Printer Friendly



This study explores two sets of hypotheses: An increment in media use for political information corresponds to (1) an increment in community consensus about social priorities (first level agenda setting); and (2) an increment in community consensus about politicians' attributes (second level agenda setting). The analysis of first level effects, which largely replicates research conducted in the United States, shows that a trend toward consensus in an agenda of issues is also present among the Spanish public. Analysis of the agendas of substantive and affective characteristics of political candidates shows that the pattern of increasing social consensus is also present at the second level of agenda setting. Increasing consensus in the affective agenda of candidates' characteristics among different population subgroups suggests that the news media, especially television, contribute to more homogeneous evaluations of rival political candidates, leveling out ideological changes within democratic societies.

Mass media have a significant influence on the focus of public attention. Over the past 25 years a wealth of evidence has accumulated about the influence of the mass media on the contents of the public agenda, especially the agenda of key issues facing a community (Dearing and Rogers 1996; McCombs and Bell 1996). In channeling the agenda towards a relative unification of opinions about the social priorities of the moment, the mass media help structure a practical approach to public issues, which are both ranked according to their relative importance and limited in number. This situation allows the community to concentrate its efforts and to focus governmental and social action. Without a compact set of issues or affairs to be resolved, a set ranked as priorities, the functionality of society would be all but impossible (McCombs 1997).

Considering the short cycles of time in which the public gives attention to specific community preoccupations and maintains them as social priorities (Downs 1972), it is indeed reasonable to think that a society cannot function without appreciable levels of consensus in what agenda setting theory calls the public agenda. One also has to add that the number of issues the government, journalists and the public alike can address at any given time is also limited. The human mind (both `individual' and `public') has only a limited capacity to process information (Miller 1965; Shaw and McCombs 1977). Therefore a considerable social consensus about which issues reach the agenda is necessary. Perhaps the very possibility of society's existence as such resides precisely in this double limitation on processing and retention.


Shaw and Martin (1992) concluded that greater consensus in the public agenda among demographic subgroups corresponds to greater exposure to the mass media.(1) Their study of North Carolina residents examined five distinct social groups, defined by age, gender, race, income, and education, in order to measure the consensus subgroup agendas as a function of frequency of contact with the news media.

The clearest tendency toward consensus as the frequency of daily newspaper reading increased was found between men and women. Similar tendencies were found with respect to differences in race and age, although no appreciable increase in consensus between income or education groups was found. This last finding was explained by relatively high levels of consensus among the education subgroups at all reading levels. Shaw and Martin also found similar patterns of consensus in limited comparisons of the same subgroups when their television viewing habits were considered. However, the television analysis was based only on survey data, while the newspaper analysis considered both the level of correspondence between subgroup agendas measured in the survey and the match of each subgroup agenda with the newspaper agenda measured in a content analysis.

Additional evidence of consensus as an outcome of media exposure was found in Taiwan (Chiang 1995), where the sample was divided according to gender, income, and education. For each of the three demographic comparisons consensus increased with newspaper reading. But the effects of television viewing found in the North Carolina study were not replicated, perhaps due to tight governmental control of television broadcasting in Taiwan.

The effects of television on community consensus still need to be tested since neither Shaw and Martin's nor Chiang's study yielded a full comparison of newspapers' and television's role in the achievement of consensus. It should be noted that newspapers are frequently regarded as more influential than television in setting the agenda of the public (Shaw and McCombs 1977; Wanta 1997). If this is so, we could hypothesize that the contribution of newspapers to consensus is greater than that of television, but particular differences may still result when studying different demographic subgroups.

This research project was designed and conducted in a seminar organized by the Department of Public Communication at the University of Navarra, under the direction of Maxwell McCombs and Esteban L6pez-Escobar. Our objective was to replicate and extend the original American research of Shaw and Martin during the May 28, 1995, elections for members of the regional parliament in the northern Spanish province of Navarra. In addition to a full comparison of newspapers and television news in a new cultural and political setting, Spain, we also sought to extend this research beyond the traditional domain of agenda setting, public issues, to additional levels of influence.

