Academic journal article Economic Inquiry

Surprise Me If You Can: The Influence of Newspaper Endorsements in U.S. Presidential Elections

Academic journal article Economic Inquiry

Surprise Me If You Can: The Influence of Newspaper Endorsements in U.S. Presidential Elections

Article excerpt

We here evaluate the heterogeneous effects of newspaper endorsements of U.S. Presidential candidates in the 100 days preceding the 2008 and 2012 elections on the probability that they win the election. Our identification strategy relies on daily variations in the winning probabilities (obtained from the Intrade prediction market) and the fact that newspapers decide their endorsements weeks before their announcement. Endorsements that are classified as surprising and consistent have the largest effect. An endorsement is surprising when the newspaper has not traditionally endorsed the candidate's party. An endorsement is inconsistent when the newspaper leans ideologically to one party but endorses a candidate from another party. (JEL L82, D7)


Most American newspapers, traditionally, explicitly endorse one presidential candidate per electoral cycle, in a printed editorial page stating the case for their support. Whether such endorsements affect electoral outcomes is still open to debate, and not simple to pin down empirically. The main difficulty is to distinguish unobserved common factors, which are correlated with both endorsements and the winning probability, from any causal effect of endorsements. A cross-election finding that candidates with more endorsements also obtain more votes does not prove that the former caused the latter. This may instead simply reveal that some candidates are better than others. Even if we consider the same candidates in multiple elections, inference remains problematic if perceived candidate quality changes across elections (see Druckman and Parkin 2008).

We here address this difficulty and quantify the causal effect of newspaper endorsements on election outcomes using a novel daily measure of the winning probability and exploiting the timing of endorsements, which are decided weeks before they are printed (Meltzer 2007). The existing literature (Chiang and Knight 2011; Puglisi and Snyder 2015) suggests a larger effect from endorsements that are unexpected. We split the latter into two groups, using well-known measures of media bias to show that not all unexpected endorsements affect winning probabilities.

Although endorsements mostly mirror the political stance of the corresponding newspapers, and are therefore expected, unexpected endorsements may come as (1) a surprise, when a newspaper traditionally endorsing one party goes for another party; equally, they may appear as (2) inconsistent, when they are at odds with the newspaper's typical discourse (political language). The measure of the (historical) "propensity to endorse democratic candidates" by Ansolabehere, Lessem, and Snyder (2006) is used to define surprise endorsements, while we appeal to the measure of "media slant" by Gentzkow and Shapiro () to capture inconsistent endorsements. For instance, while the Chicago Tribune historically endorses Republican candidates, the newspaper's content is relatively appealing to Democratic voters. Its endorsement of Barack Obama would thus be classified as a consistent surprise endorsement. These two correlated but separate measures, used in both the economics and political science literatures, allow us to distinguish between types of unexpected endorsements and their effects (see Table 1 for a summary of this classification; we will detail how we classify newspapers endorsements as being a surprise or inconsistent endorsement in Section II.A).

Our key variables come from the daily trading prices of the contract "Obama to win the election" from the prediction market site Intrade, combined with newspaper endorsement dates from "The American Presidency Project." We cover the 2008 and 2012 presidential elections. Our identification strategy relies on the daily variation in both endorsements and the winning probability. Our analysis is thus within election, so that we control for unobserved candidate characteristics. In addition, the fact that endorsements are typically decided weeks before their public disclosure (Meltzer 2007) allows us to separately determine the effects of endorsements and candidates' attributes, and eliminate concerns regarding reverse causality. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.