aDepartment of Computer and also Information Science, university of Pennsylvania, Philadelphia, PA 19104;
cOperations, Information, and also Decisions Department, university of Pennsylvania, Philadelphia, PA 19104;

Edited by Dietram A. Scheufele, college of Wisconsin–Madison, Madison, WI, and also accepted by Editorial board Member Susan T. Fiske February 21, 2021 (received for review November 8, 2019)


*

Abstract

Since the 2016 us presidential election, the deliberate spread out of misinformation online, and also on social media in particular, has created extraordinary concern, in huge part due to the fact that of its potential impacts on public opinion, politics polarization, and ultimately autonomous decision making. Recently, however, a grasp of documents have suggested that both the prevalence and also consumption that “fake news” every se is exceptionally low compared with other varieties of news and also news-relevant content. Return neither pervasiveness nor usage is a direct measure that influence, this work argues that ideal understanding that misinformation and also its effects requires a much broader view the the problem, encompassing biased and also misleading—but no necessarily factually incorrect—information the is routinely produced or magnified by mainstream news organizations. In this paper, we propose one ambitious collective research agenda to measure up the origins, nature, and also prevalence of misinformation, extensively construed, as well as its influence on democracy. We additionally sketch out part illustrative instances of completed, ongoing, or to plan research projects that add to this agenda.

You are watching: Did fake news influence the election


It is difficult to overstate the breadth and intensity of attention directed over the previous 2 y in ~ the worry of false or misleading info (also recognized as “fake news”) circulating on the internet in general and also on society media communication such as Facebook and Twitter in specific (1⇓⇓⇓⇓⇓⇓⇓⇓⇓⇓⇓–13). Follow to Google Scholar, because January 2017, more than 5,000 English language publications v “fake news” in the location have showed up in academic journals covering economics, politics science, computer and also information science, communications, law, and journalism. To put this number in perspective, fewer 보다 100 together publications appeared in all the years leading as much as the finish of 2016, when fewer than 600 publication have showed up since 2017 containing “television news” or “TV news.”

The beginning of this extraordinary rise in interest in a previously sleepy topic to be of food the 2016 us presidential election, which, in addition to other occasions that year such together Brexit, raised widespread concerns around a feasible rise that populist/nationalist political movements, boosting political polarization, and also decreasing public to trust in the media. Early reporting through journalists (14) easily focused attention on fake news turn on society media sites throughout the choice campaign. The philanthropic and scientific communities then responded with dozens that conferences and also thousands of records studying various aspects of fake news. Reinforced by ongoing mainstream media attention and also increasing congressional scrutiny of modern technology companies, the conjecture that the deliberate spread of virtual misinformation poses one urgent danger to democracy ultimately hardened into conventional wisdom (13, 15).

In the confront of this dominant narrative, a handful of writer (1, 10, 11, 16⇓–18) have said that fake news is less prevalent than breathless referrals to “tsunamis” or “epidemics” would certainly imply. In very early contribution, Allcott and also Gentzkow (1) estimated that “the median US adult read and remembered top top the stimulate of one or perhaps several fake news posts during the election period, with greater exposure come pro-Trump write-ups than pro-Clinton articles.” In turn, they estimated that “if one fake news article were about as persuasive together one TV campaign ad, the fake news in our database would have readjusted vote shares by an lot on the order of hundredths of a percentage point,” about two order of magnitude much less than essential to influence the choice outcome. Succeeding studies have actually found likewise low pervasiveness levels for fake news relative to tendency news top top Twitter (10) and Facebook (11). Finally, our very own survey of the media consumption landscape, based upon a country representative sample the TV, desktop, and mobile media usage (18), found three main results that undercut the traditional wisdom about fake news and additionally the prominence of online resources of news in general:

2) virtual news intake is a small fraction of in its entirety news consumption, i beg your pardon is dominated by TV by a variable of five to one. Even 18 to 24 y olds consume nearly twice as lot TV news than digital news. In striking comparison with the study literature’s overwhelming focus on online resources of news, we estimate that 3 in four Americans spend less than 30 s a day reading news online, if almost fifty percent consume no virtual news whatsoever.

3) Fake news is a tiny part of Americans’ details diets. Utilizing our most inclusive definition, less than 1% of consistent news consumption and less 보다 1/10th the 1% of as whole media usage could be thought about fake. Also the heaviest consumer of fake news (the 55+ age group) consume much less than 1 min the fake news every day top top average, contrasted with 106 min of continual news (94 of lock on TV) and also over 500 min of full media consumption.

