In the Eye of the Beholder: How Information Shortcuts Shape Individual Perceptions of Bias in the Media

Download data and study materials from OSF

Principal investigators:

Phil Gussin

University of California, Los Angeles


Matthew Baum

Harvard University



Sample size: 1014

Field period: 7/12/2005-7/22/2005


Research has shown that human beings are biased information processors. This study investigates an important potential example of biased information processing: whether individual perceptions of media bias are triggered by message content or individual assessments of source credibility, combined with internal characteristics, like ideology. We conduct an experiment in which subjects coded the content of a news report identified as originating from CNN, FOX or a fictional TV station. Our results suggest that ideology, political awareness, and source credibility all mediate the valence and extent of perceived media bias. Moreover, because individuals distinguish between media outlets, outlet “brand names,” and the reputations they carry, function as heuristics, heavily influencing perceptions of bias in content. This suggests that assessments of media content operate on a more nuanced level than has been captured in previous research, which has focused on the "mass media” as a whole.


H1: Relative Hostile Media Outlet Presumption (RHMOP) Hypothesis: Ideologues will tend to perceive media content from an unfamiliar outlet as morefavorable to the opposing ideological perspective than will ideologues with the opposing ideological perspective, all else equal.
H1a: Hostile Outlet Extremeness (HOE) Hypothesis: Among familiar media outlets, ideologues will perceive those that they consider hostile, but not those they consider non-hostile, as more ideologically extreme than moderates will consider those same outlets, all else equal.
H1b: Political Awareness Hypothesis: The polarization between liberal and conservative perceptions of media content will increase as political awareness increases.
H2: Heuristic Hypothesis. Individuals' perceptions regarding the ideological content of news from a given media outlet will vary in a manner consistent with their prior beliefs concerning the outlet's political orientation.
H2a: Heuristic Intensity Hypothesis: The stronger the heuristic value to an individual of the outlet label, the greater the extent to which the label will mediate that individual's perceptions concerning the ideological orientation of news content.
H2b: Familiarity Hypothesis: The heuristic value of a familiar media outlet label will mediate an individual's perceptions of media content more than a global assessment of the media as a whole.

Experimental Manipulations

Subjects were randomly assigned to a version of a news transcript designed to appear as having originated on either CNN, FOX, or KNWZ, a fictional TV news station, with all identifying information altered to insure consistency with the outlet identification (including fictional reporter names for the unfamiliar station). The transcript consisted of one quote from each candidate, as well as six comments that were, to the greatest extent possible, "balanced" (in terms of positive, negative, and neutral references to the candidates). Subjects read each comment and evaluated whether and to what extent it was favorable to or critical of Senator Kerry and President


For each of the transcript segments (six comments by journalists and one each by Bush and Kerry), we presented six response options for each candidate: "Very favorable," "Somewhat favorable," "Somewhat unfavorable," "Very unfavorable" to Kerry (Bush), "Balanced/Even handed", and "No coverage of Kerry (Bush)." Our dependent variable is the sum of all Bush codes, minus the sum for Kerry codes. The resulting scale runs from -3 to +3, with -3 representing maximum relative favorability toward Kerry, 0 representing neutrality or balance, and +3 representing maximum relative favorability toward Bush.

Summary of Results

The following is a list of our hypotheses and the proportion (%) of tests supporting each hypothesis.
H1 (RHMOP Hypothesis) was supported in 1/1 (100%) tests. H1a (HOE Hypothesis) was supported in 2/4 (50%) tests.** H1b (Political Awareness Hypothesis) was supported in 3/3 (100%) cases. The totals for H1 and Corollaries: 5/7 (71%).
H2 (Heuristic Hypothesis) was supported in 14/14 (100%) tests.
H2a (Heuristic Intensity Hypothesis) was supported in 14/15 (97%) tests. H2b (Familiarity Hypothesis) was supported in 4/4 (100%) tests.
** This figure treats liberal ratings of CNN as inconsistent with hypothesis, when, in fact, it is arguably consistent. Hence, this is a conservative estimate.


Our results are broadly consistent with previous research on the hostile media phenomenon. We extended this argument by applying the implications of recent research in social, cognitive, and evolutionary psychology on the importance of message credibility and on the role of heuristic cues to the
question of whether and when individuals will perceive media content as hostile or non-hostile. Our research suggests that the heuristic effect of outlet labels extends beyond merely conditioning whether or not an individual is prone to accept or reject a given message to influencing how individuals perceive the actual content of information. In other words, not only do citizens disproportionately counter-argue dissonant information, but they sometimes also "create" dissonance even where none actually exists. This suggests that the effects of media outlet labels on individual perceptions of bias emerge through an interaction between heuristics and biased information processing.

Additional Information

Independent Variables. For our first set of key causal variables, we interact a series of ideology dummies (liberal, moderate, conservative) with dummies for the three treatment conditions (CNN, FOX, KNWZ). Our other key causal variables are based on the distance between respondents' self-ideological ratings and their ratings of CNN and FOX, respectively, as well as, for the unfamiliar outlet, the distance between themselves and their global assessments of the ideological orientation of "the mass media." We also include controls for respondents demographic characteristics (e.g., family income, age, education, southern residency, internet access, race), political knowledge (based on a series of factual knowledge questions), and, where appropriate, overall attitude regarding media bias.


Baum, M. A., and P. Gussin. 2007. "In the eye of the beholder: How information shortcuts shape individual perceptions of bias in the media." Quarterly Journal of Political Science 3:1-31.