Ethnic Selective Exposure: A Test of Cultural-Identity Based Media Selectivity Theory

ABSTRACT

This study proposes and tests a theory of ethnic selective exposure – namely, ethnic audiences’ use of English- and ethnic-language media as a function of their most salient or strongest cultural identity among the many available to them (American, ethnic American, pan ethnic, or national origin). This identity-based media selectivity is examined with experimental data from a 2016 survey experiment and observational data from the 2006 Latino National Survey. Results consistently showed that ethnic audiences preferred to use media that are congruent with their most salient or strongest cultural identity, especially when they sought out information related to politics and public affairs. The 2016 experiment also suggests that immigrants’ everyday experience living in the nation can affect their media selection, which in turn influences their incorporation into the U.S. and thus points to the complexity of assimilation and the media’s role in it.

Acknowledgments

Study 1 was supported by the John Maxwell Hamilton Fellowship for Media & Public Affairs Research from Manship School of Mass Communication at Louisiana State University. I thank Drs. Johanna Dunaway, Kathleen Searles, Martin Johnson, James C. Garand, and Ray Pingree for their great help with the design of the stimulus materials. I also thank Drs. Johanna Dunaway, Josh Darr, and Newly Paul for their assistance in proof-editing this paper.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary material

Supplemental data for this article can be accessed on the publisher’s website.

Notes

¹ Following previous research (see Salzman, 2014), this study chooses the term “Latino” over the more classic “Hispanic” and the gender-neutral term “Latinx” to denote individuals of a Latin American cultural or ethnic identity in the United States.

² See the selective exposure literature for a full view of the impact of other self-concepts – such as age (Knobloch-Westerwick & Hastall, 2010), gender (Trepte, 2006), partisanship (Iyengar & Hahn, 2009), and nationality (Wojcieszak & Garrett, 2018) – on media selection.

³ This experiment underwent exempted review and was fully approved by Louisiana State University’s institutional review board (IRB) on February 11, 2016 (IRB# E9770). This IRB approval remained in effect until February 10, 2019.

⁴ Responses to this open-ended stimulus question were manually coded to ensure respondents in “bad experience” condition did recall bad moments of feeling excluded and those in “good experience” condition recalled good moments of feeling included in the U.S. 51 out of 225 (22.67%) were coded as “invalid treatment response,” with responses like “not sure/don’t know” or “I’ve never had any bad experience of being excluded or isolated.” For the models reported in this paper, all cases were used for analysis. Additional analyses using respondents with “valid treatment response” only were conducted and yielded similar results (see Online Supplement B).

⁵ Alternatively, Study 1 also captured participants’ most salient ethnic identity after treatment, by asking them to choose one term (i.e., American, Latino/Hispanic American, etc.) that can best describe themselves in posttest survey. Additional manipulation checks were conducted using this variable but didn’t reveal significant results, which were thus not reported in the manuscript. See Online Supplement B for details.

⁶ For an overview of Stata’s generalized structural equation models (“gsem” command) for mediation analysis, please see https://www.stata.com/manuals/semexample42g.pdf.

⁷ Secondary survey data often feature a large quantity of observations with little cost (Kiecolt & Nathan, 1985). Despite a big number of missing values on multiple variables, the 2006 LNS data still provided more than 4,600 valid observations for analysis (see Table 2), which were way more than what can be collected by a novel survey/experiment with a limited budget.

⁸ In the 2006 LNS, respondents were asked to rate the strength of their feelings of being Americans, Latinos, and national descriptors (and these questions were randomized to minimize the priming effect) first, followed by the other question tapping into their most salient ethnic identity.

⁹ Similar results were also uncovered in baseline models that did not control for confounding factors; see online supplement D.

¹⁰ For example, when asked about bad experience of being excluded in this study, one participant said most of the time he/she felt being treated unfairly at workplace.

¹¹ But note that existing literature on this line of research – as synthesized in this present study – suggested a mutual influence between one’s self-concept and his/her media preference.

Additional information

Notes on contributors

Mingxiao Sui

Mingxiao Sui is an assistant professor in the Department of Communication Studies at University of Alabama at Birmingham, Birmingham, AL. She obtained her Ph.D. degree from Manship School of Mass Communication at Louisiana State University. Her contact information is [email protected].