Dutch political parties and their personalities: Political brand personality in a complex multi-party system
Denise Haar – van de Wetering 11320478
haar.denise@gmail.com
Master’s Thesis
Graduate School of Communication Research Master Communication Science
University of Amsterdam
Dr. Marjolein Moorman 29-06-2018
DUTCH POLITICAL PARTIES AND THEIR PERSONALITIES 2
A
cknowledgementsAfter two years of research master, I am happy to hand in a thesis of which I am proud of. This thesis wouldn’t have been possible without the trust of my supervisor, Marjolein Moorman. Thank you for your confidence and your support. Also, I would like to thank Cristina, Courtney and Sander, who have been helping me by listening to my ideas,
proofreading my text or simply making time to have a cup of coffee with me. Most of all, I would like to thank my husband, Robbin-Jan, for his unconditional support and faith in me throughout my studies. I am beyond grateful for what I have been given.
You have been so, so good to me Reckless Love
Abstract
This study aims to answer how Dutch voters differentiate between political parties based on the party’s political brand personality, how political brand personality is associated with the likelihood to vote for the specified party and whether political brand personality is stable over time. With the use of Smith’s political brand personality scale (Smith, 2009), a quantitative content analysis is performed on free associations that reflected the consumer brand
knowledge. Consumer brand knowledge was retrieved by asking respondents to write down their thoughts when hearing the name of a political party in three waves around the 2017 Dutch parliamentary election. Three findings are presented. First, voters use personality dimensions to differentiate between political parties. Second, the personality dimension Honesty especially affected the likelihood to vote for each political party. For the other personality dimensions, the association with the likelihood to vote for the specified party differs between parties. Third, political brand personality is not stable over time.
Studying political brand personality in a highly complex political spectrum, such as the Netherlands, is unique. Not only do the results provide insights as to how voters
differentiate between parties in a crowded political sphere, it also contributes to the existing brand personality literature by providing an answer to the question of whether it matters if voters think about brands in terms of personality traits. Furthermore, by studying the stability of political brand personality over time, a contribution is made to the discussion of whether brand personality can be changed. The conclusions of this study allow to make practical recommendations for political strategists and it initiates to consider brand personality as a more important aspect of brand equity.
Keywords: political brands, political brand personality, consumer brand knowledge, multi-party system, brand equity
DUTCH POLITICAL PARTIES AND THEIR PERSONALITIES 4 Dutch political parties and their personalities
Political brand personality in a complex multi-party system
The word brand might evoke thoughts of products, such as Coca Cola, Nike and Volkswagen. However, branding practices are used in virtually every context (Scammell, 2007), and the word brand is appearing regularly in the political sphere (Nielsen, 2017). Nielsen and Larsen (2014) stated that the explanation for the interest in “political brands” can be found in the “changing nature of post-war Western democracies” (p. 154). Voter turnout has declined and the number of people that are a member of a political party has dropped (Mair, 2008). Party ties are weakened and voter volatility has increased (Clarke & Stewart, 1998). It is therefore not surprising that voters seek out for other cues to base their decision on: “As partisanship in the electorate has weakened, it stands to reason that voters would have to substitute other factors in their decision-making process” (Dalton, McAllister &
Wattenberg, 2002, p. 49). In this thesis, I study political brand personality of political parties as one of the factors which voters use in the electoral decision-making process. Political brand personality is defined as the “human characteristics relating to that party” (Smith, 2009, p. 212).
Based on consumer research, it has been established that consumers give brands human characteristics in order to familiarize a product in an over-communicated world (Freling & Forbes, 2005b; Downs, 1957). Brand personality enables consumers to differentiate between products (Freling & Forbes, 2005a), it affects brand preference (Banerjee, 2016), and it evokes brand loyalty (Lin, 2010). Political parties are brands “because they act as brands to consumers” (Smith & French, 2009, p. 211), which allows to transfer findings from consumer research to the political context.
As discussed, corporate brand personality is used by consumers to differentiate
between products. However, an unresolved question is whether the finding that consumers use brand personality to differentiate between corporate brands also occurs when studying
political brand personality. Therefore, I introduce the question of whether political brand personality also helps voters to differentiate between political brands.
Consumer research has indicated that brand personality allows for differentiation, especially when the products are similar (Li, Yen & Uysal, 2014) and that brand personality simplifies the consumer’s choice (Wee, 2003). Within the political context, political brand personality can function as a heuristic, which allows voters to take shortcuts and to
differentiate between political brands. Especially undecided voters might rely on brand personality as a cue to base their electoral decision on (Smith, 2009). In the decision-making process of the voter, political brand personality might thus be of influence.
In the parliamentary elections of 2017, 66% of the voters made their electoral decision only a few days before the elections (I&O Research, 2017). This indicates that a large amount of the voters were undecided. Furthermore, the Dutch political context is rather complex. With 28 political parties competing for the electoral win in 2017, it might be the case that (undecided) voters used political brand personality as shortcut to simplify their vote choice. Political brand personality might therefore be an explanatory force when studying the electoral decisions of the Dutch voters.
It is not only of interest to study how voters differentiate between the parties, but also whether it matters that individuals associate parties with certain personality traits. In other words, is the likelihood to vote for a party associated with giving human characteristics to this party? The answer to this question allows to make practical recommendations which are especially interesting for political marketeers. As stated by Plummer (2000), brand
DUTCH POLITICAL PARTIES AND THEIR PERSONALITIES 6 personality as a marketing tool is not very common, which stretches the importance of
studying this concept.
Furthermore, this study provides insights in the relationship between voting and brand personality, and recommendations can be made for the theory of brand equity. Brand equity touches upon “the differential effect of brand knowledge on consumer response to the marketing of the brand” (Keller, 1993, p. 1). Keller (1993) attributed only a minor part to brand personality in his discussion of brand knowledge. Based on this thesis it is argued that brand personality is of influence on consumer behaviour and should therefore receive a more prominent place in the brand equity theory.
Furthermore, it is important to understand if brand personality can be changed before arguing that brands should change their brand personality. Findings from studies on the stability of brand personality differ: Wee (2004) stated that brand personality is rather stable, whereas Johar, Sengupta and Aaker (2005) argued that brand personality can be modified. The contradictory findings invite to study if political brand personality is stable over a longer period of time.
In this thesis, the answer to the question what Dutch voters thought of when hearing the name of a political party is studied. Chen (2001) defined these thoughts as “free
associations” (p. 445). By studying free associations, it is possible to gain a better
understanding how a Dutch voter distinguishes between political parties. By asking voters for their thoughts when hearing the name of a party instead of asking them to rate political brands based on personality dimensions, the ecological validity of this study is assured. Furthermore, the free associations help to understand the factors that influence political decision-making, which include how a voter distinguishes between political parties and whether the political brand personality constructed in the voter’s memory is stable over time.
To answer the introduced questions, I first discuss findings from both consumer and political research that touch upon the concept brand personality. Second, I introduce the method, in which it is described how the free associations as provided by Dutch voters were coded. Third, I report on the results of how voters differentiate the political parties based on the party’s brand personality, I study the related effects, and I examine the stability of brand personality over time. Fourth, I draw conclusions based on the results and I make
recommendations for further research.
