A Spatial Model of Ideology and Political Communication:
Voter Perceptions of News Reporting
James W. Endersby
University of Missouri-Columbia
Ekaterina V. Ognianova
University of Missouri-Columbia
Ekaterina V. Ognianova
116 Walter Williams Hall
Graduate Studies Center
P.O. Box 838
Columbia, MO 65205
tel. (314) 882-2421 or (314) 442-8965
Internet: [log in to unmask]
Paper submitted to the Theory and Methodology Division of the Association
for Education in Journalism and Mass Communication, April 1, 1995
A Spatial Model of Ideology and Political Communication:
Voter Perceptions of News Reporting
The popular commercial press of the second half of the nineteenth
century and down to our own times has had as its central
immediate satisfaction of the largest number of people.
Walter Lippmann (1931)
Is Lippman's observation about American media still accurate? The
transmission of news has both an economic and a political context. The
media are economic institutions, and journalists have a motivation to
appeal to the largest possible audience. But the media consuming public
may also perceive a political bias in reporting. As perceptions of
influence consumer choices among media outlets, journalists may also have
the economic incentive to establish standards of neutrality and
Rather than analyzing ideological content of news reports or elite
assessments of political bias, this paper incorporates citizen
of news reporting as a component of a traditional model of political
ideology. The spatial theory of ideology provides political scientists
a useful model of public opinion, candidate strategy, and vote choice
(Enelow and Hinich 1984; Hinich and Munger 1994). This paper extends the
traditional spatial model to include political communication.
The paper has two objectives. First, it tests the relative placement of
journalists to politicians and voters in an ideological space. If
driven to maximize their audience, then journalists should be located near
the center of the space. Second, the paper suggests that citizen
evaluations of media reporting are a factor in the ideological space.
Political moderates should believe the media to be unbiased, while strong
ideologues should doubt the neutrality of news reporting.
This paper proposes an empirical test of the placement of journalists
within a two-dimensional ideological space. Voter perceptions of major
political and media figures, measured as feeling thermometers scores
national survey, are used to construct this spatial mapping of ideology.
Journalists and media institutions are found to lie near the center of
policy space and voter evaluations of news reporting are related to
second dimension of that space. Data support the conclusion that
communication and potential audience member assessment of news
are important features of a spatial theory of ideology.
A Spatial Theory of Ideology
Political applications of the spatial model borrowed from theories of
economic competition (Hotelling 1929; Smithies 1941). Economists
that competing firms should seek equilibrium in respects like geographical
location. For instance, if consumers choose between similar stores by
minimizing travel time, then two stores in a hypothetical small town
locate next to each other near the middle of Main Street. The economic
concept of equilibrium can be applied to the political world. The
of competing businesses toward imitation can be applied to political
parties during elections. The goal of a political party is to maximize
votes. To do so, parties and candidates purposefully take positions in
center of the distribution of voter preferences. Two parties offer
platforms to ensure that each gets half of the votes cast.
Political commentators frequently refer to politics as a left-right
continuum. In the American context, the primary dimension underlying
ideology resembles a liberal-conservative economic scale. Restricting
issues to a single policy dimension with single-peaked preferences
an equilibrium at the position of the median voter (Downs 1957; Black
1958). A voter chooses to support the candidate nearest to them on the
ideological scale. Parties and candidates should adopt platforms which
appeal to the median voter to maximize vote share. The party with policy
positions distant from the median voter will lose to a centrist party.
Unfortunately, the mathematical proof of the median voter theorem applies
only to a single underlying ideological dimension. Extension of
spatial model to multiple dimensions seems necessary since political
discourse and campaign issues seldom fall onto a single scale. Yet
empirical examinations of voter preferences suggest that the central
tendency remains; political figures are positioned near the median voter
Enelow and Hinich (1984) use survey responses on feeling thermometers from
the American National Election Studies of 1976 and 1980. Enelow
create ideological maps of the electorate and find voter ideal points
perceived candidate locations. Their findings are consistent with
cast in the elections and with predictions of candidate placement.
voter perceptions of the ideological positions of political figures
relatively stable across time (Endersby and Hinich 1992). The results
Enelow and Hinich suggest at least two dimensions within American
ideology. The first conforms to the liberal-conservative economic
dimension; they believe the second dimension includes positions on social
issues. Hinich and Munger (1994) strengthen the original voting model
establishing a theory of ideology. They also suggest the need to
incorporate political communication into the spatial model (1995).