In the vast majority of agenda setting studies to date, including Shaw and Martin (1992) and Chiang (1995), the unit of analysis on each agenda is an object, a public issue. Each of these objects in turn has numerous attributes, those characteristics and properties that sketch out the picture of each object in our heads and in the news coverage. Just as objects vary in salience, so do the attributes of each object. When members of the public and journalists describe objects, public issues, political leaders, or whatever, some attributes are emphasized, some mentioned only in passing, and others not at all (McCombs and Evatt 1995).

Both the selection of objects for attention and the selection of attributes for describing these objects are powerful agenda-setting roles. The transmission of object salience, the traditional emphasis of agenda setting research, is the first level of agenda setting. The transmission of attribute salience, the emphasis of much current research (e.g. McCombs et al. 1997b; Lopez-Escobar et al. 1998) is the second level of agenda setting.


We focused our study of the 1995 Spanish regional elections on the major issues in the city of Pamplona, the capital of Navarra, and on the images of the candidates for leader of the Navarra parliament, representing the five major political groupings: UPN (Union del Pueblo Navarro, right wing), CDN (Convergencia de Democratas Navarros, a new political party that had just splintered off from UPN), PSN (Partido Socialista de Navarra, an affiliate of PSOE, the national Spanish Workers Socialist Party), IU (Izquierda Unida, a coalition of left-wing parties), and HB (Herri Batasuna, a Basque nationalist party with an extreme left-wing ideology).

Replicating and extending Shaw and Martin (1992), we will test four hypotheses, the roles of newspaper and television news exposure in the achievement of consensus at both the first level of agenda setting-the salience of public issues--and at the second level--the salience of attributes of political candidates.

* Hypothesis 1. Increased exposure to newspapers corresponds to increased consensus among demographic subgroups regarding the public agenda of issues.

* Hypothesis 2. Increased exposure to television news corresponds to increased consensus among demographic subgroups regarding the agenda of issues.

* Hypothesis 3. Increased exposure to newspapers corresponds to increased consensus among demographic subgroups regarding the agenda of attributes of major political candidates.

* Hypothesis 4. Increased exposure to television news corresponds to increased consensus among demographic subgroups regarding the agenda of attributes of major political candidates.

Hypotheses were tested using data from a telephone survey conducted immediately after the election, as well as data from an analysis of the contents of local press and television during the electoral campaign. To define the subgroups we selected the variables of age, gender, education, and ideology or political affiliation. Two variables used by Shaw and Martin, income and race, are not included here because variations in income in Pamplona is relatively small, and the city is racially homogenous.

Ideology was added to the Shaw and Martin design because in Spain public discourse strongly emphasizes differences between the `right' and the `left,' and these differences are therefore more relevant than would generally be the case in a purely bipartisan society.


Telephone interviews lasting about 10 minutes were conducted June 1-5, 1995 with 299 residents of Pamplona randomly selected from the telephone directory. At the time of the survey 94.25 percent of the homes in Pamplona had telephones. The response rate for the survey was 60.8 percent. The balance, 39.2 percent, consists of refusals and persons not at home.

To measure the first level of agenda setting, the relative salience of different issues, we used the Gallup poll question on `the most important problem' facing the community. The replies to this open-ended question yielded six categories, which constitute the public agenda of the city of Pamplona's most important problems: (1) `Disturbances' (30 percent), which refers to street violence in the old city center of Pamplona; (2) `Unemployment' (29 percent); (3) `Traffic' (23 percent), which refers to automobile congestion in the city; (4) `Reconstruction of the old city center' (10 percent) which includes the topics of unoccupied housing, and quality of life in the city center; (5) `Pedestrianization of the city center' (4 percent), which refers to an initiative to prohibit traffic in specified areas, granting access only to pedestrians; (6) `Demonstrations by HB' (3 percent), which refers to the problems created by street demonstrations organized by Herri Batasuna.

At the second level of agenda setting, the agendas of candidate attributes, the survey asked another open-ended question: `Imagine that you had a friend who didn't know anything about the candidates for the Parliament. What would you tell your friend about ... (each candidate)?'

