers

ity]

at 0

1:43

30

Oct

ober

201

4


similar way that MPP was calculated.15 This resulted in a contextual measure based onadjusted explained variances for each political system. This variable is then used in analy-ses to control for habitual voting.

Third, another rival interpretation could be that MPP manifests the strength of socio-political cleavages in a country, an argument also made by Hallin and Mancini (2004) andKleinnijenhuis and Scholten (1989). The existence of strong socio-political cleavagesmay, on the one hand, lead to strong media-party parallelism and, on the other hand,enhance participation levels. Conversely, the existence of weak sociopolitical cleavagesmay weaken MPP and turnout levels. In order to control for this, I calculated a variableindicating the strength of sociopolitical cleavages, in a similar way that MPP wascalculated16 (see Franklin, 1992, for a similar application of explained variances for calcu-lating sociopolitical cleavages). I control MPP effects for the sociopolitical cleavages vari-able. If MPP effects hold after inclusion of this variable, the explanation of MPP being amanifestation of cleavages is eliminated.

Fourth, the interaction effect of parallelism and interest might be due to an aggrega-tion artifact. It is possible that the interaction effect is due to different distributions of theMPP variable among subgroups. This would also indicate that MPP is not a system-levelvariable. To test whether this is the case, I calculated MPP values per system, for eachlevel of interest separately. In order to conclude whether results are due to an aggregationartifact, the original MPP variable is calculated for each level of interest separately. Theseresults should be similar to the results of the aggregate MPP measure. In other words, nolarge fluctuations per group should appear. In addition, I replace the MPP variable in themultivariate analysis of participation by MPP per level. If results of this analysis are com-parable to the original results, indications of an aggregation artifact are eliminated.

Results

Table 1 describes the level of media-party parallelism for each system. It also indicates thelevel of press-party parallelism and of television-party parallelism. The table shows a con-siderable range of values: from a low MPP value of 1.0 for Germany, which indicates thatparallelism is almost nil, to a high value of 19.9 for Greece, denoting that in this countrythere exists strong parallelism between media use and party preferences. On average, MPPis 7.1; the median is 6.5. The results indicate that the dichotomy used by Hallin andMancini (2004) can be nuanced. There are large differences between countries that fallinto the same category in their study.

Media-party parallelism in Greece is very high compared to other systems. The find-ings are in line with previous studies (Dimitras, 1997; Hallin & Mancini, 2004). InGermany, Finland, and Ireland, there is nearly no media-party parallelism. The finding forIreland is according to expectations. In this country, political neutrality has come to be thetypical stance of newspapers, which are best described as catch-all media cutting acrossthe principal lines of division between the established political forces in society (Hallin &Mancini, 2004). For Finland, the findings are not surprising either. In the mid-1990s, lessthan 10% of circulation shares could be credited to the party press, which had mainly lostits role of political mobilization. Moreover, the party press tends to be read by a politicallyamorphous group (Salokangas, 1999). The findings for Germany, however, are somewhatmore unexpected. Although the bivariate associations between media exposure and partypreferences17 conform to expectations—readers of “left-wing” newspapers are signifi-cantly more likely to vote for leftist parties, and readers of “right-wing” papers are signifi-cantly more likely to do so for rightist parties (for studies on party-political leanings of

Dow

nloa

ded

by [

Ston

y B

rook

Uni

vers

ity]

at 0

1:43

30

Oct

ober

201

4


German newspapers, see, e.g., Patterson & Donsbach, 1993)—these associations are weakand do not add up to large explained variances. For example, the German newspaper Bildis known for its clear politically rightist contents, but its readers are not significantly dif-ferent from the other respondents in the sample.18 Evidently, German media audiences arerelatively amorphous groups in party-political terms. The MPP values of the other systemsare situated somewhere in between the strong level of parallelism in Greece and the weaklevel in Germany.

In general, press-party parallelism (mean=5.2) is stronger than television news–partyparallelism (mean=2.2), as expected. In the majority of countries, television-party paral-lelism is nearly absent. The exceptions are Italy, Greece, and Flanders, where a quite dis-tinct association exists between exposure to specific TV news programs, on the one hand,and party preferences, on the other. In general, however, the weak associations indicatethat television news programs try to appeal to broader publics—in party-political terms—than newspapers. PPP is strongest in Greece, Italy, Spain, and Denmark and least apparentin Germany, Portugal, Finland, and Ireland.

