Does Identity Matter? Ethnicity, Religion and Effects of Negative Campaigning on the Perception of Candidates

(1)

SOCIAL IDENTIFICATION & NEGATIVE CAMPAIGNING 1

Does Identity Matter?

Ethnicity, Religion and Effects of Negative Campaigning on the

Perception of Candidates

Master’s Thesis

Kelechi Okechukwu Amakoh 12087424

Graduate School of Communication, University of Amsterdam

Erasmus Mundus Joint Master’s degree in Journalism, Media and Globalisation Dr Alessandro Nai

May 29, 2020 7,853 (pp. 6 – 33)

(2)

Table of Contents List of Figures ... 3 List of Tables ... 3 List of Appendices ... 3 Acknowledgements ...4 Abstract ...5

Ethnicity, Religion and Effects of Negative Campaigning on the Perception of Candidates...6

Theoretical Framework ...9

The Concept of Negative Campaigning ... 9

The In-group and Out-group Dynamics ... 13

Method ... 18 Design ... 18 Stimuli ... 21 Participants ... 23 Measures ... 24 Manipulation Check ... 24 Dependent Variable ... 25 Results ... 27

Main Effects of Negative Campaigning ... 27

The Moderating Effects of Ethnicity and Religion ... 28

Discussion and Conclusion ... 30

References ... 34

Moderator Analyses ... 44

Appendix... 48

(3)

List of Figures

Figure 1: The two lenses of Negative Campaigning Figure 2: Conceptual Model Figure 3: Participants’ perception of Ahmed Balarabe and Babatunde Ademola Figure 4: Participants’ propensity to vote for Ahmed Balarabe and Babatunde Ademola

Figure 5: Policy-based Negative Campaign Message

Figure 6: Policy-based Negative Campaign Message

List of Tables

Table 1: Negative Campaign Messages and In-group and Out-group (Sponsor Candidate, Ahmed Balarabe)

Table 2: Moderator Analysis: Negative Campaign Messages and In-group and Out-group (Sponsor Candidate, Ahmed Balarabe) – with covariates

Table 3: Moderator Analysis: Negative Campaign Messages and In-group and Out-group (Target Candidate, Babatunde Ademola)

Table 4: Moderator Analysis: Negative Campaign Messages and In-group and Out-group (Target Candidate, Babatunde Ademola) – with covariates

List of Appendices

Appendix A: Profiles of hypothetical Candidates

Appendix B: Stimuli

Appendix C: Selected Sections of Survey Questions Appendix D: Example of a real negative campaign message in Nigeria

Appendix E: Front page of The Punch showing negative campaign message

(4)

Acknowledgements

I thank the Almighty God who made everything beautiful in itself and in its time. His grace was sufficient for me to complete this two-year master’s programme. It was a worthwhile experience.

To my supervisor, Dr.Alessandro Nai, thank you for your genuine interest in my topic and for guiding me through the process of completing this work. Thank you for your critical reviews and concrete suggestions. It is a testament of your enthusiasm and interest for the work to proceed on the right path. In all, the iterative process taught me persistence.

My appreciation goes to Drs. Andreas Schuck and Penny Sheets who provided guidance and suggestions on my preliminary ideas during the thesis preparatory stage. I thank the faculty members in the department: Drs. Katjana Gattermann, Mark Boukes, Damian Trilling, Knut de Swert and Prof. Claes de Vreese for their dynamic ways of teaching and imparting

knowledge. You do a great job! Also, I thank Nhu Trang for her unique administrative roles. How can I forget where it all started? The experience at Aarhus University was invaluable. Thank you to Bettina Andersen, Inger Munk, Morten Brænder and Kirsten Andersen for providing leadership and solid administrative cover for the programme. To Drs. Teke Ngomba and Henrik Bødker, your teaching combination remains the best I have ever seen – so dynamic! Thank you to Drs. Mette Skak, Tonny Knudsen, Kristina Simonsen, Ann-Kristin Kölln and Mathias Tromborg for their unique teaching roles during the programme. Also, my appreciation to members of faculty at Danmarks Medie- og Journalisthøjskole (DMJX): Drs. Roger Buch, Flemming Svith and Jeper Gaarskjær. Keep on contributing to the knowledge economy.

There is a popular Igbo proverb which says that “it takes a village to raise a child!” I am indebted to my family, friends and loved ones. To my friends who constantly sought to know how I was fairing in a new environment, Ese o! To my relatives who were unflinching in their moral support, Nagode! To my immediate family, thank you for being my support system. I thank God for placing you in my life for such a time as this! Imela, Chukwu gozie gi! Kelechi Okechukwu Amakoh

Amsterdam, The Netherlands May 24, 2020

(5)

Abstract

Across the world, politicians do not relent in ‘going negative’. Studies suggest that party affiliation as a form of voters’ social identification determines the effectiveness of negative campaigning on the perception of candidates. The literature on negative campaigning is yet to provide an understanding of the effect of negative campaigns on the perception of the

candidacy of sponsor and target politicians in states where other forms of self-identification are salient. Therefore, this study focuses on two forms of social identification salient in multi-ethnic countries: religion and multi-ethnicity. The overarching question guiding this study is to what extent do ethnicity and religiosity determine the effectiveness of negative campaigns on the perception of candidates? I start from the assumption that in-group / out-group dynamics based on religion and ethnicity determine the effectiveness of negative campaigning on the perception of politicians. To test this assumption, I conducted an online survey-embedded experiment (N=241) in Nigeria between March 26 and April 16, 2020. The stimuli for the study were hypothetical candidates and campaign messages designed to reflect the realities in the Nigerian political sphere. The findings indicate that compared to a policy-based negative message, a personal-based negative message by a sponsor candidate is more effective in reducing the perception of the target candidate when the voters are in-group members with the target candidate. This study is a conversation starter for research on social identification and negative campaigning effectiveness. Also, it sets a research agenda for future research.

(6)

Ethnicity, Religion and Effects of Negative Campaigning on the Perception of Candidates

…at the root of all politics is the universal language of conflict” (Schattschneider, 1960, p. 2)

Studies suggest that the social identification of voters determines the effect of negative campaigning on the perception of politicians. In the build-up to the Election Day, voters who identify with the sponsor politician are more likely to accept a negative campaign message (Budesheim et al., 1996; Craig & Cossette, 2017). In addition, there is a tendency for voters to dislike those who do not share their beliefs, biases or support the same political party (Iyengar et al., 2019). There is a heightened focus in negative campaigning literature on party affiliation as a form of social identification for voters. On the other hand, other forms of social identification matter more for citizens of some countries. For example, citizens of multi-ethnic countries (e.g. Nigeria, Ghana) mostly identify themselves based on ethno-regional lines and religious beliefs (Langer & Ukiwo, 2007). However, the literature on negative campaigning is yet to provide an understanding of the effect of negative campaigning on the perception of the candidacy of both sponsor and target politicians in states where other forms of self-identification are salient. This study stretches the boundary of the literature by looking at how two forms of social identification - religion and ethnicity determine the effectiveness of negative campaigning on the perception of politicians. This study expects that a personal-based negative message is not more effective than a policy-personal-based negative message in

reducing the perception of the target candidate when the voters are in-group members with the target candidate.