The data on the candidates' image were organized according to two dimensions: (a) an affective dimension, referring to the (i) `positive,' (ii) `negative' or (iii) `neutral' terms with which the candidates were described, and (b) a substantive dimension which included three descriptive categories: (i) the candidate's `ideology' and his issue positions, (ii) his `qualifications' and experience, and (iii) his personal characteristics and `personality.'

The agenda of candidate's substantive characteristics reflects the distinctive characteristics of those who aspire to elected office; characteristics that are considered more or less important by the mass media and by society. It tells us which aspects are salient when the public considers the character of politicians. Considering an affective public agenda about candidates for elected office suggests a possible contribution by the mass media to an appraisal of political leaders that in some manner is shared by individuals belonging to different groups--and even to groups supposedly in conflict, as is the case of voters leaning to the right or to the left. Comparisons of substantive descriptions of the candidates, plus the affective tone of these descriptions, both aggregated across all five candidates, will reveal how much similarity there is in various social groups' descriptions of political candidates.


Using the survey data we defined four pairs of demographic subgroups: `younger', i.e. those less than 43 years old (48 percent) or `older', i.e. those aged 43 or more (52 percent); those with a university education, whether completed or not (39 percent) and those without (61 percent); men (38 percent) and women (62 percent); those on the `right' (58 percent) and those on the `left' (42 percent).

Age groups were defined according to the median age of the population in Pamplona. The categories of `right' and `left' were defined according to the declared vote in the May, 1995 election, taking into account that the UPN (Union of the Navarran People) is ordinarily associated with the `right'; whereas the `left' generally comprises the PSN-PSOE (Navarran Socialist Party-Spanish Workers' Socialist Party), IU (United Left), and HB (Herri Batasuna). In the case of voters for the new party, CDN (Convergence of Navarran Democrats), the assignment to the `right' or `left' was based on respondents' recall of their vote in previous elections. Despite the under-representation of men in the sample, the distribution of male respondents in terms of age, ideology, media use, etc., is adequately random to test the hypotheses.


Previous studies of consensus determined media use by the frequency of exposure, i.e. the number of days a week in which a newspaper is read or a news program is watched. In contrast, exposure was measured here by the level of attention given by the reader or viewer of the news to content that is understood as political. In the survey 32 percent of the respondents affirmed that they read `all' of the political information in the press, 57 percent `little or some,' and 12 percent, `none'; 41 percent of those surveyed affirmed that they paid attention to `all' of the political information on television, 46 percent to `little or some,' and 13 percent to `none.'


To determine the first level media agenda, the news coverage of the local daily newspapers, Diario de Navarra and Diario de Noticias, and the regional TV news programs, Pamplona TV, and Telenavarra, were content analysed. Pamplona TV is a local commercial channel; Telenavarra is the regional program of TVE1, one of two channels of the Spanish state television service. The content analysis used the issue categories that emerged from the survey question of the most important issues. Content of both papers and of both TV channels was aggregated into a consolidated newspaper agenda and a consolidated TV agenda because of high correlations among the agendas and high overlap of the audiences within each news medium.

For the analysis of consensus at the second level of agenda setting, the images of the candidates in terms of both the substantive and affective dimension were content analysed in the two local newspapers and Telenavarra. Again, the newspaper agendas were consolidated to form a single newspaper agenda for the analysis. The local TV news was omitted because the survey found that it was not an important source of information about the political candidates.

The same coding sheets were used for the newspapers and television. Ten coders, all familiar with the current political environment in Navarra, but none a member or activist of any of the parties, carried out the content analysis. A preliminary coding of the data to determine reliability yielded a median Pearson's r of 0.91.(2)


The overall correlation between the agenda of the newspapers, Diario de Navarra and Diario de Noticias, and the public's agenda was .80 for the six `problems of the city of Pamplona.' Turning to the analysis of the public agenda among the various demographic subgroups, Figure 1 displays the results according to the three levels of newspaper reading. The triangular model, originally proposed by Shaw and Martin (1992), presents the values of the correlations between demographic subgroups at the base of each triangle.


