Explaining Personalization in Politics.
Is it strategy, fashionable or personality driven?
Sander Kunst, University of Amsterdam
Research Master Social Sciences
UvA ID: 10338160
Supervisor: Gijs Schumacher
Second Reader: Katjana Gattermann
Amsterdam, 16-06-2017
Abstract
Do politicians speak more about themselves and about their private lives? It is almost common wisdom nowadays that politics in Western Europe is becoming increasingly personalized. That is, political communication, the perception of it by voters and its coverage by the news media has been increasingly dominated by individuals rather than collectives such as political parties, ideology or policy. Most research up to now has focused on personalization in media reports. This way it fails to distinguish between two potential causes of personalization: 1) it is the media that drives personalization or 2) it is politicians that drive personalization. Focusing on the latter explanation, this paper will start from the premise that the extent to which politicians put themselves at the centre of their rhetoric is an indicator of personalization. We subsequently try to explain personalization: is it a specific political strategy, is it a time trend or is it the personality of the politician? To test the outlined theoretical expectations, automated text analysis has been conducted, using a unique dataset of Dutch and Danish party conference speeches between 1945 and 2015. Our results indicate that most variation in personalization is between individuals. There is a small, positive time trend, yet political strategic factors do not explain personalization.
Introduction
The role of political parties in Western Europe has undergone fundamental changes since the end of World War II. Depillarization, secularization and individualization have severely transformed the position of political parties in society. Whereas up to the 1970s mass political parties were the main representatives of well-defined groups in society, political parties are said to be increasingly ‘ruling the void’ ever since (Mair, 2013). The decline in relevance of political parties has arguably shifted power and importance from collectives, such as parties, to individual politicians over time (McAllister, 2009; Garzia, 2011).
The decline in importance of parties at the expense of individuals is called “personalization” (Rahat & Sheafer, 2007; Van Santen & Van Zoonen, 2009). According to scholars, personalization is increasingly affecting Western democracies, pointing to the media covering more individual politicians than parties and voters basing their choice predominantly on a politician’s personal traits than a party program (McAllister, 2009; Garzia, 2011; Caprara & Zimbardo, 2004). Personalization is usually not considered by scholars as desirable for democracy since the increased attention for personalities instead of parties is assumed to hamper the rational decision-making process of voters (Adam & Maier, 2010). Instead of party programs and current issues, voters arguably base their vote solely on the achievements and standpoints of individual politicians (Ibid.). This diminishes the importance of competence and professional capacities in the decision-making process. Consequently, personalization dumbs-down and reduces the complexity of the political discussion and turns elections into superficial popularity contests (Ibid.). Although politicians have allegedly become more important for the press and voters than parties, it is unclear how politicians themselves respond to these developments. In terms of their own communication, we can wonder if they also communicate more from their own personal stance than before. Hence, have politicians also put more emphasis on themselves in their communication than the party they represent? Furthermore, what could account for differences in personalised language use between politicians? Is it fashionable, strategy or personality-driven?
This paper will argue that personalization of party leaders’ communication contains two different elements: talking more about themselves instead of the collective they represent (individualization) and/or referring more to their private life (privatization). By applying a longitudinal approach, it is possible to observe and map trends over time. Using a novel dataset of Dutch and Danish party conference speeches between 1945 and 2015, this paper will argue that personalization is dependent on time, but also the individual preferences of the speaker are of great importance for a politician’s communication style.
The paper is structured as followed. First, a theoretical framework is constructed and the tested hypotheses are lined out. Next, the used research design is explained in more detail. Thereafter, by means of OLS regression the hypotheses are tested, followed by a conclusion and discussion of the results.
Theory & Hypotheses
What is personalization?Personalization is a trend or process over time, characterized by a decline in relevance of collective institutions (e.g. political parties), while leaders of these collective institutions have become more important in the eyes of the press and the public (Poguntke & Webb, 2005; Rahat & Sheafer, 2007; Karvonen, 2010; Adam & Maier, 2010; Balmas et al., 2014). While personalization seems as a straightforward concept, the literature on personalization is not univocal on how to approach it. Although most scholars seem to agree on the minimal definition of personalization as a trend over time with a larger focus on personality instead of collective institutions, the field has been hampered by a lack of clarity on what personalization exactly entails (Adam & Maier, 2010). This lack of clarity has in some cases resulted in contradicting results (See Rahat & Sheafer, 2007). However, in recent years more detailed conceptualization of different aspects of personalization have emerged (Rahat & Sheafer, 2007; Adam & Maier, 2010; Van Aelst et al., 2011; Balmas et al., 2014). In particular, institutional, media and behavioural aspects of personalization are distinguished. These are explained in more detail below.
Three general types of personalization
First, institutional personalization entails “the adoption of rules, mechanisms, and institutions that put more emphasis on the individual politician and less on political groups and parties” (Rahat & Sheafer, 2007: 207). For example, politicians in electoral systems with personal mandates (e.g presidential systems) have more incentives to personalize their campaigns than politicians who are elected from party lists (McAllister, 2009). Also in general, leaders have become more exposed to public attention (King, 2002; Poguntke & Webb, 2005; Wattenberg, 1991). In European parliamentary systems reforms have been passed, such as increasing the weight of preference voting on the final result or the number of intraparty choices, which have pushed electoral systems from party-centrism towards candidate-centrism (Renwick & Pilet, 2011). Also organizational changes within parties, such as the increasingly popular introduction of party leader primaries (Kenig, 2009; Pilet & Cross, 2014) motivate politicians to develop their personal image (Rahat & Sheafer, 2007).
Second, media personalization encompasses two facets: 1) increased media coverage of individuals instead of parties, labelled as individualization and 2) portraying politicians as “ordinary citizens” instead as purely policymakers, referred to as privatization (Rahat & Sheafer, 2007; Van Aelst et al., 2011). Instead of focussing on their policy ideas and solutions, the media has arguably shifted its interest to the private lives of politicians, such as family, personal interests, hobbies and scandals. Politicians today need to be more than policy hacks, they also need to showcase their favourite music or intimate details of their family life (Van Aelst et al., 2011). This is not only because the media forces them to. Politicians also show more of the person behind the politicians as a strategy or opportunity to create sympathy and to connect with voters to show that they are “just like them” (Holtz-Bacha, 2004: 49). In other words, politicians use their private life for electoral gains.
