The Personalization of Party Politics: The Voters' Perspective Jörgen Hermansson jorgen.hermansson@statsvet.uu.se Modern political democracy is usually understood as party democracy. That designation conveys the insight that parties constitute the nexus of most that can be reasonably required for a political system to be regarded as democratic. Against that background it is a challenge to democracy that, according to some analyses, a sort of personalization of politics is currently taking place. 1 That personalization is said to be occurring in several ways, for example: heads of government behave like presidents even in parliamentary systems; mass media focus on party leaders and their personalities rather than on their parties politics; and citizens vote for persons rather than for parties, a trend that is the concern of this paper. Such observations and hypotheses regarding an ongoing personalization of politics are consistent with the more far-reaching thesis presented by Bernard Manin in his book, The Principles of Representative Government (1997). Manin states that party democracy is slowly but surely undergoing a transformation to audience democracy, a process which he indicates or at least suggests is detrimental to fundamental democratic values (Manin 1997; Ahlbäck Öberg et al. 2007). Within a general perspective, there is much to substantiate Manin s thesis. Society as a whole has become more fluid; that applies not least to voters, who have become less faithful to political parties. The reduced significance of socio-economic circumstances, along with such psychological factors as weakened party identification, have increased the likelihood of short-term effects the so-called party-leader effect comprising one example. (Campbell et al. 1960; Miller & Shanks 1991, 1996; Evans 2004). Changes to the party system, the tendency of parties to move toward the political centre, and the trend to presidential parliamentarianism may also be assumed to contribute to the personalization of politics (Karvonen 2010; Hermansson 2010). From a democratic standpoint, however, there is no self-evident answer to the question of how to evaluate the behaviour of voters who personalize politics. Their logic and motivations need not necessarily possess a democratic quality which is inferior to the reasoning of voters for whom a specific party leader is of comparatively little importance. It is rather an empirical question as to whether or not personalizing voters feel alienated from politics, have a weaker identification with an ideology, or a lower level of participation. 1 See Karvonen 2010 for an overview.
Accordingly, the empirical question addressed in this paper is: Can person-voters be distinguished from voters whose electoral choices are to a comparatively greater degree guided by party considerations and, if so, in what ways? The objective is to gain an understanding of what it is that this form of personalization (i.e. voting for individuals rather than parties) expresses, and how that can be evaluated within a democratic perspective. The discussion begins with a theoretical presentation that suggests several hypotheses which are then tested first individually, then jointly by means of a multivariate analysis. In conclusion, the empirical results are interpreted in relation to the more overarching question of how to evaluate the personalization of politics within the perspective of democratic theory. PREVIOUS RESEARCH ON PERSONALIZATION Lauri Karvonen s The Personalisation of Politics (2010) is an appropriate point of departure for the study of politics personalization. Karvonen explores the question of whether or not there is a trend toward increasing personalization, and finds that no such tendency can be clearly documented; rather, the evidence is mixed. Karvonen s study also provides an overview of previous research on the personalization of politics. The approach of this analysis to the question has long been a subject of attention in countries such as Finland and Ireland, whose electoral systems are formed in such a way that citizens vote directly for individual candidates and only indirectly for specific parties. But no corresponding studies have been conducted on electoral systems such as that of Sweden, where votes are given primarily to parties. There, political scientists have instead focused their interest on party-leader effects, especially the question of whether leaders lift or sink their parties (Oscarsson & Holmberg 2008). It has proven difficult to document short-term party-leader effects. An extensive comparative study of nine Western democracies (including Sweden) found relatively small effects, which did not increase over time. Longer-term effects are even more difficult to detect (Aarts et al. 2011). There are, however, grounds for interpreting this apparent absence of short-term party-leader effects with some caution, given the obvious measurement problems associated with this type of study. Especially in democracies such as Sweden s, which are issue-based and party-centred (Arter 2008), there is reason to suspect that voters prefer not to acknowledge (or do not even realize) that they allow their choices to be influenced by individual personalities, as that may be perceived as less issue-based and of inferior democratic quality. Also, some studies have found a positive factual relationship between physical attractiveness and the number of votes received (Berggren et al. 2010). UNDERSTANDING THE PERSONALIZATION OF POLITICS: THREE THEORETICAL PERSPECTIVES What is it that inclines some voters to focus on a party leader rather than a party? Given the nature of political journalism, we should probably be surprised that not all voters personalize politics. However, the Swedish political system and its regulations place an emphasis on parties. The