In the original study conducted in North Carolina, there was no tendency toward consensus among education subgroups, whereas in Pamplona this is the subgroup showing the strongest tendency. But the increase in agreement is not monotonic across the three levels of readership. Rather the distinction is between readers and non-readers of political information in the newspaper. Among men and women there is a similar pattern, a sharp distinction between readers (to any degree) and non-readers. Shaw and Martin also found increased consensus among men and women.

For ideology the pattern is curvilinear, but again readers are distinctly different from non-readers. For age the pattern is also curvilinear, but the distinction between readers and non-readers is blurred. Shaw and Martin found a clearer pattern in the age groups.

In sum, the patterns for three of the four demographics analyses--education, gender and ideology--support Hypothesis 1. But among the three demographic variables used in both this study and the original North Carolina study there is an identical outcome for only one, gender.

The tendency toward convergence among demographic subgroups can also be examined by applying the `index of convergence' proposed by Shaw and Martin (1992, p. 914), a measure which takes into account the strength of each group's relationship with the media agenda (the legs of the triangles in Figure 1) as well as the degree of correlation between the subgroup agendas that was the focus of our initial analysis.(3)

Again, as Figure 2 shows, there is a clear difference between those persons who are interested in political processes and participate in them by acquiring information and those who show no interest in political information. The patterns across the levels of readership match those found in Figure 1. Hypothesis 1 is supported by the difference between those who do not read newspapers at all and those who do.


Turning to the television audience, the overall correlation of the television news agenda with the public agenda was .66, slightly lower than the match with newspapers. The analysis, proceeding in the same way as for the newspaper readers, shows the convergence of the ideological subgroups increases monotonically with attention to political news on television, supporting Hypothesis 2 even more strongly than for these subgroups among newspaper readers. There also is support for Hypothesis 2 among the education subgroups. The pattern here is identical to that found among newspaper readers, i.e. a sharp distinction between consumers and non-consumers of political information.

The pattern is also identical for the TV and newspaper audiences among the age subgroups. Unfortunately, neither pattern supports the hypotheses. Finally, the gender analysis for the television audience directly contradicts Hypothesis 2. Convergence among men and women steadily declines with the amount of attention paid to political information on television.

In sum, there is some evidence of convergence among the television audience, but the evidence is weaker than for newspaper readers. Analysis of the index of convergence replicates the four patterns. There is support for Hypothesis 2 among the ideological and education subgroups, but not among the age and gender subgroups.


For the study of consensus at the second level of agenda setting, we selected those data on the attribute agendas of both the media and the public in the elections for the parliament of Navarra. We present first the substantive dimension, the candidates' characteristics, and then the affective dimension, the appraisal of the candidates.


The overall correlation between the substantive images of the candidates in the newspapers and the public's images of these candidates is .41. In the case of television news the overall correlation is .12, essentially no match at all.

Summarizing the results of the direct comparisons--corresponding to the baseline of the triangles in Figure 1--the median correlation between the four demographic subgroups steadily increases with level of attention paid to political information in the newspaper. It is the same pattern that we saw in the level one analyses, a sharp distinction between non-readers and readers of political information. The same pattern prevails for the television audience, but the distinction between non-consumers and consumers of political information is not quite as distinct. What also stands out here to a certain extent is the high level of consensus even among non-consumers of political information in either newspaper or television, a benchmark measure of the shared political culture (results not shown).

Turning to the detailed comparisons based on the index of convergence, Figure 3 shows increased convergence in the agenda of substantive attributes among both young and old with increased consumption of political information in the newspapers and on television. There is a modest monotonic increase across all three levels of consumption, providing good support for Hypotheses 3 and 4 in terms of substantive attributes.


However, among the education, gender and ideology subgroups there is little variation across the three levels of a function to newspaper use, although it should be noted that the index of convergence is relatively high in all cases. That is to say, there is little support for Hypothesis 3 for these subgroups.

In contrast, the support for Hypothesis 4, convergence rising with increased consumption of political information on television, is strong among all four groups. The same monotonic increase noted for the age groups is also present among the education and gender groups. For the ideology groups, the distinction is between non-consumers and consumers of political information on television.


The overall correlation between the affective images of the candidates in the newspapers and the public's images of these candidates is .86. In the case of television news, the overall correlation is. 18, essentially no match at all.