Theoretical Framework Political Parties as Brands
The concept brand has become more important over the last twenty years, and is adapted “everywhere now” (Scammell, 2007, p. 177). “Branding principles have been applied in virtually every setting where consumer choice of some kind is involved, e.g. with physical goods, services, retail stores, people, organisations, places or ideas” (Keller, 2002, p. 151 in French & Smith, 2009, p. 210). One of those settings in which there is a lot of choice
involved, is the political setting. It is therefore not surprising that the word brand is more frequently used within the political context as well (Nielsen, 2017). This is mainly because political parties “act as brands” (Smith & French, 2009, p. 211).
A brand can be defined as “a distinguishing name and/or symbol intended to identify the goods or services of either one seller or a group of sellers, and to differentiate those goods and services from those of competitors” (Aaker, 1991, p. 7). In other words, a brand allows the consumer to differentiate between different products (French & Smith, 2010). If Aaker’s (1991) definition of a corporate brand is considered, political parties can be approached as a brand: parties have a name and a symbol, which allows consumers (the voters) to identify their goods (party stances and ideology) and to differentiate between the competitors (the other political parties) (Smith & French, 2009).
DUTCH POLITICAL PARTIES AND THEIR PERSONALITIES 8 A political brand can be defined as “an associative network of interconnected political information, held in memory and accessible when stimulated from the memory” (French & Smith, 2010, p. 466), and “which can be identified and differentiated from other political representations” (Nielsen, 2017, p. 126). In other words, political brands are the knowledge a voter has about a political party. This knowledge allows voters to differentiate one party from other political parties, with the use of “artefacts, names, policies, sentiments, or symbols” that represent the political party (Nielsen, 2017, p. 126). The knowledge a voter has about a political party can be conceptualised as consumer brand knowledge (Keller, 2003). Consumer Brand Knowledge
Consumer brand knowledge is the meaning of a brand stored in the consumer’s memory, which includes “all descriptive and evaluative brand-related information” (Keller, 2003, p. 596). This descriptive and evaluative information can touch upon, for example, the value a consumer attaches to the brand, the evaluations a consumer has about a brand, and the “descriptive features that characterize the brand name product extrinsically,” related to the personality of a brand (Keller, 2003, p. 596).
Political brand personality. One aspect of the consumer brand knowledge can be the brand personality, as it is something that a consumer might have stored in his or her memory which gives meaning to the brand (Freling & Forbes, 2005b). Aaker and Fournier (1995) described brand personality as “the specific set of meanings which describe the ‘inner’ characteristics of a brand. These meanings are constructed by a consumer based on behaviours exhibited by personified brands or brand characteristics” (Aaker & Fournier, 1995, p. 393). In other words, brand personality is a description of a brand made by the consumer about a brand as if the brand was a person. The consumer develops this description based on the behaviour of a brand.
The same line of thought can be applied when conceptualising political brand personality. The personality of a political brand is defined as the “human characteristics relating to that party” (Smith, 2009, p. 212). For example, the Dutch political party CDA is often perceived as reliable but also as old-fashioned (Eilander, 2012). Smith (2009) identified political brand dimensions which can be used to identify a political brand’s personality profile. Based on Smith’s operationalisation, the personality traits mentioned by Eilander (2012) would be indicators of the personality dimensions Honesty and Image. It is thus very interesting to study whether Dutch voters anthropomorphise political parties and how they use brand personality to differentiate between the political brands.
The mechanism that lies behind brand personality has to do with the theory of anthropomorphism. Anthropomorphism is defined as “attributing humanlike properties, characteristics, or mental states to real or imagined nonhuman agents and objects” (Epley, Waytz, & Cacioppo, 2007, p. 865). Freling and Forbes (2005b) stated that brand personality can be an aspect of anthropomorphism. The authors argued that consumers give brands characteristics as if the brand was a human being. An important question is then why people use humanlike characteristics to describe brands.
Freling and Forbes (2005b) introduced an interesting explanation. They asked
respondents why they give humanlike characterises to brands. The respondents answered that they “humanised their brands because using a brand that somehow seems more human reduces uncertainty and risk involved in using the brand, makes the brand more familiar, and gives the consumer a feeling of comfort” (Freling & Forbes, 2005b, p. 156-157).
Downs (1957) introduced an compelling argument when discussing why individuals anthropomorphise brands. He stated that “information is costly in the real world” (Downs, 1957, p. 149). Downs argued that classical political theories assumed that citizens are always perfectly informed about political stances and issues. It is expected that the voter makes a
DUTCH POLITICAL PARTIES AND THEIR PERSONALITIES 10 carefully considered choice between the parties. However, he stated that it is important to consider the “economic realities,” which is the fact that information can be costly (Downs, 1957, p. 149). Not only have the voters little time to compare all policies parties propose, they also might not have the information needed to make a thoughtful consideration (Downs, 1957). Therefore, voters will not be perfectly informed, which subsequently forces them to seek out for other factors on which their voting decision can be based on.
To handle a lot of information in a complex environment, voters can make use of shortcuts (Lau & Redlawsk, 2001). Nielsen and Larsen (2014) argued that a political brand can be understood as a voter heuristic, which is a shortcut or “a helping hand in a complex political world” (p. 154). Brand heuristics are based on the bits of information a consumer has stored in their memory and the lacking motivation of the consumer to learn more about the political parties. A shortcut which can be particularly helpful in a complex political sphere is the political brand personality, especially because it allows differentiation between the political parties based on their personality (Smith, 2009, p. 212). To summarise: information is costly and in a complex electoral landscape and voters cannot process all information. Instead, they can make use of shortcuts, such as brand personality, to reduce the costs of information processing.
Brand equity. Political brand personality is an aspect of brand knowledge, as
discussed in light of brand equity. Brand equity is defined as “the differential effect of brand knowledge on consumer response to the marketing of the brand” (Keller, 1993, p. 1). Brand knowledge is conceptualised as an “associative network model”, which contains the
associations a consumer has towards about a brand (Chen, 2001). Brand knowledge touches upon brand awareness, brand image, and brand recognition and recall. Brand image is characterised as the associations which a consumer links to a brand, which includes the brand’s attributes, benefits, the consumer’s brand attitude and the strength and uniqueness of
the brand associations (Keller, 1993). According to Keller, user and usage image attributes, which are the type of person who uses a product and in which situation a product is used, can produce personality traits. However, Keller did not extend the role of brand personality as part of a consumer’s brand image.
Consumer perspective. So far, most of the reviewed theories discuss brands from a consumer perspective. This perspective touches upon “how consumers learn about and are motivated by brands” (Smith & French, 2009, p. 210). This approach does not focus on what the consumer was intended to feel, but what he or she actually feels when hearing the name of the brand. It is the “perception of the brand”, constructed by the consumer (Plummer, 2000, p. 80). According to Aaker and Joachimsthaler (2000, in Smith & French, 2009), a brand can also be studied from the perspective of management, which is the development of a brand identity by branding practices in a specific marketplace. It represents how the manager of the product wants the consumer to “think and feel” about the brand (Plummer, 2000, p. 2).