From a general perspective, negative campaigning is prominent since politicians always go negative during election campaigns. The sole aim of politicians going negative is to sell their candidacy to both partisan and swing voters (Robinson & Torvik, 2009). In most cases, election campaigns promote both the candidate and political parties’ merit – strengths,

(7)

achievements and plans for governance and the perceived weakness of their opponents – failures, antecedents, plans for governance (Lau & Rovner, 2009). Whether online or offline, most incumbents prefer using a positive campaign appeal as the primary campaign strategy, while challenging candidates and political parties are most likely to prefer a higher use of negative campaign appeal/mudslinging of the incumbent and the ruling political party. These dynamics of negative campaigning has become a prominent research focus with an increased number of examples from the United States in the last three decades. The increase in US examples is due to the importance of election campaigns in the country and its hegemony in the field of political science and related academic fields studying the dynamics of negative campaigning (Nai & Walter, 2015; Walter & van der Eijk, 2019). On the other hand, there is little scientific knowledge of the dynamics of negative campaigning outside the United States. In other words, negative campaigning research with non-US examples is still in its infancy. Hence, the deliberate efforts by researchers to provide non-US examples to better enhance the perspectives of negative campaigning (Nai, 2018; Boshoff, 2014; Ridout & Walter, 2015). This gap in the literature also limits scientific knowledge on the role of other forms of social identification in negative campaigning.

The in-group and out-group dynamics is important across all forms of social identification (e.g. political party affiliations, religion, ethnicity, nationalities, and gender). A group refers to an assembly of individuals who share common characteristics within any form of social identification. (Tajfel, 1974). Negative campaigning literature on intergroup

dynamics has extensively focused on political party affiliations as a form of social

identification. This study uniquely contributes to literature as it uses the intergroup dynamics to understand how ethnicity and religion influences the effects of negative campaigning on the perception of politicians. Intergroup dynamics determine whether an attack is accepted or rejected as well as when negativity backfires on the sponsoring politician. Voters are

(8)

especially more likely to support an incumbent who has done well if they are members of the same ethnic group (Adida et al., 2017). This kind of choice is reinforced by the biases and prejudices of voters, the proximity of voters of the same ethnic group and the choice-conflict present when politicians go negative during election campaigns (Battle & Seely, 2010; Houston et. al, 1999; March & Graham, 2015). In addition, the media aggravate the choice-conflict for voters. The choice-conflict lens used in reporting politicians, who go negative, activates the “burglar alarm” in the society (Ogwezzy-Ndisika et. al, 2019; Schuck et al., 2016; Zaller, 2003). In sum, conflict is the common language of politics (Schattscheneider 1960).

The overarching question guiding this study is to what extent do religiosity and ethnicity determine the effectiveness of negative campaigns on the perception of candidates? To answer this question, this study conducts a survey-embedded online experiment among a convenience sample of the voting population in Nigeria. The choice of Nigeria is suitable for understanding the intergroup dynamics based on ethnicity and religion. Nigeria is one of the African countries with extreme ethnic, regional and religion diversity (Osaghae, 2003; Posner, 2004; Keller, 2014). Nigerian politicians constantly play the ethnic and religious cards during election campaigns (Opeibi, 2006). A recent example is relevant here. In July 2015, the newly elected president of Nigeria, Muhammadu Buhari reinforced the role of ethnicity in Nigerian politics while addressing an audience at the United States Institute of Peace. A member of the diverse audience inquired about his plans to solve the challenges facing the Niger-Delta region of Nigeria. Buhari responded thus: “I hope you have a copy of the election results. The constituents, for example, [that] gave me 97% [of the vote] cannot in all honesty be treated on some issues with constituencies that gave me 5%. I think these are political realities”. (Sahara Reporters, 2015). According to political analysts, Buhari’s reply was “jarring” (Joseph, 2015). Furthermore, the response strengthened the saliency attached to ethnicity by Nigerian

(9)

ethnicity and religion in determining the effects of negative campaigning on the perception of negative campaigning. Findings of this study provides a nuanced understanding of when negative campaigning matters as well as set a research agenda for future research.

Theoretical Framework

The theoretical framework for this study focuses on two main concepts: Negative campaigning and the in-group / out-group dynamics. In this section, I discuss the concepts' relevance and the key expectations of this study based on existing literature.

The Concept of Negative Campaigning

Negative campaigning is an age-long concept. It is as old as the beginning of elections. As early as 64 BC, Marcus Cicero who was running for the office of the Consul, the highest office in Ancient Roman society, was advised by their elder brother, Quintus to “smear these men (opponents) at every opportunity with the crimes, sexual scandals, and corruption they have brought on themselves” (Cicero & Carville, 2012, p. 25). Overall, ancient campaigns “were pretty personal” (Freeman, 2012, p. A27). Today’s politicians take a cue from Quintus’ advice during election campaigns.

The ubiquitous use of negative campaign has generated interests among

academics in the last three decades. These include the conceptualisation and effectiveness of negative campaigning on political behaviour. For the conceptualisation of negative

campaigning, there are polarised viewpoints (Johnson-Cartee & Copeland, 1991). Due to this polarisation, “negativism remains in the eye of the beholder” (Lau & Rovner, 2009, p. 290). A school of thought sees negative campaigning as three-pronged: advocacy, attack and

comparative/contrast adverts (Jamieson et al., 2000). On the other hand, scholars suggest that conceptualising negative campaigning stems from two lenses: evaluative and directional (Figure 1). From an evaluative lens, negative campaigning is every unfair, manipulative,

(10)

illegitimate and lying against the opponent. In other words, scholars see this form as going “dirty” or playing “poison” politics (Jamieson, 1992; Kamber, 1997). This lens is often

criticised as being subjective, unreliable and complicated to decipher (Lau & Rovner, 2009).

Figure 1

The two-lenses of Negative Campaigning

The directional perspective of negative campaigning refers to any campaign not positive in approach. This perspective suggests that it is negative campaigning when there are statements made by a politician against the opposition and their policies (Lau & Pomper, 2001). There is a general acceptance of this perspective of negative campaigning and its function among scholars in the fields of political science and communication science (Nai & Walter, 2015). Based on this general acceptance, I build upon this perspective of studying negative campaigning.

There are two divides in the directional perspective of negative campaigning. The first side of the divide is the sponsoring politician who makes the attack. The other side is

Negative Campaigning

Evaluative Lens

“Dirty”, “Illegitimate”, “Manipulative” politics

Directional Lens

Policy-based

_{Focuses on the policy stance of the opposition}

Personal-based

Focuses on the personality of the opposition candidate

(11)

the recipient of the attack. Otherwise referred to as the target of the attack. Therefore, it is important to assess both the sponsor and target of an attack when studying the effectiveness of negative campaigning (Lau et al., 2007). In their seminal work, Ansolabehere and colleagues suggest that the demobilisation of voters on Election Day is due to voters’ “blanket negativity towards both candidates”, i.e. sponsor and target of an attack. (Ansolabehere et al., 1994, p. 834). Here, blanket negativity refers to the general decrease of voters’ support for both candidates. On the other hand, scholars further suggest that the target of the attack records lower support due to the attack from the sponsor politician (Basil et al., 1991). However, some attacks may spell doom for the sponsoring politician. When this happens, there is a backlash effect. This effect leads to a negative feeling and perception of the sponsor of the attack (Carraro & Castelli, 2010; Garramone, 1984; Roese & Sande, 1993). In a

meta-analysis of studies on the effects of negative political campaigning by Lau and colleagues, 33 out of the 40 studies that measured backlash effect indicated that sponsors of negative

advertisement experienced a backlash effect for going negative. (Lau et al., 2007). Scholars suggest that a backlash effect takes place when the target of the attack already receives more support from voters (Merritt, 1984).

