Keywords

Introductory Comments

Adopting both a theoretical and empirical perspective we now explore the negative effects of false election information on electoral and other democratic processes. We also determine the economic and psychological dynamics at play in order to show that false election information will be difficult to address without a well-crafted legal remedy.

There is an emerging interdisciplinary literature on the effect of false information on elections which largely borrows from the established tools of economic, sociological and psychological analysis. The theoretical literature on false campaign statements, as we briefly discuss below, provides some conceptual framing for our analysis and is supported by a growing empirical literature. The theoretical literature makes claims about how the forces at play in false campaign statements may interact, and therefore how the relevant channels may be suitably regulated.

While many commentators speculate that elections are increasingly affected by false campaign statements, there are surprisingly few studies that seek to empirically test this claim. However, scholars are now starting to recognise the necessity of empirically evaluating the effect of false information on elections. Clare Wardle suggests that academia must become ‘central to [the] conversation’ in building ‘the research framework that will help us understand the scale and impact of [false information]’ (Wardle, 2018: 952).

A Theoretical Understanding: Or Should We Give Consumers What They Want?

Hunt Allcot and Matthew Gentzkow’s 2017 model of fake news production and consumption is a useful heuristic for understanding the potential incentive structures embedded in the propagation of false information (Allcot & Gentzkow, 2017). In their model, media firms observe, with differing precision, the ‘true’ state of the world, in this case, how well a given candidate will perform in office. Firms subsequently publish their signals about the state of the world and are at liberty to add bias to these reports. Consumers of these reports have heterogenous ‘priors’ about the state of the world, meaning they each have differing intuitions about the performance capabilities of electoral candidates. In this model, the consumers’ utility is defined in terms of two seemingly diverging motivations: first, to know the truth (receiving the ‘truest’ signals from media firms) and, second, having their heterogenous priors affirmed by corroborative sources.

Consumers therefore wish to know the true state of the world, while simultaneously, and sometimes paradoxically, wishing to have their biases confirmed. Given these goals, consumers then choose a news firm to maximise their utility—with frequently visited firms accruing larger sums of revenue. Biased corroborative ‘news’ is produced more frequently when: consumers’ feedback about the true state of the world is limited; when the true ‘quality’ of a media firm cannot be gauged (see Gentzkow & Shapiro, 2006); and when consumers prefer psychologically confirmatory content (Mullainathan & Schliefer, 2005).1 While it may seem far-fetched to claim that media consumers would prefer deliberately biased news, empirical studies in psychology affirm that cognitive dissonance and confirmation bias significantly affect our news consumption decisions (Pennycook & Rand, 2021).

The model suggests that there are perverse incentives within the information market to produce inaccurate and untrue information that confirms prior beliefs about the world (Braun & Eklund, 2019). These perverse incentives are even more salient in the digital age, where barriers to entry for news production and dissemination are at historic lows. For example, throughout the final three months of the 2016 US Presidential election, many influential false news stories were found to have originated from Veles, a small town in North Macedonia in which a quarter of the residents live in poverty. Notably, the primary motivator for production was monetary profit rather than electoral distortion (Hughes & Waismel-Manor, 2020). Regardless of motive, the effect on the information environment was the same.

While candidates are motivated by ideology and the desire for an electoral advantage rather than financial gain, strictly speaking, in both the candidates’ and disinformation entrepreneurs’ cases, there are seemingly inescapable incentives to produce false information because both cases engage the same psychological mechanisms. The fact that there are powerful—and arguably ineradicable—motives for producing false information (Wardle & Derakhshan, 2017) suggests that the information environment needs to be regulated in order for democracy, and its key procedural moment—elections—to function properly.

In Allcot and Gentzkow’s model, when the political information market is awash with ‘fake’ or biased news, there are a number of significant external social costs. Electors who believe fake outlets to be legitimate have less accurate beliefs. These less accurate beliefs lead to the election of poorer quality candidates or else candidates whose policies do not reflect the real interests of those who elected them. In turn, consumers become more sceptical of legitimate news producers, and these effects may be reinforced in equilibrium as demand decreases for legitimate news and fake news outlets proliferate (Allcot & Gentzkow, 2017).

Similar theoretical models suggest that, because false information changes voters’ information environment and therefore policy preferences, it affects the way in which parties vie for the support of the electorate. This false information leads to a divergence in the policy positions of the parties as they compete for the support of the disinformed and informationally divided electorate, resulting in a lower level of welfare for the electorate as a whole, as enacted policies do not promote their true welfare understood in the absence of false information (Grossman & Helpman, 2019).