This paper suggests a further application of the spatial theory from
economics and political science to mass communications. In political
applications, competing businesses are replaced by parties and candidates
and customers are replaced by voters. In a mass communications model,
businesses are replaced by media institutions and individual journalists
and the voters are replaced by the potential audience members. This
model links the political and economic models within the field of mass
communications. According to theoretical expectations, competing media
expected to take positions in the middle of consumers' (political)
preferences. The next section of the paper examines this possibility.
Journalism and Political Ideology
Journalists' needs to attract large audiences stems from the commercial
system in which American media operate (Culbertson, 1983). The
and opinions that they provide is one of the many commodities produced
distributed in a free market economic system (Altheide 1984; Ginsberg
Herman & Chomsky 1988; Entman 1989). As early as Joseph Pulitzer's time, a
newspaper was seen inevitably as a business (Baker 1954). Like any
enterprise in corporate America, the news organizations have to follow the
rules of competition and profit. For example, Reiter (1987) has argued
journalists are entrepreneurs trying to make a living by giving the
audience the information that they think it would accept. Similarly Gans
(1972, 1980, 1981) has suggested that journalists' values are those of
In the larger context of liberal capitalism, the marketplace of ideas is
just one component of the free market economy. Most American
independent from the government, market-driven, and profit-oriented
(Hallin, 1987). They need to please the audience and attract advertisers.
It is only logical that they would follow the same rules as other
commercial products competing in an open market do and that the market
mechanisms apply to them.
Not all media should seek a neutral, centrist position. Media targeting
specialized markets should move toward the center of the subset of
consumers within that market, not of the public at large. Early American
newspapers, for instance, were clearly partisan in nature and sought
to satisfy party identifiers. Media in socialist and authoritarian
also have less incentive to maintain standards of objectivity.
operated media should first satisfy the ruling elite; only in
and nations seeking popular support should seek impartial news
American print and broadcasting media depend on advertising and profit from
maximizing their consumers. In broadcasting, the addition of more viewers
or listeners does not increase the cost to producers but may
interest of advertisers (Nord, 1980). With print media, the largest
making the initial copy; any additional copy costs less and might pay off
from advertising revenues. Profit for mass media products is volume.
Shoemaker and Reese (1991) emphasize that the primary goal of privately
owned media organizations is profit. This economic goal is
major American media companies due to their stockholder ownership. The
increasing tendency in the media industry toward big monopolies,
cross-ownership, and control by outside corporations has multiplied the
economic stakes. Media giants with more human and financial resources
lose, should be more likely to follow their customers' demands and
fewer editorial, ideological, and, ultimately economic risks. That is
to say that smaller media are not also profit-driven. Few media in the
United States rely on subscription or sales for profit-making; the
source of revenue for mainstream media is advertising.
For any news organization the audience appeal "translates into higher
circulation and ratings, producing greater advertising revenue"
& Reese, 1991: 122). "We always look for programming that appeals to
everybody," says Herman Keld, programmer for CBS, regarding television
programming in general (Gitlin, 1983: 57). For news media, the target
audience is largely the same as the electorate taken as a whole.
audience share usually means a non-controversial, safe approach to news
coverage: source-domination or depending on government officials,
and think-tanks (Chomsky, 1985; Sigal, 1986; Herman & Chomsky, 1988;
Parenti, 1990), framing news stories as conflict, for example election
stories as "horse-race" (Patterson 1974; Entman, 1989; Lemert,1989;
et al., 1991), episodic coverage of events without connecting them to
larger issues or alternating "he said, she said" quotes (Kovach, 1992).