If a management perspective would been adapted in this thesis, I would have focussed on how the political parties communicate their brand personality. However, I am interested in how voters differentiate between political brands based on the party’s brand personality. The focus of this study is thus on the consumer perspective.
Measuring Political Brand Personality
To understand how a consumer perceives the personality of a brand, Aaker (1997) developed five brand personality dimensions. These dimensions were constructed for corporate brands. Each dimension has up to 11 personality traits as indicators. The
dimensions are Sincerity, Excitement, Competence, Sophistication and Ruggedness. Figure 1 provides an overview of the brand personality dimensions and their corresponding personality traits.
DUTCH POLITICAL PARTIES AND THEIR PERSONALITIES 12
Many authors have used Aaker’s Brand Personality Scale to assess the personality of various brands across different countries. For example, Ferrandi, Valette-Florence and Fine-Falcy (2000) measured the personalities of 12 French brands with the use of Aaker’s Brand Personality Scale. They concluded that the Brand Personality Scale was structured in similar fashion in France compared to the American context in which Aaker developed the scale. A second example is the application of the Brand Personality Scale to measure the brand personality of Ford in Chile (Rojas-Méndez, Erenchun-Podlech & Silva-Olave, 2004). It was concluded that Aaker’s scale was applicable to measure the brand personality of Ford, with the exception of the brand personality dimension Ruggedness. This is not surprising, as it is argued that culture has a moderating effect on the Brand Personality Scale. Ruggedness might be a more culturally specific personality dimension as compared to the other dimensions (Aaker, Benet-Martínez, & Garoleza, 2001). However, the stability of the other personality dimensions allows for the use of the Brand Personality Scale within the field of consumer studies.
Whereas Aaker’s scale has been broadly applied within consumer research, only two studies have been published in which the operationalisation is transferred to a political context (Gorbaniuk, Kusak, Kogut, & Kustos, 2015). First of all, Schneider (2004) applied the Brand Personality Scale to German politics. He concluded that the personality traits were applicable in 81% of the responding parties. A second author who applied Aaker’s operationalisation of
brand personality, is Smith (2009). He tested whether the personality dimensions were applicable when considering British political parties as brands. Based on factor analyses he concluded that, in the British context, an additional personality dimension had to be included: Uniqueness (see Figure 2). Before describing this specific personality dimension, a
description of all of Smith’s (2009) party personality dimensions is provided.
The personality dimension Honesty concerns personality traits such as reliable, wholesome and sincere. This dimension touches upon how honest and reliable a consumer perceives a brand to be. The personality dimension spirited is associated with traits such as daring, imaginative and cheerful. A brand with a high score on this dimension, would thus be a very enthusiastic and exciting brand. The personality dimension Image refers to how up-to-date a brand is. Thus, is the brand trendy, exciting and young? If so, the brand would score highly on this personality dimension. However, if a brand is perceived as outdated, it might score lower on this personality dimension. The personality dimension Leadership refers to whether a consumer perceives a brand to be successful. Personality traits associated with this dimension are confident, intelligent and hardworking. The personality dimension Toughness refers to how tough a brand is perceived by the consumer. Is the brand associated with masculinity, ruggedness and toughness? The brand might score high on this personality dimension if this is the case. Uniqueness touches upon personality traits such as unique, independent, and original. This personality dimension is especially about how a party’s Figure 2. Smith’s (2009) Modified Brand Personality Scale. Based on Smith (2009, p. 220)
DUTCH POLITICAL PARTIES AND THEIR PERSONALITIES 14 personality is different compared to the other personality dimensions of other parties (Smith, 2009).
Schneider and Smith asked respondents to rate a political party based on the
personality items. This is very interesting, because they made the assumption that voters have the brand personality stored as consumer brand knowledge. However, it is uncertain whether this assumption is correct. To increase the ecological validity of this study, the consumer brand knowledge is studied by asking respondents to write down their thoughts when hearing the name of a political party. These free associations are coded based on the political brand personality scale, as provided by Smith.
Political Brand Personality as Differentiator
Brand personality as differentiator. Scammell (2007) argued that “the task of brand research lies in discovering how differentiators operate in consumers’ perceptions and in finding patterns of differentiation” (p. 180). In other words, it should be understood how consumers distinguish between brands. Especially when multiple brands are competing within the same sphere, it is essential that these brands are “sufficiently differentiated from each other” (Till, Baack & Waterman, 2015, p. 96). Brand differentiation is thus a tactic to fight the severe competition (Thomas & Sekar, 2008). Furthermore, brand differentiation allows democracy to function. According to Popkin, Gorman, Phillips and Smith (1976), voters balance the costs and benefits of voting. When the parties “converge”, the distinctiveness of the parties decreases which subsequently decreases the incentive to vote. Because all parties look alike, it does not matter which one of them wins the election. However, when parties “diverge”, the distinctiveness of the parties increases and “more voters will find it worthwhile to vote” (Popkin et al., 1976, p. 799). Brand differentiation is thus essential in a political context in which multiple parties are competing for the electoral win.
Brand personality allows differentiation between political parties and subsequently make brands more attractive for consumers (Thomas & Sekar, 2008). Especially when the characteristics of products or the brands are very similar, brand personality allows consumers to differentiate (Haigood, 1999; Li et al., 2014).
When differentiating political brands based on their brand personalities, the
consumer’s choice is simplified (Wee, 2003). Freling, Crosno and Henard (2010) argued that brand personality can be used as a marketing tool to differentiate between products which subsequently affects the growth of a brand (Doyle, 1989). Brand personality helps to
differentiate and subsequently affects the consumer’s preferences (Freling & Forbes, 2005a). Furthermore, it allows to build an emotional connection with the brand (Akin, 2011). To understand the importance of brand differentiation, the complex political system of the Netherlands is briefly discussed.
The Netherlands and its complex political spectrum. Based on studies conducted on Western European electoral behaviour, it is argued that the choice a voter makes is affected by “social cleavages, ideological divisions, strong (sometimes institutionalized) party-group ties, non-dimensional issues, and pre-election coalition behaviour among the parties” (Whitten & Palmer, 1996, p. 231). Until the late 1960, these cleavages are reflected by the pillarisation of the Dutch society, in which voters almost never changed their party preference.
After 1960, the pillars disappeared and the Dutch elections became “the most volatile of Western Europe” (Van Der Meer, Lubbe, Van Elsas, Elff & Van Der Brug, 2012, p. 333). According to Pellikaan, De Lange and Van Der Meer (2016), the Dutch political landscape is unstable because of electoral fragmentation and mainstream parties changing their positions. Furthermore, the decision for which party to vote for in the 2017 elections was made last minute; 66% of the Dutch citizens indicated that they were not sure for which party to vote
DUTCH POLITICAL PARTIES AND THEIR PERSONALITIES 16 for only two days before the election (I&O Research, 2017). In a political landscape with 28 political parties competing for the electoral win, and voters making up their mind last minute, it is of great importance to understand how voters make electoral decisions and how they differentiate between the many political parties. As discussed, brand personality might be an important aspect of brand differentiation. Therefore, I raise the following question:
RQ1: How do voters differentiate between Dutch political parties based on the party’s perceived personality dimensions?
