It's Not That We Don't Know, It's That We Don't Care: Explaining Why Selective Exposure Polarizes Attitudes

Abstract

Today audiences can select content that is consonant with their political notions, and they take advantage of this opportunity. Such partisan selective exposure to media outlets has been identified as one of the reasons for attitude polarization. Little research attention, however, has been devoted to the processes underlying this phenomenon. This article advances the understanding of selective exposure by testing whether its effect on attitude polarization is attributable to people's familiarity with arguments that reinforce or challenge their own opinions. A sample of politically engaged extremists (N = 440) was subjected to a structured interview. As hypothesized, users of partisan media were more familiar with arguments reinforcing their views and, as a result, more prone to polarization. Knowledge of challenging arguments, however, was not affected by consumption of partisan media, nor did it promote depolarization. Results are discussed in light of two conflicting theories: the persuasion and the motivational-reasoning model.

Notes

¹The discussion regarding the existence of a consistent ideology among the mass public is too extensive to be encompassed by this study. Yet it is worth noting that Converse's work has been challenged, both methodologically and theoretically (e.g., Dalton, 2000; S. Feldman & Zaller, 1992; Jacoby, 1995). Feldman and Zaller (1992), for instance, disagree with Converse's suggestion that the public does not understand the meaning of Left and Right yet agree that the political attitude among the public may not be captured by a simple conceptualization of a left-right dimension.

²It is important to note that reverse causality is also possible, that is, that stronger attitudes may lead to a preference for like-minded content. However, Stroud (2010) found the effect of selective exposure on polarization to be stronger and more consistent than the effect of polarization on selective exposure.

³In Israel military service and reserve duty are mandatory, the latter till the age of 45 to 70, depending on the position and rank; and indeed during the Disengagement Plan many reserve soldiers were drafted.

⁴For instance, in September 2010 a Supreme Court judge stated that he would have refused to evacuate settlements had he been called to do so (Ma'ariv, 09.08.10).

⁵Mainly, key demographics were compared, as well as a model testing the effects of demographics and exposure to mainstream media on attitudes toward refusal (a representative sample from 2009 was used, as well as a sample of settlers from the same survey). The same regression model was tested on a representative sample from 2004, with the additional variable of familiarity with arguments against refusal.

⁶The designation of a media outlet as partisan was based on previous research (e.g., Soffer, 2011) as well as on public perceptions. For instance, based on content analysis, previous research demonstrated the presence of dovish views in Haarerz editorials and news section (Korn, 2004). The only broadsheet newspaper in Israel, Haaretz has low circulation, and most of its readers are left wing. Its right-wing counterparts are Makor Rishon and Hatzofe, as well as the radio station Arutz 7, whose audience is homogenously right wing.

⁷In addition, the model presented here controls for the correlation between familiarity with reinforcing and challenging arguments. Thus, any tendency for social desirability is captured by this correlation, as well as a more general perception of one's knowledge. This allows a more accurate attribution of the remaining effect to the different types of knowledge. Examples for arguments in favour of and against refusal are: Refusal creates an additional burden on other members of my military unit; If I refuse, then the left-/right-wing refusal also becomes legitimate; In a democratic society people have the right to act according to their conscience, if they are willing to pay the price for their actions.

⁸SPSS macro was supplied by Andrew Hayes upon request.

⁹Other variables were also found significant (self-endorsed values: F(1, 403) = 6.3, p < .05; conservative values: F(1, 403) = 5.3, p < .05; religious beliefs: F(1, 403) = 18.3, p < .01).

Note. All coefficients were extracted using Hayes's (2013) syntax for assessing moderated mediated models. Gender = women as baseline; political leaning range from 1 (extreme right) to 7 (extreme left).

*p < .05. **p < .01. †p < .10.

¹⁰Such small effect is typical for mediated models and is also a result of the large number of control variables.

Additional information

Notes on contributors

Shira Dvir Gvirsman

Shira Dvir Gvirsman (Ph.D., Hebrew University, 2011) is a Lecturer in the School of Communication at Netanya Academic College. Her research interests include public opinion, polarization, and the psychological processes underline media effects.