Weaponising personality through propaganda: the effects of personality-congruent micro-targeted foreign propaganda on political stability

56  Download (0)

Full text


Weaponising personality through propaganda:

the effects of personality-congruent micro-targeted foreign propaganda on political stability

Neill Bo Finlayson (12697915) Master’s Thesis

Communication Science (Research Master’s) Graduate School of Communication Science,

University of Amsterdam Supervisor: Dr. Tom Dobber

Word count: 7,498 July 1, 2022


Propagandists now have an array of sophisticated tools at their disposal to influence foreign adversary audiences. One such technique that has proved particularly popular with nefarious actors seeking foreign interference is personality-based micro-targeting. Yet, questions remain about the true effectiveness of micro-targeted propaganda in achieving propagandists’

ultimate goal: political instability. This paper aims to examine whether propaganda messages that are congruent to a person’s personality more effective in raising levels of political instability. Relying on self-congruity theory, the experiment (N = 200) manipulated the emotionality of propaganda messages to ascertain they are more effective for people with dark personality traits. The main analysis results provided no evidence to support this proposition. However, a robustness check, using a more nuanced set of variables, did find that people high on the Dark Triad are more susceptible to influence in relation to their levels of trust in government if they are exposed to anger-based (i.e., personality-congruent)

propaganda messages. The results also showed that this effect is not dependent on the extent to which participants cognitively process a message. These findings have important

theoretical and societal implications that contribute to salient debates on the threats posed by micro-targeted propaganda.

Keywords: propaganda, micro-targeting, foreign interference, political instability.




As communication technologies have continued to develop at pace in the 21st century, so too have propaganda strategies. Political parties and state actors around the world are

increasingly using social media and computational tools to “shape public attitudes, opinions, and discourses at home and abroad” with a view to influence “the outcomes of elections, disrupt diplomatic efforts, and undermine peacebuilding efforts” (Bradshaw & Howard, 2018, p.23, 29). The most famous example of this came when the Russian government was adjudged to have interfered in the 2016 US Presidential election in a “sweeping and

systematic fashion” through disinformation campaigns on social media (Mueller, 2019, p.1), while similar attempts at foreign interference were documented during the 2019 EU

elections (European Commission, 2020, p.3). Russia’s invasion of Ukraine in 2022 has highlighted once again that propaganda remains a vital tool in the armoury of states looking to exert foreign influence as both sides have utilised contemporary techniques to not only win the hearts and minds of the watching world but to destabilise and undermine opposition support (Gilbert, 2022; Harwell & Lerman, 2022; Warzel, 2022). The question is: how effective, and thereby dangerous, are these new techniques?

Propaganda assumes a “pejorative meaning” in modern vernacular, associated primarily with manipulation and distortion (Till, 2021, p.1363). It is defined as: “communication designed to manipulate a target population by affecting its beliefs, attitudes, or preferences in order to obtain behaviour compliant with political goals of the propagandist” (Benkler et al., 2018, p.29). Propaganda is thus “misleading by design” and, although not necessarily underpinned by an ideological agenda, it has an inherently nefarious aim (Vamanu, 2019, p.199). This paper focuses on foreign propaganda - campaigns directed at domestic audiences by foreign actors - because it poses more complex regulatory and normative challenges than domestic campaigns, while posing a greater threat to the global political



order (Ó Fathaigh et al., 2021). Primarily, such efforts are considered to be a form of vertical propaganda: top-down, subversive communication efforts conducted by

governments or states (Ellul, 1973; Fairfield, 2018). Another important distinction is that this study focuses on black propaganda (Becker, 1949; Jowett & O’Donnell, 2006),

whereby the perpetrators’ role is obfuscated through false sources used to peddle misleading information - as opposed to white propaganda whereby the perpetrator is identifiable as the source (Gelders & Ihlen, 2010; Lukito, 2020).

From Twitter bots to troll factories (Linvill & Warren, 2020; Zannettou et al., 2019), propagandists now have an array of tools at their disposal to help ‘firehose’ black propaganda at audiences (Paul & Matthews, 2016; Shao et al., 2018). One such tool is micro-targeted advertising on social media, which has transformed the efficiency and impact of propaganda campaigns (Bradshaw & Howard, 2018; Johnson, 2021; Till,

2021). Considered the “holy grail” of political campaigning (Tufekci, 2014), micro-targeting is defined as “a strategic process intended to influence voters through the direct transmission of stimuli, which are formed based on the preferences and characteristics of an individual”

(Papakyriakopoulos et al., 2018, p.2). Rather than broadcasting uniform, top-down

propaganda messages to mass audiences, as has been the case for the last century (Wainless

& Berk, 2020), propagandists can now “send the ‘right’ propaganda message to the ‘right’

person”, tailoring the tone and content of their messages to the particular interests, fears or beliefs of specific groups in society (Ó Fathaigh et al., 2021, p.857). As such, micro-

targeting has the potential to amplify the effects of propaganda messaging, especially when compared to more static traditional propaganda techniques (Dobber et al., 2021).

Psychometric profiling, as a micro-targeting technique, is particularly potent. This involves mining large swathes of personal data on social media to infer users’ personality traits and



target them with messages tailored to these specific traits (Zarouali et al., 2020).

Psychometric micro-targeting was mastered, and made infamous, by Cambridge Analytica - the political consultancy firm associated with the surprise victories of the Leave campaign in the UK Brexit referendum and Trump in the 2016 US election (Confessore, 2018; Grassegger

& Krogerus, 2017). The key to Cambridge Analytica’s success was “finding emotional triggers for each individual voter” and to then target those most persuadable (Cadwalladr, 2017). This manipulation and exploitation of people’s deep-seated emotional blindspots became attractive to nefarious actors seeking to “stoke grievances and incite social conflict”

(Ribeiro et al., 2019, p.140), particularly state actors, such as Russia (Jakubowski, 2019).

While the objectives of micro-targeted campaigning for the Trump and Leave campaigns focused on vote choice and mobilisation, Kremlin propaganda is primarily designed to

“distract, divide, and demoralise” adversary populations (Linvill & Warren, 2020, p.464).

This is all with an aim to “destabilise a society and a state through massive psychological conditioning” (Bayer et al., 2019; Russell, 2016, p.2). Despite this threat, it remains unclear how effective these micro-targeting tactics actually are in achieving these goals.

Academic research has hitherto provided few answers to this question of effectiveness. In fact, micro-targeted propaganda generally is “a problem that has received scant attention in literature” (Ó Fathaigh et al., 2021, p.856). Although personality-based micro-targeting can be influential on consumer behaviour (Matz et al., 2017), there is scarce empirical evidence of its effectiveness in a political context (Papakyriakopoulos et al., 2018). What little research that does exist on political micro-targeting tends to focus on domestic campaigns, measuring effects on voter turnout and behaviour (Haenschen & Jennings, 2019). However, what marks micro-targeted propaganda out as different to ordinary political campaigning – and

potentially more dangerous – is that the intention is to not simply influence voter turnout or behaviour but cause wider political instability within foreign nations through covert means



(Linvill & Warren, 2020; Paul & Matthews, 2016). Yet, no such research exists yet that tests the relationship between personality-based micro-targeted propaganda and political

instability, which, considering the gravity of the threat it poses, is alarming. The consequence is that as debate continues to rage over who is ‘winning’ the information war between Russia and Ukraine, global powers remain naïve about the true efficacy, and thereby threat, of the propaganda techniques being used. Thus, the overarching research question for this paper is as follows:

RQ: To what extent does exposure to personality-based micro-targeted propaganda by foreign actors increase levels of political instability?