Political Brand Personality and Voting Intention
After establishing how voters differentiate between political parties based on the political brand personality, it is interesting to examine whether it matters if voters think of political brand in terms of personality traits. Freling and Forbes (2005a) studied the effects of brand personality. They found that when a brand has as a strong and favourable personality dimension, no matter which one of the personality dimensions, the brand was perceived as more favourable, unique, strong and congruent compared to brands without a brand personality.
The impact of brand personality was also studied by Banerjee (2016), who concluded that brand preference can be affected by brand personality. Furthermore, Lin (2010) argues that a having a distinct brand personality is associated with more brand loyalty. Brand personality affects the consumer’s feelings about a brand, the “self-expression it allows, the relationship it facilitates and the simplification of brand choice it enables” (Freling & Forbes, 2005b, p. 150). Furthermore, brand personality allows the consumer to be confident, which subsequently might stimulate the consumer to purchase a product from the brand (Doyle, 1989).
Eisend and Stokburger-Sauer (2013) conducted a meta-analysis on the consequences of brand personality. It appeared that the brand personality dimensions Sincerity and
Competence, as defined by Aaker (1997), have a strong effect on “brand attitudes and brand commitment” (Eisend & Stokburger-Sauer, 2013, p. 214). Freling and Forbes (2005b) also looked at the effects of brand personality, and they concluded that emotional gratification can be achieved when a brand has a positive and strong brand personality. This may subsequently lead to a higher likelihood to purchase the brand. A negative brand personality however, has negative consequences for the brand and it negatively affects purchase intention.
Building upon the discussed results, it would be expected that political brand
personality also has an effect on the behaviour of the consumers, which in this case is the vote they cast on election day. However, very little is known about the effects of political brand personality. We know that political brands affect voting behaviour (Nielsen & Larsen, 2014), and that voters assign parties personality traits because this makes it easier to see whether a party resembles to them (Nielsen, 2017). However, there is no answer to the question of whether it matters that voters attach humanlike characteristics to political parties. Therefore, I ask the following question:
RQ2: How is vote choice associated with political brand personality?
Political Brand Personality and Campaign Development
Officially, election time in the Netherlands started on the 20th of February 2017 and
ended on election day, which was the 15th of March 2017 (Dienst Publiek en Communicatie,
2017). During these four weeks, the political parties aired political commercials, their politicians participated in various debates and the party’s stances were shared across social media platforms. During electoral campaigns, voters are motivated to (re)consider their party
DUTCH POLITICAL PARTIES AND THEIR PERSONALITIES 18 choice. Political campaigns inform citizens about the political parties and subsequently can motivate citizens to go cast their vote (Geers, Bos & De Vreese, 2017). It is thus interesting to see whether political brand personality is stable throughout the electoral campaign.
Johar, Sengupta and Aaker (2005) studied the stability of brand personality, and they found that consumers update their inferences about brand personality based on new
information. More specifically, when consumers with a positive brand personality in mind when thinking about a brand were exposed to information that referred to negative brand personality traits, the consumer developed a more negative brand personality. When hearing the name of a brand, individuals who do not have a brand personality in mind update their image of a brand based on “evaluative implications,” and not on “trait-related inferences” (Johan et al., 2005, p. 467). In other words, when consumers have a brand personality in mind when hearing the name of a brand, their brand personality is updated by exposure to
information with reference to negative brand personality traits. Based on this study, it may be plausible that voters update their brand personality based on new information, such as
political campaigns.
However, Wee (2004) studied the stability of corporate brand personality and he concluded that brand personality dimensions are rather stable. Similar results are presented by other authors (Aaker, 1996; Geuens, Weijters & De Wulf, 2009). Based on these findings, it could thus be argued that political brand personality is rather stable. The contradicting
findings on the question of whether brand personality is stable, raises the following question:
Method Research Design and Method
To answer the three research questions, a quantitative content analysis was performed on free associations, which reflect the consumer brand knowledge. The free associations were retrieved from an online self-completed survey, conducted by the University of Amsterdam and TNS NIPO. The respondents were part of the Dutch TNS NIPO panel, and they were randomly selected from this database to participate in the study (N = 2,144).
The respondents answered the following question: “What do you think about when you read the name of the following political party [name of the party]?”. By asking what individuals think about a political party with an open question, it allows for the retrieval of all related and relevant memories of the respondent based on this free association (French & Smith, 2010). In this study, the respondents were asked what they think of when hearing the name of a party for the seven most outstanding parties from both the far-left and far-right side of the Dutch political spectrum. The parties are the VVD (Volkspartij voor Vrijheid en
Democratie), PvdA (Partij van de Arbeid), SP (Socialistische Partij), CDA
(Christen-Democratisch Appèl), GroenLinks, D66 (Democraten 66) and PVV (Partij voor de Vrijheid). The respondents answered this question three times. Wave 1 was conducted before the electoral campaign started, in October 2016. Wave 2 was conducted during the electoral campaign, in the beginning of March 2017. On the 15th of March 2017, the elections took
place. Right after the elections, between the 17th and 21st of March 2017, wave 3 was
conducted. By asking for free associations in three waves, it is possible to make a longitudinal comparison based on the brand personalities the respondents expressed. The respondents who did not answer the free associations are removed from the data set. After removal, the data set includes 1,994 respondents.
DUTCH POLITICAL PARTIES AND THEIR PERSONALITIES 20 The respondents gave a total of 33,512 free associations. This means that the
respondents wrote down about 17 associations (M = 16.81) per person. Divided over the seven parties and three waves, it can be concluded that the respondents wrote down about 0.8 associations for each party. For each wave, the number of associations a respondent could give per party was limited to five. If cases where the respondent gave more than five associations per party, only the first five associations were coded.
The respondents had an average age of 48 (M = 48.16, SD = 17.36). The youngest participant was 18 years old and the oldest participant was 91 years old. The mean age of the respondents is slightly higher than the mean age of Dutch inhabitants, which is 41.6 (CBS, 2018). However, it should be noted that the mean of the respondents might be slightly older due to the fact that only respondents older than 18 were allowed to participate in the study (Moorman, 2015).
The respondents were almost perfectly divided in terms of gender: 51.3% of the participants were female (N = 1,022). This division is very similar to the gender distribution in the Netherlands: 50.4% of its inhabitants are female (CBS, 2018). Most of the respondents have an MBO 2, 3 or 4 degree (34.6%), an HBO or university bachelor degree (26.1%), or an MBO 1 degree (13.1%). The sample is slightly higher educated compared to the Dutch population (Onderwijs in Cijfers, n.d.). This finding also occurred in the research of Behr, Bellgardt and Rendtel (2005), who stated that Dutch higher educated individuals are more likely to participate in panels. It is therefore not surprising that the level of education of the respondents in the sample is somewhat higher compared to the level of education of the Dutch population. I will include information about the respondent’s gender, age and education as control variables, as it is expected that these demographic features affect voting behaviour (Caprara, Barbaranelli & Zimbardo, 1999).