As has actually been suggested elsewhere (13), these results on their own do not conclusively show that fake news does not have coherent effects on windy opinion, politics polarization, and also trust in institutions. That is possible, for example, that even extremely low prices of exposure to fake or misleading news might have outsized effects, at least on part people, or that equivalent amounts of online and also television news consumption have various impacts. Nonetheless, this results carry out strongly imply that study on the origins, nature, prevalence, and consequences of misinformation should take a much wider view of the topic 보다 outright false info disseminated on social media or even online (16). In particular, there space at least three factors for taking such a wider view.

First, while the is possible that exposure to fake news has more impact 보다 an indistinguishable amount that exposure to genuine news, or that digital news has much more impact than tv news, it is equally possible that the contrary is true. For example, recent job-related has uncovered that subjects rate mainstream publication as much more trustworthy 보다 fake or very partisan web page irrespective the their own partisanship (12), and that consideration reduces id in false headlines however not in true ones, again regardless of of partisan alignment (19). Likewise, if television intake can it is in dismissed as more “passive” 보다 reading, direct comparisons between television and online news and also advertising consistently find better recall that televised contents (20⇓–22) particularly for low-involvement consumers (23). Ultimately, questions of affect are empirical questions and also answering those concerns will call for making comparisons between different varieties of content and different modes of production.

Second, fake news sites room not the just sources the false information: The tendency media can also promulgate falsehoods simply by report on them (24). In the lead-up to the 2003 Iraq War, because that example, a big majority of media institutions uncritically repetitive the administration’s false insurance claim that they possessed unequivocal proof that Saddam Hussein possessed tools of mass damage (25, 26). In august 2009, as soon as Sarah Palin wrote in a blog write-up that the Affordable care Act would develop “death panels,” the case was repetitive in over 700 mainstream news articles even after ~ it was debunked by a variety of fact-checking institutions (27, 28). More recently, an evaluation of Russian disinformation efforts during the 2016 presidential election concluded that these efforts similarly succeeded in getting to the public mostly via the credulous reporting of tendency media outlets (29). Return the motivations and mechanisms driving misinformation in tendency media different from web page that purposely promote falsehoods, the results may be many times greater; thus, a proper bookkeeping of the prevalence of false information requires a wide consideration the potential sources.

Third, misinformation is a much wider phenomenon 보다 outright falsehoods. There are plenty of ways to command a reader (or viewer) to with a false or unsupported conclusion that perform not require saying anything that is unambiguously false (30). Presenting partial or biased data, quoting sources selectively, omitting alternative explanations, improperly equating unlike arguments, conflating correlation v causation, making use of loaded language, insinuating a case without actually making the (e.g., by quoting someone else making it), strategically notified the presentation that facts, and even simply an altering the headline deserve to all manipulate the reader’s (or viewer’s) impression without their awareness. These methods are pervasive in mainstream professional journalism (see, e.g., ref. 31) and also are not limited to politics topics, although that is often the emphasis of research study on media prejudice (32, 33). Inaccurate and also misleading coverage is additionally pervasive in other locations of journalism (34), consisting of important domains for publicly opinion and democracy together as health and wellness (35, 36), science (37), and business (38).

For all three reasons, studies of the ubiquity of misinformation and also its impact on democratic decision make must take on a much more comprehensive conception of the trouble that includes biased and also potentially misleading info that is embedded in mainstream news content throughout all significant modes of manufacturing (24, 29). Unfortunately, study of this scope and scale is hindered by 3 interrelated yet distinct obstacles. First, research study on misinformation and also its effects is right now dependent on datasets that room idiosyncratic, one-off, and also often tiny in scale, rendering comparisons across different modes of media consumption, various sample populations, and also different time periods an overwhelming to make. Second, much of the pertinent data are tough to collect, either because they room scattered throughout thousands of locations in various formats, or are regulated by personal companies (Google, Facebook, Twitter, Microsoft, media companies, etc.) who face huge disincentives and restricted upsides to sharing data with scholastic researchers (39). Third, the relevant academic research is scattered throughout several disciplines (e.g., economics, marketing, politics science, communications, psychology, sociology, computer system science, and also network science), each through its own set of theoretical frameworks, embraced methodologies, and also publishing venues. Collating and reconciling results throughout these disciplinary boundaries is challenging and regularly leads to inconsistent or incoherent conclusions (40).

See more: Death By Firing Squad In Us, Can Inmates Choose Death By Firing Squad

Addressing this shortcomings in present data and research techniques will call for a significant effort come coordinate clinical communities, data resources, and also academic–industry collaborations. Back we space not the very first to speak to for such an effort (see, e.g., refs. 4, 5, and also 39), ours proposal differs from vault instantiations in that it is clearly focused on the require for common research infrastructure as well as the opportunities for collaboration and partnership the such an infrastructure may create. In the next section, we describe our proposal at a high level, break it into four distinct yet mutually reinforcing objectives. Us then illustrate the potential of ours proposed method with a choice of in-progress and also planned research study projects, and also some instances of windy outreach and engagement tasks that we think will boost the research.