The second step in the analysis is to look at the relationship between parallelism andelectoral participation. According to expectations, the first model (see Table 2) indicates thatMPP affects turnout levels positively.19 Subsequently, turnout is regressed on MPP, control-ling for two variables that have a considerable impact on turnout levels in European elec-tions. These two factors are the presence of compulsory voting laws (in Belgium andLuxembourg) and the concurrence of European elections with national, regional, or localelections (in Belgium, Ireland, Italy, Luxembourg, and Spain) (Niedermayer, 1990). Not

Table 1Press-party parallelism (PPP), television-party parallelism

(TPP), and media-party parallelism (MPP) per system, 1999

Political system PPP TPP MPP

Greece 16.6 6.0 19.9Italy 7.2 8.0 14.4Spain 8.0 3.5 10.5Flanders 4.3 6.9 10.0Denmark 8.3 0.6 9.1France 5.4 3.2 7.5Britain 6.9 0.5 7.1Sweden 6.0 0.6 6.8Portugal 1.9 3.0 6.2Netherlands 5.2 0.8 5.7Austria 4.4 1.0 5.3Luxembourg 3.2 1.5 4.5Wallonia 2.2 −0.7 2.4Finland 1.9 −0.0 1.8Ireland 1.4 0.1 1.6Germany 0.7 0.2 1.0

M 5.2 2.2 7.1Mdn 5.3 0.9 6.5

Note. 0 = no parallelism, 100 = maximum parallelism.

Dow

nloa

ded

by [

Ston

y B

rook

Uni

vers

ity]

at 0

1:43

30

Oct

ober

201

4


considering these factors might result in a misattribution of explanatory power to the MPPvariable. The second model represented in Table 2 presents a regression analysis for elec-toral participation, controlling for these two variables. It shows that MPP still has a signif-icant positive effect on voting turnout levels. The b value of 1.59 indicates that, onaverage, turnout will rise by 1.59% if the MPP variable increases 1%. As expected, con-current national elections and the presence of compulsory voting laws have a positiveimpact on turnout. The three variables explain 88% of the variation in turnout.

By modeling electoral participation at the individual level, one can control for factorssuch as media use and political attitudes. Modeling participation at the individual levelalso allows testing the hypothesized interaction effect of MPP with political interest. Table 3presents the results of a logistic regression for electoral participation. This model controlsfor variables at the individual level, and at the contextual level for compulsory voting andconcurrent national elections. This model includes MPP and other contextual characteris-tics along with sociodemographic, attitudinal, and media use variables, which I inserted ascontrols. The strength of parallelism affects participation positively and significantly, asexpected. In systems with strong media-party parallelism, citizens are—ceteris paribus—more inclined to turn out and vote than in systems with weak levels of MPP. The positive,significant parameter of political interest indicates that participation increases as interestrises. The (negative) effect for the interaction of MPP and interest, however, implies thatthe effects of MPP diminish as interest increases. The effects of MPP are largest for lowlevels of interest.

Concerning the controls, age and attendance at religious services have a positive sig-nificant effect on participation, as could be expected (Oppenhuis, 1995). Three other vari-ables, political interest (e.g., Brady, Verba, & Schlozman, 1995), party attachment (e.g.,Schmitt, 1990), and internal political efficacy (e.g., Almond & Verba, 1963; McLeod,Scheufele, & Moy, 1999), affect participation positively, according to expectations. Personswho think the EU is a “good thing” also turn out in larger numbers, compared to the rest.This latter variable is important to control for, as former studies have shown that pro-European attitudes can increase participation in European elections (e.g., Niedermayer,1990). The findings of this study support this conclusion. The positive, significant parameterof the variable indicates that people with pro-European attitudes are more prone to vote.

The results also indicate that compulsory voting and concurrent national electionsaffect participation positively, which is in conformity with the results shown in Table 2.

Table 2OLS regression of electoral participation on systemic characteristics, EP elections 1999

Model 1 Model 2

B SE B SE

MPP 1.240 1.124 1.593* .443Concurrent national elections 17.334* 5.565Compulsory voting 32.333* 6.978Constant 45.566* 9.654 30.498* 4.092Variance explained (R2/adjusted R2) .080/.014 .882/.853N 16 16

*p < .01.