Although there is already a substantive body of literature which focusses on personalization in the media, the empirical evidence for personalization in the media is at best mixed (Kriesi, 2012; Van Aelst et al., 2012). This is mostly because personalization is not a universal trend. Based on more recent cross-country longitudinal studies, personalization seems to vary strongly across countries. For example, the strongest evidence for media personalization is provided by Kriesi (2012) who can only find a general trend towards more personalization in the Netherlands in his comparative study of 6 European countries (France, UK, Germany, Austria, The Netherlands, Switzerland). The different levels in media personalization are dependent on the country-specific context such as the electoral system and regime type. Moreover, some types of media, such as tabloid newspapers and infotainment news shows, spend more time on human-interest stories because of their sensationalistic angle, showing higher levels of personalization (Adam & Maier, 2010; Vliegenthart et al., 2011).
With more personalization, also voters may perceive elections more as a competition between individuals rather than political parties (Rahat & Sheafer, 2007). In that case voters may be increasingly swayed by the politicians’ personality instead of his/her policies (Caprara & Zimbardo, 2004; Caprara et al., 2006; Caprara, 2007; Bittner, 2011; Garzia, 2011). More specifically, voters vote for a politician because they experience a match between their own personality traits and those of the politician. In other words, people are more likely to support a candidate who mirrors their own traits and personality (Caprara & Zimbardo, 2004). This of course speaks directly to the normative assumptions of personalization, in which voters are presumed to increasingly base their vote on a ‘feeling’ instead of a ‘rational’ consideration.
Media personalization may occur because journalists produces news with a more personal and human-interest touch than before. But do politicians themselves also communicate their
personal stories more than before? This third aspect of personalization - behavioural personalization - refers to politicians (in general) and leaders (more specifically) becoming and acting increasingly independent from their political party (Rahat & Sheafer, 2007; Poguntke & Webb, 2005; Michels, 1915; Katz & Mair, 1995). One of the possible elements of behavioural personalization is their language use. It has become apparent leaders nowadays increasingly need to be able to talk both about policy and their own personal life (Holtz-Bacha & Van Zoonen, 2000). Since people are increasingly connected to a party because of the leader than the party program (Caprara & Zimbardo, 2004; Garzia, 2011), politicians have invested in creating a more personal and emotional bond between leader and constituency (Holtz-Bacha, 2004; Garzia, 2011). By using more personal language, leaders can try to appear more ordinary and familiar. Being aware of the fact that people prefer to hear about persons rather than party programs, politicians can adapt their communication to this public preference. Therefore, by striking a more personal tone, leaders can “humanise” themselves and appear to be “just like anybody else” (Holtz-Bacha, 2004: 49).
Almost no other existing study looks at politicians’ personalization of language use (for exception see Pettitt, 2012). Political communication of leaders could become more personalised in two ways: 1) referring more to themselves instead of their collective (individualization) and/or 2) talking more about their private life (privatization). If politics is becoming more personalised, leadership rhetoric would also need to become more personalised. Most studies on personalization argue it is because of changing organizational settings of political parties, which place leaders in a more dominant position in their party (Poguntke & Webb, 2005; Garzia, 2011; Katz & Mair, 1995). On the other hand, based on a case study of Israel, Rahat & Sheafer (2007) argue in favour of a causal chain model, claiming that personalization in the institutional setting leads to the media covering more individuals, which pushes politicians to become more individualistic in their behaviour. Therefore, there are several possible factors which could contribute to behavioural personalization of politicians, prompting them to refer more to themselves than their collective.
Behavioural personalization: When do politicians get personal?
The literature on personalization argues personalization is a trend over time (Rahat & Sheafer, 2007; Karvonen, 2010). This is primarily because personalization has been linked to the growing individualization of Western societies (Garzia, 2011). People increasingly see themselves foremost as individuals instead of being part of a collective identity (Bauman, 2001).
Figure 1: Use of singular and plural pronouns over time (Google NGram)
Individualization could also affect the way people (in general) and leaders (specifically) communicate. Using Google Ngram, it is possible to see a change in language use over time. Google Ngram consists of word lists of 4% of all books ever printed in the English language (Michel et al., 2011). Figure 1 shows the use of singular and plural pronouns. The graph shows the percentage of plural pronouns seems to remain constant, while the percentage of singular pronouns seem to take off half-way the 1970s. Based on this graph it seems it also has become more common to use singular pronouns. Therefore, politicians will have over time used more personalization in their political rhetoric. This leads to the first hypothesis:
H1: Over time, leaders will use more personalized language
One of the most profound institutional changes due to individualization, depillarization and secularization in the last few decades is the role political parties fulfil within society. The erosion of cleavage politics has caused tremendous changes to the ways parties organize and operate (Katz & Mair, 1995; Garzia, 2011). In the so-called “Golden Age” of party democracy, political parties represented well-defined groups within society (Katz & Mair, 1995). Voters were linked to these parties based on their socio-demographic background (Lipset & Rokkan, 1967; Mair, 2008).
These mass parties acted as the main political representative for each collective identity. As parties were the main link between civil society and the state, they formulated and defended the shared interests of each collective (Duverger, 1964). Furthermore, people were connected to the party via the interests it represented, not the leadership. As a result, parties were essentially organized from the bottom-up, which restricted leaders in their exercise of power (Katz & Mair, 1995).
At the same time, in almost every European country, partisanship and party loyalty has declined since the 1970s (Franklin et al., 1992; Dalton & Wattenberg, 2002; Franklin, 2010). Due to different socio-economical and socio-cultural processes, groups became more heterogeneous, making it harder to identify collective identities. Voters have experienced “a shift away from a style of electoral decision-making based on social group and/or party cues toward a more individualized and inwardly oriented style of political choice” (Dalton, 1996: 346). People feel less attached to one specific party, but act more as political consumers, becoming less predictable. Because of this changing environment in which parties cannot fall back on well-defined constituencies, mass parties needed to readapt to expand their electoral appeal beyond the group they traditionally represented (Mair, 2008; Garzia, 2011). Consequently, parties turned into so-called catch-all parties to attract a broader segment of society (Kirchheimer, 1966; Katz & Mair, 1995). To cater to the many different interests within society, parties started to ‘shed their ideological feathers’ and converged to the centre.
Since the relative differences between parties have become much smaller, it is argued that politics is increasingly about valence issues (Stokes, 1963; Stokes, 1992; Clarke & Whitten, 2009). Valence issues are those on which nearly everyone agrees: good healthcare, strong economy, less crime etc. Therefore, vote choice is argued to be less about which party shares voters’ ideological positions, but more about who they think is most competent at achieving these widely-shared goals. To decide who is most capable, voters make judgements about the party leadership (Clarke & Whitten, 2009; Caprara & Zimbardo, 2004). As party leaders have become more important for the electoral success of the party, parties have increasingly become top-steered with only minor space for input by members (Katz & Mair, 1995).