Despite the salience of political news, reporting seldom is detailed or
analytical. Norms of objectivity, the deliberate detachment of
journalists' personal political beliefs from their news stories, has become
a standard of American journalism. Those norms equate to a moderate and
nonpartisan political ideology (Gans 1972; Ognianova and Endersby
Modeling ideology and perceptions of bias in mass communications requires
national survey data of potential voters and consumers.
evaluate political and media figures. Feeling thermometer scores may
as summary measures for evaluations of candidates and other elites.
thermometer ratings provide a measure of distance between the
and the political figure enabling researchers to produce a map of the
policy space. This study uses data collected during the 1974 American
nal Election Study conducted by the Center for Political Studies at
University of Michigan (Miller et al., 1975). Though the National
Study is conducted biannually, the 1974 survey is the only occasion in
which respondents are asked to rate a variety of media figures along
political elites on the feeling thermometers.
The 1974 American National Election Study surveyed 1575 randomly selected
American adults for personal interviews conducted from November
5, 1974 to
January 31, 1975. The survey provides comprehensive information about
respondents' political attitudes and voting behavior during the period of
Watergate and the 1974 Congressional Elections. Included in the survey
questionnaire are a series of items asking the respondent to rate
and media figures on a feeling thermometer from 0 to 100. A score of 0 i
ndicates that the respondent feels "cold" or negative toward the
group; a 100 means the respondent has a "warm" or positive feeling for
person or group. A score in the middle, 50, indicates the respondent
particular feelings one way or the other. Political figures rated by the
respondents included Hubert Humphrey, Ted Kennedy, Richard Nixon, and
George Wallace; political groups included Democrats, Republicans,
conservatives and liberals. In addition, the survey asked respondents to
rate three television news anchors, John Chancellor, Walter Cronkite,
Howard K. Smith, along with TV commentators and newspaper editors in
general. Table 1 reports the mean and standard deviations of the
thermometer scores for the filtered sample described below. Television
reporters scored somewhat higher in the aggregate; politicians lower.
survey instrument also requests information about mass media
and attitudes toward the media, including the performance of the media
the Watergate coverage.
Summary Statistics of Thermometer Scores
Person or Group
Edward "Ted" Kennedy
Howard K. Smith
n = 1025
Among the full sample, the respondents said they followed public affairs
most of the time (39 percent), read a daily newspaper (72.6
frequently watched the television networks national news shows in the
evening with anchors such as Cronkite, Chancellor, and Reasoner-Smith
(58.6 percent). Out of the three early evening news broadcasts, CBS was
watched the most (44.9 percent), followed by NBC (32 percent), with ABC
watched least (21.3 percent). Respondents often read more than one
r (perhaps local and regional). The majority also said they read
about politics frequently, with 39.1 percent reading frequently about
national politics and 41.5 percent reading frequently about local
The party identification of the respondents was distributed as follows:
18.4 percent were strong Democrats, 20.5 percent were weak
percent were Independents leaning toward the Democrats, 13.7 percent
Independents with no preference for the Democratic or the Republican
9.3 percent were Independents leaning toward the Republican party, 13.9
percent were weak Republicans, and 9.2 percent were strong
the liberal-conservative continuum, the majority of the respondents
percent) viewed themselves as middle of the road, followed by slightly
conservative (17.9 percent), conservative (16.1 percent), liberal (13.3
percent), slightly liberal (10.6 percent), extremely conservative and
extremely liberal (2.2 and 2.1 percent respectively).
The methodology does not allow missing data so respondents who did not rate
all thirteen figures listed above were removed from the sample. In
addition, a filtering procedure was devised to divide thermometer
into nine intervals; individuals who gave similar ratings to 90
more of the evaluated political figures and media personalities were
seriously responding to the survey and were also removed from the
After the omission of missing data and the execution of the filtering
edure, the sample contains 1025 respondents.
