Davis and Dunaway (2016) showed that mass partisan-ideological sorting in the United States is consistently associated with citizens' perceptions of party polarization, and that media fragmentation is related to sorting mainly among the politically interested, using aggregated national and state media-penetration metrics and repeated cross-sections (ANES 1986-2008). They left open the causal core of this result: whether changing media environments and elite polarization actually produce sorting over time, through the formation of perceived party polarization, with what downstream affective and behavioral consequences. The five directions they flag - finer-grained exposure measurement at designated-market-area or individual level; panel or longitudinal causal identification; tracing how perceived party polarization forms and is distorted; replicating the conditional Internet effects; establishing consequences in causal frameworks - have each attracted substantial partial progress in the 2017-2026 literature, but no study yet combines fine-grained exposure measurement with longitudinally identified causal estimates of the full pathway, so the problem remains open.

Background

Mass partisan-ideological sorting is the growing alignment of citizens' partisan identity (which party they identify with and how strongly) with their ideological self-placement (liberal to conservative); a fully sorted citizen holds matching and strong partisan and ideological identities. Sorting matters because aligned identities are thought to harden partisan animus and disciplined party competition. Two families of explanations dominate. One points to the high-choice media environment: the multiplication of cable, satellite, and Internet outlets since the 1980s lets politically interested citizens selectively expose themselves to congenial, partisan-tinged news, which is hypothesized to sharpen the perceived distance between the parties and thus sort the audience. The other points to elite polarization: as Democratic and Republican members of Congress drift apart in roll-call voting (commonly summarized by DW-NOMINATE scores, which place each legislator on a liberal-conservative scale computed from their recorded votes), clearer elite cues are said to teach citizens which ideology goes with which party. Davis and Dunaway (2016, Public Opinion Quarterly 80(S1):272-297) synthesized both accounts using the American National Election Studies (ANES) time-series surveys pooled over 1986-2008, a sorting index combining the match and strength of partisan and ideological identities, respondent-level perceived party polarization (placement of the Republican party minus placement of the Democratic party on the seven-point liberal-conservative scale), and contextual media measures: the national average number of television channels per household, national internet users per 100 persons (International Telecommunication Union/World Bank series), and irregular state-year internet estimates from the Current Population Survey; supplementary analyses used the Cooperative Congressional Election Studies (CCES), a large annual opt-in-Internet survey, including self-reported home high-speed Internet access. Their findings: sorting is consistently connected to perceived party polarization; objective elite polarization shows little consistent direct effect but matters more for the politically interested; and media fragmentation is positively related to sorting only among respondents highly interested in news and politics. The authors flag five unresolved problems arising from this design: media exposure measured only as aggregated national and state penetration, which may mask heterogeneity across designated market areas (DMAs, the Nielsen-defined local television markets that determine what stations a household can receive) and within them across individuals; the absence of panel or time-series identification, since repeated cross-sections cannot distinguish causal influence from contemporaneous co-movement of media environments, elite polarization, and sorting; the unexplained formation and distortion of perceived party polarization, which often fails to track objective elite behavior; unreplicated supplementary findings such as the conditional effect of home high-speed Internet in the CCES; and the unknown downstream behavioral and affective consequences of sorting, including affective polarization (out-party dislike and distrust studied as a phenomenon distinct from policy disagreement). These five gaps jointly block one causal question: whether, how, and for whom the changing media environment and elite polarization actually produce mass sorting and its consequences.