As differences between parties have become smaller (Mair, 2008), there is more room for the party leadership to be decisive (Garzia, 2011). Since leaders are becoming more important for the final electoral result of the party, we would expect leaders to try and create a more personal bond with voters. Consequently, we would expect leaders to adopt a more personalized style, talking more about themselves than their party. This leads to the second hypothesis:
H2: As the ideological proximity of a party to the centre becomes smaller, personalization will increase.
To counter the growing influence of the party leadership, parties have implemented different measures to give members a larger say in party matters. One of those antidotes is widening the selectorate by giving members a direct vote in internal leadership elections. This method is adopted by a growing number of parties (Kenig, 2009; Pilet & Cross, 2014). However, this idea has been heavily criticized, because it would shift power away from knowledgeable and involved delegates to uniformed and uninterested members (Katz & Mair, 1995; Blyth & Katz, 2005, Schumacher & Giger, 2017). Therefore, there is an inherent danger in democratizing the selection of the party leader for party discipline (Rahat & Hazan, 2001). Leaders who have been elected by a more inclusive selectorate “owe their loyalty to their voters in the candidate selection process, and not only to their party. Such candidates are no longer assured of a future in politics by being loyal team players; instead they must stand out and be recognized - not by the party leaders but by their inclusive selectorate” (Ibid., 2001: 313).
Thus, this new way of selecting leaders forces candidates to adopt a more personal style of politics, resulting in “a tendency to act in a manner that largely disregards the group(s) with which the candidate is associated” (Ibid., 2001: 313). Because of this shifting loyalty and the pressure to showcase their accomplishments and vision, leaders will refer, as a consequence, more to themselves than their party. This leads to the third hypothesis:
H3: Leaders who have been elected via primaries will use more personalization.
Besides more profound structural institutional changes, also short-term changes can change the style used by the leader. Previous research showed that poor performance of parties can cause change to the specific issues parties emphasise and/or their position on them (Budge, 1994; Somer-Topcu, 2009; Riker, 1982; Carmines & Stimson, 1989). Parties who have lost during an election or end up in the opposition have received a clear signal from the public that their party platform is not resonating. To change the dynamic of party competition, opposition parties will introduce new issues (Riker, 1982; Riker, 1986). On the other hand, government parties know that their platform works and will be hesitant to change it. To consolidate its position, government parties will stick to their ‘winning’ platform, as changing it bears too much risk losing its current electoral support (Carmines & Stimson, 1989). Thus, opposition parties will likely undergo a change, while government parties stick to their guns.
This motivation may not only affect party platforms, it could also impact their general strategy, including the way they seek to communicate with voters. Consequently, leaders could change their style of communicating to attract voters (e.g. putting themselves more at the centre of their message). As noted, leadership images are important for voters (Bittner, 2002; Garzia, 2011). This is expected to be more important for opposition parties. Government parties put a larger emphasis on past achievements, attack opponents to a lesser extent and highlight their top-candidate less than opposition parties (Dolezal et al., 2016). Therefore, parties who are in opposition could, in response to their electoral defeat, change their strategy, resulting in a larger emphasis on their top candidate. As a result, leaders of opposition parties will most likely refer more to themselves than their parties, compared to leaders of government parties. This leads to the fourth hypothesis:
H4: Leaders from opposition parties will use more personalization compared to leaders of government parties.
Up until now, changes in language use over time as well as institutional and strategic factors have been considered to explain personalization. These are the typical explanations of personalization in the political science literature. The latter part of this section turns to the psychological literature which perceives language use as more of an individual trait rather than something that is strategically manipulated. The personality of the speaker could influence his or her language use. Research on personality in politics predominantly focusses on the Big Five personality traits: openness to experience, extraversion, conscientiousness, neuroticism and agreeableness. Psychologists studying personality in politics found many patterns between Big Five personality traits and language habits (see for example Pennebaker & King, 1999 or Pennebaker, 2011). For example, people high on neuroticism are more likely to use the first-person singular in their communication and evoke more negative emotions (Pennebaker & King, 1999). On the other hand, people high on openness to experience use longer words and use more tentative words.
Besides personality, the use of pronouns can also fluctuate because these words have a certain function. By talking more in terms of ‘I’, ‘My’ and ‘Mine’, it is a way for politicians to emphasize their attachment to the topic they are talking about (Pennebaker, 2011). The use of singular pronouns has featured heavily in online debates during the presidency of Obama, mocking his campaign slogan as “Yes I Can”, because of his high rate of self-referencing (Pettitt, 2012). Moreover, the intensive use of first person pronouns by now-president Donald Trump in the presidential campaign of 2016 has often been linked to narcissism (Ahmadian et al., 2017). In his analysis of conference speeches by Labour party leaders, Pettitt (2012: 132) argues the
difference in rhetoric is informed by how leaders perceive the role of individual in the political domain to be: “Brown and Kinnock are examples of collectivists who put themselves in the background rhetorically. Brown changed his rhetoric when he became leader, which seems to have gone against his basic political personality, and indeed he was never comfortable in the role as party leader and Prime Minister (…) Foot was the eternal supporter of the radical and the dissenter and hence consistently used a very self-referential rhetoric”. Also, American President Lincoln used the first-person plural more often as he did not believe in individualistic politics (Field, 2011: 51).
Therefore, personalization is dependent on the preferences in rhetoric by the individual leaders, informed by their personality. This difference will result in some leaders being more self-referential than others. This results in the fifth hypothesis:
H5: The level of personalization is driven by the preferences of the individual leader
Research Design
Case SelectionThe study will cover two countries, the Netherlands and Denmark between 1945 and 2013. Because of their strong party parliamentary tradition, both present a least likely case for personalization to occur (McAllister, 2009). However, the personalization-thesis predicts personalization to also occur in parliamentary democracies (Kriesi, 2012). Thus, if personalization is present, this should also be visible in The Netherlands and Denmark. Signs of media personalization are especially present in the Netherlands. (Ibid., 2012). However, this evidence is not conclusive, since Vliegenthart et al. (2010) did not find a clear trend towards more personalized press coverage in the Netherlands. In an analysis to Dutch election posters, Vliegenthart (2012) finds an increasing presence and prominence of party leaders and the party logo, and a decreasing focus on ideology. This means personalization can also be observed in more subtle ways. In the following sections the methods, data and operationalization of the variables will be discussed in more detail.