Spatial models of politics construct an ideological map containing
perceived positions of political figures as well as the most preferred
points of the voters. An empirical model of the electorate can be
using the Cahoon-Hinich method of point and range location, and a spatial
mapping from the survey responses can be constructed. The model
calculations of the approximations of the policy distance from voter
points to each of a set of known political figures. This
jm, of voter m for candidate j can be represented as follows:
Tjm = cjm - [||xj-zm||2 + aVj]1/r ,
where xj is a vector of the jth candidate's ideal point in the ideological
space, zm i the vector of the mth voter's ideal point in the
space, vj is a scalar representing the valence position (the nonpolicy
effects on candidate j's evaluation), a is the sensitivity of the valence
dimension, r is the measure of sensitivity of voters to candidates
their ideal points, and cjm is a scalar denoting influence of
nonsystematic variances on Tjm. The scalar term representing the
characteristics of a candidate, aVj, was constrained to be zero;
data for accurate estimation of these nonpolicy values is not available.
The methodology used here is identical to that developed by Cahoon and
Hinich (1976), Cahoon, Hinich, and Ordeshook (1978), Hinich (1978),
Enelow and Hinich (1984: ch. 9). An estimation of the positions of the
candidates in the ideology space is determined as follows: The
scores for each of the candidates and the relevant groups are adjusted
that a zero indicates a voter's preferred position and a 100 is the
distance from that ideal point. The differences between candidates and vo
ters are raised to the second power as a measure of squared Euclidean
distance. In addition, each of the scores is subtracted from a single
person or group, in this case, the Democrats. The selection of this pivot
The adjusted scores compose the covariance matrix for a factor analysis
that results in an estimate of the political figures' position in
respondents' ideology space. The factor analysis is followed by two
separate regressions. The first regression eliminates the error terms. The
second regression estimates the remaining parameters for the model and
ideal points of the respondents in the policy space. This allows the
creation of a map that places both the political figures and the
respondents' ideal points in the ideological space.
Results: Political and Media Figures in the Ideology Space
Figure 1 presents the estimated positions of politicians and journalists in
a two-dimensional ideological space. A two dimensional model explains 62
percent of the variance in the 1974 scores. The range of the
respondents is not included in Figure 1 and is much larger than indicated
by this graph. The full distribution of respondents, shown in Figure
relatively elliptoid. The median and mean voter in each dimension is
located at the origin. The numeric value of the ideal points only
demonstrates the relative placement between positions; the magnitude has no
meaning in and of itself.
Placement of politicians, parties and ideological groups in the policy
space meets general expectations of their locations. Political
aligned primarily on a left-right continuum from Ted Kennedy on the left
to Richard Nixon on the right. Only the position of George Wallace
suspect, yet he is on the right side of the liberal-conservative
All of the political figures and groups lie to one side of the median
on the second dimension, perhaps exhibiting the electorate's dissati
sfaction with politics in 1974. Though the politicians are shifted to one
side of the y-axis, they are still near the center of the
Relative locations can be determined by comparing Figure 1 and Figure 2.
Journalists, however, are located exactly where we would expect in a media
market. The three television anchors are clustered near the
origin in the
space. Groups of journalists, TV commentators and newspaper editors
located nearby. The hypothesis that journalists should be located near
center of the space finds strong support among ratings of the 1974
Yet the placements so far have no meaning in and of themselves, only
relative affective positions between respondents and perceived
of well-known individuals and groups. Can meaning be given to the
dimensions? Enelow and Hinich find that the first dimension in this space
corresponds to policy preferences along a liberal-conservative
and that the second dimension is related to preferences on social
such as abortion and drug use. To allow interpretation of the dimensions,
self-placement of individuals on policy questions needs to be
into the analysis.