Summarizing the results of the direct comparisons of the four demographic subgroups, results show a familiar pattern of support for the hypotheses. For both the newspaper audience and the television audience, there is a sharp distinction between non-consumers and consumers of political information. Although the pattern is the same, the median correlations are consistently weaker than their counterparts for the substantive dimension of second level agenda setting (results not shown).

Turning to the detailed comparisons based on the index of convergence, Figure 4 shows a strong pattern of support for the second level agenda setting hypotheses. Looking first at the newspaper data, there is a monotonic increase in the index of convergence corresponding to levels of reading among men and women. The familiar distinction between consumers and non-consumers is apparent among the education and ideology subgroups. Only among the age groups is there lack of support for Hypothesis 3. The negative monotonic trend there is directly contrary to the hypothesis.


In the television audience there is support for Hypothesis 4 in all four demographic comparisons. Among the age, gender, and ideology groups there is a monotonic increase in the index of convergence corresponding to increased use of TV for political information. The trend is not monotonic in the education subgroups, but the familiar distinction between non-consumers and consumers is readily apparent. Although the overall level of support is higher for the television hypothesis than for the newspaper hypothesis, the values of the index of convergence for the affective dimension are consistently lower in the television data.


The degree of support for the general proposition that there is an increase in consensus corresponding to increased use of news media for political information varies considerably across the 24 settings examined here. But overall our hypotheses about consensus and media use are supported in 17 of the 24 sets of comparisons displayed in Table 1. The total of 24 is obtained from four demographic groups times six different agendas (the various combinations of levels and media).
TABLE 1 A summary of the support for the consensus hypotheses

              Consensus at first     Consensus at second
               level of agenda         level of agenda
                  setting                 setting

                                     Substantive dimension

groups        Newspapers     TV      Newspapers     TV

Age                                   [check]     [check]
Education      [check]     [check]                [check]
Gender         [check]                            [check]
Ideology       [check]     [check]                [check]

              Consensus at second
                level of agenda

              Affective dimension

groups        Newspapers     TV

Age                        [check]
Education      [check]     [check]
Gender         [check]     [check]
Ideology       [check]     [check]

Comparing the two levels of agenda setting, the agenda of issues and the agenda of attributes, the degree of support slightly favors the agenda of attributes. Five of the eight patterns examined for level one and 12 of the 16 patterns examined for level two support the proposition.

Within level two, there is stronger support for agenda setting at the affective dimension than at the substantive dimension-seven of eight comparisons vs. five of eight. Even though the values of the index of convergence for the ideology groups in Figure 4 are smaller than those found in Figure 2 (the agenda of issues) or Figure 3 (the agenda of substantive attributes), the clear trend of an increase in the convergence of the tone of candidate descriptions is a remarkable finding in a political setting where, typically, positions on the right and left are firmly held. In short, despite firm partisan views there is a movement toward similar affective pictures of the candidates corresponding to use of the news media for political information.

For comparisons of newspapers and television, the levels and dimensions of agenda setting also must be taken into account. At level one, three comparisons support Hypothesis 1 (newspapers) and only two comparisons support Hypothesis e (television) At level two, only one comparison supports Hypothesis 3 (newspapers at the substantive dimension) and all four comparisons support Hypothesis 4 (television at the substantive dimension). For the affective dimension, three comparisons support Hypothesis 3 and all four comparisons support Hypothesis 4. That is to say, across both levels of agenda setting, there is more evidence for a converging effect of television than for a similar effect for newspapers.

Among the four demographic subgroups the clearest distinctions are found among the education and ideology subgroups, five out of six comparisons in both cases. The weakest pattern is found among the younger and older subgroups, only three out six comparisons confirm the hypotheses. For gender, four of six comparisons support the proposition.

Of course, all of these comparisons must also be considered in terms of the actual values of the index of convergence, which vary considerably from setting to setting.