Theoretical framework

In order to grasp the effectiveness of personality-based micro-targeting, it is first important to consider self-congruity theory. The crux of self-congruity theory is that individuals respond better to messages that are congruent to their own sense of self-concept (Sirgy, 1982, 1985).

The notion of self-concept comprises many dimensions and conceptualisations (Markus &

Kunda, 1986; Sirgy, 1982), one of which is an individual’s personality (Aaker, 1999). Indeed, research shows that personality – defined as someone’s “enduring psychological tendencies”

(Mondak et al., 2010, p.98) – is a key driver of individual attitudes and behaviour (Nai, 2019). Therefore, according to self-congruity theory, messages that are congruent to a

person’s personality will be more effective because they exhibit greater personal relevancy to that person (Matz et al., 2017; Xue & Phelps, 2013). Even though it has been deployed in several high-profile political campaigns (Zuiderveen Borgesius et al., 2018), there is a lack of research on the use of personality-congruent advertising in a political context

(Papakyriakopoulos et al., 2018). One recent study by Zarouali et al. (2020) did, however,



find that people were more inclined to change their behaviour or attitude when exposed to a political advert congruent to their personality.

This use of self-congruity theory by Zarouali et al. (2020) relates to the concept of canalisation in propaganda theory. Canalisation is the idea that for propaganda to be

successful it must exploit pre-existing attitudes, norms and values to channel information to the target based on their predispositions (Lazarsfeld & Merton, 1948). Today, this can be achieved through psychometric profiling (Till, 2021), meaning that social media has

essentially become “a canalisation machine” for propagandists (Young & McGregor, 2020).

There is, however, little research into the role of personality in this process. The

aforementioned study by Zarouali et al. (2020) operationalised personality using extraversion which, although insightful, is just one of many relevant personality traits (e.g., Cooper et al., 2013; Gerber et al., 2013). A more suitable set of traits for the study of murkier political campaigning, such as black propaganda, is the Dark Triad (Nai & Maier, 2021a).

The Dark Triad, propaganda and affect congruity

The Dark Triad comprises three “socially nefarious” personality traits (Nai & Maier, 2021a, p.271): narcissism, psychopathy and Machiavellianism (Furnham et al., 2013; Paulhus &

Williams, 2002). Each trait in the triad relates to different types of emotional deficits (Jonason & Krause, 2013). Psychopathy encapsulates callousness, limited empathy and impulsivity; narcissism is linked with feelings of superiority, ego-reinforcing behaviours and low empathy; while Machiavellianism is characteristic of cynicism, manipulation and

strategically-oriented behaviour in pursuit of self-beneficial goals (Gøtzsche-Astrup, 2021;

Jonason & Krause, 2013; Paulhus & Williams, 2002). Most importantly, however,

individuals high in the Dark Triad tend to be more susceptible to manipulation (Hart et al., 2021) – the key cognitive deficit that propagandists seek to exploit (Benkler et al., 2018).



Propaganda tends to deploy emotive rhetoric to maximise the affective arousal felt by its targets (Gelfert, 2018), which can increase the persuasiveness of a message (Brader, 2005;

Nai & Maier, 2018; Wirz, 2018). Of course, persuasiveness also depends on the emotionality of the message and personality of the audience. For instance, people with Dark Triad traits are unable to process negative expressions appropriately due to a “selective emotional

empathy dysfunction” (Ali et al., 2009, p.760), and are therefore more persuaded by negative and fearful messaging (Galais & Rico, 2021). Furthermore, people high on the Dark Triad tend to convey negative emotions more than others online (Preotiuc-Pietro et al., 2016), while they associate strongly with negative emotions, such as anger and contempt (Pavlović &

Franc, 2021). Considering this, it is logical to assume that people high on the Dark Triad will be particularly receptive to propaganda messaging that evokes fear and anger. The question then is to what end?

Propaganda and political instability

The intention of foreign propaganda is “to mislead, polarise and destabilise” adversaries for political gain (Bayer et al., 2019, p.139). Because the stability of any political entity rests in the “continuity of those elements by which that object is identified” (Dowding & Kimber, 1983, p.237), any force that threatens the continuity of established elements in a political system may be considered a threat to political stability. Two such threats are affective polarisation and low trust in government.

Affective polarisation

Affective polarisation – that is, the mutual animosity felt between partisan groups - is now considered “one of the main concerns for the health and quality of contemporary

democracies” in the world (Hernández et al., 2021, p.1). It refers to the extent to which people feel warmth towards their partisan in-group and animosity towards out-groups



(Druckman & Levendusky, 2019; Wagner, 2021). Affective polarisation intensifies political competition and amplifies conflict, which breeds contempt for the democratic process (Citrin

& Stoker, 2018; Rodon, 2022) and diminishes trust in politics (Torcal & Carty, 2022) – two harbingers for political instability (Listhaug & Ringdal, 2008). Furthermore, affective polarisation is associated with increased intolerance, incivility and a decline in political collaboration (Hobolt et al., 2021; Layman et al., 2006); which are all key indicators of an unhealthy democracy (MacKuen et al., 2010).

Any manner of competing identities or societal rifts that cause in-group versus out-group polarisation - such as ethnicity, nationality or religion - can rouse affective polarisation (Torcal & Comellas, 2022). Such cleavages are often exploited by politicians through the use of emotional messaging, which works to embolden supporters and harden opposition (Brader, 2005, 2006). Negative messaging, via online media in particular, induces even greater

opinion polarisation as individuals ‘double-down’ on their beliefs (Asker & Dinas, 2019;

Kim & Kim, 2019). Considering that propaganda is “a discourse built around powerful dichotomies”, such as ‘good vs. evil’ and ‘us vs. them’, aimed at triggering affective arousal (Vamanu, 2019, p.206), and given that negative emotional messaging is positively associated with polarisation, the following is expected:

H1a: Exposure to a propaganda message that elicits negative affect (i.e., fear or anger) will increase levels of affective polarisation, as compared to the control message.

Previous attempts to examine the moderating effects of personality on affective polarisation have mainly focused on the Big Five Model of personality traits (e.g., Satherley et al., 2020).

The Dark Triad has received comparatively little academic attention on the subject, despite evidence linking these traits to polarising behaviour more generally. For instance, people high



on the Dark Triad are more likely to engage in uncivil and aggressive behaviour online (Frischlich et al., 2021; Petit & Carcioppolo, 2020) and tend to have stronger intentions to engage in political violence due to increased partisan identities (Gøtzsche-Astrup, 2021).

Furthermore, the Dark Triad has been found to positively interact with perceived group relative deprivation to predict radicalisation (Pavlović & Franc, 2021) and is positively associated with increased outgroup threat perceptions (Galais & Rico, 2021) – key features of affective polarisation. Since self-congruity theory suggests that messaging which is congruent to a person’s personality is more influential than incongruent messages, and those with Dark Triad traits are particularly susceptible to negative emotional messaging, the following is expected:

H2a: Exposure to a propaganda message that elicits negative affect (i.e., fear or anger) will be more influential in increasing affective polarisation for participants high in Dark Triad traits than those low in Dark Triad traits.