Dow

nloa

ded

by [

Ston

y B

rook

Uni

vers

ity]

at 0

1:43

30

Oct

ober

201

4


The model explains 27.0% of individual-level variation in participation. The model with-out the main and interaction effect of parallelism accounts for 23.7% of the variance inparticipation. This implies a nonnegligible increase in explanatory power when adding themain and interaction effect of parallelism to more traditional explanatory variables. Inconclusion, media-party parallelism has a positive and significant effect on voter turnout.MPP has a larger effect on electoral participation for persons with low levels of politicalinterest than for persons with high levels of interest.

A variable that includes the effects of all newspapers and television news programsthat one is regularly exposed to20 is included in the model. It serves as a control for theparallelism (interaction) effects by avoiding the misattribution of explanatory power to thecontextual (MPP) variable that in reality belongs to individual media exposure variables.The effect of the individual-level media exposure variable by itself is not of interest forthis article, and it does not tell us anything about any effects of specific media exposure onvoter participation, however, its use as a control was required to arrive at a well-specifiedmodel.

How much of the MPP effect is due to television, and how much is due to the press? Ireplicated the analyses of Table 3 with the main and interaction effect of parallelism sepa-rately for television and press, based on the values of TPP and PPP reported in Table 1.For press-party parallelism (PPP), this resulted in a pseudo R2 of .26. The effects of theparameters were nearly identical to those shown in Table 3. Moreover, the effects of PPPand interest by PPP on participation were exceed similar: .11 (SE=.01, p < .01) and −.04(SE=.01, p < .01), respectively.

Table 3Logistic regression of electoral participation on individual and systemic

characteristics, EP elections 1999

b (SE)

Sociodemographic variablesAge .018 (.002)*Attend religious services .207 (.025)*

Attitudinal and media exposure variablesIndividual media exposure 3.714 (.246)*Political interest .487 (.037)*Party attachment .305 (.031)*Internal efficacy .358 (.033)*EU good thing .264 (.057)*

MPP and other contextual variablesMPP .116 (.007)*MPP by interest −.028 (.009)*Compulsory voting 2.761 (.166)*Concurrent national elections .956 (.074)*

Constant 34.553 (3.883)*Variance explained (McFadden pseudo R2) 0.270Variance explained without MPP and

MPP by interest0.237

N 8988

*p < .01.

Dow

nloa

ded

by [

Ston

y B

rook

Uni

vers

ity]

at 0

1:43

30

Oct

ober

201

4


To test the effect of TPP, I estimated the same model of Table 3 with the values ofTPP. The results are quite contrary to expectations; pseudo R2 is .28, and the effect of TPPis rather strong, namely .35 (SE=.02, p < .01). The interaction effect with interest is notsignificant (b=−.03, SE=.02). According to these results, it seems that parallelism in thetelevision news environment is just as important for getting out the vote as press-partyparallelism. Although the extent to which television-party parallelism exists is small inmany countries, its effect on voter participation is positive and significant. The interactionwith political interest, however, is not significant.21

Robustness Tests and Validating Analyses

Jack-knife Test

In order to eliminate the chance that the results are solely dependent on specific (outlying)media systems, I conducted a jack-knife test on the model in Table 3. This model was esti-mated 16 times, each time excluding one country from the analyses. The results are simi-lar in all cases; pseudo R2s are similar, as are the size and direction of MPP and the sizeand direction of the interaction effect of interest and MPP. The interaction effect loses sig-nificance when either Germany or Greece are left out of the analysis (t=−.90 if Germanyis left out and −1.10 if Greece is left out). However, the direction of the effect remains thesame. The results indicate that no specific (outlying) system distorts the coefficients. Theoverall conclusion is that the effects presented in Table 3 are highly robust.

Habitual Voting

The results might partly represent differences in the role that habits play in the act of vot-ing, as explained above. In contexts where MPP is high, party preferences might be stableon the basis of a routine or tradition. Instead of an MPP effect, it might be this routine fac-tor that impacts participation positively. In order to test for such a spurious habitual votingeffect, I calculated a variable indicating the strength of habitual voting per context.Weighted explained variances for this variable—which is calculated as the associationbetween party preference scores and party attachments for each party separately—varyfrom 3.05% in Wallonia to 13.25% in Denmark. If the variable is included in the regres-sion equation in Table 2, the MPP effects are not affected. This result indicates that theMPP effect is not spuriously caused by habitual voting.