In sum, the theory of the mass media agenda setting offers a useful perspective--whose implications have still not been completely developed--for examining the influence of news on the public mind. Most of the research until now has been centered in the first level of agenda setting, namely an agenda of issues. This investigation, which focused on the maintenance of consensus within society, deals with both the first as well as the second level of the agenda. It addresses not only the transfer of issues, but also the transfer of characteristics and appraisals--attributes--from the mass media to the public. Increasing consensus in the attribute agenda of candidates' characteristics among different population subgroups suggests that the news media make an important contribution to the political environment in democratic societies, leveling out divergent appraisals of political actors that are a reflection of deep ideological cleavages.

(1) The idea that heavy media use evens out differences between socio-demographic (or ideological) groups draws somewhat on the notion of mainstreaming introduced by Gerbner et al. 1980.

(2) Additional details from the content analysis and survey, including full descriptions of the issue and image agendas, are presented in McCombs, Llamas, Lopez-Escobar and Rey (1997a).

(3) a and b being the two correlation coefficients for media and the two subgroups with a [is greater than or equal to] b, and c being the correlation between the two subgroups, the index is computed by (a + b) - (a - b) + c.


Chiang, C. (1995): `Bridging and Closing the Gap of our Society: Social Function of Media Agenda Setting, A Case Study in Taiwan'. Unpublished master's thesis, University of Texas at Austin.

Dearing, J. and Rogers, E. (1996): Agenda Setting, Thousand Oaks, CA; Sage.

Downs, A. (1972): `Up and Down with Ecology: The Issue Attention Cycle', Public Interest, 28, 38-50.

Gerbner, G., Gross, L., Morgan, M., and Signorielli, N. (1980): `The "Mainstreaming" of America: Violence Profile No. 11', Journal of Communication, 30, 10-29.

Lopez-Escobar, E., Llamas, J., McCombs, M., and Le Reyon, F. (1998): `Two Levels of Agenda Setting Among Advertising and News in the i995 Spanish Elections', Political Communication, 15, 225-38.

McCombs, M. (1997): `Building Consensus: The News Media's Agenda Setting Roles', Political Communication, 14, 433-44.

McCombs, M. and Bell, T. (1996): `The Agenda Setting Role of Mass Communication'. In: Salwen, M. and Stacks, D. (eds.) An Integrated Approach To Communication Theory and Research. Mahwah, NJ, Lawrence Erlbaum, pp. 93-110.

McCombs, M. and Evatt, D. (1995): `Los temas y los aspectos: Explorando una nueva dimension de la agenda setting', Comunicacidn y Sociedad, 8, 7-32.

McCombs, M., Llamas, J., Lopez-Escobar, E., and Rey, F. (1997a): `Candidate Images in Spanish Elections: Second-level Agenda-setting', Journalism & Mass Communication Quarterly, 74, 703-172

McCombs, M., Shaw, D., and Weaver, D. (eds.) (1997b): Communication and Democracy: Exploring the Intellectual Frontiers in Agenda Setting Theory, Mahwah, NJ, Lawrence Erlbaum.

Miller, G. (1965): `The Magic Number Seven Plus or Minus Two'. Psychological Review, 63, 81-97.

Shaw, D. and Martin, S. (1992): `The Function of Mass Media Agenda Setting', Journalism Quarterly, 69, 902-20.

Shaw, D. and McCombs, M. (eds.) (1977): The Emergence of American Political Issues, St. Paul, MN, West Publishing.

Wanta, W. (1997): `The Messenger and the Message: Differences among the Mass Media'. In McCombs, M., Shaw, D., and Weaver, D. (eds.) Communication and Democracy: Exploring the Intellectual Frontiers in Agenda Setting Theory, Mahwah, NJ, Lawrence Erlbaum, pp. 137-51.


Esteban Lopez-Escobar is Professor of Public Opinion and Communication Theory, School of Communication, University of Navarra, 31080 Pamplona, Spain. Tel: +34 948 425600; fax: 7 34 948 425636;
COPYRIGHT 1998 Oxford University Press
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1998 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Lopez-Escobar, Esteban; Llamas, Juan Pablo; McCombs, Maxwell
Publication:International Journal of Public Opinion Research
Geographic Code:4EUSP
Date:Dec 22, 1998

Terms of use | Privacy policy | Copyright © 2020 Farlex, Inc. | Feedback | For webmasters