Coding of Free Associations
Six coders received a two-day training by the author of this paper. As preparation, the coders coded random samples of associations from a training set. By calculating the
intercoder reliability, it was possible to retrieve the associations that caused disagreement. The intercoder reliability was calculated with the use of Krippendorff’s Alpha (Hayes & Krippendorff, 2007). With the use of Krippendorff’s Alpha, the observed disagreement is divided by the expected disagreement between the coders (De Swert, 2012). Krippendorff’s Alpha can be tested for variables on nominal, ordinal, ratio and interval levels. Furthermore, the number of coders does not interfere with the results of the Alpha level (Lombard, Snyder-Duch & Bracken, 2006). The “smallest alpha for the data to be accepted as reliable” is 0.67 (Krippendorff, 2011, p. 105).
During the training sessions, the associations with low levels of agreement were discussed. Each coder shared their interpretation of the association. The coding of the brand personality was done by the three coders with the highest intercoder reliability, which resulted in a reliable data set (α = 0.68)1.
Variables
Political brand personality is coded based on the scale Smith (2009) provided. This scale includes six different Personality Dimensions: Honesty (1), Spirited (2), Image (3), Leadership (4), Toughness (5) and Uniqueness (6). Each personality dimension has personality traits, which function as items to measure the personality dimensions. The dimensions and their related traits as provided by Smith, are in English. However, the free associations the respondents provided were in Dutch. Therefore, Smith’s personality traits
1 Coding based on Smith’s personality scale was challenging, which was reflected by the intercoder reliability.
Therefore, it was decided that the data of the coders with the highest intercoder reliability (the author of this paper and two coders) was kept. The author of this paper and one of these coders recoded the unreliable data. This solution was most appropriate considering the scope of this study. However, in further research is recommended on how Smith’s personality scale can be transferred to other political systems.
DUTCH POLITICAL PARTIES AND THEIR PERSONALITIES 22 were translated to Dutch. Furthermore, the antonyms were included as well. It was difficult to code only for the exact translation of the personality traits, so the coders were free to have a slightly broader interpretation. For example, when a respondent wrote down “liar”, it was coded for the personality dimension Honesty. If a respondent wrote down “whining”, or “old fashioned”, it was coded for the dimension Image. For the complete codebook, please consult Appendix A.
Second, the coders indicated whether a free association was positive, neutral, or negative. By also coding for the evaluation of the free association, it is possible to make a distinction between positive and negative personality dimensions.
A third variable which is used in the analyses, is the political preference of the respondents. This is measured based on the likelihood that the respondent will vote for the specified party, from very unlikely (1) to very likely (10). The political preference is measured in all three waves.
Results Differentiation Based on Personality Dimensions
The first research question touches upon the differentiation of political parties based on the personality dimensions. I will present the results per political party. A complete overview of the distribution of personality dimensions can be found in Appendix B.
VVD. Many respondents mentioned negative personality traits related to Honesty when hearing the name of the VVD. This dimension includes personality traits such as unreliable, dishonest and lying. A one-way ANOVA supports the thought that the VVD has significantly more mentions of negative personality traits associated with Honesty (FW (6; 27,823) = 49.93, p < .001) compared to the other parties. For example, hearing the name of the VVD (M = 0.10, SD = 0.32) evoked more negative personality traits related to Honesty
compared to the SP (M = 0.03, SD = 0.17). The difference is significant (Mdifference = 0.07, p <
.001).
When the distribution of positive personality traits is considered (Figure 4), it becomes clear that hearing the name of the VVD evoked relatively many mentions of personality traits associated with Leadership. Personality traits associated with this dimension are secure, successful and stable. Compared to the PvdA, SP, GroenLinks and PVV, the VVD (M = 0.02, SD = 0.13) received significantly more mentions of such personality traits (Fw (6; 27,829) = 20.28, p < .001). For example, the mean difference between the VVD and PVV (M = 0.01, SD = 0.07) is significant (Mdifference = 0.01, p < .001). To conclude: hearing the name of the VVD
evoked negative personality traits related to Honesty and positive personality traits related to Leadership.
PvdA. If the distribution of the negative personality dimensions is considered (Figure 3), it is clear that for each personality dimension, the PvdA had the most or second most mentions of negative personality traits. This might indicate that the voter had an overall negative personality in mind when thinking of this political party as compared to the other parties. Based on the distribution of positive personality dimensions, it can be concluded that the voters did not associate this party with specific personality dimensions. The party thus had an overall negative personality profile.
SP. The SP did not have a distinctive personality profile based on the distribution of the positive and negative personality dimensions. The analyses of variance show similar results: the political party did not stand out on particular personality dimensions.
DUTCH POLITICAL PARTIES AND THEIR PERSONALITIES 24
CDA. Many respondents mentioned negative personality traits associated with the personality dimension Image. This dimension includes personality traits such as old
fashioned, complaining, and being opposed to everything. Furthermore, hearing the name of this party evoked relatively many mentions of negative personality traits associated with the dimension Spirited. This dimension touches upon personality traits such as being cowardly, disorganised and monotonous. A one-way ANOVA supports the observation that the CDA received significantly more mentions of negative personality dimensions related to Spirited (FW (6; 27,823) = 5.32, p < .001) and Image (Fw (6; 27,823) = 21.23, p < .001) compared to the other parties. For example, respondent mentioned more negative personality traits
associated with Image when hearing the name of the CDA (M = 0.04, SD = 0.20) compared to the VVD (M = 0.01, SD = 0.09). The difference is significant (Mdifference = 0.03, p < .001).
Figure 3. Relative distribution of the negative personality dimensions over the political parties
Furthermore, the CDA had a relatively greater mentions of positive personality traits associated with Honesty. A one-way analysis of variance indicates that, indeed, the CDA is significantly different from the other parties based on this personality dimension (FW (6; 27,824) = 26.34, p < .001). The post-hoc tests indicate that the CDA had more mentions of personality traits associated with Honesty compared to the other parties, with the exception of D66. For example, the respondents mentioned more positive personality traits associated with Honesty when hearing the name of the CDA (M = 0.04, SD = 0.20) compared to the PvdA (M = 0.01, SD = 0.08). The difference is significant (Mdifference = 0.03, p < .001).
GroenLinks. When comparing the parties based on the distribution of positive personality dimensions, it appears that GroenLinks had relatively numerous mentions of the personality dimensions of Image and Spirited. A one-way analysis of variance supports the argument that GroenLinks significantly differed from the other parties when these personality dimensions are considered (Spirited (+): FW (6; 27,824) = 18.29, p < .001; Image (+): FW (6; 27,824) = 16.52, p < .001).