Sociopolitical Cleavages

Adding to the regression equation in Table 2 a variable that measures the strength ofsociopolitical cleavages does not result in a significant effect, nor does it affect the MPPeffects (tsociopolitical cleavages=2.28, p < .05; tMPP=17.60, p < .01; tMPP by interest=−3.17, p <.01; R2=.270). This finding indicates that media-party parallelism is a contextual charac-teristic that cannot be reduced to other social or political cleavages in a country.

Aggregation Artifact

MPP distributions between levels of interest do not differ significantly, indicating thatthere is no aggregation artifact and that MPP is a truly aggregate-level (contextual)variable. A replication of the analysis supports this: The model reported in Table 3 is

Dow

nloa

ded

by [

Ston

y B

rook

Uni

vers

ity]

at 0

1:43

30

Oct

ober

201

4


reestimated with MPP measured for different levels of political interest, instead of theaggregate MPP variable. When MPP in the logit model in Table 3 is replaced by “MPP perinterest level,” the results do not differ much from those reported in Table 3. Explainedvariance of that model is .26, and the model estimates do not deviate much from the esti-mates of Table 3. More importantly, the main effect and interaction effect of MPP arenearly the same as reported in Table 3. The b values and standard errors are respectively,.10 and .01 (p < .01) for the main effect and −.03 and .01 (p < .01) for the interactioneffect. The fact that the results do not differ when MPP is measured for each group sepa-rately supports the notion of MPP being a system characteristic, and not an aggregationartifact.

Conclusion

This article has broadened the concept of press-party parallelism (Seymour-Ure, 1974) tomedia-party parallelism, and introduced a way to measure it. The measure reflects theextent to which party preferences can be explained by media usage. The results showed aconsiderable degree of variation between countries. In general, the association betweennewspaper reading and party preferences (press-party parallelism) is stronger than therelationship between TV news viewing and party preferences (television-party parallel-ism). H3a is accepted. Television news programs try to appeal to a broad public, whereasnewspapers are more likely to focus on narrower groups, at least in some countries.

Media-party parallelism has a considerable impact on electoral participation. H1 isaccepted. This effect is even stronger for those who are not interested in politics. Thesefindings support H2. Levels of party parallelism are, in most systems, much lower for tele-vision news than for newspapers, but the effect on participation is significant for both.H3b is rejected. This indicates that—contrary to expectations (e.g., Schönbach, 1983)—television news is just as important as newspapers in mobilizing citizens to take part in anelection in contexts where news programs reflect partisan leanings. Television news doesmatter for political participation, although there is no evidence that such an effect is stron-ger for less politically involved persons.

According to Hallin and Mancini (2004), journalistic practice in many Europeancountries tends to change toward “Anglo Saxon” and nonpartisan styles. This wouldinvolve a strong separation of news and commentary, a diminished emphasis on partisanideas and ideals, and an increased emphasis on neutral information and nonpartisan enter-tainment. There seems to be a trend toward catch-all-ism, favoring internal diversity overexternal (Hadenius & Weibull, 1999). Newspapers tend to blur their ideological identitiesand connections in order to appeal to as broad an audience as possible (Hallin & Mancini,2004). The existence of such trends would diminish the partisanship of newspapers, which inturn tends to depress voter participation.

At the same time, voting patterns seem to become more volatile with an increasingnumber of floating voters. A process of cognitive mobilization decreases the functionalneed for partisan cues to guide voting behavior and mobilize political involvement (Dalton,1996). Short-term election campaigns are becoming more important than long-termpartisan endorsements for getting out the vote (Dalton, McAllister, & Wattenberg, 2000;Lilleker, 2006). However, long-term partisan endorsements remain important for mobiliz-ing citizens who are not politically interested, as demonstrated in Table 3. Therefore, adecline of partisan leanings of newspapers (and media in general) may have detrimentaleffects on participation rates of the least-connected citizens, increasing the already-existinggap between strongly and weakly interested segments of society.

Dow

nloa

ded

by [

Ston

y B

rook

Uni

vers

ity]

at 0

1:43

30

Oct

ober

201

4


Notes

1. This study focuses on press-party parallelism and effects on participation in Sweden during1979–2002.

2. I will focus on political interest since that variable offers the best comparable measure in thisdata set.

3. Party preferences are treated as dependent variables in this study, while media exposure istreated as independent. This does not imply that I have expectations about the causality of the twovariables. I am just interested in the multiple relationships between these variables and take R2 as avalid indicator for that. For this purpose, I could also have treated media exposure as dependent andparty preferences as independent variables, which would yield about the same results.

