Influence of Source Liking and Personality Traits on Perceptions of Bias and Future News Source Selection

Abstract

This study examined news selection intentions that followed judgments of story bias and the extent to which those intentions were influenced by source liking. The study also examined the extent to which two personality traits, argumentativeness and need for cognition, affected perceptions of story bias and intent to select an offending source in the future. Participants were more likely to say they would return to the source of a “biased” story if they liked and had selected the source previously than they were if the source was unknown. Individuals high in argumentativeness were less likely than those low in argumentativeness to view a story with negative information about their group as biased. Need for cognition was weakly related to intent to return to an offending source. Implications for consumption of counterattitudinal information and source selection in the current news environment are discussed.

Notes

∗p < .05 (one-tailed).

∗∗p < .01 (one-tailed).

† = p < .10 (one-tailed).

¹ However, the study employed the survey method, making it difficult to determine if skepticism of news media led to varying consumption levels or if varying consumption levels led to increased skepticism.

² Initial conceptualizations of the hostile media effect involved a comparison of a partisan group's perception of story bias to a nonpartisan group's perception. Because we were interested in moderators of bias judgments (i.e., article source and argumentativeness) and in decisions that follow those judgments, we did not think it necessary to include an oppositional group of participants, and we chose an article designed to elicit strong perceptions of bias among our participants. Research on the HME suggests the article would have been perceived as less biased against our participants/students by a group of participants at a rival or other school. Additionally, our participants should have perceived the story as less biased against their group if the story had not mentioned their school specifically—however, the question of interest here was whether they would consider it less biased when it appeared in a more liked and used news source.

³ Our operationalization and use of the term bias as a key variable here is based on previous hostile media effect studies that attempted to tap perceptions of the unfairness of news content toward a particular group (c.f., Giner-Sorolla & Chaiken, 1993; Perloff, 1989; Vallone, Ross, & Lepper, 1985). We acknowledge, however, that bias is often considered a marker of the larger construct of news credibility (Gaziano & McGrath, 1986; Meyer, 1988; West, 1994) and suggest that future studies attempt to reconcile the implications of various operationalizations of such closely related concepts.

⁴ It should be noted that participants had not visited the Web site in order to view the stimulus and control stories; the stories had been printed and appeared as though they had been downloaded from the sites. Therefore, it could be argued that the item that asked participants if they would be likely to read a story on the given site again, thus, implying that they selected the article and visited the site in the first place, was not a completely valid measure of future selection intent. It is possible that this item was, therefore, somewhat confusing to participants. However, we believe the item's strong correlation with the item measuring likelihood of reading a story by the given reporter again suggests that the item was interpreted consistently with our intent.

⁵ Although the presence of any liking score might seem problematic for a fictitious new site it is likely that participants generated an attitude toward the fictitious site based on attitudes toward the broader category of news sources in general after rating several sites (Schuman & Presser, 1981). What is important for the current study is the significant difference in liking scores between the known (CNN.com) and unknown source.

⁶ Even though source by bias interactions were found for future source selection, the collapsing of source variables for the path analysis was considered appropriate to determine the relationship between personality variables, bias ratings, and future selection across news sources. Further, regression analyses (conducted in the same manner as those for H1) found no interaction of news source and argumentativeness (mean centered) on bias ratings, t = .53, p > .10 (two-tailed) [argumentativeness predicted lower bias ratings β = −.20, p < .05 (one-tailed); CNN as source did not predict bias ratings, t = 1.87, p > .05 (two-tailed), R ² = .04, F(3, 159) = 2.31, p > .05 (two-tailed)]. Additionally, regression analyses found no interaction of news source and NFC (mean centered) on future source selection, t = .32, p > .10 (two –tailed) [NFC did not predict future selection t = .77, p > .10 (two-tailed); CNN as source positively predicted future selection, β = .30, p < .001 (two-tailed), R ² = .11, F(3, 158) = 6.18, p < .001 (two-tailed)].

⁷ Although the sample size for the current study is smaller than ideal for using path analysis, structural equation modeling scholars have acknowledged that smaller samples can be used with models that are fairly simple in nature, include highly reliable latent variables, and that are evaluated using fit indices shown to be less sensitive to variations in sample size (Holbert & Stephenson, 2002). The models in the current study fit these first two criteria and were evaluated with two of the fit indices and cut off criteria recommended for smaller sample sizes (Hu & Bentler, 1998; 1999): comparative fit index at a minimum of CFI > .95, and standardized root mean square residual (SRMR) at a minimum of < .08. Chi-square estimates have been reported for additional information and are considered acceptable when the observed value is not significant at the 95% confidence level (Schumaker & Lomax, 1996).

⁸ Although the amount of variance explained here is low, one should consider that this analysis combined cases from both the CNN.com and allnews.com stimulus article conditions. Tests of H1 suggest that future selection intentions were less affected by bias perceptions in the CNN.com condition than in the allnews.com condition, perhaps accounting for a rather weak relationship between bias perceptions and future selection in this model.