The post-hoc tests indicate that indeed GroenLinks had more mentions of positive personality traits associated with the dimensions Image and Spirited compared to the other parties. For example, the respondent mentioned more positive personality traits associated with Spirited when hearing the name of GroenLinks (M = 0.02, SD = 0.13) compared to the PvdA (M = 0.00, SD = 0.02). The difference is significant (Mdifference = 0.02 p < .001).
D66. The distribution of the positive personality dimensions indicates that this party had an overall positive brand profile. The party did not stand out on specific personality dimensions.
DUTCH POLITICAL PARTIES AND THEIR PERSONALITIES 26
PVV. The respondents mentioned many negative personality traits associated with the dimensions Leadership and Toughness when hearing the name of the PVV. Leadership touches upon personality traits such as dumb, unstable and passive. Toughness refers to personality traits such as weak, sad and capricious. A one-way ANOVA supports this observation: the PVV had more mentions of negative personality traits associated with Toughness and Leadership compared to most of the other parties (Toughness (-): FW (6; 27,829) = 10.32, p < .001; Leadership (-): FW (6; 27,829) = 26.97, p < .001). For example, hearing the name of the PVV (M = 0.03, SD = 0.17) evoked the mention of more negative personality traits associated with Toughness compared to the VVD (M = 0.01, SD = 0.09). The difference is significant (Mdifference = 0.02, p < .001).
To conclude: voters had a distinct negative personality in mind when hearing the names of CDA, VVD and PVV. The PvdA did not stand out on one of the dimensions, but has an overall negative brand personality. Furthermore, GroenLinks had a positive and
Figure 4. Relative distribution of the positive personality dimensions over the political parties
distinctive brand personality based the personality dimensions Image and Spirited. Also, the CDA was perceived to be more Honest compared to the other parties. D66 had an overall positive brand profile but the party did not stand out on one of the dimensions. The SP did not emerged based on one of the dimensions. Furthermore, hearing the name of the party did not evoke the mention of many personality traits. This might indicate that the voters did not attribute humanlike characteristics to this party.
Voting Behaviour and Personality Dimensions
The second research question touches upon the query of whether the likelihood to vote for a party differs when the voter thinks in terms of personality or not. To answer this research question, the likelihood to vote for the specified party is predicted based on the personality dimensions. Based on the indication of whether the association is positive or negative, the dimensions can be transformed based on these sentiments. With the use of a linear regression, the association between the likelihood to vote for the specified party and the personality dimensions is examined2. Based on the F-values, it can be concluded that the regression
models successfully predict the likelihood to vote for the specified parties. The R2 of the
models ranges between 0.04 and 0.19, which indicates that the models explain up to 19% of the variation in the likelihood to vote for the specified party based on the personality
dimensions and the control variables. Because of the many significant results, only the most striking results per political party will be discussed. Table 1 visualises all significant
associations. Please consult Appendix C for an overview of all regression models.
2 It is unfortunately not possible to perform an analysis of variance with the personality dimension as grouping
variable and the likelihood to vote for the specified party as dependent variable because the likelihood to vote is measured for each political party separately.
VVD. Mentioning negative personality traits related to Honesty is significantly associated with a lower likelihood to vote for the VVD, and mentioning positive personality traits is significantly associated with a higher likelihood to vote for the VVD. For example, the mention of a positive personality trait associated with Honesty is positively associated with the likelihood to vote for the VVD in Wave 1 (b = 2.55 t = 4.35 p < .001) Furthermore, mentioning positive personality traits associated with Leadership also increases the likelihood to vote for this party (e.g. wave 1: b = 1.97, t = 3.41, p = .001).
PvdA. Negative personality traits linked to Honesty are negatively associated with the likelihood to vote for the PvdA (e.g. wave 1: b = -1.12, t = -6.38, p < .001). When the
respondent mentioned a positive personality trait related to Honesty, there is a positive association between this personality dimension and the likelihood to vote for the PvdA (e.g. wave 2: b = 2.01, t = 3.02 p = .003). Additionally, there is a positive association between the mention of a positive personality trait associated with Leadership and the likelihood to vote for the PvdA (e.g. wave 1: b = 1.24, t = 3.31, p < .001).
SP. The likelihood to vote for the SP is negatively related to the mention of a negative personality trait associated with Honesty (e.g. wave 2: b = -1.38, t = -4.24, p < .001), whereas the mention of a positive personality trait for this dimension is positively associated with the likelihood to vote for this party (e.g. wave 1: b = 2.09, t = 4.61, p < .001). Furthermore, mentioning a negative personality trait regarding Image is negatively associated with the likelihood to vote for the SP (e.g. wave 1: b = -1.02, t = -2.74, p = .006). Similar results appear for Toughness: mentioning a negative trait related to this dimension is negatively associated with the likelihood to vote for the SP (e.g. wave 1: b = -1.78, t = -2.47, p = .014).
CDA. Mentioning negative personality traits linked to Honesty is negatively associated with the likelihood to vote for the CDA (e.g. wave 2: b = -1.62, t = -6.71, p < .001), and mentioning positive personality traits of this dimension is positively associated
DUTCH POLITICAL PARTIES AND THEIR PERSONALITIES 30 with the likelihood to vote for this party (e.g. wave 3: b = 2.23, t = 6.09, p < .001).
Furthermore, mentioning negative personality traits related to Image is negatively associated with the likelihood to vote for this party (e.g. wave 2: b = -1.20, t = -3.25, p = .001), whereas mentioning positive personality traits associated with Leadership (e.g. wave 2: b = 1.37, t = 3.20, p = .001) and Spirited (e.g. wave 2: b = 3.45, t = 3.06, p = .002) is positively associated with the likelihood to vote for the CDA.
GroenLinks. The likelihood to vote for GroenLinks is positively associated with the mention of a positive personality trait for Honesty (e.g. wave 1: b = 2.51, t = 4.90, p < .001). The mention of a negative personality trait related to this personality dimension is, however, negatively associated with the likelihood to vote for GroenLinks (e.g. wave 1: b = 1.44, t = -4.51, p < .001). Furthermore, the likelihood to vote for GroenLinks is positively associated with the mention of a positive personality trait linked to Image (e.g. wave 2: b = 1.20, t = 3.61, p < .001), whereas the mention of a negative personality trait for this dimension is negatively associated with the likelihood to vote for this party (e.g. wave 2: b = 1.69, t = -3.18, p = .002).
D66. Mentioning a negative personality trait related to the personality dimension Honesty is negatively associated with the likelihood to vote for D66 (e.g. wave 3: b = -1.67, t = -4.98, p < .001), whereas mentioning a positive personality trait of this dimension is
positively associated with the likelihood to vote for D66 (e.g. wave 2: b = 1.10, t = 4.39, p < .001). Similar results appear for the personality dimension Image: the mention of a positive personality trait associated with this dimension is positively associated with the likelihood to vote for this party (e.g. wave 1: b = 1.01, t = 2.04, p = .041), whereas a negative personality trait linked to this dimension is negatively associated with the likelihood to vote for D66 (e.g. wave 1: b = -1.95, t = -3.77, p < .001). Also, the mention of negative personality traits related
to Uniqueness is negatively associated with the likelihood to vote for D66 (e.g. wave 1: b = -1.32, t = -4.19, p < .001).