Trust in government

Government trust is a foundational component of democracy (Listhaug & Ringdal, 2008;

Spiteri & Briguglio, 2018). Defined broadly as the public’s confidence in the government to

“not abuse their privileged positions of power” (Lühiste, 2006, p.476), recent research on the matter has been motivated by the perceived decline of public trust in politics (van der Meer, 2017; Webster, 2018). There is debate as to whether this phenomenon is being driven by increasingly negative emotional political campaigning. Early research found no such link (Lau et al., 1999; Martinez & Delegal, 1990), but subsequent research has found that negative messaging does relate to political distrust (Gotlieb et al., 2017; Lau et al., 2007). However, because audiences “interpret new political information through partisan lenses”, the

relationship between negative messaging and political trust likely depends more on message



content, an individual’s partisan identity or their media habits (Klein & Robison, 2020, p.59).

Therefore, it is unreasonable to expect that propaganda messaging will have uniform effects on political trust across the public. The following is expected:

H1b: Exposure to a propaganda message that elicits negative affect (i.e., anger or fear) will have no significant effect on levels of trust in government, as compared to the control message.

A frequently cited predictor of political trust is personality (Mondak, 2010; Mondak et al., 2010), although research on this has primarily been limited to the Big Five traits (e.g.,

Mondak & Halperin, 2008). Yet, inferences can be made from Dark Triad theory to speculate about the influence of dark personalities on political trust. For instance, people high on Dark Triad traits tend to score low on agreeableness (Jones & Paulhus, 2014), which is one of the strongest correlates of high levels of political trust (Mondak & Halperin, 2008). Some facets of the Dark Triad also relate positively to gullibility and a general trust dysfunction, while some correlate negatively with trust in benign authorities (Hart et al., 2021). Overall, the literature suggests that narcissists’ sense of entitlement means they likely perceive themselves as victims of authority and are thereby less trusting; Machiavellians’ cynicism means they are inherently distrusting of others; while the “impulsive nonconformity” of psychopaths draws them to anti-establishment sentiments (Galais & Rico, 2021, p.3). The following is expected:

H2b: Exposure to a propaganda message that elicits negative affect (i.e., fear or anger) will be more influential in decreasing trust in government for participants high in Dark Triad traits than those low in Dark Triad traits.

Mediating role of message elaboration

The extent to which someone is motivated to process a message (i.e., message elaboration) is an important factor to consider when assessing the effectiveness of political messaging. In



line with the principles of self-congruity theory, research shows that personalising a message improves message elaboration due to the increased perception of self-relevance in the

message (Sahni et al., 2018). This means that when a message is perceived to be congruent to a recipient’s sense of self, they cognitively engage with the message more than if it is

perceived to be incongruent (Wheeler et al., 2005; Zarouali et al., 2020). Therefore, higher message elaboration can increase the persuasiveness of the message to positively influence recipients’ attitudes and behaviours (Petty & Cacioppo, 1986; Wheeler et al., 2008; Zarouali et al., 2020). In particular, negative emotions increase message elaboration and thus

persuasion (Nai & Maier, 2018), especially for people high on the Dark Triad traits (Galais &

Rico, 2021). Therefore, it is logical that a propaganda message that is congruent to a participant’s personality (i.e., a negative message received by someone high in the Dark Triad) will lead to an increase in message elaboration, and thereby persuasiveness, which will in turn increase levels of political instability. Thus, the following is expected:

H3: When a propaganda message is congruently matched with someone’s personality, this will lead to an increase in message elaboration which in turn will lead to higher levels of political instability (i.e., higher affective

polarisation and lower trust in government).

Based on the preceding theoretical framework, the following model is proposed:

Figure 1: Visualisation of the hypothesised moderated mediation model.



Methods and Data Sample

A non-random convenience sample (N = 200) was used for this study. Participants were recruited from the social network of the author, social media groups and online survey

sharing platforms. The solicitation of participants was premised on participants being resident in the United Kingdom, the Netherlands or Germany - the three countries for which the stimulus context applies. The majority of participants reside in the UK (66.5%), more than in the Netherlands (27%) and Germany (6.5%). The sample is relatively young, with 45% in the 25-34 age category and 17.5% in the 18-24 category, while the majority were female, making up 55.8% of the sample. The sample was aligned to the political left (M = 3.52, SD = 1.72), with 75% placing themselves left of centre. Participants were also well educated, with 79%

having at least a bachelor's degree and 42.5% with a master’s degree (M = 5.32, SD = 1.02).

The sample was fairly evenly split between the experimental conditions (ncontrol = 68; nfear = 64; nanger = 68).

Design, Procedure and Stimuli

To test the above hypotheses, this study utilises a 2 (personality: high on Dark Triad vs. low on Dark Triad) X 3 (message emotionality: fear vs. anger vs. control) between subjects design. Small to moderate effect sizes are expected to be found (Roczniewska & Higgins, 2019; Zarouali et al., 2020). As such, an a priori power analysis was conducted using G*Power based on a significance level of α =.05, an effect size of η2 =.06, and a statistical power of (1-β)=.80 (Cohen, 1988). This minimum sample size required to reliably detect main and interaction effects is 179.

The experiment was conducted through a self-administered online survey, using Qualtrics software (see Appendix B). Upon reading an introduction to the survey and signing the



informed consent (see Appendix C), participants were asked for their sociodemographic characteristics. They were then asked questions measuring Dark Triad personality traits.

After this, participants were randomly assigned to one of three stimuli with differing emotionality: a fear appeal, an anger appeal and a control statement. The stimuli (see Appendix A) were presented as mock Facebook posts published by a fictional organisation named the ‘Moscow Forum for Democracy’. All features (e.g., graphics, engagement

statistics, format and size) were kept constant across the three conditions, apart from the main text. The statement text related to the recent sanctions imposed on Russia by European

nations due to the Ukraine conflict, while also referring to the Covid-19 pandemic, casting doubts over the integrity of western European governments, in particular the UK, the Netherlands and Germany.

The emotionality of each statement was manipulated by altering key words or phrases, based on message framing literature. The fear condition evoked feelings of fear, worry, uncertainty, irrationality and anxiety, with a focus on threatening or frightening hazards (Petersen, 2010;

Steenbergen & Ellis, 2006); the anger condition evoked outrage and aversion, with a focus on negative threats, blame and unfavourable moral evaluations (Petersen, 2010; Steenbergen &

Ellis, 2006); while the control condition used non-emotive words or phrases to achieve a more neutral emotionality. The stimuli are text-only which is sufficient for this study, especially considering audio or visual cues make little difference to stimuli effects or participant attention (Groh et al., 2022; Hameleers et al., 2020). A timer of 20 seconds was put on each stimulus to ensure participants spent sufficient time considering the statement.

Participants were then asked a series of questions relating to trust in government, affective polarisation and message elaboration, in an order randomised for each participant. The order of all items in each question battery was also randomised. Finally, participants had to answer



the manipulation check before being presented with a debriefing statement (see Appendix C) at the end of the survey.

Stimuli pre-test

A pre-test was conducted for the stimuli using a small convenience sample (N = 15).