The 1974 survey asked respondents to identify their partisanship and their
political ideology. Separating respondents according to their
self-placement on each of the scales allows creation of a median score
each category in each dimension. Those median scores are plotted
positions of the political and media individuals and groups in Figure
Here the first dimension is given a clear interpretation. Partisanship
related to voter placement along the x-axis. Ideological orientation
related to positions on the x-axis even more strongly than political
identification. Self-placement on the liberal-conservative spectrum
corresponds to this first dimension, thus we can give meaning to this
Interpretation of the second dimension in spatial models of ideology has
consistently been a problem for political scientists. The "social
characterization is often used, but the second dimension (the y-axis
shown here) comprises more of a bundle of issues that are not easily
on the left-right economic scale. The inclusion of media figures allows
more clear representation of the second axis in the context of
communication. The 1974 surveys also asked respondents questions on
performance of the television and newspaper reporting. Figure 4 plots
median responses to two of the questions on coverage of Watergate and
assessment of television news.
Responses on these questions do not strictly follow the dimension of
partisanship and economic ideology. Though not perpendicular to the
median responses for each category correspond to voter placement on the
y-axis. Individuals with a positive evaluation of the news media lie
closest to the origin and to the placements of the journalists. Those with
negative evaluations tend to lie in the northwest quadrant, close to
perceived position of President Nixon. The second dimension in this
is related to evaluation of the role of the media in political reporting
(though not exclusively so).
Interpretations and Conclusions
Selection of 1974 election as the context of the survey is somewhat
unfortunate as evaluations of media and political figures was likely
intensified during the Watergate era. Yet expectations of greater
ideological emphasis of news organizations was not found. On the contrary,
respondents placed journalists in the center of the policy space.
The fit of the model excluding the media personalities and groups is
somewhat better than the model reported here (82 percent of the
explained). Though the two models are not directly comparable, additional
placement of journalists produces somewhat greater error. Yet the null
assumptions, if ideological content is not important to evaluation of
figures, would suggest that the method would place journalists and
politicians in a random and uninterpretable fashion. The findings here are
reasonable placements and, with the addition of survey responses on
questions, locations of both politicians and journalists take on
the context of the political environment of 1974.
Both of the primary research objectives are met in this analysis. Major
media commentators are located near the center of ideological
of the general public. Journalists are evaluated as moderate by the
even in the controversial years of the mid-1970s. Yet evaluation of
journalists also takes on political content. Relative placement of
politicians and journalists in part is related to perceptions of media
responsibility and fairness.
The spatial model of ideology provides a successful means to analyze the
role of the national media in public affairs. Mainstream media
seek the middle of the ideological preferences. Yet that strategy of
journalists, at least during the Watergate era, seems to take on meaning.
Evaluation of media performance is related to a second dimension of
ideological preferences, and journalists are rated in comparison to the
institutions on which they report. The findings here suggest that
journalists successfully seek to be perceived as ideologically neutral, yet
that search also has an ideological content.
Altheide, D. (1984). Media hegemony: A failure of perspective. Public
Opinion Quarterly, 48, 476-490.
Baker, R. T. (1954). A history of the Graduate School of Journalism of
Columbia University. New York: Columbia University Press.
Black, D. (1958). The theory of committees and elections. Cambridge:
Cambridge University Press.
Cahoon, L., Hinich, M. J., & Ordeshook, P. C. 1975. "A Statistical
Multidimensional Scaling Method Based on the Spatial Theory of Voting."
P. C. Wang (Ed.), Graphical Representation of Multivariate Data (pp.
243-278). New York: Academic Press.
Chomsky, H. (1985, October). The bounds of thinkable thought. The
Culbertson, H. M. (1983, June). Three perspectives on American journalism.
Journalism Monographs, 83.
Downs, A. (1957). An economic theory of democracy. New York; Harper
Endersby, J. W., & Hinich, M. J. (1992). The Stability of Voter
Perceptions: A Comparison of Candidate Voter Positions Across Time Using
the Spatial Theory of Voting. Mathematical and Computer
Enelow, J, M., & Hinich, M. J. (1984).The spatial theory of voting: An
introduction. Cambridge: Cambridge University Press.