Data
To test the outlined hypothesis, we use a novel dataset consisting of Dutch and Danish party conference speeches between 1945 and 20151. Each speech is treated as a unit of analysis.
1 Most of the data was collected by Gijs Schumacher in context of the Party Conference Research Group.
Table 1: Overview of included speeches
Because of the timespan, it is possible to observe and map trends over time between parties. Party conferences are traditionally intended to approve and explain the decisions made by the party elite. Moreover, it gives an opportunity to get interaction with members and to determine the course of the party. Therefore, these conferences attract a variety of different people: from party elites to grassroots members. The highlight of these gatherings are usually the speeches made by the party leadership. However, increasingly these congresses do not only serve as a check on the party leadership. Because of the media attention these meetings generate, parties know portions of the speech will be broadcasted, creating visibility to a wider audience. As a result, conferences are increasingly used by parties to showcase their unity to voters. Table 1 presents the speeches included for each country and each party.
Dependent Variable Personalization
The dependent variable for this study is behavioural personalization. In this paper, behavioural personalization is considered to be a style or strategy used by political figures. Talking more about themselves instead of the collective is a way of establishing a more personal and emotional bond between leader and followers. This involves referring more to themselves than their collective (individualization) and/or talking more about their private life (privatization)
To measure behavioural personalization, it is operationalized in two ways. Individualization is operationalized as the ratio between the counted number of self-references in a speech (I, me, mine) and the number of references to a collective the speaker represents (we,
The Netherlands Denmark
Party # Years Party # Years
ARP 74 1946-1970 DF 33 1997-2014 CDA 72 1977-2013 EL 15 1989-2014 CHU 2 1975 SD 228 1945-2006 CPN 6 1946-1988 SF 56 1970-2000 D66 89 1966-2011 V 201 1956-2014 GroenLinks 27 1990-2013 533 KVP 46 1951-1980 PvdA 174 1946-2013 PVV 9 2005-2012 SP 15 1998-2012 VVD 113 1949-2012 627
us, our). More elaborate denominators have been tried, including party names, but these all had correlations of around 0.9 with the used denominator (see Appendix A). Consequently, a higher ratio indicates more personalization. Privatization is operationalised as the ratio of word groups such as “my family”, “my childhood”, “my wife” etc to the total number of words in the speech. In that way, we can consider whether individualization and privatization have increased over time A logarithmic function is applied, creating a more normally distributed dependent variable.
The ratios for individualization and privatization is obtained using a simple dictionary method, implemented in the R Package Quanteda (Benoit et al., 2017). Using this dictionary, the computer counts the words, resulting in less a labour-intensive task. Moreover, because the computer counts the words for the researcher, the chance of counting mistakes is close to zero, resulting in a higher reliability contrary to relying on human coders.
Independent Variables
Distance to ideological centre
Distance to the ideological centre is measured by using data from the Manifesto Group (MG) Manifestos are the most authoritative statements through which parties communicate their policy preferences to citizens at each campaign (Bäck et al., 2011). The MG approach relies on the sentence-by-sentence coding of manifestos. The MG method does not consider the position parties take on issues, but only the salience they attach to them. For each party the left-right positions is determined by means of summing and subtracting different issue categories, placing parties on scale from -100 (extreme left) to +100 (extreme right)2 (Laver & Budge, 1992). Next, for each election year the centre of gravity of the party system is determined by using the formula provided by Gross & Sigelman (1984). The centre of gravity is calculated as:
(𝑉# 𝑇
%
#&'
∙ 𝑝# )
with T as the sum of the vote share at the election, V a party’s vote share and p a party’s left-right position. For each speech of every party at each election the distance has been calculated between
2(per104 + per201 + per203 + per305 + per401 + per402 + per407 + per414 + per505 + per601 + per603 +
per605 + per606) - (per103 + per105 + per106 + per107 + per403 + per404 + per406 + per412 + per413 + per504
+ per506 + per701 + per202).
its left-right position and the overall centre of gravity. Because this measure can only be done for election years, the measure is used for follow-up years as well until a new election takes place. A possible point of criticism of this method is the idea that parties already start to shift positions between elections. However, parties conform themselves to the party manifesto as their policy basis for the next 4 years. As a result, dramatic shifts in parties’ left-right position between elections are unlikely.
Government – Opposition
A dummy variable indicated whether the speech was delivered during time in opposition or in government.
Primaries
For each speech, a dummy variable indicates whether the speaker was appointed to a leadership position by means of the one-member one-vote principle (a primary) or by a different method, such as a vote by delegates at a party conference, or an appointment by a specific party council (Giger & Schumacher, 2017). For many parties, leadership primaries have only been introduced in the last few years. Only D66 in the Netherlands already has had leadership primaries since the 1960s.
Preferences of the speaker
To test whether the speakers’ own preferences are of influence on the rate of personalization in a speech, a dummy variable for each individual speaker is created.
Does personalization increase over time?
Before considering which factors have contributed to personalization of politicians’ language use, the time trend will first be examined by looking both at the levels of individualization and privatization in the Netherlands and Denmark. Figure 2 shows the level of individualization (ratio of self-references to collective references) per decade in the Netherlands and Denmark.
To get a better insight into what causes changes to the ratio of self-references to collective-references, both elements of the ratio need to be further examined. In Figure 3 the ratio of self-references to the share of total words per decade for the Netherlands and Denmark is displayed. As can be seen in the figures, for the Netherlands there is not a clear-cut increase in the share of self-references over time. For Denmark, the share of self-referencing sharply decreases after the 1960s. Since the 2010s, the share of self-referencing goes up again.
Figure 2: Individualization per decade
The Netherlands Denmark
Note: Dots are means, bars are 95% confidence intervals. I performed t-tests comparing personalization scores in decade to decade t-1; * = p<0,1; ** = p<0.05; *** = p <0.01,
Figure 3: Ratio of self-references per decade
The Netherlands Denmark
Note: Dots are means, bars are 95% confidence intervals. I performed t-tests comparing personalization scores in decade to decade t-1; * = p<0,1; ** = p<0.05; *** = p <0.01,
Figure 4 shows the ratio of collective-references to the total number of words per decade. For the Netherlands, the share of collective references has decreased from the 1960s up until the 00s, after which it rose again since the ‘00s. Consequently, an increase of individualization in leadership speeches over time in the Netherlands is predominantly a result of a decline in the share of collective references. For Denmark, the share of references seems to be increasing steadily in the last few decades. Hence, the decline in the individualization ratio seems to be primarily because of a constant increase in collective references.