electoral system encourages citizens to vote for parties, with the possibility of voting for individual candidates provided as a secondary option. Thus, both the granting of mandates and demands for political responsibility are directed to parties. To provide a conceptual framework for the possible causes of personalization, this discussion considers three general theoretical perspectives which variously interpret personalization as a consequence of voters regarding themselves as (a) unaffiliated, (b) clients, or (c) responsible citizens. The unaffiliated-voter perspective is based on sociological theories regarding post-modern society, in combination with research that indicates a weakening of previous links between political parties and various subgroups of society. (Giddens 1990; Bauman 2006; Franklin et al. 1992). Within that context, it is reasonable to assume that many voters today feel no connection with a political party. They find it more difficult to identify with any particular party, have little or no sense of group belonging, and may even feel that politics does not concern them. This results in societies where traditional identities are weakened and where connections of individuals to politics and parties gradually erode. By personalizing politics, such unaffiliated voters may find a way to experience at least something that makes politics and voting seem meaningful. Within the second theoretical perspective, voters are likened to clients in a marketplace (Downs 1957). Presented with several alternatives, the client finds it difficult to choose; and if the various service providers (parties) seem similar, the client/voter makes a decision based on external packaging. This theoretical framework is related to research which indicates that many voters do not restrict themselves to a single party, but can consider voting for one among two or more similar parties (Holmberg & Oscarsson 2004; Oscarsson & Holmberg 2008, Kapitel 2). This makes for a greater variety of choices, and greater demands for making conscious choices especially for voters in the political centre where the battle for marginal voters is most intense, with multiple similar parties to choose from. Both perspectives voters as unaffiliated and as clients are problematical from the standpoint of democracy. Non-affiliation clearly violates one of Robert Dahl s criteria for an ideal democracy, that of informed understanding. The client perspective does not represent an equally clear violation; but it conflicts with Dahl s criterion in that it is not conditioned on informed understanding, and may even involve a lack of it. The voter as client may feel uncertain when the available alternatives appear to be too similar, and as a result may choose on the basis of packaging and other superficial factors, even when he or she in fact may not wish to do so. The third perspective is related to the trend toward presidentialization that can be observed in parliamentary systems (Karvonen 2010; Hermansson 2010). The theoretical basis is the idea of republican citizenship, especially in its USA form (Pettit 1997). The conception of the voter is related to citizenship values that are also familiar within a parliamentary context such as that of Sweden; but it differs from party democracy in that personalization is regarded as legitimate. In systems like that of Sweden, politics are considered to be tied to parties. But that is not a conceptual necessity; a mandate can as well be granted to an individual as to a party. In this context, a shift toward presidentialization would involve a weakening of party democracy as we know it, but not to a weakening of democracy in general. From this point of view,
personalization would mean that the voter assumes responsibility as a citizen who participates in the making of a crucial decision about the nation s future. The prime ministerial candidates appear as representatives of various options to choose from, due not least to the fact that they are individuals with certain personal characteristics and thereby more or less qualified to govern the country. To this one may add an additional perspective that is not related to any of the foregoing theoretical frameworks. Regardless of which of those three more general models best describes the basis of political personalization, the act of voting often involves making a decision that is difficult and consequently tends to be postponed. For that reason, one might suspect that citizens who personalize their choices may be over-represented among those who make their final decisions at the later stages of election campaigns; personalization would be most evident among uncertain voters. From these four different perspectives can be derived a number of hypotheses which can be tested empirically. Non-affiliation among voters with low party identification According to the theory of post-modern society, voters are increasingly unaffiliated. This nonaffiliation involves a risk that, instead of basing choices on political substance with parties as bearers of ideologies and substantive issues voters focus on the packaging and presentation of political messages. As a result, when citizens are urged to choose among political parties in a parliamentary election, they personalize their choices. A reasonable assumption is that younger generations are more influenced than older generations by the conditions of post-modernism. The younger should thus display a higher rate of non-affiliation than the older; and if the interpretation suggested here is correct, the personalization of politics should also be more prevalent among the younger. The first hypothesis can be formulated as follows: H1: Younger voters assign greater importance to party leaders than do older voters. In addition, it is here assumed that this hypothesized non-affiliation can take two different forms: weak party identification, and low interest in politics. It is assumed that loyalty to groups, including political parties, is weakening in post-modern societies. Identification with parties is also declining. By definition, voters who strongly identify with a party identify themselves with that party, but not necessarily with the party leader. In Sweden, a voter has been a Social Democrat or a Conservative as a consequence of one s social background, and party leaders has been of secondary importance. That suggests the following hypothesis: H2: Voters with weak party identification assign greater importance to party leaders than do voters with strong party identification.
