NEWS SELECTION

One overarching theme in the study of political communication addresses how journalists, editors, and news outlets choose which events to report and, likewise, which to ignore (Gans Reference Gans2005, Graber Reference Graber2007, Entman Reference Entman2012, Lippman Reference Lippman1922). One concern has been with the tendency of news outlets to favor “infotainment,” or stories selected to grasp the audience’s interest with entertaining content over stories that are more substantively important (Baum Reference Baum2002, Zaller Reference Zaller1999).

Although infotainment-style reporting has increased (Slattery, Doremus and Marcus Reference Slattery, Doremus and Marcus2001), scholars struggle to understand how much news decisions are driven by the entertainment value of stories as opposed to their importance. First, it is generally difficult to convince news outlets to admit how much their news decisions are based on entertainment value. Second, however, because an unlimited and a not clearly measurable number of events that journalists could report exists, researchers lack a baseline measure to represent the stories available to be reported on any given day. This makes a comparison between stories that are reported to stories that could be reported difficult. In this article, we attempt to overcome this problem by first creating a baseline of events that could be reported each day and then by comparing how often the entertaining stories in that baseline are reported to the less entertaining stories in our baseline. We define a news story as “entertaining” if it includes a dog as a major actor.

Are events reported because they feature a dog? If a dog is a prominent part of an event, how much more news attention will that event receive compared to events that do not feature dogs? We attempt to shed light on these questions because the answers are of prime importance in understanding political communication (and not because a journal article featuring dogs will be found entertaining by our colleagues and therefore be more widely read, circulated, and cited). First, traditional models of journalism (e.g., Entman Reference Entman, Overholser and Jamieson.2005) suggest that dogs are not inherently important to democratic citizenship and that a dog’s involvement should not make journalists more likely to report that event. If we find that events featuring dogs are reported more often than those without dogs, this would reflect the role of entertainment value relative to the value of societal importance in news decision making. Our results, therefore, provide leverage for understanding the entertainment preferences of audiences.

Second, scholars suggest that entertainment-oriented news segments appeal to audiences that otherwise might be inattentive (Baum Reference Baum2003). Drawing in wide audiences with entertainment (e.g., cuddly dogs) may increase citizen knowledge and understanding of important contemporary issues and therefore be good for democracy (Baum Reference Baum2002).

Third, some scholars argue that dogs bestow political advantages (Mutz Reference Mutz2010, cf. Jacobsmeier and Lewis Reference Jacobsmeier and Lewis2013), whereas others suggest that dogs are used strategically by politicians to garner public attention (Maltzman et al. Reference Maltzman, Lebovic, Saunders and Furth2012). Therefore, our results also may suggest strategies for political actors seeking news attention.

RESEARCH DESIGN

We identified a baseline of events that represents the potential news stories that could be reported on any given day. To establish the baseline, we used stories from the US “paper of record,” the New York Times (NYT) National News Desk section. Because stories in this section are national in nature, they all could be reported by other news organizations regardless of geography. We do not claim that stories in the NYT National News Desk section are the only stories available for reporting or that they represent the “correct” news agenda. Instead, we used the NYT national stories as a sample baseline of events that could be reported on each day by other newspapers.

To select comparison papers for the purpose of measuring how widely different events are dispersed among the nation’s newspapers, we began with the top 100 US newspapers by circulation in 2012, and we included all of those archived by Lexis-Nexis Academic for the years spanning from 2000 to 2012. ^{Footnote 5} Table 1 lists the 10 newspapers meeting these criteria. Each paper is a major outlet and has access to the events covered in the NYT National News Desk section.

Table 1 Newspapers in the Study

Stories in the NYT National News Desk section provide a reasonable sample of reportable events each day, and our goal was to measure how many of these 10 other papers cover the same events. If an event in the NYT was reported in all 10 papers in our sample, we considered that event to be more widely dispersed among the nation’s newspapers than an event reported by the NYT but by only one other paper. Our design did not intend to measure the agenda-setting influence of the NYT over other papers (e.g., Shaw and Sparrow 1999) but rather the independent influence of a baseline set of events over the news decisions of our sample of 10 papers. With this said, our design did allow for the possibility that a story was introduced to other papers by the NYT because we included the day-after contents of the non-NYT papers. ^{Footnote 6}

This design relied on a simple model of the news process. First, an event happens. The likelihood of that event being reported in any given newspaper is a function of how important and interesting reporters and editors at that newspaper perceive it to be. If events in our NYT baseline featuring dogs were reported more in our sample of 10 papers than events in the baseline without dogs, this indicated that dogs lead events to receive more coverage.