The second element of personalization is privatization, meaning politicians refer more to their own private life. In Figure 5, the ratio of references of word groups referencing to the politicians’ private life to the total numbers of words are displayed per decade. As can be seen, in both countries there is an increase in privatization. Therefore, in the Netherlands there is both an increase in individualization as well as privatization, while for the Denmark only privatization has increased over time.
Figure 4: Ratio of collective references per decade
The Netherlands Denmark
Note: Dots are means, bars are 95% confidence intervals. I performed t-tests comparing personalization scores in decade to decade t-1; * = p<0,1; ** = p<0.05; *** = p <0.01,
Figure 5: Privatization per decade
The Netherlands Denmark
Note: Dots are means, bars are 95% confidence intervals. I performed t-tests comparing personalization scores in decade to decade t-1; * = p<0,1; ** = p<0.05; *** = p <0.01,
Do institutional, strategic and personal characteristics explain
personalization?
Individualization
The next step is to find out which factors besides time caused changes to the level of personalization, thus individualization and privatization. Four regression analysis will be conducted with individualization and privatization as the dependent variables.
The results for the regression analysis with individualization as the dependent variable for the Netherlands are displayed in Table 2. In the first model the distance of the speaker’s party to the ideological centre is added, resulting in a significant coefficient. This means when parties move closer to the centre, individualization will increase. In the second model, whether a speaker has been elected by a primary is added to the model. This also yields a significant result, meaning leaders who have been elected to their position by means of a primary use more individualization in their speeches. In the third model, whether a speaker was part of the government or the opposition is included. This yields an insignificant result, meaning there is no difference in the level of individualization between speakers in the opposition and in government. In the fourth model, time is added as a covariate. The significant positive coefficient confirms what has been shown earlier: over time, speakers refer more to themselves. In the final model, we also control
for the personality of the speaker by including dummies for each politician. This gives a much higher adjusted-R2 of 28%! Hence, the personal language preferences of the speaker explains much variation.
Table 2: Regression analysis Individualization for The Netherlands Dependent variable: Individualization (1) (2) (3) (4) (5) Distance to Centre -0.006*** -0.005*** -0.005*** -0.003* 0.001 (0.001) (0.001) (0.001) (0.002) (0.003) Elected by Primary 0.178*** 0.182*** 0.140** -0.200 (0.043) (0.043) (0.044) (0.112) Gov.- Opp. 0.024 0.044 0.031 (0.028) (0.028) (0.039) Time 0.003*** 0.007** (0.001) (0.003) Party Leader 0.079** 0.027 0.031 0.020 0.051 (0.027) (0.030) (0.030) (0.030) (0.059) Constant 0.552*** 0.530*** 0.509*** 0.370*** 0.193 (0.026) (0.026) (0.035) (0.049) (0.181)
Speaker dummies? No No No No Yes
Observations 625 625 625 625 625
R2 0.047 0.073 0.074 0.098 0.539
Adjusted R2 0.044 0.069 0.068 0.091 0.285
Moreover, time remains significant, signalling politicians have also over time become more self-referential. This result can be interpreted in two ways: (1) politicians have become more self-referential regardless of personality or (2) more self-referential politicians emerge in top positions now than before.
Table 3: Regression analysis Individualization for Denmark Dependent variable: Individualization (1) (2) (3) (4) (5) Distance to Centre -0.0001 0.0005 0.0002 0.0004 -0.0001 (0.001) (0.001) (0.001) (0.001) (0.001) Elected by Primary -0.191** -0.196** -0.185** 0.071 (0.070) (0.071) (0.070) (0.130) Gov.- Opp. -0.017 -0.041 -0.019 (0.025) (0.025) (0.028) Time -0.003*** -0.004*** (0.001) (0.001) Party leader 0.047 0.077* 0.075* 0.051 -0.021 (0.028) (0.030) (0.030) (0.030) (0.043) Constant 0.235*** 0.221*** 0.235*** 0.375*** 0.534** (0.029) (0.029) (0.035) (0.051) (0.202)
Speaker dummies? No No No No Yes
Observations 533 533 533 533 533
R2 0.005 0.019 0.020 0.045 0.666
Adjusted R2 0.002 0.014 0.013 0.036 0.460
The results for individualization as dependent variable for Denmark are displayed in Table 3. In the first model, distance to centre is added, resulting in an insignificant result. The inclusion of whether the speaker was elected by means of a primary, results is a negative significant result. Hence, party leaders in Denmark who have been elected to that position via a primary use less personalization. The inclusion in the third model of whether the party of the speaker was in government or opposition at the time yields an insignificant result. As seen before, in the fourth model, time is a significant predicator, confirming over time personalization in Denmark has decreased. Hence, political leaders have started to talk more in collective terms and less about themselves. In the final model, a dummy variable for each speaker is included. This raises the adjusted-R2 to 0,46! As was the case in the Netherlands, the preferences of the speaker explain the most variation in the level of personalization. Moreover, whether the speaker was elected via a direct primary is now also insignificant, hence there is no difference. The same dilemma as in the Netherlands is also present in Denmark, but the other way around. Hence (1) Danish politicians have become less self-referential regardless of personality or (2) more collective-referential politicians emerge in top positions now than before.
Privatization
The results for the Netherlands with privatization as the dependent variable is displayed in Table 4. In model 1, the distance to the centre is added, resulting in a significant coefficient. In model 2, elected by means of primary is added, confirming leaders who have been elected via a primary make more references to their private life. Surprisingly, in model 3 it becomes clear leaders from government parties talk less about their private life than leaders from opposition parties. In model 4, time is added as a covariate, rendering all other variables insignificant. The positive coefficient indicates privatization has gone up over time. However, when the dummy variables for each speaker, nothing is significant anymore. The increased adjusted – R2 confirms this model is a far better fit. Hence, this confirms how also for privatization the personality of the speaker is most important.
The result for Denmark are displayed in Table 5. Distance to centre, elected by primary and whether the party was in opposition or government does not have an impact according to the first 3 models. This even results in a negative Adjusted R2, confirming a very bad model fit. The inclusion of time in model 4 yields a significant result, indicating privatization has increased in Denmark over the years. Even when the dummies for each speaker are included in model 5, time is still significant. The sharp increase in the adjusted R2 point to a much better model fit.
Thus, also for Denmark the personality of the speaker has a large impact on whether people refer to their private life.