4. Regular exposure was defined as once a week or more.5. Scale variables would yield more specific information; unfortunately, this information was

not available in the data set.6. Substantial audience was defined as 5% of the sample or more.7. Regional and local newspapers and TV news broadcasts are not included, for they do not

have a substantial reach in the samples. Excluding these media, however, is not likely to harm theresults, as there is evidence that local or regional newspapers usually have less partisan contentsbecause they usually try to maximize their audiences (Dalton, Beck, & Huckfeldt, 1998; Hallin &Mancini, 2004). This is also likely to be true for local TV news.

8. Weighting is done according to the percentage of votes obtained at national parliamentelections because these are people’s major frame of reference with respect to the party system(Oppenhuis, 1995).

9. Practically, it may even take a value below zero if the relationship between media use andparty preferences is smaller than one would expect on the basis of chance.

10. The data utilized were originally collected for the European Election Study Workgroup, con-sisting of Cees van der Eijk, Klaus Schoenbach, Hermann Schmitt, Holli Semetko, Wouter van derBrug, Mark Franklin, Sören Holmberg, Renato Mannheimer, Jacques Thomassen, and BernhardWessels. Fieldwork was carried out by a consortium of European survey organizations co-ordinated byIPSOS (Hamburg). This study has been made possible with grants from the University of Amsterdam,the Dutch National Science Foundation (NWO, the Netherlands), The Bundespresseamt (Bonn), theCIS (Madrid), the University of Mannheim, the ISPO Institute (Milan), and Trinity College (Hartford,Connecticut). Neither the original collectors of the data nor their sponsors bear any responsibility forthe analyses or interpretations presented here. The data are distributed by Steinmetz Archive, Amsterdam,the Netherlands, and associated data archives. See also http://www.europeanelectionstudies.net.

11. Sample sizes were approximately 500 for the smaller countries and 1,000 for the biggercountries. The exception was Italy (N=3,708).

12. Data are weighted so that they generate a turnout and distribution of party choice that isidentical to the actual results of the June 1999 European election in the respective countries. Subse-quently, all country samples are weighted to an equal size so that the effective number of cases isequal for each of the systems.

13. One might object that there is a form of circularity in the analyses: People who do not votewill score low on the “probability to vote” questions, and they will score “zero” for participation. How-ever, there is no such circularity problem in this case, as the probability to vote questions are used forcalculating the aggregate variable of MPP (which is the same for everybody in one country).

14. Due to a lack of cases, Northern Ireland is missing in the results.15. The party preference score is regressed on the strength of party attachment for those who

voted for this party at the last national parliamentary election. This is successively done for eachparty. The adjusted explained variances are weighted according to the size of the party in question atthe last national parliamentary elections.

16. There were 4 dummies for self-reported social class, 2 dummies for urbanization, 4 dum-mies for religion, 4 dummies for attendance at religious services, and 1 dummy for membership in atrade union.

Dow

nloa

ded

by [

Ston

y B

rook

Uni

vers

ity]

at 0

1:43

30

Oct

ober

201

4


17. Tables with regression coefficients explaining party preferences by media use are notincluded here. They can be obtained from the author on request.

18. This is so except for the beta coefficients of reading Bild on the preference of the Greenparty (−.098, significant at the 1% level) and that of the extreme right party Republikaner (.124, sig-nificant at the 1% level). However, due to the relatively small sizes of these parties (respectively,6.7% and 1.8% of the votes at the last national parliamentary election), the effects do not elevate the(weighted) measure of MPP much.

19. The effect is not significant. However, I deal not with a sample but with the entire popula-tion in the analyses of Table 2, which makes statistical significance redundant.

20. A pairwise regression per system is conducted for electoral participation on regular expo-sure to different newspaper titles and television news programs. For each case, an unstandardizedpredicted value (y-hat) is calculated. Subsequently, the mean of the y-hats (per system) is subtractedfrom the unstandardized predicted value. This results in a variable with a mean of zero.

21. This may be due to a lack of power for this variable, as the variation in the TPP variable isvery small.

References

Almond, G. & Verba, S. (1963). The Civic Culture. Political Attitudes and Democracy in FiveNations. Princeton: Princeton University Press.

