One-Track Minds? Cognitive Needs, Media Diet, and Overestimation of Public Support for One's Views

Abstract

This study suggests that diversity of media diet mediates the relation between cognitive needs and accurate perception of public opinion. It was hypothesized that people with high need for cognition will expose themselves to various media outlets that support and that oppose their views; as a result, they will develop a more accurate perception of public opinion. In contrast, people with high need to evaluate will show preferences to like-minded media outlets and, consequently, will perceive their opinion to be more popular than it actually is. The hypotheses were tested using a survey conducted with a sample of 450 Israeli participants. The results confirmed the hypotheses regarding need for cognition, but the results regarding need to evaluate pointed to cognitive biases other than selective exposure.

NOTES

FUNDING

This material is based upon work supported by the Israeli Ministry of Science under Grant No. 3-8784. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author and do not necessarily reflect the views of the Israeli Ministry of Science.

Notes

1. Nir's (2011) full model actually suggested four different types of political thinking. However, given that the only differences found in her analysis were between partisan thinkers and rational thinkers, the current study reviews only these two types. In addition, the analytical strategy taken here was one of moderation rather than actually dividing the sample into four categories according to reasoning style.

2. Israeli Arabs (who comprise approximately 18% of the Israeli population) were not included in the current study.

3. Ultraorthodox Jews refrain from using the Internet for religious reasons: The content available is considered a temptation that may entice individuals away from their righteous path.

4. Respondents' own opinions on the issue were obtained first. Yet, changing the order of the questions had not produced significant difference in previous studies (Fabrigar & Krosnick, 1995).

5. Obtaining a specific point estimate served an additional purpose beyond eliciting a majority or minority estimate. We used these estimates because we wanted to examine respondents' ability to provide accurate estimates of opinion distributions.

6. The political orientation of outlets was based on expert rating, which was followed by a reliability test (Krippendorff's α of .81). The list of right-wing outlets included online outlets (Channel 7, Latma, Makor Rishon, Yesha News, Rotter.net, Srugim, Kr8, and Hazofe), print outlets (Israel Hayom, Makor Rishon, Yated Neeman, and BeSheva), and radio outlets (Channel 7, Kol Hai Radio, and Galei Israel). The list of left-wing outlets included online outlets (Haaretz.co.il, Yesh Gvul, Lo Nistom, Ha Small Ha Leumi, Hagada Ha Smalit, Haoketz, Ha Haverim Shel George, Ha Televizia Ha Hevratit, Magazine Hakibus), unspecified “other online publications of social or human rights organizations,” the print version of Haaretz, and Kol Hashalom radio station.

7. Interestingly, a significant interaction was found between the two needs when predicting attitude extremity (as stems from Nir's [2011] model). Both needs were associated with stronger attitudes (need for cognition: b = 0.07, SE = 0.02, p < .01; need to evaluate: b = 0.15, SE = 0.02, p < .01). Yet, the interaction between the two was negative (b = −0.05, SE = 0.01, p < .01), suggesting that those high in need to evaluate and also high in need for cognition hold weaker attitudes in comparison to those with high need to evaluate and low need for cognition. Strength of attitudes also mediated the relation between need to evaluate and crosscutting exposure (b = 0.04, SE = 0.01, 95% CI [0.001, 0.011]). Those high in need to evaluate showed stronger attitudes, which lead to less crosscutting exposure. This mediated effect was moderated by need for cognition: at the higher levels of need for cognition, the mediated relation was weaker (low need for cognition—one standard deviation below the mean, b = −0.01, SE = 0.00, 95% CI [−0.035, −0.003]; high need for cognition—one standard deviation above the mean, b = −0.01, SE = 0.00, 95% CI [−0.018, −0.001]). However, these effects were not found for like-minded content. Nor did they significantly influence public opinion perceptions.