It is necessary to consider the focus of an attack. The focus here refers to the type of attacks hurled at a targeted politician by a sponsor politician and can be either personal-based (trait-based) or policy-based (issue-based) attacks (Brooks & Geer, 2007; Hopmann et al., 2018). When an attack is personal-based, the sponsoring politician focuses on the personality (character, personal life, appearance) of the opposition candidate. A policy-based attack aims at criticising the policy stance (plans for sectors of the economy,

antecedents, of the target politician (Geer, 2006; Auter & Fine, 2016). Therefore, I conceptualise negative campaigning as every sponsored attack by a politician against the personality (character, antecedents, personal life, appearance, accomplishments, and

(12)

educational background) or policy stance (such as economy, education, job creation, social amenities, antecedents as an incumbent) of a targeted politician.

There are major risks involved when politicians go negative. Negativity tends to hurt all candidates involved (Fridkin & Kenney, 2004). Whether it is a policy-based or

personal-based attack, there is a general tendency for a decrease in the perception of the candidates by voters. When politicians go negative, voters are likely to have blanket negativity towards the candidates who go negative (Ansolabehere et al., 1994). When this happens, there are unintended and intended effects for both candidates. An unintended effect of negative campaigning is detrimental to the candidacy of the sponsor politician. The backlash effect is an unintended consequence for the sponsor politician (Garramone, 1984; Jasperson & Fan, 2002; Roese & Sande, 1993). When this happens, voters are more likely to have a negative feeling towards the sponsor of the negative message than the target candidate. A strong negative campaign message against the target is more likely to elicit a backlash effect for the sponsor candidate (Garramone, 1984). Sponsor candidates who make use of personal-based attack messages are more likely to experience a backlash effect (Min, 2004). Based on previous literature, I hypothesize that:

Hypothesis I (H1): In comparison to a personal-based negative message, a policy-based

negative message will cause a lesser backlash effect for the sponsor candidate.

An intended effect of a negative campaign message is to deplete the support base of an opposing candidate. Politicians go negative as Election Day approaches with the sole

intention of edging out rival candidates. Sponsor politicians target their opponent with the aim of reducing the perception of voters towards the target candidate (Jasperson & Fan, 2002; Damore, 2002). Previous literature have examined the impact of both personal-based and policy-based negative campaign messages on candidates (Fridkin & Kenney, 2004). A

(13)

a voter’s perception of the candidate than a policy-based negative campaign message (Homer & Batra, 1994). Based on these literature, I hypothesize that:

Hypothesis (H2): In comparison to a policy-based negative message, a personal-based

negative message is more effective in reducing voters’ perception of the target candidate.

The second concept relevant to this study is the in-group /out-group dynamics. This in determining the acceptance of an attack by the in-group members of the sponsoring politician or the rejection of an attack by the in-group members of the targeted politician.

The In-group and Out-group Dynamics

The in-group and out-group dynamics is important across all forms of social identification (e.g. political party, religion, ethnicity, nationalities, and gender). It is a relevant way of understanding how forms of social identification influence political behaviour. In most cases, the discourse of in-group and out-group dynamics in some literature has been restricted to one form of social identification – political party affiliations (Budesheim et al., 1996; Craig & Cossette, 2017). Thereby limiting the extent to which in-group and out-group dynamics influence on negative campaigning effects. In multi-ethnic states (e.g. Nigeria, Ghana), ethnicity and religious affiliations are two most forms of self-identification by citizens (Langer & Ukiwo, 2007). When there is a choice-conflict situation, self-identification

becomes active and influences how voters’ decide and perceive the candidates. Therefore, this study looks at the role of ethnicity, religion and the effects of negative campaigning on the perception of politicians.

Literature on intergroup dynamics refers to the negative prejudices meted by members of an in-group to out-group members. These prejudices are either deliberate or otherwise (March & Graham, 2015). Negative political campaign rhetoric strengthens prejudices of people against out-group members, thereby leading to an affective polarization

(14)

of the society (Iyengar et al., 2012). The way citizens process the information received through negative political rhetoric also deepens polarization in the society. In addition, citizens’ prior attitude affects how they process information about a politician, thereby reinforcing their prejudices (Adida, et.al, 2017;Taber & Lodge, 2006). People are more willing to accept information that confirms their prior beliefs and attitudes (Taber & Lodge, 2006). It is easier to accept any information that reinforces what citizens already know than otherwise. On the other hand, they are most likely to refute new information that negates their preconceived ideas and notions. An individual only accepts new information when such information reaffirms the social identity and prejudices of an individual (Adida, et. al, 2017). It is apparent that societies where religion and ethnicity categories are prominent means of social identity, individuals will be motivated to reason and process information along religious and ethnic lines.

Selective information processing is due to the trust and support that exists within members of an in-group, thereby influencing how they vote on Election Day (Adida, 2015; Robinson, 2020). Findings in the literature suggest that in the build-up to the Election Day, in-group members are more likely to accept a negative campaign message if it is from an in-group politician (Budesheim et al., 1996). In addition, in-group members are more likely to support a politician from their group who goes on an attack spree against the out-group

politician. This act reinforces their perceived superiority in the electoral contest and renews their support for the sponsoring politician. The in-group members will continually have a positive perception of the sponsor candidate. For the target politician, no amount of attacks will dwindle the support of his candidacy by in-group members. The trust and support provided by in-group members of a targeted politician are constant (Adida, 2015). This implies the tenacity to keep trusting and providing unflinching support for politicians (both sponsor and target) by their in-group members. In addition, it helps in reinforcing citizens’

(15)

beliefs and biases and is unlikely to change when there is a piece of negative information about their preferred politician.

Voters are more likely to reward the good performance of an incumbent if they are members of the same ethnic group as well as punish bad performance if the incumbent is a non-coethnics (Adida et. al., 2017). Besides bad performance, negative information about an incumbent can be a reason for voters who are non-coethnics to withdraw their support. While voters of the same ethnic group with the incumbent will always favour information that favours the incumbent. In addition, scholars suggest that this is peculiar to states with ethnicity as a form of social identity. Adida and colleagues’ findings align partly with previous finding in literature about the importance on the role of ethnicity in determining voting patterns (Battle & Seely, 2010). Similarly, the 1992 and 1996 elections in Ghana witnessed massive support for the candidacy of Jerry Rawlings by the Ewe ethnic group (93% and 95% support respectively). This unprecedented support witnessed in Ghanaian politics was because of Rawlings claim of being a member of the Ewe group because his mother hails from there. In addition, this strategic support by the Ewes increased their prominence in Ghanaian politics (Scarritt & Mozaffar, 1999).