Weak party identification does not exclude the possibility of interest in politics. It is, however, likely that low political interest leads to weak party identification; and it is reasonable to assume that, all else being equal, personalization is more likely to be evident among voters who are relatively uninterested in politics. For such voters, parties have little significance for voting behaviour; and it may also be assumed that, all else being equal, party leaders have greater significance, suggesting the following hypothesis H3: Voters with low political interest assign more importance to party leaders than to parties. Difficult choices in the clients marketplace When politics is regarded as a marketplace, three problems for voters can be discerned: heavier competition for voters in the political centre, increased need to gather information when there are many alternatives, and greater difficulty in choosing among alternatives. The premise of the so-called median voter theory is that the electorate is distributed along a normal curve, with the great majority of voters clustered around the centre of the left-right political spectrum. That presents a strong incentive for parties to move to the centre in order to attract the median voters, who are thus presented with several similar alternatives and are as likely to choose the one as the other. It may be assumed that, all else being equal, party leaders will be more important to median voters than to those who are further from the political centre, suggesting the following hypothesis: H4: Voters whose ideological preferences place them in the political centre are more likely than others to assign importance to party leaders. According to the similar parties model, [ partistallsmodellen?] the voter s choice predicament is similar to that of a client in a marketplace. But instead of an assortment of products and services, the voter is presented with an assortment of similar parties to choose from. Most voters in Sweden can consider more than one party as their election choice. That, together with the fact that more short-term factors may determine choice of party, supports the assumption that the significance of party leaders is greater, all else being equal, if it is short-term rather than long-term factors that determine voter choices. On the other hand, according to adherents of the similar parties model, the voter has more candidates to choose from which means that, all else being equal, party leaders have greater potential significance than would be the case if the voter were committed to a single party. Thus the following hypothesis: H5: Voters who have multiple alternatives to choose from base their final decisions on their assessments of party leaders. Finally, it should also be noted that when voters feel there is little distinction between the available alternatives, they must make their choices on other grounds than party politics. It may therefore be assumed that in such situations, all else being equal, personal factors will play a greater role, suggesting the following two hypotheses: H6: Voters who perceive little difference between the parties will assign greater importance to party leaders than do voters who perceive significant differences.
H7: Voters who perceive little difference between party blocs will assign greater importance to party leaders than do voters who perceive significant differences between the blocs. Citizens presented with a choice of president As regards the Swedish form of government, researchers have found a long-term trend toward presidentialization of the parliamentary system, with the positions of both the administrative government and the prime minister being strengthened. This has led to greater attention being focused on the individual who forms the government, and it is therefore hardly surprising that parties, voters and mass media have become increasingly interested in the individuals who lead parties. If it is correct that voters perceive elections as involving a choice between two government alternatives, each led by its presidential candidate, the following hypothesis would be supported: H8: Voters who perceive an election as a contest for governmental power, with crucial significance for the nation s economy and publicly financed general welfare system, assign greater important to party leaders than do voters who do not share that perception. Also related to the issue of presidentialization is the question of whether or not voters perceive a difference between political blocs. To the extent that voters perceive such differences as minor or non-existent, the personal characteristics of prime ministerial candidates (the most prominent party leaders) should weigh more heavily. Citizens may wish to take responsibility, but receive inadequate guidance from party platforms. In that case, voters may pay more attention to the prime ministerial candidates and other individual representatives of political blocs than to the various ideological alternatives presented. The following hypothesis is related to both the citizen-voter and client-voter perspectives: H7: Voters who perceive little ideological difference between political blocs assign greater importance to party leaders. Tendency to postpone difficult decisions It follows from the foregoing that theories of non-affiliation, presidentialization and citizen voters all suggest hypotheses concerning the increased significance of party leaders. There are also other factors involved, as indicated by this hypothesis: H9: Voters who make their electoral decisions at a late stage of a campaign assign greater importance to party leaders than do those who decide earlier. RESEARCH ON VOTING FOR PARTY LEADERS The question addressed in the following section concerns which voter characteristics are associated with an increase in the importance of party leaders for electoral decisions at the expense of parties.