Table 4: Regression analysis Privatization for The Netherlands Dependent variable: Privatization (1) (2) (3) (4) (5) Distance to Centre -0.00002*** -0.00001** -0.00002*** -0.00001 -0.00000 (0.00000) (0.00000) (0.00000) (0.00000) (0.00001) Elected by Primary 0.0003* 0.0003* 0.0001 0.0003 (0.0001) (0.0001) (0.0001) (0.0004) Gov.- Opp. -0.0002* -0.0001 -0.0001 (0.0001) (0.0001) (0.0001) Time 0.00001*** 0.00000 (0.00000) (0.00001) Party Leader 0.0002* 0.0001 0.0001 0.00003 0.0002 (0.0001) (0.0001) (0.0001) (0.0001) (0.0002) Constant 0.001*** 0.001*** 0.001*** 0.0001 0.0003 (0.0001) (0.0001) (0.0001) (0.0002) (0.001)
Speaker dummies? No No No No Yes
Observations 625 625 625 625 625
R2 0.031 0.039 0.048 0.088 0.518
Adjusted R2 0.028 0.035 0.042 0.081 0.251
Table 5: Regression analysis Privatization for Denmark Dependent variable: Privatization (1) (2) (3) (4) (5) Distance to Centre 0.00000 0.00000 0.00000 0.00000 -0.00000 (0.00000) (0.00000) (0.00000) (0.00000) (0.00001) Elected by Primary -0.0002 -0.0002 -0.0002 0.001 (0.0003) (0.0003) (0.0003) (0.001) Gov.- Opp. -0.0001 0.00005 -0.00003 (0.0001) (0.0001) (0.0001) Time . 0.00001*** 0.00002** (0.00000) (0.00001) Party Leader 0.0001 0.0001 0.0001 0.0002 0.0002 (0.0001) (0.0001) (0.0001) (0.0001) (0.0002) Constant 0.0003** 0.0003** 0.0004* -0.0003 0.002* (0.0001) (0.0001) (0.0001) (0.0002) (0.001)
Speaker dummies? No No No No Yes
Observations 533 533 533 533 533
R2 0.002 0.003 0.003 0.037 0.553
Adjusted R2 -0.002 -0.003 -0.004 0.028 0.278
Note: Standard error between parenthesis; *p<0.05, **p<0.01,***p<0.001
Survival of the fittest?
For both the Netherlands and Denmark there seems to be considerable difference between speakers when looking at individualization and privatization. To test this further, F-tests are conducted. The F-test compares the between variance to the within variance, hence, a F-value higher than 1 indicates there is more variance between than within groups. In case of the Netherlands, for individualization (F = 2.07; p = 0.00) and privatization (F = 1.96; p = 0.00)
there is considerably more variance between speakers than within each speaker. Also for Denmark there is more variance between than within speakers for individualization (F = 3.16; p = 0.00) and privatization (F = 1.95; p = 0.00).
This can be illustrated further by taking a closer look at the party leaders in the Netherlands. In contrast to Denmark, politicians in the Netherlands have become more self-referential in their communication over time. How can this be explained? An important figure in the post-WW II period is Willem Drees (PvdA). In the analysis, Drees has an Individualization mean of 0.21, indicating he uses a lot of collective terms in relation to more self-referential terms. It is well-known Drees lived a very simple and humble live. By people close to him he was described as boring and a clerk (Daalder et al., 1988). Furthermore, Drees believed strongly in the ideals of social democracy, putting large efforts into creating a better life for the poor. Therefore, Drees can be a described as a collectivist who believed he was the servant of an ideology. This is what we see for other post-WW II leaders in the Netherlands as well. Being subordinate to an ideology or ideal is what is also ascribed to early leaders of the ARP, such as Schouten (0.14) and Berghuis (0.47) (Bosscher, 2013; Drexhage-Zijlstra, 2013). Also, one of the most prominent faces of Communist Party after the war, Paul de Groot (0.05), who was a fierce Stalinist, dreaming of communist utopia in the Netherlands is clearly putting ideology first (Stutje, 2000). What seems to unite many of the figures in Dutch politics just after the end of World War II is the infinite belief in an ideal, putting themselves rhetorically in the background (Voerman, 2007).
This seems to have changed dramatically in recent decades. As Voerman (2007) argues, although there is still much doubt within political communication about whether personalization is actually taking place, politicians increasingly believe it is. For example, D66-leader Van Mierlo (1.19) argued in 1991 political parties should transform to platforms via which individuals could spread and discuss their ideas (Voerman, 2007). As a self-fulfilling prophecy, the pressure is on politicians to showcase themselves more to voters. A prominent example of this idea is PvdA-leader Wouter Bos (1.23). Bos never made it a secret he believed politicians should show more of their personal side to strengthen democracy and to close the gap between politicians and voters (Bos, 2005). Although his successor Cohen (0.63) was far more reserved and did not feel at ease in a personality focussed environment, which some consider to be the reason for his quick downfall (Willems, 2012), Diederik Samsom (1.39) took on a different approach and embraced personalization to show voters ‘who they are voting for’ (Van den Berg, 2012). Very telling for this general trend is the large increase in books published by politicians around elections (Voerman, 2007). From averaging 1 book every 3 years in 1965 to nowadays averaging 3 books every year since 1965. For example, current leaders such as Buma (0.80), Roemer (1.43) and
Pechtold (0.90) all have their own book. These books are predominantly used to show the person behind the politician (Voerman, 2007). Therefore, it seems Dutch politics is perhaps turning into a ‘survival of the fittest’ scenario, in which politicians can only be successful if they feel comfortable to involve their own persona in the political process.
Conclusion
The declining relevance of political parties has arguably shifted power and importance to individual politicians and leaders, leading to so-called “personalization of politics”. This paper has looked deeper into possible personalization in the used language by party leaders. By arguing personalization of leadership communication consists of two different elements, individualization and privatization, it has become clear personalization is not a universal (positive) trend.
By using a novel dataset of Dutch and Danish conference speeches between 1945 and 2015, this paper confirmed although politics might get more personalised, this is not directly visible in the language use of politicians. For example, although individualization in The Netherlands has increased over time, in Denmark it actually decreased. What could explain the large increase in Denmark in collective terms remains unclear. Perhaps the Netherlands is already a more individualised society than Denmark. The exact reason remains unclear and should therefore be looked into in future research.