Ethnicity and religion are salient forms of social identification in African countries (Koter, 2019; Langer & Ukiwo, 2007). Politicians and their campaign strategists play these cards to attract the trust and support of their in-group members. For example, in Nigeria, politicians constantly use ethnic and religious cards during election campaigns across the country (Okolie & Okoedion, 2019; Opeibi, 2006). Ethnic politics has been a form of political mobilization in Nigeria since the 1990s (Osaghae, 2003). In addition, religion has played a prominent role in politics. As succinctly put by Matthew Hassan Kukah, the current Bishop of the Roman Catholic Diocese of Sokoto state, Nigeria, “religion is a vital tool for [the] legitimisation of a leader. The only thing that politicians want is that they are interested

(16)

in a religious leader standing as a member of the praying wing of their [political] party” (Channels Television, 2018). A recent example is relevant here. In the build-up to the 2015 presidential elections, Nigerians woke up to a rude shock. The front pages of two national newspapers: The Punch and The Sun had election campaign adverts sponsored by one of the state governors of the then ruling political party, Peoples Democratic Party (PDP), Ayodele Fayose. The advert was against the presidential candidate of the main opposition party, All Progressives Congress (APC), Muhammadu Buhari. The advert’s use of a Biblical verse (religion) and the term “Northern Presidency” (ethnicity) to justify the eligibility of a candidate caused a ruckus and elicited divergent reactions within the polity. While some political analysts suggested that the advert was an avenue for the APC to provide a “smart counter-narrative”, others noted that the advertisement was in extremely "bad taste” (Adeseri & Nnochiri, 2015; Adibe, 2015; Ezeamalu, 2015; Unegbu, 2015). The schism in public opinion about the advert and the wanton negativity during the campaigns, Nigeria’s 2015 presidential election campaign was “the most challenging in the history of Nigeria” (Ogwezzy-Ndisika et al., 2019, p.239).

This study focuses on how the in-group and out-group dynamics based on religious beliefs and ethnicity of voters determines the effectiveness of negative campaigning on the perception of candidates (Figure 2). When politicians go negative, the intergroup dynamics becomes active. A negative message by the sponsor candidate activates the trust and support for the target candidate by members of the target’s in-group (Adida, 2015;

Robinson, 2020). The negative message causes the in-group members of the target to feel that their identity as a group is under threat. On the other hand, the in-group members of the sponsor candidate are more likely to accept the negative message against the target candidate. The acceptance of the message reinforces the in-group member’s beliefs, biases and

(17)

in-SOCIAL IDENTIFICATION & NEGATIVE CAMPAIGNING 17

group member of the sponsor, neither policy-based nor personal-based negative messages will lead to a backlash effect for the sponsor candidate.

Based on these previous literature, I hypothesize the following:

Hypothesis 3 (H3): Neither a policy-based nor a personal based negative message will cause

a backlash effect for the sponsor candidate if the voters are in-group members of the sponsor candidate.

Hypothesis 4 (H4): In comparison to a policy-based negative message, a personal-based

negative message is not more effective in reducing the perception of the target candidate when the voters are in-group members with the target candidate.

Figure 2

(18)

Method

Design

To test the hypotheses of this study, I conducted an online survey-embedded experiment on a convenience sample of 241 participants, between Thursday, March 26 and Thursday, April 16, 2020. I adopted a two-pronged method of recruiting participants for the study. First, I shared recruitment messages on four WhatsApp groups consisting of those who obtained at least a degree from the University of Lagos, Nigeria. The groups serve as a virtual meeting hub for alumni members to relive memories and contribute positively to national and international discourse. As a member of these WhatsApp groups, I was able to share the survey link, respond to inquiries, provide more information on the study, note feedback from participants and send reminders for more participation from group members. Second, I explored a snowballing method of recruiting more participants for the study. I encouraged respondents to share the anonymous survey link within their circle of influence (friends, family, colleagues and acquaintances). There were no incentives for participants.

Through a randomisation process, participants saw either a policy-based or a personal-based negative campaign message. I created the messages to reflect a typical

Nigerian political campaign rhetoric (Gaines et al., 2007). I used two hypothetical candidates in creating the messages - “Ahmed Balarabe” represented the current ruling political party, All Progressives Congress (APC) while “Babatunde Ademola” represented the current main opposition political party in Nigeria, Peoples Democratic Party (PDP). I created their profiles to fit the typical political trajectory of a Nigerian politician and aid the study of the intergroup dynamics. Ahmed Balarabe, a Muslim who hails from the Fulani ethnic group (a group in the Northern part of Nigeria), sponsored both the policy-based and personal-based attacks in the study. While Babatunde Ademola, a Christian from the Yoruba ethnic group (a group in the Southern part of Nigeria) was the target of both attacks hurled by Ahmed Balarabe.

(19)

To ascertain the in-group and out-group dynamics, I grouped participants based on the similarity of their ethnic group and religious belief with that of the candidates. In Nigeria, there are 374 ethnic groups, with three groups forming the majority (Hausa/Fulani, Yoruba and Igbo). The multiplicity of ethnic groups and its saliency in African politics is because of the colonisation of Africa (Ekeh, 1975; Mustapha, 2006). Until date, there is the North-South dichotomy when it comes to Nigerian politics. Hence, ethnic groups within the North are likely to come together as an in-group if the politician contesting is from any ethnic group in the North and vice-versa. The formation of this in-group arises from several political horse-trading, negotiations and compromise. This study adopts the popular dichotomous view of ethnicity in Nigeria: North and South (Mustapha, 2006) in classifying the participants into in-group and out-in-group. This is still a relevant dynamics in Nigerian politics today. All the ethnic groups from the Southern states of Nigeria formed the South, while all the ethnic groups from the Northern states formed the North. This was necessary to reflect the reality of the ethnic politics obtainable in Nigeria as of today. Hence, to account for the in-group and out-group dynamics, every participant who identified with a Southern ethnic group and Christianity as religion I assigned them as an in-group member with the target candidate, Babatunde Ademola and out-group member of the sponsor candidate, Ahmed Balarabe. For participants who identified with Islam as a religion and any Northern ethnic group, I assigned them as in-group members with the sponsor candidate, Balarabe and the out-group members of the target candidate, Ademola.

With an experimental setup, I was able to empirically test how social identities moderate the effect of negative campaigns (policy-based and personal-based attacks) on the perception and support of both politicians (Sponsor / Target). Second, this method is cost-effective. Participants only needed a working internet-enabled computer device and time to answer the survey. Third, using an online survey experiment method bridged the distance

(20)

between my present location (The Netherlands) and the participants in Nigeria (Krupnikov & Findley, 2018). Fourth, conducting an online survey helped to reduce the social desirability biases that could occur from asking sensitive information about voting behaviour, ethnic and religious beliefs in an interviewer mediated survey (Kreuter et al., 2008; du Toit, 2016). Critics of survey experiment argue that the real-world political stimuli presented by

researchers to subjects in a survey experiment do not occur all at once in real-world politics (Gaines et al., 2007). The three limitations of the survey experiment method include modesty of treatment, scale, and measurement (Sniderman, 2018). However, this method is relevant and has constantly evolved in the fields of media and communication studies, public opinion and political science (since the 80s and 90s respectively) (Gamson & Modigliani, 1989; Sniderman & Piazza, 1993). Therefore, an experimental setup is advantageous for this study and outweighs the critiques.

The questionnaire was divided into five sections. Participants answered basic

demographic questions on gender, age, level of education, ethnic group and religious beliefs. The second part of the questionnaire had questions aimed at ascertaining the political interest, opinion on voting and three questions on Nigerian politics. Before being exposed to a

condition (a campaign message from a hypothetical politician), participants read a paragraph on how campaigns in Nigeria are regulated by the Independent National Electoral

Commission (INEC) with the following instruction “To demonstrate that you have read this instruction, please select only "other" and type "candidate-centered campaign" in the text box. What type of campaign activities do candidates engage in (check all that applies)?” This attention check was aimed at seeing the extent to which participants paid attention to the instruction. After the exposure to a condition, participants were presented with a manipulation check. The manipulation check was to ensure that the conditions performed as intended in the study (Reinard, 2006). Contrary to arguments that these checks are a threat to a study’s

(21)

validity, studies show that there is no evidence to back up such arguments (Kung et al., 2018). Respondents were asked about how they felt after reading the condition, their perception and propensity to vote for the hypothetical sponsor and target politicians. Finally, the participants were debriefed on the reasons for the study and their exposure to hypothetical candidates and situations.