The empirical data are from the Swedish national election of 2010. The relevant question was formulated in Survey 2010 2 as follows: Regarding the party you voted for in the parliamentary election, which did you prefer more the party, or the party leader (or spokesperson)? Answer by ticking the box on the scale below which best describes your viewpoint. Respondents could answer on an 11-point scale, from 0 10 where 0 indicated the strongest preference for party and 10 for party leader. Point 5 on the scale indicated both equally. A subgroup s mean value on the scale is henceforth referred to as its degree of personalization. Personalization: party leaders popularity in relation to their parties The foregoing discussion regarding the importance of personal factors has suggested a number of hypotheses. The objective is to identify possible causes of the tendency for individual voters to assign greater importance to party leaders than to their respective parties when electoral decisions are made. Table 1 shows the popularity of party leaders relative to their parties among affiliated voters. 2 Survey 2010 bla bla bla. Berätta om denna enkätundersökning ****
Table 1. Party leaders popularity among affiliated voters, 2010 (per cent, plus-minus index, mean value) Party leader (party)* Stronger preference for party leader Equal preference for party and party leader Stronger preference for party Per cent Number of respondents +/ index: party leader Degree of personalization (mean) Reinfeldt (M) 42 42 16 100 409 +26 5,97 Wetterstrand/Eriksson (MP) 26 45 28 101 117 2 5,06 Hägglund (KD) 19 44 37 100 79 18 4,52 Olofsson ( C) 16 37 47 100 89 31 4,09 Björklund (FP) 19 28 53 100 126 34 3,90 Åkesson (SD) 14 34 52 100 65 38 3,44 Sahlin (S) 8 27 64 99 398 56 3,02 Ohly (V) 8 21 71 100 82 63 2,78 All respondents 23 35 42 100 1371 19 4,29 Note: *Party abbreviations: (M) Conservatives, (MP) Greens, (KD) Christian Democrats, (C) Centre, (FP) Liberals, (SD) Swedish Democrats, (S) Social Democrats, (V) Left Party A 11-point scale from 0 ( party ) to 10 ( party leader ), with midpoint 5 ( both equally ). Answers 0-4 coded as stronger preference for party and 6-10 as stronger preference for party leader. Sample included respondents who voted for parties not represented in the parliament after 2010 election. As indicated by the bottom line of Table 1, most Swedish voters assign greater importance to a party than to a party leader. That is also indicated by the plus-minus index (the percentage that prefers the party leader minus the percentage that prefers the party) of minus 19. However, a significant portion of voters nearly a quarter regard party leaders as more important than their respective parties; and over a third regard party leaders and their parties as equally important. The most popular party leaders were the Conservatives Reinfeldt and the Greens tandem spokespersons, Wetterstrand and Eriksson. Among affiliated voters, Reinfeldt is by far the most popular; 42 per cent of Conservative voters regard him as more important than his party. He is also the only party leader with a positive plus-minus index (+28 %), nearly 30 per cent greater than the next most popular among affiliated voters, Wetterstrand/Eriksson. Least popular in relation to their respective parties were the Left s Lars Ohly and the Social Democrats Mona Sahlin. In both case, roughly eight per cent of affiliated voters considered them more important than their parties. The plus-minus index for both was very low minus 56 per cent for Sahlin and minus 62 for Ohly. The focus of the following section is on differences in personalization, as reflected for example in the mean value of 5.97 for Conservative voters but only 3.02 per cent for Social Democratic voters. How does personalization vary among differed voter subgroups?