Participants were all shown the three stimuli in a randomised order before being asked to evaluate the level of fear, anger and negativity exhibited in each stimulus (see Appendix D).

Results showed that the fear condition (M = 5.67) exhibited higher levels of fear than the anger condition (M = 4.47) but this difference was not significant, p=.29. Similarly, the anger condition (M = 7.80) exhibited higher levels of anger than the fear condition (M = 7.13), but to an insignificant degree, p=.051. The manipulated conditions (fear and anger) did, however exhibit a significantly higher level of negativity (M = 8.23) than the control condition (M = 6.01), p<.05. Therefore, the pre-test was only partially successful. The stimuli were revised to make the differences in emotionality between the fear and anger stimuli more pronounced.


The exact question wording and answer options for each measure is in Appendix B.

Dark Triad

Dark Triad personality traits were measured through participant self-reporting. Although this weakens measurement validity and reliability, it is a well-established method to measure personality and the alternatives are no less limiting (Zarouali et al., 2020). A validated battery of 12 items, known as the “Dirty Dozen”, was deployed (Jonason & Webster, 2010; Nai &

Maier, 2021b), with responses recorded on a seven-point answer scale (from ‘Strongly disagree’ to ‘Strongly agree’). A principal axis factoring analysis showed that the 12 items loaded onto three factors, based on Kaiser’s criterion (Eigenvalues > 1), which explained 58.8% of the variance in all items. This likely reflects the three dimensions of the Dark Triad.



The three traits are often combined to create one single construct (e.g., Moshagen et al., 2018) - a precedent which is followed in this study. Therefore, the 12 items were aggregated to form a single-factor Dark Triad index variable, which was deemed just sufficiently reliable (Cronbach’s α =.79; M = 2.97, Mdn = 2.92, SD =.79).

A median split was performed to dichotomise the Dark Triad variable to ‘low’ (n= 104) and

‘high’ (n= 96) groups using a split-point of 2.92 on a seven-point scale. Although not hugely distant from the mid-point (3.5), a strictly conservative interpretation of a median split would dictate that the groups created by the split here are not sufficiently meaningfully. This may lead to an increase in Type I errors, a loss of power and decreased validity (Aiken & West, 1991; DeCoster et al., 2011; Rucker et al., 2015). However, competing literature shows that such median splits have a negligible effect on strength of associations and likelihood of Type I error, while posing no real threat to statistical robustness (Farrington & Loeber, 2000;

Iacobucci et al., 2015a). In any case, performing a median split in an experimental design is deemed “completely legitimate” (Iacobucci et al., 2015b, p.690). Therefore, the main analysis will be conducted first using a median split, followed by a robustness check without the median split.

Political instability

Political instability is operationalised here through two proxy dependent variables: affective polarisation and trust in government.

Affective Polarisation

The measurement of affective polarisation used here attempts to tap into the horizontal societal cleavage between those who trust the government and those who do not. This is a novel approach to measuring affective polarisation and has not been validated in extant literature. However, this approach is justified on three fronts: affective polarisation can stem



from non-traditional partisan cleavages (Torcal & Comellas, 2022); polarisation is linked with a growing divide in trust in government (Torcal & Carty, 2022); and levels of political trust among the public have diverged significantly since the Covid-19 pandemic (Rieger &

Wang, 2021). A feeling thermometer - the most widely used measure of affective polarisation (Iyengar et al., 2019) - was used here, based on the one used by Iyengar et al. (2012). Two items asked participants to self-report how warm they felt towards those who tend to trust the government and those who tend to distrust the government on a 0 to 100 scale (‘Very cold’ to

‘Very warm’). The difference in scores between the two items represents a participant’s overall affective polarisation score on a 10-point scale, (M = 2.65, SD = 1.83).

Trust in government

Participants’ level of trust in government was measured using a validated battery of questions, based on the study by Im et al. (2014), which asked them to rate the extent to which they agree (from ‘Strongly disagree’ to ‘Strongly agree’) with statements relating to their levels of government trust. Deviating slightly from the Im et al. (2014) battery, the items were presented as statements, rather than questions, and responses were recorded on a seven- point scale. The items formed a reliable scale (Cronbach’s α =.85; M = 3.02, SD = 1.26).

Message elaboration

Four items, designed and validated by Wheeler et al. (2005), were used to gauge the self- reported level of cognition that participants exerted when reading and thinking about the stimulus message. The items were reworded into first-person statements, rather than

questions, and a seven-point answer scale was used (‘Strongly disagree’ to ‘Strongly agree’).

However, the four items formed an unreliable scale (Cronbach’s α =.68; M = 3.51, SD =.73) and so the third item was removed to create a more reliable scale, (Cronbach’s α =.72; M = 3.64, SD =.78). Although still below the recognised reliability threshold for Cronbach’s α, the



three-item scale was deemed sufficiently reliable for the purposes of this study given its validated reliability in extant literature.

Data preparation and quality control

All incomplete responses were removed from the sample. A response was considered complete if the participant answered all questions up to and including the final manipulation check for negativity, the last survey item that required a response. Respondents from outwith the UK, Netherlands or Germany were excluded from the sample. Similarly, respondents that straight-lined responses (had zero variance) in the trust in government and affective

polarisation batteries, which also included reverse-coded items, were excluded from the sample. Participants who completed the survey in less than three minutes1 were also

excluded. For the data analysis, the conventional threshold for statistical significance of a p- value smaller than .05 is adopted.

Randomisation check

To ensure sample demographics were evenly split between the experimental groups, a randomisation check was performed using sex and age. Two one-way analysis of variance (ANOVA) tests showed that neither the difference in mean sex between the conditions, F(2, 196) = 2.03, p>.05, nor the mean age, F(2, 197) =.09, p>.05, was significant. Thus, the sample is sufficiently randomised across the conditions.

Manipulation check

A manipulation check was conducted by asking participants at the end of the survey to rate the extent to which the statement they read exhibited feelings of fear or anger and to what extent it was negative. A series of independent t-tests showed that the fear condition (M =

1Less than three minutes was deemed an insufficient amount of time to reliably engage with the survey following survey previews and practice runs, conducted by the researcher.



6.43, SD = 2.67) exhibited a greater sense of fear than the anger condition (M = 6.10, SD = 2.22), but this difference (Mdifference =.33) was not statistically significant, t(129) =.78, p=.44, 95% CI [-.51, 1.18]. Similarly, the anger condition (M = 7.82, SD = 1.23) exhibited a greater sense of anger than the fear condition (M = 7.47, SD = 1.92), but again the difference

(Mdifference =.35) was not statistically significant t(217) = -1.25, p =.21, 95% CI [-.91, .20].

This means that respondents did register a difference in emotionality between the two conditions, just not to a statistically significant degree. In terms of negativity, there was a statistically significant difference between the degree of negativity in the fear condition compared to the control, t(125) = -3.52, p<.01, 95% CI [-1.45, -.41], as well as the anger condition, t(130) = -2.28, p =.02, 95% CI [-1.20, -.08]. This means negativity was

successfully manipulated in the experimental conditions. It is therefore fair to conclude that the experiment was partially successful. However, the perceived similarity in emotionality between the two experimental conditions may adversely affect the likelihood of uncovering significant effects in the analysis.