Stimuli

Participants saw either a policy-based or a personal-based attack messages. These hypothetical campaign messages are the independent variables for this study. Based on existing literature, I operationalize negative campaigning focus to be personal-based or policy-based (Brooks & Geer, 2007; Hopmann et. al, 2018). A personal-based attack by a sponsor candidate focuses on the personality of the opposition candidate, while a policy-based attack criticises the policy stance of the target politician (Auter & Fine, 2016; Geer, 2006).

I created the two campaign messages based on the campaign messages of Nigerian politicians during elections. For the policy-based campaign message of Ahmed Balarabe, the focus was on lifting “100 million Nigerians from poverty through agriculture.” The attack element in the message was the comparison of Balarabe’s plan to alleviate poverty through agriculture and Ademola’s use of stomach infrastructure. According to Balarabe, adopting stomach infrastructure as a means of poverty alleviation is “a waste of the nation’s scarce resources”. The term “Stomach infrastructure” became salient in Nigerian politics during the 2014 governorship election campaign in Ekiti State. The term refers to the hand out of state funds for short-term benefits. This is at the expense of implementing fiscal policies for a long-lasting impact on the economy. This was a major winning campaign tool used by the

opposition political party in Ekiti state, Peoples Democratic Party (PDP) against the then incumbent Governor, Dr Kayode Fayemi (Jeyifo, 2014; Omilusi, 2019). Since then, stomach

(22)

infrastructure has remained a relevant tool used by Nigerian politicians to campaign and win the votes of the citizens. The saliency and relevance of the concept of stomach infrastructure, the extent of poverty and the role of agriculture in the country informed the creation of the policy-based attack campaign. This was necessary to consider because a negative campaign base on a non-salient or irrelevant topic will not gain traction among voters (Min, 2004; Craig & Cossette, 2017).

The personal-based campaign message was modelled according to the negative campaigning witnessed during the primaries of 2019 Lagos Governorship election. The then incumbent Governor, Mr. Akinwunmi Ambode of All Progressives Congress (APC) was fighting' a political battle to get his party’s nomination for a second term. However, his chances were bleak and there was a stiff opponent, Mr. Babajide Sanwo-olu, who had the ‘blessings’ and support of the party leaders in the State. During a press conference by Ambode, he went negative accusing Sanwo-olu of having been “arrested for spending fake dollars in a night club in America”. In the same vein, the personal-based attack I created highlighted Balarabe accusing Ademola of spending fake pounds in the United Kingdom and lacked the integrity to lead the country. I embedded these campaign messages in the interfaces of both Twitter, a popular microblogging website and WhatsApp, a popular messaging

platform. I did this to control for how participants perceive the campaign messages.

Participants were primed to imagine that there is a sudden call for a new presidential election in Nigeria scheduled to hold on May 29, 2020 with the two candidates as the leading candidates before reading the candidates’ profiles and the campaign message. The priming of participants was based on previous studies conducted in a non-election year (Craig et al., 2014). After the priming, participants had to read their profiles for at least 30 seconds before exposure to either of the campaign messages.

(23)

Participants

Two hundred and forty-one (241) participants completed the survey-experiment within the data collection period. After removing those who failed the attention check, the final analysis used responses from 144 participants. The participants were all Nigerians between the ages of 18 and 60 years (M=29.58, SD=11.14). There were more males (56. 9%) than females (41%) in the sample, while three participants “prefer not to say” (2.1%). In addition, most participants in the survey possess a Bachelor’s Degree (39.6 %).

In terms of ethnicity, there were more participants who identified with Southern ethnic groups (Yoruba, Igbo, Anang, Efik, Esan, Ibibio, Ijaw, Ikwerre, Isoko, Ugep, Ukwuani, Yagba, Urhobo) (93.1%) than the Northern ethnic groups (Fulani, Hausa, Igala, Idoma, Tiv, Yagba, Ron, Kache, Jju) (6.9%). In addition, most participants identified with Christianity as their religion (85.4%). Other religions represented in the sample include Islam (13.2 %), African Traditional Religion (0.7%) and “Others” (0.7%). In terms of the in-group and out-group dynamics, there was a skewed representation. The target candidate, Babatunde Ademola had more in-group members (84%) than the sponsor candidate, Balarabe Ahmed (13.9%). While the sponsor candidate, Balarabe Ahmed had more out-group members (84%) than the target candidate, Babatunde Ademola (13.9%). The skewedness in the in-group and out-group variables are taken into account in the analysis section of this study. In addition, most participants (42.4 %) had a “moderate amount” level of interest in politics on a 5-point scale (1= “None at all” and 5 = “A great deal” M=3.27 SD=1.10).

(24)

Measures

Manipulation Check

To ensure the both negative campaign message was perceived as intended, I included a manipulation check section (Reinard, 2006). Participants indicated the extent to which they agree or disagree that the campaign message was harsh, negative, funny, positive, emotional, credible, populist and informative. This was measured on a 7-point Likert scale (1=“Strongly Disagree” and 7=“Strongly Agree”). To check what they thought about the conditions, I conducted Independent samples t-tests. Participants exposed to the conditions for personal-based attacks perceived it as significantly more negative (M = 5.99, SD = 1.28) than those exposed to the policy-based attacks (M = 5.22, SD = 1.64, t (137) = 3.10, p = .002). The tests showed that participants clearly agreed that the campaign messages were both negative. Participants exposed to the personal-based attack had a higher perception of the campaign message as harsh (M =5.24, SD = 1.68) than those exposed to the policy-based message (M = 5.05 SD = 1.70). However, it was not statistically significant (t (133) =.67, p = .503).

Participants exposed to both messages neither agreed nor disagreed that the message was funny. Those who saw the policy attack ranked higher (M = 3.94 SD = 1.87) than the personal-based group (M = 3.17, SD = 1.80), but it was not statistically significant (t (132) =

2.45, p=.016). Participants exposed to the personal-based attack disagreed that the message

was positive (M=2.33, SD=1.26) more than those exposed to the policy-based attack (M=2.94, SD=1.54), but it was not statistically significant (t (121) = 2.51, p=.014). The personal-based attack message was ranked as being more emotional (M=3.80, SD=2.10) than the policy-based message (M=2.67, SD=1.48), but it was not statistically significant (t (136)

=-1.52, p=.132). In terms of credibility, the personal-based attack message ranked higher

(M=2.95, SD=1.38) than the policy-based attack message (M=2.67, SD=1.48) but it was not statistically significant (t (130) =-1.12, p=.263). In terms of being populist, the policy-based

(25)

message ranked higher (M=4.53, SD=1.78) than the personal-based message (M=4.16,

SD=1.62), but it was not statistically significant (t (129) =1.28, p= .204). Participants rated

the personal-based message as more informative (M=3.71, SD=1.85) than the policy-based message (M=3.22, SD=1.83), but it was not statistically significant (t (137) =-1.56, p=.122).

Dependent Variable

In this study, I am interested in participants’ perception of the hypothetical candidates based on the campaign message they were exposed to. Based on my theoretical framework, I test the variable for both the sponsor and target candidates.