Berelson, B. R., Lazarsfeld, P. F., & McPhee, W. N. (1954). Voting: A study of opinion formation ina presidential campaign. Chicago: University of Chicago Press.

Brady, H. E., Verba, S., & Schlozman, K. L. (1995). Beyond Ses: A Resource Model of PoliticalParticipation. American Political Science Review, 89 (2), 271–294.

Brants, K., & Van Kempen, H. (2002). The ambivalent watchdog: The changing culture of politicaljournalism and its effects. In R. Kuhn & E. Neveu (Eds.), Political journalism: New challenges,new practices (pp. 168–186). London: Routledge.

Brynin, M., & Newton, K. (2003). The national press and voting turnout: British General Electionsof 1992 and 1997. Political Communication, 20, 59–77.

Cappella, J. N., Price, V., & Nir, L. (2002). Argument repertoire as a reliable and valid measure ofopinion quality: Electronic dialogue during campaign 2000. Political Communication, 19, 73–93.

Dalton, R. J. (1996). Citizen politics: Public opinion and political parties in advanced Westerndemocracies. Chatham, NJ: Chatham House.

Dalton, R. J., Beck, P. A., & Huckfeldt, R. (1998). Partisan cues and the media: Information flows inthe 1992 presidential election. American Political Science Review, 92, 111–126.

Dalton, R. J., McAllister, I., & Wattenberg, M. P. (2000). Consequences of partisan dealignment. InR. J. Dalton & M. P. Wattenberg (Eds.), Parties without partisans: Political change inadvanced industrial democracies (pp. 37–63). Oxford, England: Oxford University Press.

Dimitras, P. E. (1997). Greece. In B. Stubbe Østergaard (Ed.), The media in Western Europe: TheEuromedia handbook (pp. 98–109). London: Sage.

Franklin, M. (1992). The decline of cleavage politics. In M. Franklin, T. Mackie, & H. Valen (Eds.),Electoral change: Responses to evolving social and attitudinal structures in Western countries(pp. 383–405). Cambridge, England: Cambridge University Press.

Ha, S., & McCombs, M. (2004). Who accepts the news?: News coverage of presidential campaigns, vot-ers’ information processing ability, and media effects susceptibility. Austin: University of Texas.

Hadenius, S., & Weibull, L. (1999). The Swedish newspaper system in the late 1990s: Tradition andtransition. Nordicom Review, 20, 129–152.

Hallin, D. C., & Mancini, P. (2004). Comparing media systems: Three models of media and politics.Cambridge, England: Cambridge University Press.

Iyengar, S., & Kinder, D. R. (1987). News that matters: Television and American opinion. Chicago:University of Chicago Press.

Iyengar, S., Peters, M. D., & Kinder, D. R. (1982). Experimental demonstrations of the “not-so-minimal”consequences of television news programs. American Political Science Review, 76, 848–858.

Dow

nloa

ded

by [

Ston

y B

rook

Uni

vers

ity]

at 0

1:43

30

Oct

ober

201

4


Katz, E., & Lazarsfeld, P. F. (1955). Personal influence: The part played by people in the flow ofmass communications. Glencoe, IL: Free Press.

Kleinnijenhuis, J. (1990). Op zoek naar nieuws: Onderzoek naar journalistieke informatieverwerk-ing en politiek. Amsterdam: VU Uitgeverij.

Kleinnijenhuis, J., & Scholten, O. (1989). Veranderende verhoudingen tussen media en politiekepartijen. Acta Politica, 24, 433–460.

Krosnick, J. A., & Kinder, D. R. (1990). Altering the foundation of support for the president throughpriming. American Political Science Review, 84, 497–512.

Lazarsfeld, P. F., Berelson, B., & Gaudet, H. (1944). The people’s choice: How the voter makes uphis mind in a presidential campaign. New York: Columbia University Press.

Lilleker, D. G. (2006). Key concepts in political communication. London: Sage.McLeod, J. M., Becker, L. B., & Byrnes, J. E. (1973). Another look at the agenda-setting function of

the press. Communication Research, 1, 131–166.McLeod, J. M., Scheufele, D. A., & Moy, P. (1999). Community, communication and participation:

The role of mass media and interpersonal discussion in local political participation. PoliticalCommunication, 16, 315–336.

Miller, W. L. (1991). Media and voters: The audience, content, and influence of press and televisionat the 1987 general election. Oxford, England: Clarendon Press.