Entman, R. M. (1989). Democracy without citizens: Media and the decay of
American politics. New York: Oxford University Press.
Gans, H. J. (1972). The famine in mass media research. American Journal of
Sociology, 77, 697-705.
Gans, H. J. (1980).Deciding what's news: A study of CBS Evening News, NBC
Nightly News, Newsweek and Time. New York: Random House.
Gans, H. J. (1981). Values in the news. In I. Robertson (Ed.), The social
world (pp. 40-48). New York: Worth Publishers.
Ginsberg, B. (1986).The captive public: How mass opinion promotes state
power. New York: Basic Books.
Gitlin, T. (1983). Inside Prime Time. New York: Pantheon Books.
Hallin, D. C. (1987). Hegemony: The American newsmedia from Vietnam to El
Salvador, a study of ideological change and its limits. In D. L.
(Ed.), Political communication research: Approaches, studies,
(pp. 3-25). Norwood, NJ: Ablex.
Herman, E. S., & Chomsky, N. (1988). Manufacturing consent: The political
economy of the mass media. New York: Pantheon Books.
Hinich, M. J., & Munger, M. C. (1994). Ideology and the theory of political
choice. Ann Arbor, MI: The University of Michigan Press.
Hotelling, H. (1929). Stability in competition. Economic Journal, 39,
Kovach, B. (1992, March 20). Speech at the third annual conference on
computer-assisted journalism. Indianapolis, IN: National Institute for
Lemert, J. B. (1989). Criticizing the media: Empirical approaches. Newbury
Park, CA: Sage.
Lippmann, W. (1931). Two revolutions in the American press. The Yale Review
, 20, 433.
Miller, W. E., Miller, A. H., & Kilne, F. G. (1975). The CPS 1974 American
national election study. Ann Arbor, MI: Inter-University Consortium
Neuman, W. R., Just, M. R., Crigler, A. N. (1991). Common knowledge: News
and the construction of political meaning. Chicago: University of
Nord, D. P. (1980). An economic perspective on formula in popular culture.
Journal of American Culture, 3, 17-31.
Ognianova, E. V., & Endersby, J. W. (1995). Objectivity revisited: Ideology
and political communication. Unpublished paper.
Parenti, M. (1990, May). Pack pluralism. Lies of Our Times, 15.
Reiter, H. L. (1987). Parties and elections in corporate America. New York:
St. Martin's Press.
Shoemaker, P. J. & Reese, S. D. (1991). Mediating the message.: Theories of
influences on mass media content. New York: Longman.
Sigal, L. V. (1986). Sources make the news. In R. K. Manoff & M. Schudson
(Eds.), Reading the news: A Pantheon guide to popular culture. New
Smithies, A. (1941). Optimum location in spatial competition. Journal of
Political Economy, 49, 423-439.
 As Entman (1989) notes, even without the economic pressures behind
them, media would still have the motive of pleasing their aud
could bring prestige, a sense of reaching and inf
luencing large groups of
people, and of fulfilling the missio
n of public service. Non-commercial
media also cater to their
audience's interests and in this respect, could
the scope of traditional economic analysis.
 The first eigenvalue of 13
2.0 explains 44.3 percent of the variance,
the second eigenva
lue of 52.3 adds 17.5 percent, and the R-squared value
e first regression is .978. The R-squared value for the second
regression which estimates relative spatial coordinates of candidates and
voters is .965, quite high for the construction of the map.
 Using another method to derive voter and candidate positions, Poole a
Rosenthal (1984) and Rabinowitz (1978) find spatial locations for
Presidential candidates that are located on the periphery of the
ideological space. That is, that candidates are among the mos
individuals in the electorate. This position seems
poorly supported by
other evidence, empirical or theoretical.
The researchers do not support
interpretations of locations
by other survey data. Peripheral locations
seem a function of
both the method and the small number of candidates used.