In measuring participants’ perception about the hypothetical candidates, a battery of five statements were asked on a 7-point Likert scale (1=“Strongly Disagree” and 7= “Strongly Agree”). The statements are some of the ways used by people to describe political figures. Participants were asked to state to what extent they agreed or disagreed about the competence, leadership skills, trustworthiness, likeability and ability to represent their interest if voted. I conducted a factor analysis to see how well the items measured the perception of the candidates. All items loaded on one factor for the perception for Ahmed Balarabe (Sponsor candidate) (Eigenvalue = 3.6; Cronbach’s α = .90; M= 3.4; SD= 1.18; N =136) and Babatunde Ademola (Target candidate) (Eigenvalue = 3.6; Cronbach’s α = .88; M= 4.2 SD= 1.02; N =134). The Cronbach’s α of .90 & .88 for Ahmed and Babatunde respectively indicate that the perception scale has an adequate level of inter-item reliability. On average, participants had a positive perception of Babatunde Ademola than Ahmed Balarabe (Figure 3).

In addition, I measured the propensity to vote for the hypothetical candidates

assuming there is a sudden call for a new presidential election in Nigeria scheduled to hold on May 29, 2020. With the aid of a slider, participants had to indicate on a scale of 0 to 100 how

(26)

certain they were likely to vote for the candidates (0 = "most certainly NOT" vote for the candidate and 100 = "most certainly vote" for the candidate). On average, participants were more likely to vote for Babatunde Ademola (N= 121; M =57.26; SD= 26.1) than Ahmed Balarabe (N = 123; M = 37.15; SD = 26.1) (Figure 4).

The results of a Pearson Correlation analysis show that there is a relationship between participants’ perception and propensity to vote for both candidates. The relationship was positive, moderate in strength and statistically significant for both Ahmed Balarabe (r (121) = 46, p < .001) and Babatunde Ademola (r (119) = 41, p < .001). When the propensity of a candidate being voted is high, such candidate ranks positive on the perception scale.

Figure 3

Participants’ perception of Ahmed Balarabe and Babatunde Ademola

Note. The numbers in the chart represent the mean for the two conditions 4.2 3.4 1 2 3 4 5 6 7 Babatunde Ademola Ahmed Balarabe P er ce p tion Candidates

(27)

Figure 4

Participants’ propensity to vote for Ahmed Balarabe and Babatunde Ademola

Note. The numbers in the chart represent the mean for the two conditions.

Results

Main Effects of Negative Campaigning

The first clear result shows that the hypothetical campaign messages did not have a direct effect on participants’ perception of the hypothetical candidates. To test the H1, which expects that in comparison to a personal-based negative message, a policy-based negative message will cause a lesser backlash effect for the sponsor candidate, I conducted a series of independent samples t-tests. Contrary to the expectations and previous literature, there was no significant difference between participants exposed to the policy-based message (N= 63) was not significantly different (M = 3.23; SD = 1.11) than that of participants in the personal-based group (N = 73) (M = 3.61; SD = 1.21) t (134) = - 1.86, p = .064, d= 0.33. The

57.25 37.15 0 10 20 30 40 50 60 70 80 90 100 Babatunde Ademola Ahmed Balarabe P rop en sity to V ote Candidates

(28)

average support of the sponsor politician for participants were exposed to the policy-based message (N = 58) was not significantly different (M = 34.17; SD = 26.59) than that of

participants in the personal-based group (N = 65) (M = 39.82; SD = 25.56) t (121) = -1.20, p = .233, d= 0.22. In sum, the results indicate that there was no backlash effect for the sponsor candidate for going negative.

To test H2, which expects that in comparison to a policy-based negative

message, a personal-based negative message is more effective in reducing voters’ perception of the target candidate, I conducted an independent samples t-test. Contrary to the

expectations and previous literature, there was no significant difference between participants exposed to the personal-based negative message (N = 72; M=4.07; SD= 1.15) and those exposed to the policy-based negative message (N = 62; M=4.30; SD= .83) t (132) = 1.31, p = .193, d= 0.23. In terms of propensity to vote, there was no significant difference between participants exposed to the personal-based message (N = 64; M=59.73; SD= 24.90) than those exposed to the policy-based message (N = 57; M=54.47; SD= 27.28) t (119) = -1.11, p = .270,

d= 0.20. The results indicate that there was no significant difference between the two types of

negative campaign messages. Therefore, there was no direct effect of the messages on the perception of the target candidate.

The Moderating Effects of Ethnicity and Religion

To test H3, which expects that neither a policy-based nor a personal based negative message will cause a backlash effect for the sponsor candidate if the voters are in-group members of the sponsor candidate, I conducted a moderator analysis. The model while controlling for gender, age and education was not statistically significant F (6, 126) = 1.79, p

= .106, R2 = .08. Contrary to expectation, the interaction effect between messages and the in-group / out-in-group was not significant b =-.021, t=-.04, p = 0.971.

(29)

To test H4, which expects that in comparison to a policy-based negative

message, a personal-based negative message is not more effective in reducing the perception of the target candidate when the voters are in-group members with the target candidate, I conducted a moderator analysis. The model while controlling for gender, age and education was not statistically significant F (6. 124) = 1.41, p = 0.215, R2 = .06. There was a significant interaction effect between the campaign messages and the ingroup / outgroup members b = -1.07, t=-2.11, p = 0.037.

The results of the moderator analyses above are likely affected by the skewed distribution of the in-group / out-group variable. More participants identified with the target candidate, Babatunde Ademola as in-group members (84%) than Balarabe Ahmed (13.9%). Therefore, I zoom in on the sample of the in-group members for the target candidate,

Babatunde Ademola (84%). I conducted a simple linear regression for only in-group members of the target candidate, Babatunde Ademola. Contrary to expectation, the results indicate that in comparison to a policy-based negative message, a personal-based negative message by the sponsor candidate is more effective in reducing voters’ perception of the target candidate if the voters are in-group members with the target candidate b* = -.16, t (109) = 14.66, p < .001,

F (1, 109) = 2.745, p = .100 with an R2 of .025.

To examine how in-group members of the target candidate perceive the sponsor candidate, I conducted a simple linear regression. The results indicate that a personal-based negative message from the sponsor candidate is more effective in increasing how an in-group member of the target candidate perceives the sponsor candidate than a policy-based negative message. b* = .18, t (111) = 7.55, p = <.001, F (1, 111) = 3.523, p = .063 with an R2 of .031.

(30)

Discussion and Conclusion

So, to what extent do religiosity and ethnicity determine the effectiveness of negative campaigning on the perception of candidates? Findings indicate that in comparison to a policy-based negative message, a personal-based negative message by the sponsor candidate against the target candidate, voters’ perception for the target candidate reduces if they are in-group members with the target candidate. This is contrary to previous literature based on party affiliations as a form of social identification (Budesheim et al., 1996; Iyengar et al., 2019). The influence of religion and ethnicity in moderating the effectiveness of negative campaigning is limited. There is an iota of in-group and out-group dynamics at play when ethnicity and religion are forms of social identification. However, these dynamics play out to an extent (Battle & Seely, 2010; Adida, 2015). The findings also suggest caution in ascribing an overarching power to in-group and out-group dynamics in influencing the effect of

negative campaigning.