Newton, K., & Brynin, M. (2001). The national press and party voting in the UK. Political Studies,49, 265–285.

Niedermayer, O. (1990). Turnout in the European elections. Electoral Studies, 9, 45–50.Norris, P., Curtice, J., Sanders, D., Scammel, M., & Semetko, H. A. (1999). On message: Communi-

cating the campaign. London: Sage.Oppenhuis, E. (1995). Voting behavior in Europe: A comparative analysis of electoral participation

and party choice. Amsterdam: Het Spinhuis.Patterson, T. E. (1998). Political roles of the journalist. In D. Graber, D. McQuail, & P. Norris (Eds.), The

politics of news: The news of politics (pp. 17–32). Washington, DC: Congressional Quarterly Press.Patterson, T. E., & Donsbach, W. (1993). Press-party parallelism: A cross-national comparison. Paper

presented at the annual meeting of the International Communication Association, Washington, DC.Rokkan, S., & Torsvik, P. (1970). The voter, the reader and the party press: An analysis of political

preference and newspaper reading in Norway. In S. Rokkan, A. Campbell, P. Torsvik, &H. Valen (Eds.), Citizens, elections, parties (pp. 397–416). New York: David McKay.

Roncarolo, F. (2002). A crisis in the mirror: Old and new elements in Italian political communica-tion. In R. Kuhn & E. Neveu (Eds.), Political journalism: New challenges, new practices(pp. 69–91). London: Routledge.

Salokangas, R. (1999). From political to national, regional and local: The newspaper structure inFinland. Nordicom Review, 20, 77–105.

Schönbach, K. (1983). Das unterschätzte Medium: Politische Wirkungen von Presse und Fernsehenim Vergleich. München: K. G. Saur.

Schmitt, H. (1990). Party Attachment and Party Choice in the European Elections of June 1989. Across-national comparative analysis of the post-electoral surveys of the European Voters Study1989. International Journal of Public Opinion Research, 2(2), 169–184.

Semetko, H. A. (1996). The media. In L. LeDuc, R. G. Niemi, & P. Norris (Eds.), Comparing democ-racies: Elections and voting in global perspective (pp. 254–279). Thousand Oaks, CA: Sage.

Seymour-Ure, C. (1974). The political impact of mass media. London: Constable.Van der Eijk, C., & Franklin, M. (1996). Choosing Europe? The European electorate and national

politics in the face of union. Ann Arbor: University of Michigan Press.Van Kempen, H. (2006). Press-party parallelism and its effects in Sweden: A longitudinal study

1979–2002. Scandinavian Political Studies, 29, 406–422.Voltmer, K. (2000). Structures of diversity of press and broadcasting systems: The institutional context of

public communication in Western democracies. Berlin: Wissenschaftszentrum für Sozialforschung.Weibull, L. (1995). Media diversity and choice. In K. E. Gustafsson (Ed.), Media structure and the state:

Concepts, issues, measures (pp. 149–170). Göteborg: Mass Media Research Unit Publications.

Dow

nloa

ded

by [

Ston

y B

rook

Uni

vers

ity]

at 0

1:43

30

Oct

ober

201

4


Appendix A: General Survey Information

Response rates (%) for the net sample were as follows: Finland, 41.2; Sweden, 31.0;Denmark, 59.0; United Kingdom, 49.0; Ireland, 29.4; Germany, 49.3; Austria, 50.4;Netherlands, 30.2; Belgium, 37.0; Luxembourg, 30.1; France, 44.0; Portugal, 44.5;Greece, 28.0; Spain, not appropriate (quota study); Italy, not appropriate (Internetstudy).

Detailed reports of fieldwork, coding and weighting, response rates, and so can forthbe obtained from http://www.europeanelectionstudies.net.

Appendix B: Question Wording and/or Explanation of Variables

Variable Wording or Explanation

Electoral participation, EP elections 1999

A lot of people abstained in the European Parliament elections on June 10 while others voted. Did you cast your vote? Yes (1) no (0).

Age Age at time of interview.Attendance it

religious services

How often do you attend religious services: several times a week (5), once a week (4), a few times a year (3), once a year or less (2), or never (1)?

Individual media exposure

A variable constructed from a series of separate observations regarding exposure to newspaper titles and TV news programs. See Note 20.