Results Main analysis

The main analysis was conducted through two factorial ANOVAs. The independent variables were propaganda message type (control; fear; anger) and personality (low Dark Triad; high Dark Triad). Two separate analyses were run for each dependent variable: affective

polarisation and trust in government.

The first ANOVA found that the direct effect of propaganda message exposure on affective polarisation was not statistically significant, F(2, 194) =.58, p =.561, nor was the direct effect of personality on affective polarisation, F(1, 194) =.21, p =.651. The interaction effect of propaganda message and personality on affective polarisation was also not statistically significant, F(2, 194) = 1.01, p =.365. It also has a very weak effect size, η2 =.01. All



assumptions were met: Levene’s F-test was not statistically significant, F(5, 194) = 1.34, p = .250, so equal variance in the population can be assumed. Hypotheses H1a and H2a must be therefore be rejected.

A separate ANOVA revealed that the direct effect of propaganda message exposure on trust in government was not statistically significant, F(2, 194) = 2.63, p=.074, which offers support for H1b: exposure to a propaganda message with negative affect does not have a significant effect on trust in government. The analysis also showed that the direct effect of personality on trust in government was not statistically significant, F(1, 194) = 315, p=.575.

The interaction effect of propaganda message and personality on trust in government was also not statistically significant, F(2, 194) = 1.00, p=.371. It also has a very weak effect size, η2 =.01. All assumptions were met and Levene’s F-test was not statistically significant, F(5, 194) =.96, p =.443.

Robustness check

A series of regression analyses were carried out, as part of a robustness check on the main analysis, using the continuous Dark Triad variable and dummy variables for the experimental treatment conditions. The regression models predicting the direct effect of a fear-based propaganda message, and the interaction effect of personality, on both affective polarisation, F(3, 196) = 1.05, p =.374, and trust in government, F(3, 196) =.91, p =.439, were not statistically significant. Similarly, the regression model predicting the direct effect of an anger-based message, and the interaction effect of personality, on affective polarisation was not statistically significant, F(3, 196) = 1.11, p =.346. Thus, H1a and H2a are rejected.

However, the regression model predicting the effect of an anger-based message, and the interaction effect of personality, on trust in government was statistically significant, F(3, 196)

= 3.58, p =.015. The model can thus be used to predict levels of trust in government.



However, the strength of the prediction is small, with exposure to an anger-based message and personality explaining just 5% of the variance in trust in government (R2 =.05). As shown in Table 1, exposure to an anger-based message, compared to the control message, has a statistically significant direct effect on trust in government, b =.42, b* =.16, t = 2.25, p =.026, 95% CI [.05, .79], even when controlling for the interaction effect, b = 1.87, b* =.70, t = 2.54, p =.012, 95% CI [.42, 3.32]. This does not support H1b. The analysis also showed a statistically significant direct effect of Dark Triad scores on trust in government, b =.27, b*

=.17, t = 1.99, p =.049, 95% CI [.00, .54]. This means that for every one-point increase in Dark Triad scores a participant’s level of trust in government is likely to increase by .27 units. This contradicts the results of the main analysis.

Table 1: Results of the regression analyses

The interaction effect between exposure to an anger-based propaganda message and Dark Triad scores on trust in government is also statistically significant, b = -.48, b* = -.58, t = - 2.04, p =.043, 95% CI [-.95, -.02]. This means that for every unit increase in Dark Triad scores, participants exposed to the anger-based condition will likely have an additional decrease of .48 units in trust in government compared to those exposed to the control condition. In other words, a one unit increase in Dark Triad scores is linked with a .27

increase in trust in government, but for those exposed to the anger-based condition, it is a .21



decrease. As shown in Figure 2, these results suggest that people higher on the Dark Triad are likely to have significantly decreased levels of trust in government than those low on the Dark Triad after exposure to an anger condition, as compared to exposure to the control condition. Thus, the effect of exposure to an anger-based message in lowering trust in government is significantly stronger among participants high on Dark Triad traits compared to those low on Dark Triad traits. Therefore, although the main analysis does not support H2b, the regression analysis does partially support H2b but only regarding anger-based and not fear-based messaging.

Figure 2: Scatter plot of the interaction effect

Moderated mediation analysis

The hypothesised moderated mediation model was tested using the PROCESS macro model number 7 (Hayes, 2013) with a multi-categorial approach (Hayes & Preacher, 2014). As such, k-1 dummy variables were created and inserted into the moderated mediation model,



along with the continuous Dark Triad variable and message elaboration. Two separate analyses were run to account for the two dependent variables.

With affective polarisation as the dependent variable, the results of the first analysis showed that the index of moderated mediation for the first model, with the fear-based message as the treatment condition, was statistically non-significant (b= -.006, S.E.=.062; 95% CI [-.129 to .147]). The second model, using the anger-based message as the treatment condition, also found a non-significant index of moderated mediation (b= -.012, S.E.=.063; 95% CI [-.130 to .134]). For both models, zero falls within the confidence intervals which means that there is not a statistically significant moderating effect of personality on the indirect effect of neither fear-based nor anger-based messaging on affective polarisation via message elaboration.

With trust in government as the dependent variable, the results of the second analysis showed that the index of moderated mediation for the first model in this analysis, using the fear-based message as the treatment condition, was statistically insignificant (b =.003, S.E.=.036; 95%

CI [-.078 to .078]). The second model, with the anger-based message as the treatment

condition, also revealed a non-significant index of moderated mediation (b =.007, S.E. =.036;

95% CI [-.080 to .082]). Therefore, there is not a statistically significant moderating effect of personality on the indirect effect of neither fear-based nor anger-based messaging on trust in government via message elaboration. Altogether, these results do not support H3.


The proliferation of political micro-targeting on social media has spawned new opportunities for foreign propagandists to disrupt and deceive adversary audiences (Till, 2021). Yet, despite the threats posed by micro-targeted propaganda to society, it is an issue that receives little academic study (Ó Fathaigh et al., 2021). The aim of this paper was to examine the causal effects of personality-congruent micro-targeted propaganda on political instability



using an experimental design. In answer to the research question, the main analysis provides no evidence that exposure to personality-congruent propaganda messaging increases levels of political instability. However, the robustness check, conducted using regression analyses, does provide some evidence of its effectiveness.

First, the robustness check revealed a significant direct effect of the anger-based message on trust in government. Somewhat surprisingly, this shows that exposure to an anger-based propaganda message can in fact increase trust in government, while holding Dark Triad traits constant. This result confounds theoretical expectations that negative messaging correlates with distrust in government (Gotlieb et al., 2017; Lau et al., 2007), while also suggesting that negative messaging can have uniform effects on political trust, contrary to expectations. This discovery adds further complexity to this already incongruous line of research.

Second, the robustness check found a statistically significant interaction effect between the anger-based message and personality on trust in government. This supports the assertion that people high on the Dark Triad are more susceptible to influence in relation to their levels of trust in government if they are exposed to anger-based propaganda messages, as compared to neutral messages. In other words, personality-congruent propaganda messaging can be more effective than incongruent messaging in lowering government trust. This finding extends the body of research on personality-congruent micro-targeting beyond domestic political

campaigning and into the realm of foreign propaganda - a form of political campaigning that is arguably more threatening and corrosive to political stability (Linvill & Warren, 2020; Paul

& Matthews, 2016). Not only that but it follows Zarouali et al. (2020) in reaffirming the application of self-congruity theory to micro-targeting research (Sirgy, 1982, 1985). This is especially important for the study of propaganda as it proves that the concept of canalisation in propaganda theory remains relevant in examining media effects in the social media age



(Lazarsfeld & Merton, 1948; Till, 2021). This, of course, warrants further academic investigation.