Problem Statement

The problem posed by Davis and Dunaway (2016) is to replace their associational, aggregate-level evidence with a causally identified account of mass partisan-ideological sorting in the United States. Concretely, the unresolved question is: to what extent do changing media environments (expanding cable, satellite, and Internet availability, and the partisan reach of outlets within them) and elite party polarization, operating through citizens' perceived party polarization, causally produce increasing alignment of partisan and ideological identities in the mass public, for whom (by political interest or awareness), and with what downstream affective and behavioral consequences? The question is posed for the period over which mass partisan-ideological sorting has increased, not only for the source's 1986-2008 ANES window; whether an answer is identified in the cable era, during the broadband rollout, or in the current social-media environment is a matter of transport between media eras (see the difficulties below), not a change of scope. The source's own design leaves five limitations that bear on this question: (1) media exposure and availability measured only as national and state penetration aggregates, rather than as designated-market-area outlet reach or individual-level consumption records; (2) no longitudinal, panel, or quasi-experimental observation of partisan identity, ideological identity, and perceptions over time that would separate causal influence from contemporaneous co-movement (and from the reverse path by which sorting shapes perceptions); (3) the formation and distortion of perceived party polarization by media coverage, campaigns, and partisan bias remain untraced; (4) supplementary findings such as the conditional home high-speed Internet effects in the CCES analyses remain unreplicated under more representative measurement; and (5) the downstream interparty-animus, affective-polarization, and behavioral consequences of sorting remain unestablished within causal frameworks. These directions all bear on one objective - a causal account of the sorting pathway, namely a causally identified, rather than merely associational, estimate of the pathway from changing media environments and elite party polarization, through perceived party polarization, to mass partisan-ideological sorting, including the interest- and awareness-conditional heterogeneity the question asks about, together with the downstream affective and behavioral consequences.

References

  1. Nicholas T. Davis; Johanna L. Dunaway, "Party Polarization, Media Choice, and Mass Partisan-Ideological Sorting", Public Opinion Quarterly 80(S1):272-297 (2016), DOI 10.1093/poq/nfw002, https://doi.org/10.1093/poq/nfw002
  2. Nicholas T. Davis, "Perceptions of elites and (asymmetric) sorting", Research & Politics 5(2) (2018), DOI 10.1177/2053168018777099, https://doi.org/10.1177/2053168018777099
  3. Levi Boxell; Matthew Gentzkow; Jesse M. Shapiro, "Greater Internet use is not associated with faster growth in political polarization among US demographic groups", Proceedings of the National Academy of Sciences 114(40):10612-10617 (2017), DOI 10.1073/pnas.1706588114, https://doi.org/10.1073/pnas.1706588114
  4. Yphtach Lelkes; Gaurav Sood; Shanto Iyengar, "The Hostile Audience: The Effect of Access to Broadband Internet on Partisan Affect", American Journal of Political Science 61(1):5-20 (2017), DOI 10.1111/ajps.12237, https://doi.org/10.1111/ajps.12237
  5. Shanto Iyengar; Yphtach Lelkes; Matthew Levendusky; Neil Malhotra; Sean J. Westwood, "The Origins and Consequences of Affective Polarization in the United States", Annual Review of Political Science 22:129-146 (2019), DOI 10.1146/annurev-polisci-051117-073034, https://doi.org/10.1146/annurev-polisci-051117-073034
  6. Philipp Lorenz-Spreen; Lisa Oswald; Stephan Lewandowsky; Ralph Hertwig, "A systematic review of worldwide causal and correlational evidence on digital media and democracy", Nature Human Behaviour 7:74-101 (online 2022, print 2023), DOI 10.1038/s41562-022-01460-1, https://doi.org/10.1038/s41562-022-01460-1
  7. Joshua Robison, "Elite party disunity negatively predicts mass partisan-ideological sorting", Electoral Studies 85:102659 (2023), DOI 10.1016/j.electstud.2023.102659, https://doi.org/10.1016/j.electstud.2023.102659
  8. Michael Heseltine; Bernhard Clemm von Hohenberg; Ericka Menchen-Trevino; Tomasz Gackowski; Magdalena Wojcieszak, "Effects of Over-Time Exposure to Partisan Media and Coverage of Polarization on Perceived Polarization", Political Communication 42(3):432-453 (online 2024-11-04), DOI 10.1080/10584609.2024.2423093, https://doi.org/10.1080/10584609.2024.2423093
  9. Daniel S. Lane; Melody Chen; Yifei Wang, "An “Identity Turn” in political communication?: testing the relationship between media use and identity alignment in the United States", Journal of Communication 76(4):290-302 (2026; online 2025), DOI 10.1093/joc/jqaf026, https://doi.org/10.1093/joc/jqaf026
  10. Rongbo Jin; Philip Edward Jones, "Political Awareness and Sorting", American Politics Research 54(3):347-351 (2026), DOI 10.1177/1532673X261420777, https://doi.org/10.1177/1532673X261420777