Political interest Political interest score constructed by a scale consisting of two questions: interest in politics and interest in election cam-paign. Low (1), medium (2), high (3).

Party attachment Do you feel yourself to be very close to a political party (3), fairly close (2), or merely a sympathizer (1)? Or not at all? (0).

Internal efficacy So many people vote in elections that my vote does not matter. (1) Strongly disagree, (2) disagree, (3) agree, (4) strongly agree.

EU good thing Generally speaking, do you think that [your countrys] membership in the European Union is a good thing (1), a bad thing (0), or neither good nor bad (0)? (dichotomy)

MPP Contextual measure for the homogeneity of media use groups. See section “Measuring MPP.”

MPP by interest (MPP score) by (political interest score).Party preference

scoreThe following question was asked for each of the political par-

ties in a country: We have a number of parties in [name of your country] each of which would like to get your vote. How probable is it that you will ever vote for the following parties? Please specify your views on a 10-point scale where 1 means “not at all probable” and 10 means “very probable.” (Follows separate item for each party)

(Continued)

Dow

nloa

ded

by [

Ston

y B

rook

Uni

vers

ity]

at 0

1:43

30

Oct

ober

201

4


Appendix C: Calculation of the MPP Variable

The calculation of the MPP variable is explained for the case of Austria in 1999.

Y1=a+b1X1+b2X2+b3X3+b4X4+b5X5+b6X6+b7X7+b8X8+b9X9+b10X10+b11X11+b12X12, in which

Y1=score on party preference scale scale for the “SPÖ” party (10-point scale);X1=regular readership of Kronen-Zeitung (dummy);X2=regular readership of Kleine Zeitung (dummy);X3=regular readership of Kurier (dummy);X4=regular readership of Der Standard (dummy);X5=regular readership of Die Presse (dummy);X6=regular readership of Täglich Alles (dummy);X7=regular readership of Salzburger Nachrichten (dummy);X8=regular readership of Oberösterreichische Nachrichten (dummy);X9=regular exposure to Zeit im Bild 1 (dummy);X10=regular exposure to Zeit im Bild 2 (dummy);X11=regular exposure to Zeit im Bild 3 (dummy);X12=regular exposure to ARD/ZDF-Nachrichten (dummy).

The subsequent regressions for Austria include the same independent variables, andrespectively OVP, FPO, Grüne, LIF, and CSA as Y2 through Y6—the dependent vari-ables. The adjusted R2s—the amount of variance for the party preference score of each

Appendix B: (Continued)

Variable Wording or Explanation

TV news program exposure

(Open question) Which channels or television news programs do you watch regularly? Coded per program (0) no regular expo-sure; (1) regular exposure.

Newspaper exposure

(Open question) Which newspaper or newspapers do you read regularly? Coded per newspaper (0) no regular exposure; (1) regular exposure.

Compulsory voting

Compulsory voting law in effect? (1) Belgium and Luxembourg; (0) all other countries.

Concurrent national elections

(1) Belgium, Ireland, Italy, Luxembourg, Spain; (0) all other countries.

Self-reported social class

If you were asked to choose one of these five names for your social class, which would you say you belong to—the working class, the lower middle class, the middle class, the upper middle class, or the upper class?

Urbanization Would you say you live in a rural area or village, in a small or middle size town, or in a large town?

Religion Do you consider yourself as belonging to a particular religion? (If yes:) Which one?

Membership in trade union

Are you a member of a trade union? Yes (1); no (0).

Dow

nloa

ded

by [

Ston

y B

rook

Uni

vers

ity]

at 0

1:43

30

Oct

ober

201

4


party explained by the media exposure dummies—are weighted by their size in parliament(last national parliamentary election result). For Austria in 1999, the following are theresults per party.

Party Adjusted R2 Size of partya Size by adj. R2

SPO .026 38.1 0.9906OVP .029 28.3 0.8207FPO .085 21.9 1.8615Grüne .143 4.8 0.6864LIF .152 5.5 0.836CSA .016 0.0 0.0000Total for parties distinguished 98.6b 5.1952MPP 5.27 (5.1952/98.6×100)

aIn percentage of votes gained at last national parliamentary election before EP election in June1999.

bThe difference between this total and 100% represents “other” parties.

Dow

nloa

ded

by [

Ston

y B

rook

Uni

vers

ity]

at 0

1:43

30

Oct

ober

201

4

media-party parallelism and its effects: a cross-national comparative study

Documents