The saliency of ethnicity in Nigerian politics is hampered by the uneven distribution of ethnic groups in Nigeria is a factor to consider. Out of the 374 ethnic groups in Nigeria, there are three major ethnic groups: Yoruba, Hausa /Fulani and Igbo (Mustapha, 2006; Langer & Ukiwo, 2007). Two of the three major ethnic groups (Yoruba and Igbo) are situated in the Southern part of Nigeria, while the Hausa/Fulani ethnic group is in the Northern part of the country. Due to ethnic fractionalization, intergroup dynamics is delicate. Politicians

constantly ensure they satisfy every region of the country when selecting members of his campaign team, most importantly his running mate. A politician seeking to become the president and hails from the North and is a Muslim must ensure that his running mate is chosen from the South and is a Christian and vice-versa. The only exception to this unwritten but salient rule was in 1993. The presidential candidate of the Social Democratic Party (SDP), Moshood Kashimawo Abiola, was a Muslim and a Southerner; while his running mate,

(31)

Babagana Kingibe was also a Muslim from the North (Miles, 2000). However, this study was limited, as it did not provide participants with these additional dynamics, thereby limiting the external validity and generalizability of the findings. There was no mention of the running mate of the hypothetical candidates nor where details of their shadow cabinet members provided. This was necessary to control for how different negative campaign messages are interpreted in the absence of other conflicting dynamics.

This study is not without limitations. One of the limitations was recruiting participants during a pandemic. The fear and anxiety associated with the pandemic limited the study from recording more diverse participation. During the data collection stage of this study, Nigeria was coming to terms with the pandemic, as there were recorded cases of the coronavirus (COVID-19) disease coupled with the ‘shaky’ initial response from the government. Therefore, telling people to participate in a study focused on providing a nuanced

understanding of political communication in Nigeria was a hard sell. Second, there were few complaints of network challenges while accessing the survey. This limitation is common for online studies conducted in areas with limited network coverage (Langer et al., 2017). The network issues account for the dropout rate recorded during the recruitment of participants. Third, there was no incentive for responding to the survey. The incentive would have gone a long way in retaining the attention of potential respondents even in the face of network challenges.

Overall, this study contributes to the literature on negative campaigning by teasing out some nuances related to other forms of social identification and negative campaigning effectiveness. This study is a conversation starter and provides a platform for future research to be explored. Future research should explore beyond the use of hypothetical candidates and campaign messages. This study should be replicated with real candidates and

(32)

within an election year. Future online survey studies in areas where the network is generally poor should consider providing an incentive for participation. Also, there is a need for comparative studies of negative campaigning in countries with diverse ethnic groups and religious beliefs. This will provide an avenue to understand the role ethnicity and religion plays with relation to the negative campaigning effectiveness.

There are four areas of interest to push this discussion of social identification and its influence on the effect of negative campaigning. In multi-ethnic states, it is interesting to examine how minor and major ethnic groups negotiate when deciding the candidate (in-group member) to support on the Election Day. This will provide a nuanced understanding of the negative campaigning and the role of ethnicity and religion as salient forms of social identification. Second, it will be an added knowledge to literature to study the effect of responses to negative campaign messages from target candidates on the dynamics of negative campaigning. This study restricted its scope to only attack messages of the sponsor candidate. The study of attacks and responses from both sponsor and target candidates can serve a secondary layer of understanding to the intergroup dynamics based on religion and ethnicity as the major forms of social identification.

A third dimension worth exploring is the effectiveness of different online channels used in disseminating negative campaign messages by politicians and their effect on the intergroup dynamics. In addition, there is need to examine effect of written negative

campaign messages via online platforms and spoken negative campaign messages at political rallies. In multi-ethnic states in sub-Saharan Africa (e.g. Nigeria), politicians spend resources in recruiting people to spew negative campaigns online (Hitchen et al., 2019). On the other hand, political rallies remain the largest form of mobilizing voters in sub-Saharan Africa

(33)

(Paget, 2019). Therefore, this aspect of the research will unpack the dynamics and effects of negative campaigning through political rallies and online communication channels.

(34)

References

Adeseri, L. & Nnochiri, I. (2015, January 21). Fayose under fire over anti-Buhari advert.

Vanguard. https://www.vanguardngr.com/2015/01/fayose-fire-anti-buhari-advert Adibe, J. (2015, January 27). Fayose’s Advert: Offensive Or Hate Speech? [Conference

presentation]. Adapted from a paper presented at a roundtable on hate speech organized by the Kukah Centre, Abuja. https://saharareporters.com/2015/01/29/fayose’s-advert-offensive-or-hate-speech-jideofor-adibe

Adida, C. L. (2015). Do African voters favour coethnics? Evidence from a survey experiment in Benin. Journal of Experimental Political Science, 2(1), 1-11.

https://doi.org/10.1017/XPS.2014.26

Adida, C., Gottlieb, J., Kramon, E., & McClendon, G. (2017). Reducing or reinforcing in-group preferences? An experiment on information and ethnic voting. Quarterly Journal of

Political Science, 12(4), 437-477 https://doi.org/10.1561/100.00017018

Ansolabehere, S., Iyengar S., Simon, A. & Valentino, N. (1994). Does Attack Advertising Demobilize the Electorate? American Political Science Review, 88(4): 829–838.

https://doi.org/10.2307/2082710

Auter, Z.J., Fine, J.A. (2016). Negative Campaigning in the Social Media Age: Attack Advertising on Facebook. Political Behaviour 38, 999–1020.

https://doi.org/10.1007/s11109-016-9346-8

Basil, M., Schooler, C., & Reeves, B. (1991). Positive and negative political advertising: Effectiveness of ads and perceptions of candidates. In F. Biocca (Ed.), Television and

political advertising, Vol. 1. Psychological processes; Vol. 2. Signs, codes, and images (p.

245–262). Lawrence Erlbaum Associates, Inc.

Battle, M. & Seely, J.C. (2010). It's All Relative: Modeling Candidate Support in Benin,

(35)

Bogardus, E. (1947). Measurement of Personal-Group Relations. Sociometry, 10(4), 306-311.

https://doi.org/10.2307/2785570

Boshoff, J. J. (2019). Negativity and Political Debates: The 2014 National and Provincial Elections in South Africa. Communicatio, 45(2), 1-17.

https://doi.org/10.1080/02500167.2019.1626458

Brooks, D.J & Geer, J.G (2007). “Beyond Negativity: The Effects of Incivility on the Electorate.”American Journal of Political Science 51(1): 1–16.

https://doi.org/10.1111/j.1540-5907.2007.00233.x

Budesheim, T. L., Houston, D. A., & DePaola, S. J. (1996). Persuasiveness of in-group and out-group political messages: The case of negative political campaigning. Journal of

Personality and Social Psychology, 70(3), 523. https://doi.org/10.1037/0022-3514.70.3.523

Carraro, L., & Castelli, L. (2010). The implicit and explicit effects of negative political campaigns: Is the source really blamed? Political Psychology, 31(4), 617-645.

https://doi.org/10.1111/j.1467-9221.2010.00771.x

Channels Television. (2018, December 10). Bishop Kukah Advises against Mixing Religion

and Politics |Roadmap 2019|[Video]. YouTube. https://youtu.be/Jz0JpfMvzHc?t=132 Cicero, Q. T., & Carville, J. (2012). Campaign Tips from Cicero: The Art of Politics, from the

Tiber to the Potomac. Foreign Affairs, 91, 18-28. https://www.jstor.org/stable/23217963 Craig, S. C., & Cossette, P. S. (2017). Whose Issue Is It Anyway... and Does It Really Matter?