It may be inferred from these findings that the initial dichotomisation of the Dark Triad, via a median split, created two groups that did not meaningfully represent ‘high’ and ‘low’ in the sample. This is supported by the results of the robustness check which, by harnessing greater nuance through the continuous variable, appears to more validly measure the effects present in the sample. With this in mind, the results of the robustness check should take precedence.

However, the main analysis results should not be dismissed outright and must be borne in mind when interpreting the findings of this study. The lack of any statistically significant results in the moderated mediation analysis shows that the effects found in the robustness check are not contingent on the extent to which participants cognitively process a message.

Therefore, in answer to the research question, this study does provide some causal evidence that exposure to specific personality-congruent messages (anger appeals for people high on the Dark Triad) can increase certain aspects of political instability (lower trust in

government). That said, the incongruous results of the main analysis and robustness check do mean the results of this study are somewhat ambiguous. Therefore, further academic

investigation is required to validate the findings.

From a wider theoretical perspective, this study contributes to the embryonic body of literature on personality-based political micro-targeting (Ó Fathaigh et al., 2021). The findings provide some evidence that supports the theoretical framework which, relying on many facets of Dark Triad theory (Hart et al., 2021; Jones & Paulhus, 2014; Mondak &

Halperin, 2008), posited that negative messaging may be more effective in increasing political instability for people with dark personalities. Prior research on this subject either analysed more ubiquitous theoretical models of personality traits, such as the Big Five (e.g.,



Mondak & Halperin, 2008), or analysed political micro-targeting in a domestic election context (e.g., Haenschen & Jennings, 2019). What is particularly novel about this study, and the findings, is that it finds an interaction between the intrinsically manipulative properties of propaganda messaging with more malevolent personality traits (Nai & Maier, 2021a; Till, 2021) – an unsurprising effect but one which has thus far escaped empirical scrutiny.

Aside from the theoretical relevance of this study, the findings here underline the threat that micro-targeted foreign propaganda poses to society. It suggests that micro-targeted

propaganda which exploits people’s personality and emotional triggers can be influential in destabilising adversary populations – the raison d’êtreof foreign propagandists (Bayer et al., 2019). The societal implications of this prompts further empirical investigation into micro- targeted propaganda, especially since the efficacy of these micro-targeting techniques will only continue to improve. The many different theoretical and methodological strands to this study may provide fruitful avenues for future research. For instance, affective polarisation and trust in government are just two dimensions of political instability and, considering the raft of factors which may constitute a threat to political stability (Dowding & Kimber, 1983), there is tremendous scope for exploring the different ways in which micro-targeted

propaganda impacts political stability. Similarly, the Dark Triad is just one of many sets of personality trait models (e.g., Gerber et al., 2013; Zarouali et al., 2020), which leaves scope for further research into how other personalities may be susceptible to propaganda effects.


There are, of course, some limitations to this study. First, although convenience samples can be representative of populations (Mullinix et al., 2015), the sample here was

disproportionately young, educated and politically left-leaning, which limits the

generalisability of the findings. Also, a significant portion of the sample was recruited using



survey sharing forums, social media groups and websites which may include disingenuous or hurried responses. Although the sample size was greater than that required by the power analysis, a larger, more heterogeneous sample would have improved its representativeness and perhaps allowed for more stringent data cleaning measures.

Second, the emotionality of the messages in the stimulus could have been manipulated better.

As shown by the results of the manipulation check, the differences in emotionality between the anger and fear conditions registered by participants was not statistically significant.

Although this was not fatal to the experiment, since participants did still register a difference, it does detract from the validity and reliability of the experiment.

Third, the subject of the stimuli content - a “consortium” or “cartel” of “western governments in Europe” - is a rather abstract concept, unlike “the EU” for example. As such, the stimuli context was perhaps too intangible to garner feelings of in-group and out-group partisanship among the sample. The decision to use “western governments” as the subject of the stimuli was partially due to sample recruitment being restricted to the UK, Netherlands and

Germany, and there are few recognisable entities for which all three are now members.

Recruiting participants from a more discernible body – such as a single country or the EU – would improve the generalisability of the results and validity of the stimuli.

Fourth, there were also some limitations with the measures in this study. The cleavage operationalised here to measure affective polarisation (“those who trust/distrust the government”) is perhaps not manifest enough to elicit natural feelings of polarisation, especially compared to more obvious cleavages, such as Democrats and Republicans for example (e.g. Iyengar et al., 2019). Also, the number of zero scores (35 out of 200) and the low mean (M = 2.65) suggests that participants may have been confused by this measure, which may explain why no statistically significant effects were found on this dependent



variable. The message elaboration measurement was also flawed in that it did not have a sufficiently high Cronbach’s alpha, even after removing an item. This too may help explain the insignificant findings in the moderated mediation analysis. Finally, self-reporting is far from a perfectly valid and reliable measure of personality but, as mentioned, it was the best available method (Zarouali et al., 2020). Future research should ensure the validity of these measures are improved.

In conclusion, the results of this study paint a very nuanced picture of the effectiveness of personality-based micro-targeted foreign propaganda. It is argued here that personality- congruent propaganda messaging can be more effective in increasing political instability than incongruent messaging. However, the findings of this study suggest that this is only the case for specific message types (anger appeals) and on specific dimensions of political stability (trust in government). Therefore, it would be presumptuous to conclude that personality- congruent propaganda messaging has uniform effects on political stability across the board, as set forth by the research question. Nevertheless, by providing some evidence of the potency of micro-targeted propaganda, the findings of this study do little to allay fears over the threat that such techniques pose to political stability.


Aaker, J. L. (1999). The Malleable Self: The Role of Self-Expression in Persuasion. Journal of Marketing Research, 36(1), 45–57. https://doi.org/10.1177/002224379903600104 Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and interpreting interactions

(pp. xi, 212). Sage Publications, Inc.

Ali, F., Amorim, I. S., & Chamorro-premuzic, T. (2009). Empathy deficits and trait emotional intelligence in psychopathy and Machiavellianism. Personality and Individual Differences, 47(7), 758–762.



Asker, D., & Dinas, E. (2019). Thinking Fast and Furious: Emotional Intensity and Opinion Polarization in Online Media. Public Opinion Quarterly, 83(3), 487–509.


Bayer, J., Bitiukova, N., Bárd, P., Szakács, J., Alemanno, A., & Uszkiewicz, E. (2019).

Disinformation and propaganda – impact on the functioning of the rule of law in the EU and its Member States. European Parliament’s Committee on Civil Liberties, Justice and Home Affairs.

https://www.europarl.europa.eu/RegData/etudes/STUD/2019/608864/IPOL_STU(201 9)608864_EN.pdf

Becker, H. (1949). The Nature and Consequences of Black Propaganda. American Sociological Review, 14(2), 221–235. https://doi.org/10.2307/2086855 Benkler, Y., Faris, R., & Roberts, H. (2018). Network Propaganda: Manipulation,

Disinformation, and Radicalization in American Politics. New York, NY: Oxford University Press.