DATA

To identify the stories in the New York Times, we performed a Lexis-Nexis Academic search of all NYT National News Desk stories containing the word “dog” or “dogs” between January 1, 2000, and December 12, 2012. This produced 2,590 results. Within these 2,590 national news stories, we identified those that involved a dog as an important actor in the story. For example, we retained stories about drug-sniffing dogs and seizure-alert dogs. We omitted stories in which dogs were incidental to the reporting of the story—for example, when a subject of the report was observed as having a dog. From this initial sample of 2,590 National News Desk articles, we identified 18 in which a dog was an important actor (table 2). To serve as a comparison, we used the other articles appearing in the NYT National News Desk section on the same day as each of the 18 dog stories. ^{Footnote 7} This provided 334 non-dog stories. ^{Footnote 8}

Table 2 Articles about Dogs in the New York Times National News Desk Section, 2000–2012

Some stories appearing in the NYT National News Desk section cover historic or otherwise important events, whereas other stories focus on seemingly less newsworthy occurrences. We assumed that more newsworthy events will be published more prominently and with more page space. To control for the relative newsworthiness of the events covered in our NYT baseline, we accounted for both the prominence and the length of the NYT articles. Our measure of prominence was 1 minus the quotient of the page that the article appeared on and the highest number of pages among all NYT national news articles published that day. To account for page space, we used the word count of the story. The median word count for NYT articles was 560 and the word-count variable used in our analysis was raw word count as a share of the median word count. A word-count score of 2, for example, indicated an article that was twice as long as the median NYT article.

For our 10 comparison newspapers, we compiled all national news stories published in these papers on the date that each of the dog stories appeared in the NYT as well as on the date following its publication. We then determined how many of the other 10 newspapers in which the same story of each of the NYT national news articles appeared. ^{Footnote 9} The possible range of this measure was [0, 10] and we observed scores ranging from 0 to 9. About 25% of the stories received a score of 1 or more. This seemingly low number was not surprising: whereas we expected some overlap across newspapers, there also was an expectation that local coverage would vary because different audiences have different demands for information (Branton and Dunaway Reference Branton and Dunaway2009). As such, there was significant variation across newspapers, even in the area of national news. This finding should assuage the concerns of media critics who believe that news outlets have become too homogeneous in recent decades.

ANALYSIS

Using the NYT as a baseline representing the events available for reporters to cover, we wanted to know how much events involving dogs disperse across newspapers relative to other national news events. The measure we used for this was the incident-rate ratio estimated using a Poisson regression. ^{Footnote 10} Each story reported in the NYT National News Desk section (including and not including a dog) comprised one case. The analysis contained 352 cases (i.e., 18 dog and 334 non-dog stories).

Our dependent variable was the number of times a story was reported by the 10 newspapers in our sample. Our independent variable of interest was a dichotomous variable indicating whether the story prominently featured a dog (i.e., coded 1 for yes). If dog stories were reported more often than non-dog stories, the coefficient should be positive. We included our controls for prominence (i.e., Prominence in NYT) and page space (i.e., NYT Article Word Count). We also included a fixed effect for the date to account for broader over-time variation (e.g., number of events available to reporters and temporal location in the news cycle) and for possible idiosyncrasies in the Lexis-Nexis indexing of newspapers in the study.

Table 3 shows that dogs lead events to be reported more widely. The positive and significant coefficient for our dog-story indicator suggested that events featuring dogs were more likely to be reported by other newspapers in our study than the non-dog events. These findings came after controlling for prominence and length. Our estimates suggested that a dog story on the last page of the NYT national news report proliferated at 2.6 times the rate of a non-dog NYT article on the same page. By comparison, a story that made the front page of the NYT also was covered by other papers at 3.1 times the rate of a story published on the last page of the NYT national news report. In other words, having a canine subject in a national news event produced coverage of the story that was 80% as large as the effect of the difference between being NYT front-page and back-page worthy.