Issue Ownership and Negative Campaigning. World Affairs, 180(3), 72-96.

https://doi.org/10.1177/0043820017750209

Craig, S. C., Rippere, P. S., & Grayson, M. S. (2014). Attack and response in political

campaigns: An experimental study in two parts. Political Communication, 31(4), 647-674.

(36)

Damore, D. F. (2002). Candidate strategy and the decision to go negative. Political Research

Quarterly, 55(3), 669-685. https://doi.org/10.1177/106591290205500309

Deutskens, E., De Ruyter, K., Wetzels, M., & Oosterveld, P. (2004). Response rate and response quality of internet-based surveys: An experimental study. Marketing letters, 15(1), 21-36. https://doi.org/10.1023/B:MARK.0000021968.86465.00

Ekeh, P. P. (1975). Colonialism and the two publics in Africa: A theoretical statement.

Comparative studies in society and history, 17(1), 91-112.

https://doi.org/10.1017/S0010417500007659

Ezeamalu, B. (2015, May 8). Why we published Ayo Fayose’s ‘death’ advert on Buhari. The

Sun. https://www.premiumtimesng.com/news/top-news/182617-why-we-published-ayo-fayoses-death-advert-on-buhari-the-sun.html

Freeman, P. (2012, August 31). The Attack Ad, Pompeii-Style. New York Times (1923-Current File), p. A27. http://search.proquest.com/docview/1705719803

Fridkin, K. L., & Kenney, P. J. (2004). Do negative messages work? The impact of negativity on citizens’ evaluations of candidates. American Politics Research, 32(5), 570-605.

https://doi.org/10.1177/1532673X03260834

Gaines, B. J., Kuklinski, J. H., & Quirk, P. J. (2007). The logic of the survey experiment reexamined. Political Analysis, 15(1), 1-20. https://doi.org/10.1093/pan/mpl008 Gamson, W. A., & Modigliani, A. (1989). Media discourse and public opinion on nuclear

power: A constructionist approach. American journal of sociology, 95(1), 1-37.

https://doi.org/10.1086/229213

Garramone, G. M. (1984). Voter responses to negative political ads. Journalism Quarterly, 61, 250-259. https://doi.org/10.1177/107769908406100202

Geer, J.G. (2006). In Defence of Negativity. Attack Ads in Presidential Campaigns. University of Chicago Press

(37)

Gervais, B. T. (2015). Incivility online: Affective and behavioral reactions to uncivil political posts in a web-based experiment. Journal of Information Technology & Politics, 12(2), 167-185. https://doi.org10.1080/19331681.2014.997416

Hitchen, J., Fisher, J., Hassan, I., & Cheeseman, N. (2019). Whatsapp and Nigeria's 2019

Elections: Mobilising the People, Protecting the Vote. Centre for Democracy and

Development (CDD). https://www.africaportal.org/publications/whatsapp-and-nigerias-2019-elections-mobilising-people-protecting-vote/

Homer, P. M., & Batra, R. (1994). Attitudinal effects of character‐based versus competence‐ based negative political communications. Journal of Consumer Psychology, 3(2), 163-185.

https://doi.org/10.1016/S1057-7408(08)80003-4

Hopmann, D.N, Vliegenthart, R & Maier, J. (2018). The effects of tone, focus, and incivility in election debates, Journal of Elections, Public Opinion and Parties, 28(3), 283-306,

https://doi.org/10.1080/17457289.2017.1394310

Houston, D. A., Doan, K., & Roskos-Ewoldsen, D. (1999). Negative political advertising and choice conflict. Journal of Experimental Psychology: Applied, 5(1), 3.

https://doi.org/10.1037/1076-898X.5.1.3

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, 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.

https://doi.org/10.1093/poq/nfs038

Jamieson, K.H. (1992). Dirty Politics: Deception, Distraction and Democracy. Oxford University Press.

(38)

Jamieson, K.H, Waldman,P. & Sherr, S. (2000). Eliminate the Negative? Categories of Analysis for Political Advertisements. In Crowded Airwaves: Campaign Advertising in Elections, Thurber, J., Nelson, C. & David D. (Eds), 44–64. Brookings Institution. Jasperson, A.E & Fan, D. P. (2002). An Aggregate Examination of the Backlash Effect in

Political Advertising: The Case of the 1996 U.S. Senate Race in Minnesota. Journal of

Advertising, 31(1), 1-12 https://doi.org/10.1080/00913367.2002.10673656

Jeyifo, B. (2014, June 29). Kayode Fayemi: Not What Went Wrong but What Went Right.

The Nation.

Johnson-Cartee, K.S., Copeland, G. & Johnson, E.E. (1991). Negative Advertising: Coming of Age. Lawrence Erlbaum.

Joseph, R (2015, July 28). American-Nigerian cooperation: An uncertain start to the Buhari era. Brookings Institution. https://www.brookings.edu/on-the-record/american-nigerian-cooperation-an-uncertain-start-to-the-buhari-era

Kamber, V. (1997). Poison Politics: Are Negative Campaigns Destroying Democracy? Perseus.

Keller, E. J. (2014). Identity, citizenship, and political conflict in Africa. Indiana University Press.

Kenny, C. B. (1994). The microenvironment of attitude change. The Journal of Politics, 56, 715– 728. https://doi.org/10.2307/2132189

Koter, D. (2019) Presidents’ Ethnic Identity and Citizens’ National Attachment in Africa,

Nationalism and Ethnic Politics, 25(2), 133-151.

https://doi.org/10.1080/13537113.2019.1598234

Kreuter, F. Stanley Tourangeau, P.R. (2008). Social Desirability Bias in CATI, IVR, and Web Surveys: The Effects of Mode and Question Sensitivity, Public Opinion Quarterly, 72(5), pp. 847–865, https://doi.org/10.1093/poq/nfn063

(39)

Krupnikov, Y., & Findley, B. (2018). Survey Experiments: Managing the Methodological Costs and Benefits. In Rae Atkeson, L. and Alvarez, RM (edt.): The Oxford Handbook of Polling and Survey Methods. Oxford University Press.

https://doi.org/10.1093/oxfordhb/9780190213299.013.32

Kung, F. Y., Kwok, N., & Brown, D. J. (2018). Are attention check questions a threat to scale validity? Applied Psychology, 67(2), 264-283. https://doi.org/10.1111/apps.12108

Langer, A., Meuleman, B., Oshodi, A. G. T., & Schroyens, M. (2017). Can student populations in developing countries be reached by online surveys? The case of the National Service Scheme Survey (N3S) in Ghana. Field Methods, 29(2), 154-170.

https://doi.org/10.1177/1525822X16663702

Langer, A., & Ukiwo, U. (2008). Ethnicity, religion and the state in Ghana and Nigeria: perceptions from the street. In Horizontal Inequalities and Conflict (pp. 205-226). Palgrave Macmillan.

Lau, R., & Pomper, G. M. (2001). Effects of negative campaigning on Turnout in U.S. Senate elections 1988-1998. The Journal of Politics, 63(3), pp. 804-819.

https://doi.org/10.1111/0022-3816.00088

Lau, R.R & Rovner, I.B. (2009). Negative Campaigning. The Annual Review of Political

Science 12:285–306 https://doi.org/10.1146/annurev.polisci.10.071905.101448

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.

https://doi.org/10.1111/j.1468-2508.2007.00618.x

March, D. S., & Graham, R. (2015). Exploring implicit in-group and out-group bias toward Hispanics. Group Processes & Intergroup Relations, 18(1), 89-103.