For both countries, political strategic factors did not have an impact on politicians’ language use. However, what seemed to be of great importance for both countries in terms of determents of language use is the preference of the speaker. In the end, language use is a personal issue which cannot be easily manipulated by changing institutional settings or strategic considerations. Therefore, whether a politician likes to talk about themselves or their private life is in the end dependent on whether the speaker is comfortable with it. However, as was shown in case of the Netherlands, it might be possible we are facing a ‘survival of the fittest’ scenario. Although scholars are still in doubt about whether personalization is taking place, the idea has spread in society that it is. As politicians feel a pressure to close the gap between politician and voters, they are stimulated to showcase themselves more to be on a more even level with the ‘average person on the street’. As a possible consequence, only politicians who showcase themselves more have a higher chance of ending up in leadership positions.
This work fits into the larger body of personalization research, which also points to personalization as being not a universal trend across parliamentary systems, but largely dependent on the context (Karvonen, 2010; Kriesi, 2012). Personalization of politicians’ communication seems to vary between countries. Moreover, the results are in line with the work of Pettitt (2012)
who also argues personalization of politician’s language use is dependent on personal factors instead of changing party organization or other factors which have shifted more influence to the party leadership.
Follow-up research should focus more on doing biographical studies of party leaders to pay closer attention to the link between the personality of the politician and language use. Using a more qualitative research design could generate richer and more detailed data, which is beyond the scope of this paper. Moreover, the data used here is from party conferences, which are predominantly policy driven meetings. Therefore, it would be fruitful to analyse speeches in different settings as well, such as election campaign meetings.
References
Adam, S., & Maier, M. (2010). ‘Personalization of Politics A Critical Review and Agenda for Research’, Annals of the International Communication Association, 34(1): 213-257.
Ahmadian, S., Azarshahi, S., & Paulhus, D. L. (2017). ‘Explaining Donald Trump via communication style: Grandiosity, informality, and dynamism’, Personality and Individual
Differences, 107: 49-53.
Balmas, M., Rahat, G., Sheafer, T., & Shenhav, S. R. (2014). ‘Two routes to personalized politics: Centralized and decentralized personalization’, Party Politics, 20(1): 37-51.
Bauman, Z. (2001). The individualized society. Cambridge: Polity.
Benoit, Kenneth et. al. (2017). ‘quanteda: Quantitative Analysis of Textual Data’. R package version: 0.9.9-65.
Berg, van den, J. Th. J. (2012). ‘Voor en tegen personalisering’. Retrieved from Parlement.com.
Bittner, A. (2011). Platform or personality?: the role of party leaders in elections. Oxford: Oxford University Press.
Blyth, M., & Katz, R. (2005). ‘From catch-all politics to cartelisation: The political economy of the cartel party’, West European Politics, 28(1): 33-60.
Bos. W. (2005, February 25). ‘Voor een beetje populisme. De democratie vitaliseren’. Retrieved from Groene.nl
Bosscher, D.F.J. (2013). ‘Schouten, Johannes (1883-1963)’, In Biografisch Woordenboek van Nederland. Huygens Instituut voor Nederlandse Geschiedenis.
Budge, I. (1994). ‘A New Spatial Theory of Party Competition: Uncertainty , Ideology and Policy Equilibria Viewed Comparatively and Temporally’, British Journal of Political Science, 24(4): 443–67.
Caprara, G. V., & Zimbardo, P. G. (2004). ‘Personalizing politics: a congruency model of political preference’, American Psychologist, 59(7): 581-594.
Caprara, G. V., Schwartz, S., Capanna, C., Vecchione, M., & Barbaranelli, C. (2006). ‘Personality and politics: Values, traits, and political choice’, Political Psychology, 27(1): 1-28.
Carmines, E.G., Stimson, J.A. (1989). Issue Evolution. Race and the Transformation of American Politics. Princeton: Princeton University Press.
Clarke, H. D., & Whitten, G. D. (2013). ‘Hard choices in hard times: Valence voting in Germany (2009)’, Electoral Studies, 32(3): 445-451.
Daalder, H., Cramer, N., & Bank, J. T. M. (1988). Willem Drees. Amsterdam: De Haan.
Dalton, R. J. (1996). Citizen Politics. Chatham. NJ: Chatham House.
Dalton, R. J., & Wattenberg, M. P. (2002). Parties without partisans: Political change in advanced
industrial democracies. Oxford: Oxford University Press.
Dolezal, M., Ennser-Jedenastik, L., Müller, W. C., Praprotnik, K., & Winkler, A. K. (2016). ‘Beyond salience and position taking How political parties communicate through their manifestos’, Party Politics: 1-13.
Drexhage-Zijlstra, I. (2013). ‘Berghuis, Wiert Pauwel (1911-1989)’, In Biografisch Woordenboek van Nederland. Huygens Instituut voor Nederlandse Geschiedenis.
Duverger M. (1964). Political Parties: Their Organization and Activity in the Modern State. London: Methuen Books.
Field, P. S. (2011). ‘Abraham Lincoln and the first-person plural: A study in language and leadership’, American Nineteenth Century History, 12(1): 49-75.
Franklin, M., Mackie, T., & Valen, H. (1992). Electoral change: Responses to evolving social and attitudinal. structures in Western countries. Cambridge: Cambridge University Press.
Franklin, M. N. (2010). ‘Cleavage research: A critical appraisal’, West European Politics, 33(3): 648-658.
Garzia, D. (2011). ‘The personalization of politics in Western democracies: Causes and consequences on leader–follower relationships’, The Leadership Quarterly, 22(4): 697-709.
Gross, D. A., Sigelman, L. (1984). ‘Comparing Party Systems: A Multidimensional Approach’, Comparative Politics, 16(4): 463–479.
Holtz-Bacha, C., & Norris, P. (2000). "To Entertain, Inform and Educate": Still the Role of Public
Television in the 1990s? Joan Shorenstein Center on the Press: Politics and Public Policy.
Holtz-Bacha, C. (2004). ‘Germany: How the Private Life of Politicians got into the Media’, Parliamentary Affairs, 57(1): 41-52.
Karvonen, L. (2010). The personalisation of politics: a study of parliamentary democracies. ECPR Press. Katz, R. S., & Mair, P. (1995). ‘Changing models of party organization and party democracy: the emergence of the cartel party’, Party politics, 1(1): 5-28.
Kenig, O. (2009). ‘Democratization of party leadership selection: Do wider selectorates produce more competitive contests?’, Electoral Studies, 28(2): 240-247.
King, A. (2002). Leaders' personalities and the outcomes of democratic elections. Oxford: Oxford University Press.