Brader, T. (2005). Striking a Responsive Chord: How Political Ads Motivate and Persuade Voters by Appealing to Emotions. American Journal of Political Science, 49(2), 388–

405. https://doi.org/10.1111/j.0092-5853.2005.00130.x

Brader, T. (2006). Campaigning for Hearts and Minds: How Emotional Appeals in Political Ads Work. Chicago, IL: University of Chicago Press.

https://press.uchicago.edu/ucp/books/book/chicago/C/bo3640346.html Bradshaw, S., & Howard, P. N. (2018). The global organization of social media

disinformation campaigns. Journal of International Affairs, 71(1.5), 23–32.

Cadwalladr, C. (2017, May 7). The great British Brexit robbery: How our democracy was hijacked. The Guardian. https://www.theguardian.com/technology/2017/may/07/the- great-british-brexit-robbery-hijacked-democracy



Citrin, J., & Stoker, L. (2018). Political Trust in a Cynical Age. Annual Review of Political Science, 21(1), 49–70. https://doi.org/10.1146/annurev-polisci-050316-092550 Cohen, J. (1988). Statistical Power Analysis for the Behavioral Sciences (2nd ed.).

Routledge. https://doi.org/10.4324/9780203771587

Confessore, N. (2018, April 4). Cambridge Analytica and Facebook: The Scandal and the Fallout So Far. The New York Times.

https://www.nytimes.com/2018/04/04/us/politics/cambridge-analytica-scandal- fallout.html

Cooper, C. A., Golden, L., & Socha, A. (2013). The Big Five Personality Factors and Mass Politics. Journal of Applied Social Psychology, 43(1), 68–82.


DeCoster, J., Gallucci, M., & Iselin, A.-M. R. (2011). Best Practices for Using Median Splits, Artificial Categorization, and their Continuous Alternatives. Journal of Experimental Psychopathology, 2(2), 197–209. https://doi.org/10.5127/jep.008310

Dobber, T., Metoui, N., Trilling, D., Helberger, N., & de Vreese, C. (2021). Do (Microtargeted) Deepfakes Have Real Effects on Political Attitudes? The International Journal of Press/Politics, 26(1), 69–91.


Dowding, K. M., & Kimber, R. (1983). The Meaning and Use of ‘Political Stability’*.

European Journal of Political Research, 11(3), 229–243.


Druckman, J. N., & Levendusky, M. S. (2019). What Do We Measure When We Measure Affective Polarization? Public Opinion Quarterly, 83(1), 114–122.




Ellul, J. (1973). Propaganda: The Formation of Men’s Attitudes. New York, NY: Vintage Books.

European Commission. (2020). Communication on the European Democracy Action Plan (COM(2020) 790 final).


Fairfield, P. (2018). Lords of mendacity. In C. G. Prado (Ed.), America’s post-truth phenomenon: When feelings and opinions trump facts and evidence (pp. 153–164).

Santa Barbara: Praeger.

Farrington, D. P., & Loeber, R. (2000). Some benefits of dichotomization in psychiatric and criminological research. Criminal Behaviour and Mental Health, 10(2), 100–122.


Frischlich, L., Schatto-Eckrodt, T., Boberg, S., & Wintterlin, F. (2021). Roots of Incivility:

How Personality, Media Use, and Online Experiences Shape Uncivil Participation.

Media and Communication, 9(1), 195–208.

Furnham, A., Richards, S. C., & Paulhus, D. L. (2013). The Dark Triad of Personality: A 10 Year Review. Social and Personality Psychology Compass, 7(3), 199–216.


Galais, C., & Rico, G. (2021). An unjustified bad reputation? The Dark Triad and support for populism. Electoral Studies, 72, 102357.


Gelders, D., & Ihlen, Ø. (2010). Government communication about potential policies: Public relations, propaganda or both? Public Relations Review, 36(1), 59–62.

Gelfert, A. (2018). Fake News: A Definition. Informal Logic, 38(1), 84–117.




Gerber, A., Huber, G., Doherty, D., Dowling, C., & Panagopoulos, C. (2013). Big Five Personality Traits and Responses to Persuasive Appeals: Results from Voter Turnout Experiments. Political Behavior, 35. https://doi.org/10.1007/s11109-012-9216-y Gilbert, D. (2022, March 4). Millions of Leftists Are Reposting Kremlin Misinformation by

Mistake. Vice. https://www.vice.com/en/article/wxdb5z/redfish-media-russia- propaganda-misinformation

Gotlieb, M. R., Scholl, R. M., Ridout, T. N., Goldstein, K. M., & Shah, D. V. (2017).

Cumulative and long-term campaign advertising effects on trust and talk.

International Journal of Public Opinion Research, 29(1), 1–22.


Gøtzsche-Astrup, O. (2021). Dark triad, partisanship and violent intentions in the United States. Personality and Individual Differences, 173, 110633.


Grassegger, H., & Krogerus, M. (2017, January 28). The Data That Turned the World Upside Down. Vice. https://www.vice.com/en/article/mg9vvn/how-our-likes-helped-trump- win

Groh, M., Sankaranarayanan, A., & Picard, R. (2022). Human Detection of Political Deepfakes across Transcripts, Audio, and Video. ArXiv E-Prints.


Haenschen, K., & Jennings, J. (2019). Mobilizing Millennial Voters with Targeted Internet Advertisements: A Field Experiment. Political Communication, 36(3), 357–375.


Hameleers, M., Powell, T. E., Van Der Meer, T. G. L. A., & Bos, L. (2020). A Picture Paints a Thousand Lies? The Effects and Mechanisms of Multimodal Disinformation and



Rebuttals Disseminated via Social Media. Political Communication, 37(2), 281–301.


Hart, W., Breeden, C. J., & Lambert, J. (2021). Exploring a vulnerable side to dark

personality: People with some dark triad features are gullible and show dysfunctional trusting. Personality and Individual Differences, 181, 111030.


Harwell, D., & Lerman, R. (2022, March 1). How Ukrainians have used social media to humiliate the Russians and rally the world. Washington Post.

https://www.washingtonpost.com/technology/2022/03/01/social-media-ukraine- russia/

Hayes, A. F. (2013). Introduction to mediation, moderation, and conditional process analysis: A regression-based approach. New York, NY: Guilford Press.

Hayes, A. F., & Preacher, K. J. (2014). Statistical mediation analysis with a multicategorical independent variable. British Journal of Mathematical and Statistical Psychology, 67(3), 451–470. https://doi.org/10.1111/bmsp.12028

Hernández, E., Anduiza, E., & Rico, G. (2021). Affective polarization and the salience of elections. Electoral Studies, 69, 102203.


Hobolt, S. B., Leeper, T. J., & Tilley, J. (2021). Divided by the Vote: Affective Polarization in the Wake of the Brexit Referendum. British Journal of Political Science, 51(4), 1476–1493. https://doi.org/10.1017/S0007123420000125

Iacobucci, D., Posavac, S., Kardes, F., Schneider, M., & Popovich, D. (2015a). Toward a more nuanced understanding of the statistical properties of a median split. Journal of Consumer Psychology, 25, 652–665.