PVV. The likelihood to vote for the PVV is negatively associated with the mention of negative personality traits regarding Honesty (wave 1: b = -0.63, t = -2.44, p = .015).
Mentioning positive personality traits linked to this dimension however, is positively associated with the likelihood to vote for the PVV (wave 3: b = 5.02, t = 6.52, p < .001). Furthermore, mentioning positive personality traits of Toughness is positively associated with the likelihood to vote for this party (wave 2: b = 5.98, t = 2.83, p = .005), whereas the mention of negative personality traits related to this dimension is negatively associated with the
likelihood to vote for the PVV (wave 1: b = -0.56, t = -2.07, p = .038).
To conclude: mentioning positive personality traits which refer to Honesty positively predicts the likelihood to vote for the party, whereas mentioning negative personality traits referring to Honesty is associated with a lower likelihood to vote for the party. These findings can be generalized for all seven parties. The predictability of the likelihood to vote for the specified party differs for the other personality dimensions, but it is evident that positive traits are associated with a higher likelihood to vote for the specified party whereas negative traits are associated with a lower likelihood to vote for the specified party.
Personality Dimensions over Time
The third research question touches upon the subject of how the use of personality traits to describe a party changes as the election campaign evolves. To test the differences, a one-way analysis of variance is performed in which the six personality dimensions are divided into positive and negative categories, and compared over the three waves. Since I am interested in the effect of time, only the respondents who gave free associations for each party in all three waves are included in the analyses (N = 533). Table 2 shows the distribution of the personality dimensions over the three waves and it shows whether the difference between the
DUTCH POLITICAL PARTIES AND THEIR PERSONALITIES 32 waves is significant based on the one-way ANOVA, which is discussed in the following paragraphs.
A one-way ANOVA indicates that for four personality dimensions there are
significant differences between the waves. Please consult Appendix D for the non-significant results. First, both positive and negative personality traits related to Honesty differ between the waves (Honesty (-): Fw (2; 11,190) = 19.53, p < .001) and Honesty (+): Fw (2; 11,190) = 7.13, p = .001). Post-hoc tests indicate that for the negative personality traits associated with Honesty, there is a significant difference between wave 1 and 3 (Mdifference = 0.02, p < .001)
and between wave 2 and 3 (Mdifference = 0.03, p < .001). These results indicate that over time,
negative personality traits associated with Honesty became less prominent in the memory of the voter. Second, positive personality traits related to Honesty significantly differ between wave 1 and 2 (Mdifference = -0.01, p = .013) and between wave 1 and 3 (Mdifference = -0.01, p =
.002). The results indicate that during the campaign, positive personality traits associated with Honesty became more prominent for the voter.
Third, the positive personality dimension Spirited differs between the waves (Fw (2; 11,190) = 3.61, p = .027). The Games-Howell post-hoc test tests indicate that the number of positive personality traits linked to Spirited significantly differs between wave 1 and 3
(Mdifference = -0.01, p = .021). Compared with wave 1, personality traits linked to Spirited were
mentioned more often when hearing the political brand’s name right after the elections. Fourth, the positive personality dimension Image significantly differs between the waves (FW (2; 11,190) = 6.91, p = .001). The mention of positive personality traits related to Image differs between wave 1 and 3 (Mdifference = -0.01, p = .007). Traits associated with
Image were thus more prominent right after the elections as compared to when the electoral campaign had not taken off yet.
T abl e 2 C ha nge C ha nge T ot al c ha nge W ave 1 w ave 2 W ave 2 w ave 3 W ave 1 w ave 3 H one st y ( -) 180 221 111 ▲ 41 ▼ 110*** ▼ 69*** H one st y ( + ) 56 92 99 ▲ 36* ▲ 7 ▲ 43** S pi ri te d ( -) 31 23 26 ▼ 8 ▲ 3 ▼ 5 S pi ri te d ( + ) 13 20 31 ▲ 7 ▲ 11 ▲ 18* Im age ( -) 74 88 97 ▲ 14 ▲ 9 ▲ 23 Im age ( + ) 7 22 26 ▲ 15* ▲ 4 ▲ 19** L ea de rs hi p ( -) 143 126 121 ▼ 17 ▼ 5 ▼ 22 L ea de rs hi p ( + ) 46 47 56 ▲ 1 ▲ 9 ▲ 10 T oughne ss ( -) 64 55 64 ▼ 9 ▲ 9 T oughne ss ( + ) 5 6 12 ▲ 1 ▲ 6 ▲ 7 U ni que ne ss ( -) 75 69 53 ▼ 6 ▼ 16 ▼ 22 U ni que ne ss ( + ) 2 0 6 ▼ 2 ▲ 6 ▲ 4 N ot e. * ≤ .05, ** ≤ .01, *** ≤ .001 W ave 1 W ave 2 W ave 3 O ve rv ie w di st ri but ion of the pe rs onal ity di m ens ions ov er the thr ee w av es inc ludi ng t he s igni fic ant di ffe re nc es be tw ee n t he w av es bas ed on a one -w ay anal ys is of v ar ianc e
DUTCH POLITICAL PARTIES AND THEIR PERSONALITIES 34 Conclusion and Discussion
The aim of this study was to understand how Dutch voters differentiate between political brands based on political brand personality, whether it matters if voters differentiate political brands based on political brand personality and whether political brand personality can be changed. In this study, a consumer perspective was adapted, in which it was assumed that political parties are brands because “they act like brands to consumers” (Smith & French, 2009, p. 211). The main focus of this study was on the consumer brand knowledge, which is the meaning of a brand as it is stored in the memory of the consumer.
The focal point of this study is unique in this field of research. Whereas political parties are acknowledged to be brands, little research has been done on how voters use political brand personality to distinguish between parties, how political brand personality is associated with voting and whether political brand personality is stable over time. By studying brand personality from a consumer perspective, it is possible to answer these three questions. Discussion of Findings
Based on the distribution of the personality dimensions, it can be concluded that voters differentiate between political parties based on the political brand personality. For example, the VVD is associated with, on the one hand, being dishonest and being unreliable. This association might be caused by the various scandals party members of the VVD have faced in the last few years. On the other hand, the party has more mentions of personality traits related to the dimension Leadership, which indicates that voters perceive the VVD to be a successful and stable party. This might be a result of the party governing for quite a while now with a rather stable number of seats in parliament. Another interesting example of how voters differentiate between parties, is the CDA. On the one hand, the party is negatively associated with the personality dimension Image, which reflects being outdated and being opposed to everything. On the other hand, hearing the name of this party evoked significantly more
mentions of personality traits associated with the dimension Honesty. A party can thus have a personality profile with both positive and negative personality dimensions. One political party which has a positive personality profile, is GroenLinks. This party has significantly more mentions of personality traits associated with Spirited and Image, which indicates that the voters perceive the party to be young, energetic and cheerful. These personality dimensions may have been affected by the new young and spirited party leader of GroenLinks, Jesse Klaver.
