Replicating the discovery, scrutiny, and decline model of media coverage in presidential primaries

ABSTRACT

Media coverage has long been thought crucial to shaping the electoral fortunes of presidential primary candidates in the post-reform era, making how the media allot coverage a topic of paramount importance. Sides and Vavreck (2013. The Gamble: Choice and Change in the 2012 Presidential Election. Princeton, NJ: Princeton University Press) make a notable contribution to the study of media coverage in primaries with their “discovery, scrutiny, and decline” (DSD) model. This model, based on the 2012 Republican primary, suggests that the media’s preference for novelty leads to a cyclical identification of new and interesting candidates, a surge in coverage of that candidate, and a culminating drop of coverage back to baseline levels. But the generalizability of the DSD model beyond the 2012 GOP primary has not yet been thoroughly tested. This paper conducts such a test using the Presidential Primary Communication Corpus (PPCC) which contains news stories by The New York Times and the Washington Post of each candidate in the nine primaries from 2000 to 2020. The evidence is most supportive of the DSD model in the 2008 and 2012 Republican primaries and the 2004 and 2020 Democratic primaries but less supportive in the remaining five. This paper concludes with a discussion of why some campaigns don’t match the DSD model’s expectations.

Disclosure statement

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

Notes

1 A subsequent look at the 2016 primaries shows that coverage of Donald Trump was so extensive that it left little oxygen for others, although several candidates did appear to rise and fall in prominence on the periphery (Sides, Tesler, and Vavreck 2018).

2 A narrative that picked up more steam when Donald Trump similarly threatened to pursue the Reform Party nomination.

3 Gore spent the first 12 weeks as the sole candidate in the race, therefore he is the subject of 100% of the articles in the corpus in those weeks. To prevent this from distorting the trend lines, I omit these weeks from the data.

4 His announcement was accompanied by a media cycle of theorizing that the retired general from Little Rock was the unofficial Clinton-backed candidate.

5 I omitted the first four weeks of the 2008 Democratic series as only marginal candidates were in the race at that time.