Kirchheimer, O. (1966). ‘The Transformation of Western European Party Systems’, In LaPalombara, J., & Weiner, M. (Ed.), Political Parties and Political Development (pp.177–200). New Jersey: Princeton University Press,
Kriesi, H. (2012). ‘Personalization of national election campaigns’. Party Politics, 18(6): 825-844. Laver, M. J., & Budge, I. (1992). ‘Measuring policy distances and modelling coalition formation’, In Budge, I., & Laver, M.J. (Ed.), Party policy and government coalitions (pp. 15-40). Palgrave
Macmillan UK.
Lehmann, P., Matthieß, T., Merz, N., Regel, S., Werner, A. (2016): Manifesto Corpus. Version: 2016b. Berlin: WZB Berlin Social Science Center.
Lipset, S. M., & Rokkan, S. (1967). Party systems and voter alignments: Cross-national perspectives. New York: Free press.
Mair, P. (2008). ‘The challenge to party government’, West European Politics, 31(1-2): 211-234. Mair, P. (2013). Ruling the void: The hollowing of Western democracy. Verso Books.
McAllister, I. (2009). ‘The personalization of politics’, In Dalton, R.J., & Klingemann, H.D. (Ed.) The Oxford handbook of political behavior (pp. 571-588). Oxford: Oxford University Press. Michel, J. B., Shen, Y. K., Aiden, A. P., Veres, A., Gray, M. K., Pickett, J. P. & Pinker, S. (2011). ‘Quantitative analysis of culture using millions of digitized books’, Science, 331(6014): 176-182.
Michels, R. (1915). Political Parties: A Sociological Study of the Oligarchical Tendencies of Modern Democracy. New York: Hearst’s International Library Co.
Mughan, A. (2000). Media and the presidentialization of parliamentary elections. Chicago: Springer. Noonan, P. (1998). On speaking well. Harper Collins.
Pennebaker, J. W., & King, L. A. (1999). ‘Linguistic styles: language use as an individual difference’, Journal of personality and social psychology, 77(6): 1296-1312.
Pennebaker, J. W. (2011). ‘The secret life of pronouns’, New Scientist, 211(2828): 42-45. Pettitt, R. T. (2012). ‘Me, myself and I: ‘Self-referencing in Labour Party conference leaders' speeches’, British Politics, 7(2): 111-134.
Pilet, J. B., & Cross, W. (2014). The selection of political party leaders in contemporary parliamentary
democracies: a comparative study. Routledge.
Rahat, G., & Hazan, R. Y. (2001). ‘Candidate selection methods an analytical framework’, Party
Rahat, G., & Sheafer, T. (2007). ‘The personalization (s) of politics: Israel, 1949–2003’, Political
Communication, 24(1), 65-80.
Renwick, A., & Pilet, J. B. (2011). The personalization of electoral systems: Theory and European evidence. In ECPR General Conference (pp. 25-27).
Riker, W.B. (1982). Liberalism Against Populism. A Confrontation between the Theory of Democracy and the Theory of Social Choice. Prospect Heights: Waveland Press.
Riker, W.B. (1986). The Art of Political Manipulation. New Haven: Yale University Press.
Schoonvelde, M., Schumacher, G., & Bakker, B. (unpublished). Measuring Political Psychology Constructs from Text. What we should and shouldn’t do. Unpublished.
Schumacher, G., & Giger, N. (2017). ‘Who Leads the Party? On Membership Size, Selectorates and Party Oligarchy’, Political Studies, 65(1): 162-181.
Somer-Topcu, Z. (2009). ‘Timely Decisions: The Effects of Past National Elections on Party Policy’, Journal of Politics (71): 238–248.
Stokes, D. E. (1963). ‘Spatial models of party competition’, American political science review, 57(2): 368-377.
Stokes, D. (1992). Valence Politics. Oxford: Clarendon Press.
Stutje, J. W. (2000). De man die de weg wees. Leven en werk van Paul de Groot. Amsterdam: De Bezige Bij.
Van Aelst, P., Sheafer, T., & Stanyer, J. (2012). ‘The personalization of mediated political communication: A review of concepts, operationalizations and key findings’, Journalism, 13(2): 203-220.
Van Santen, R., & Van Zoonen, L. (2009). ‘Stand van de wetenschap: Popularisering en personalisering in politieke communicatie’, Tijdschrift voor Communicatiewetenschap, 37(2): 155-176. Van Zoonen, L., & Holtz-Bacha, C. (2000). ‘Personalisation in Dutch and German politics: The case of talk show’, Javnost-the Public, 7(2): 45-56.
Vliegenthart, R., Boomgaarden, H.G., Boumans, J. (2010) Changes in political news coverage: Personalization, conflict and negativity in British and Dutch newspapers. In: Brants, K. & Voltmer, K. (Ed.) Political Communication in Postmodern Democracy (pp. 92 – 110). London: Palgrave Macmillan.
Vliegenthart, R. (2012). ‘The professionalization of political communication? A longitudinal analysis of Dutch election campaign posters’, American Behavioral Scientist, 56(2): 135-150.
Voerman, G. (2007). ‘De ontstuitbare opmars van het private in de politiek’, In Renders, H., & Voerman, G. (Ed.), Privé in de politieke biografie (pp. 15-23). Amsterdam: Boom/Biografie Instituut.
Wattenberg, M. P. (1991). The rise of candidate-centered politics: Presidential elections of the 1980s. Harvard University Press.
Webb, P., & Poguntke, T. (2005). ‘The presidentialization of contemporary democratic politics: Evidence, causes and consequences’, In Poguntke, T., & Webbe, P. (Ed.) The Presidentialization of Politics: A Comparative Study of Modern Democracies (pp. 1-25). Oxford: Oxford University Press. Willems, M. (2012, February 20) ‘Cohen weg als politiek leider PvdA – ‘onvrede gegroeid’”. Retrieved from NRC.nl
APPENDIX A
To test for the usefulness of the individualization ration, for the Netherlands, 3 other denominators have been tried and these different ratios have been correlated with the used ratio.
Ratio1 = (Ik, Mij, Mijn) (Wij, Ons, Onze)
Ratio2 = (Ik, Mij, Mijn) (Wij, Ons, Onze, Partijnamen en afkortingen) Ratio3= (Ik, Mij, Mijn) (Wij, Ons, Onze, Partij)
Ratio4 = (Ik, Mij, Mijn) (Wij Ons, Onze, Partij, Partijnamen en afkortingen)
X Y Correlation
Ratio1 Ratio2 0,91
Ratio1 Ratio3 0,89
Ratio1 Ratio4 0,87