Other commentators suggest that competitive campaigns provide a ‘short-term incentive to make defamatory statements in order to secure victory, regardless of the long-term potential for damages after an election’ (Rowbottom, 2012: 510), and these statements may harm a candidate’s reputation and standing well after the election period. According to Rowbottom, false statements can not only distort the entire electoral outcome but even ‘short circuit channels of accountability’ for incumbents (Rowbottom, 2012: 512). They do this in the sense that, if voters would otherwise have voted for the opposition, but are convinced to retain the incumbent by a false campaign statement, then the democratic accountability mechanism—the election—has been effectively compromised: ‘A politician responsible for deeply unpopular policies or guilty of some wrongdoing may avoid the penalty at the ballot box if a false statement about the opponent convinces enough people to vote for [them]’ (Rowbottom, 2012: 510–516). The harm to democracy does not end here. False campaign statements can contribute to ‘lower participation and increased cynicism’ towards elections and even democracy in general (Dardis et al., 2008; Rowbottom, 2012: 517; Yoon et al., 2005).

William Marshall has argued that there are four cardinal harms that false campaign speech can inflict. First, it can distort the electoral process by causing electors to vote in ways they otherwise would not. Second, false statements can ‘lower the quality of discourse and debate’ by causing campaigns to degenerate into cycles of attack and denial rather than serious policy dialogue, with considerable resources being devoted to responding to falsehoods (Marshall, 2004: 294). Third, false statements lead to voter alienation and distrust in the political process more broadly. Fourth and finally, there are individual effects of false campaign statements insofar as they can ‘inflict reputational and emotional injury upon the attacked individual’ (Marshall, 2004: 296). While prima facie only an individual harm, such reputational damage may put off qualified candidates from seeking office and impose an unwarranted reputational cost on incumbent political leaders (May, 1992). All of this has implications for democracy in general. As Brennan J noted in the 1964 US Supreme Court decision Garrison v Louisiana: ‘[T]he use of the known lie as a tool is at once at odds with the premises of democratic government and with the orderly manner in which economic, social, or political change is to be effected’ (Garrison v Louisiana (1964) 379 U.S. 64, 75 (Brennan J for the Court)).

Before proceeding it should be noted that while theoretical research is obviously constrained in its ability to explain real-world electoral phenomena, it nevertheless provides a good start. Indeed, while the empirical literature on false electoral information is fledgling, we are observing a gradual convergence between the predictions of the theoretical literature, and the results of empirical studies. We expect the interplay between these literatures to be fruitful to our understanding of false electoral information over the long term.

An Empirical Understanding

The impact of false information campaigns is long term and appears to be growing stronger in each successive election cycle as an increasing proportion of the electorate is exposed to election advertising through digital media. Two conspicuous examples are the 2016 US Presidential Election (Allcot & Gentzkow, 2017; Bovet & Makse, 2019) and the Brexit Referendum (Henkel, 2021) that same year, the outcomes of which were either affected or altered by false information campaigns. The dissemination of false news stories generally favoured Donald Trump in the United States and the ‘Leave’ vote in the United Kingdom, reaching a significant proportion of the electorate in both cases. Similarly, using panel surveys of voters, Fabian Zimmerman and Matthias Kohring’s 2020 study of the 2017 German Federal election found that exposure to stories that were deliberately and verifiably false alienated voters from the governing party and drove them ‘into the arms of right-wing populists’ (Zimmermann & Kohring, 2020: 215).2

Even more worrying is the fact that exposure to false information during election campaigns coincided with lower levels of trust in news media and even politics itself. Another study drawing on panel surveys throughout the 2018 US midterm elections found that exposure to false information two weeks before the election could ‘significantly predict the changes in political cynicism immediately after the election day’ (Jones-Jang et al., 2020: 1). Although it is apparent that false information has an effect—albeit one that is difficult to measure—on voters’ political preferences, these studies also underscore that exposure to false information has a longer-term corrosive effect on the democratic fabric. As more and more people use unreliable digital sources for their political news, the damaging impact of false information will undoubtedly increase even in a democracy as relatively strong and stable as Australia’s. More people than ever are now getting their ‘news’ from sources published on the internet, especially from sources that are not even news outlets; for example, over 50% of Australians now turn to social media platforms for ‘news’ (Hughes, 2021). But false information does not need to be a mass phenomenon to cause harm because precision disinformation can have enormous pay-offs providing it is directed at the right voters; this can be achieved by the use of data enhancement and audience segmentation techniques that deliver custom-made disinformation to very narrow and specific audiences. In any case, it is fast becoming a mass phenomenon, and the exponential spread of disinformation, rumours and conspiracy theories is corroding public trust in government institutions, polarising the electorate and stoking populist, anti-democratic and extremist sentiment (Zimmerman & Kohring, 2020). It is also ‘causing significant damage to the public sphere and shifting the acceptable bounds of political debate’ (Roose & Khalil, 2020).

Notes

  1. 1.

    For a more recent empirical assessment of the motivations for spreading disinformation see Buchanan (2020). Note that those most likely to share false information were those who thought it likely to be true, or who had pre-existing attitudes consistent with it.

  2. 2.

    Note, however, that Michele Cantarella et al.’s (2019) study of the 2013 and 2018 Italian general elections found that exposure to false information had a negligible effect on populist voting (Cantarella et al., 2019).