Table 3 Poisson Regression Analysis of Proliferation of New York Times National News Desk Stories

Another way to interpret the effect of dogs is to consider page space—that is, more important news stories have more space (i.e., more words) in the paper. A NYT national news story of median length (i.e., 560 words) was covered by other papers at 1.1 times the rate of coverage of a very short NYT news item. A dog story was covered at 2.6 times the rate of coverage of a non-dog very short NYT news item. Thus, other factors being equal, a dog story of median length will be covered by the nation’s newspapers at the same rate as a non-dog story that warrants about three times the median word count in the NYT.

CONCLUSION AND EXTENSIONS

Politics often is discussed in canine terminology: “Blue Dog Democrats” and “dog-whistle politics” come immediately to mind. In addition, media critics have long categorized models of journalism as “watch dog,” “lap dog,” “guard dog,” “attack dog,” “junkyard dog,” “guide dog,” and “sled dog”—not to mention the “man-bites-dog” style of reporting. Given the propensity of journalists to report on dogs, these monikers for journalism seem quite apt.

As media scholar Timothy Cook (2005, 89) argued, “The search for newsworthy stories and newsworthy events does not equally favor all political actors and all issues.” Our results suggest that journalists favor dogs and events featuring dogs. Because dogs are inherently entertaining, and because “soft” entertainment-style news brings in audiences (Baum Reference Baum2003), the sizable dog effect that we identified is indicative of the degree to which journalists are driven to report events based on entertainment value. Thus, our findings provide support for profit-driven or economic models of news media (Hamilton Reference Hamilton2004, Uscinski Reference Uscinski2014).

Although we stand by our quantitative results, an example might be illustrative. Following Hurricane Katrina, there were numerous events about which news outlets could have reported: widespread destruction, human suffering, displacement of an entire city, effects of massive social and racial inequality, government incompetence, and so forth. When the NYT reported on New Orleans dog shelters (September 8, 2005), the National News Desk section also reported 17 other Katrina-related stories. Among those stories, the dog story was tied for seventh in terms of its prominence. Two of the six more-prominent articles dispersed (i.e., one to two papers, the other to one paper). However, four articles with greater prominence and greater word count than the dog-shelter story did not disperse to any of our other 10 papers. It is interesting that one of the stories that did not disperse was a report on human survivors moving from temporary shelters to newly available apartments; another was a report about the Democrats in Congress criticizing the Bush administration. Some observers may believe that one or both of those articles was more newsworthy than the dog story, which apparently was the judgment of the NYT editorial team because it allocated them better placement and more space. This case is anecdotal and potentially idiosyncratic, but our point is that the NYT and one other news outlet had potentially more compelling stories to cover regarding Katrina but instead chose to report about dogs. This raises the question of whether coverage of the shelter story is simply an idiosyncratic quirk or whether it is consistent with a general pattern on the part of the news media to cover dog stories (instead of other worthy events) when they are available.

It may seem rather crass of journalists to report events simply because a dog is an actor in the story; however, there may be value in such a practice. “Softening” news to make it more interesting and palatable (in this case, with a dog) may make audiences more likely to read and have an interest in the news (Baum Reference Baum2002). Thus, dogs may attract audiences to news stories that they otherwise would have ignored. A suburban housewife in Upstate New York might not care much about a hurricane 1,000 miles away; however, when the plight of dogs trapped in an abandoned shelter attracts her attention, she learns about other aspects of the hurricane aftermath as well.

Even the academy may not be immune to the “dog effect.” Just as flagship academic journals experience crazes during which a bumper crop of articles on subjects such as social capital and field experiments are published, there well may be a dog craze afoot in political science. For example, since 2010, PS: Political Science and Politics has published three articles that prominently feature dogs: “The Dog That Didn’t Bark: The Role of Canines in the 2008 Campaign” (Mutz Reference Mutz2010); “Unleashing Presidential Power: The Politics of Pets in the White House” (Maltzman et al. Reference Maltzman, Lebovic, Saunders and Furth2012); and “Barking Up the Wrong Tree: Why Bo Didn’t Fetch Many Votes for Barack Obama in 2012” (Jacobsmeier and Lewis Reference Jacobsmeier and Lewis2013). Furthermore, the one article among these three that was in print long enough to garner citations received more than all of the issue’s other featured articles, and it was the subject of a response. ^{Footnote 11} We agree with Maltzman et al. (2012, 399) that dogs are likely to become a valuable research program within political science. And why not? With their fuzzy ears and uncanny ability to foster high citation counts, political scientists may well be learning to do what journalists have done for years: dog it.