It can be concluded that Dutch voters give political parties distinctive personality profiles based on their political brand personality. This conclusion is not surprising, considering the finding that brand personality can be an important aspect when consumers have to differentiate between brands (Freling & Forbes, 2005a).
The second question touched upon whether it matters that voters anthropomorphise political parties. Based on the results, it can be concluded that the likelihood to vote for a party is affected by political brand personality. The personality dimension Honesty appeared to be of especial influence for all parties: perceiving the brand as honest increased the
likelihood to vote whereas perceiving the brand as dishonest decreased the likelihood to vote. Other studies confirm the importance of Honesty and its related personality traits (e.g.
Gorbaniuk et al., 2015; Li et al., 2014). Being reliable is a very important factor for a political party and politicians, even more than for commercial brands (Smith, 2009). It is therefore not surprising to see that Honesty has an influence on the voter’s electoral decisions.
The third question asked whether personality dimensions are stable throughout the electoral campaign. It can be concluded that the mention of both positive and negative personality traits associated with Honesty change throughout the campaign. Negative personality traits associated with this dimension are becoming less frequent when the
DUTCH POLITICAL PARTIES AND THEIR PERSONALITIES 36 more frequent. This might indicate that voters put more emphasis on how honest a political brand is instead of mentioning how dishonest they perceive a brand to be.
Furthermore, personality traits associated with Image and Spirited become more prominent as consumer brand knowledge when the campaign develops. This might indicate that throughout the campaign, the voter puts more emphasis on how up-to-date a political party is. Based on this conclusion, it is recommended that strategists consider how positive personality traits are presented by the political brand during the electoral campaign, especially because it was found that positive personality dimensions can affect the likelihood to vote for the specified party. These results provide new arguments in the discussion of whether brand personality is stable or not and it allows to recommend further studies on how voters update the political brand personality throughout the campaign (Johar, Sengupta & Aaker, 2005). Limitations and Further Research
Transferring a brand personality scale which has been developed based on British politics to Dutch political parties was rather challenging. Certain personality traits did not seem to correspond with the consumer brand knowledge as provided by the respondents, which resulted in a broader interpretation of the personality dimensions by the coders. It is highly recommended that studies of consumer brand knowledge are conducted inductively in order to retrieve political brand dimensions which are applicable in the Netherlands, as it is expected that there are personality dimensions mentioned by the voters which could not be coded based on the political brand personality scale as provided by Smith (2009).
As discussed, it is expected that voters use brand personality as a shortcut to reduce the costs of information processing. However, this thesis did not focus on the question whether mentioning a personality trait was the result of peripheral processing of information. Processing of information can be discussed based on the various studies Petty and Cacioppo (1986) performed to formulate the Elaboration Likelihood Model (ELM). This model explains
that low-involved individuals are more likely to rely on heuristic cues whereas high-involved individuals are more likely to be persuaded based on the content of a message (Petty & Cacioppo, 1979). Based on this study, it would be expected that low-involved voters would rely on brand personality as a heuristic. It would thus be interesting to see whether the mention of personality traits is associated with lower levels of involvement, as would be expected based on the work of Petty and Cacioppo.
It would be highly interesting to study changes in political brand personality over a longer period of time. Electoral campaigns are not only limited to the few weeks before elections, but are becoming more permanent (Blumler & Kavanagh, 1999). This indicates that consumer brand knowledge is not only influenced during campaigns but that voters build on the political brand personality throughout the years. By having a better understanding of how political brand personality changes over a longer period of time, it might be possible gain a better understanding of how political brand personality is constructed.
Another interesting direction of research would be in mapping how voters differentiate between all Dutch political parties that competed in the elections based on their political brand personalities. This is especially interesting because political parties such as the VVD, PvdA and CDA are moving towards the centre (Pennings, 2005), which indicates that parties are becoming more similar. The trend that parties are becoming more identical highlights the importance of studying how the voters differentiate between political brands.
Implications
As mentioned, brand personality can be discussed from the perspective of the brand equity theory. This theory touches upon the effect of brand knowledge of a consumer (Keller, 1993). Keller discusses brand personality as consequence of user and usage image attributes. However, the brand equity theory does not seem to recognise brand personality as an
DUTCH POLITICAL PARTIES AND THEIR PERSONALITIES 38 recommended to reconsider how consumer brand knowledge is studied and to give a more prominent role to brand personality because it has been shown that brand personality affects how consumers react to political brands.
This study not only fuels a scientific debate about consumer brand knowledge but it also allows for practical advice for campaign strategists and political marketeers. First of all, voters differentiate based on political personality. In a crowded political spectrum, it is highly recommended to develop a positive and differentiating brand personality as it affects voting behaviour. Especially Honesty appeared to be of influence for each political party. Therefore, it is suggested to carefully consider how a party communicates its reliability, its honesty and its friendliness. Furthermore, by mapping political brand personality profiles, it became clear that the PVV, CDA, VVD and PvdA have negative personality profiles. It is highly
recommended to work on these negative profiles by communicating the party’s positive personality traits. Also, it is generally suggested to highlight the party’s positive personality traits as it positively impacts electoral outcomes.
To conclude: political brand personality has an impact on the likelihood to vote for a party and it allows voters to differentiate between parties. It is therefore highly recommended to construct a positive and distinguishing brand personality. Further studies are necessary to expand our understanding of how political brand personality persuades voters.
References
Aaker, D. (1991). Managing brand equity: Capitalizing on the value of a brand name. New York: The Free Press.
Aaker, D. (1996). Measuring brand equity across products and markets. California Management Review, 38(3), 102–120.
Aaker, J., Benet-Martínez, V. & Garoleza, J. (2001). Consumption symbols as carriers of culture: a study of Japanese and Spanish brand personality constructs. Journal of Personality and Social Psychology, 81(3), 492–508.
Akin, M. (2011). Predicting consumers’ behavioral intentions with perceptions of brand personality: A study in cell phone markets. International Journal of Business and Management, 6(6), 193–206. http://doi.org/10.5539/ijbm.v6n6p193
Banerjee, S. (2016). Influence of consumer personality, brand personality, and corporate personality on brand preference. Asia Pacific Journal of Marketing and Logistics, 28(2), 198–216. http://doi.org/10.1108/APJML-05-2015-0073
Behr, A., Bellgardt, E. & Rendtel, U. (2005). Extent and determinants of panel attrition in the European Community Household Panel. European Sociological Review, 21(5), 489–512. http://doi.org/10.1093/esr/jci037
Blumler, J. & Kavanagh, D. (1999). The Third Age of political communication: Influences and features. Political Communication, 16(3), 209–230.
http://doi.org/10.1080/105846099198596
Caprara, G., Barbaranelli, C. & Zimbardo, P. (1999). Personality profiles and political parties. Political Psychology, 20(1), 175–197.
CBS (2018). Bevolking: Kerncijfers. Retrieved 31-05-2018, from:
https://opendata.cbs.nl/statline/#/CBS/nl/dataset/37296ned/table?ts=1527759539455 Chen, C. (2001). Using free association to examine the relationship between the