Iacobucci, D., Posavac, S. S., Kardes, F. R., Schneider, M., & Popovich, D. (2015b). The Median Split: Robust, Refined, and Revived. Journal of Consumer Psychology, 25(4), 690–704. https://doi.org/10.2139/ssrn.2663369

Im, T., Cho, W., Porumbescu, G., & Park, J. (2014). Internet, Trust in Government, and Citizen Compliance. Journal of Public Administration Research and Theory, 24(3), 741–763. https://doi.org/10.1093/jopart/mus037

Iyengar, S., Lelkes, Y., Levendusky, M., Malhotra, N., & Westwood, S. J. (2019). The Origins and Consequences of Affective Polarization in the United States. Annual Review of Political Science, 22(1), 129–146. https://doi.org/10.1146/annurev-polisci- 051117-073034

Iyengar, S., Sood, G., & Lelkes, Y. (2012). Affect, Not Ideology: A Social Identity Perspective on Polarization. Public Opinion Quarterly, 76(3), 405–431.


Jakubowski, G. (2019). What’s Not to Like?: Social Media as Information Operations Force Multiplier. Joint Force Quarterly, 3, 8-17.

Johnson, R. (2021). Information warfare: Theory to practice. In The World Information War.

London, UK: Routledge.

Jonason, P. K., & Krause, L. (2013). The emotional deficits associated with the Dark Triad traits: Cognitive empathy, affective empathy, and alexithymia. Personality and Individual Differences, 55(5), 532–537. https://doi.org/10.1016/j.paid.2013.04.027 Jonason, P. K., & Webster, G. D. (2010). The dirty dozen: A concise measure of the dark

triad. Psychological Assessment, 22(2), 420–432. https://doi.org/10.1037/a0019265 Jones, D. N., & Paulhus, D. L. (2014). Introducing the Short Dark Triad (SD3): A Brief

Measure of Dark Personality Traits. Assessment, 21(1), 28–41.




Jowett, G., & O’Donnell, V. (2006). Propaganda and Persuasion. Washington, DC: SAGE.

Kim, Y., & Kim, Y. (2019). Incivility on Facebook and political polarization: The mediating role of seeking further comments and negative emotion. Computers in Human

Behavior, 99, 219–227. https://doi.org/10.1016/j.chb.2019.05.022

Klein, E., & Robison, J. (2020). Like, Post, and Distrust? How Social Media Use Affects Trust in Government. Political Communication, 37(1), 46–64.


Lau, R. R., Sigelman, L., Heldman, C., & Babbitt, P. (1999). The Effects of Negative Political Advertisements: A Meta-Analytic Assessment. The American Political Science Review, 93(4), 851–875. https://doi.org/10.2307/2586117

Lau, R. R., Sigelman, L., & Rovner, I. B. (2007). The Effects of Negative Political

Campaigns: A Meta-Analytic Reassessment. Journal of Politics, 69(4), 1176–1209.


Layman, G. C., Carsey, T. M., & Horowitz, J. M. (2006). Party Polarization in American Politics: Characteristics, Causes, and Consequences. Annual Review of Political Science, 9(1), 83–110. https://doi.org/10.1146/annurev.polisci.9.070204.105138 Lazarsfeld, P. F., & Merton, R. K. (1948). Mass communication, popular taste, and organized

social action. In L. Bryson (Ed.), The communication of ideas. New York, NY:


Linvill, D., & Warren, P. (2020). Troll Factories: Manufacturing Specialized Disinformation on Twitter. Political Communication, 37, 1–21.


Listhaug, O., & Ringdal, K. (2008). Trust in political institutions. In H. Ervasti & H. Ervasti (Eds.), Nordic Social Attitudes in a European Perspective. Cheltenham, UK: Edward Elgar Publishing.



Lühiste, K. (2006). Explaining trust in political institutions: Some illustrations from the Baltic states. Communist and Post-Communist Studies, 39(4), 475–496.


Lukito, J. (2020). Coordinating a Multi-Platform Disinformation Campaign: Internet Research Agency Activity on Three U.S. Social Media Platforms, 2015 to 2017.

Political Communication, 37(2), 238–255.


Lukito, J., Suk, J., Zhang, Y., Doroshenko, L., Kim, S. J., Su, M.-H., Xia, Y., Freelon, D., &

Wells, C. (2020). The Wolves in Sheep’s Clothing: How Russia’s Internet Research Agency Tweets Appeared in U.S. News as Vox Populi. The International Journal of Press/Politics, 25(2), 196–216. https://doi.org/10.1177/1940161219895215

MacKuen, M., Wolak, J., Keele, L., & Marcus, G. E. (2010). Civic Engagements: Resolute Partisanship or Reflective Deliberation. American Journal of Political Science, 54(2), 440–458. https://doi.org/10.1111/j.1540-5907.2010.00440.x

Markus, H., & Kunda, Z. (1986). Stability and malleability of the self-concept. Journal of Personality and Social Psychology, 51(4), 858–866. https://doi.org/10.1037/0022- 3514.51.4.858

Martinez, M. D., & Delegal, T. (1990). The irrelevance of negative campaigns to political trust: Experimental and survey results. Political Communication, 7(1), 25–40.


Matz, S. C., Kosinski, M., Nave, G., & Stillwell, D. J. (2017). Psychological targeting as an effective approach to digital mass persuasion. Proceedings of the National Academy of Sciences, 114(48), 12714–12719. https://doi.org/10.1073/pnas.1710966114 Mondak, J. J. (2010). Personality and the Foundations of Political Behavior. Cambridge,

UK: Cambridge University Press.



Mondak, J. J., & Halperin, K. D. (2008). A Framework for the Study of Personality and Political Behaviour. British Journal of Political Science, 38(2), 335–362.


Mondak, J. J., Hibbing, M. V., Canache, D., Seligson, M. A., & Anderson, M. R. (2010).

Personality and Civic Engagement: An Integrative Framework for the Study of Trait Effects on Political Behavior. American Political Science Review, 104(1), 85–110.


Moshagen, M., Hilbig, B., & Zettler, I. (2018). The Dark Core of Personality. Psychological Review, 125. https://doi.org/10.1037/rev0000111

Mueller, S. C. R. S. (2019). Report on the Investigation into Russian Interference in the 2016 Presidential Election (p. 448). US Department of Justice.


Mullinix, K. J., Leeper, T. J., Druckman, J. N., & Freese, J. (2015). The Generalizability of Survey Experiments. Journal of Experimental Political Science, 2(2), 109–138.


Nai, A. (2019). The Electoral Success of Angels and Demons: Big Five, Dark Triad, and Performance at the Ballot Box. Journal of Social and Political Psychology, 7(2), 830–

862. https://doi.org/10.5964/jspp.v7i2.918

Nai, A., & Maier, J. (2018). Perceived personality and campaign style of Hillary Clinton and Donald Trump. Personality and Individual Differences, 121, 80–83.


Nai, A., & Maier, J. (2021a). Is Negative Campaigning a Matter of Taste? Political Attacks, Incivility, and the Moderating Role of Individual Differences. American Politics Research, 49(3), 269–281. https://doi.org/10.1177/1532673X20965548




Related subjects :