Personalization as an expression of uncertainty The survey results disclose an increased tendency to make party choices at a late stage of a political campaign (Oscarsson & Holmberg 2008). Many voters seem to find it difficult in decide. A reasonable hypothesis is that it is among uncertain voters that personalization is most prevalent. All else being equal, party leaders can be assumed to have comparatively great importance for such voters, especially in the final stages of a campaign. If so, party leaders should be more important to voters who make their party choice toward the end of a campaign. But that hypothesis is not confirmed by the results, which instead indicate the reverse; see Table 2. However, these data do not meet standard criteria of statistical significance. Table 2. When did you decide which party you would vote for in the parliamentary election? (Mean values) Degree of personalization Number of respondents Long before the election 4.33 691 About one month before 4.57 178 Weeks before the election 4.46 167 Final week before the election 4.13 213 On election day 4.03 153 Is personalization an expression of non-affiliation? The first of the above-noted theoretical perspectives relates personalization to what is usually referred to as post-modern society. The traditional identities of individuals are weakened, and their links to politics and political parties become less self-evident. As a consequence they feel unconnected. If so, personalization should be more prevalent among younger voters. In particular, it should function as a substitute for both party identification and political interest. If the theory of post-modern society is correct, non-affiliation and non-identification should be more prevalent among younger voters, and it could also be expected that they would display a greater tendency to personalize politics.
Mean Figure 1. Personalization in various age groups. (Mean values) 6 5 4 3 2 1 0 18-20 21-25 26-30 31-40 41-50 51-60 61-65 66-75 76-85 Age group However, it turns out that the results do not confirm the hypothesis that younger voters assign greater importance to party leaders than do older voters. On the contrary, the tendency to personalization is significantly more pronounced among older voters. If the second hypothesis is correct, personalization should be most prevalent among three subgroups: voters with weak links to the parties they vote for, those with weak party identifications, and tactical voters who actually prefer another party than the one they vote for; see Table 3. Table 3. Personalization in subgroups with varying degrees of party identification (Mean values) Degree of Party Identification Strong identification Moderate identification No identification Conservatives 6.17 5.87 5.64 Greens 4.81 5.18 5.00 Christian Democrats 4.80 4.37 4.10 Centre 4.18 4.09 3.87 Liberals 2.88 4.05 4.11 Sweden Democrats 4.33 3.19 2.72 Social Democrats 3.14 2.98 2.78 Left Party 2.55 2.97 2.71 Total 4.46 4.38 3.94 Note: The question about party identification followed a question which asked the respondent to state a party affiliation The follow-up question was: Do you regard yourself as a strong supporter of that party? The response alternatives were: Yes, very strong, Yes, moderately supportive, and No.
The results shown in Table 3 do not support the hypothesis about non-affiliation. For most parties, increases in degree of personalization are associated with increased party identification. That does not apply to the Liberals, the Greens or the Left Party, however. One possible explanation of the difference is that it may be related to party culture. The Liberals and the Left are similar in that they have individualistic cultures which encourage a critical attitude toward the party leadership (Barrling Hermansson 2004). The Barrling Hermansson study concerns members of parliament and does not permit generalization to voters; but it does suggest the possibility that the same attitude may be common among party voters. It should be possible to test the hypothesis of personalization as an expression of non-affiliation by dividing the electorate into subgroups with varying levels of interest in politics; see Table 4. Table 4. Generally speaking, how interested are you in politics? (Mean values) Personalization Number of respondents Very interested 4.34 139 Fairly interested 4.35 700 Not especially interested 4.27 464 Not at all interested 3.85 67 Personalization is roughly the same, regardless of interest level, which means that the hypothesis is not supported. On the contrary, degree of personalization is lowest among those who state they are not at all interested in politics. From this it may be concluded that, while the theory of post-modern society may provide an insightful diagnosis of increased non-affiliation in general, that is not reflected in an increased tendency to personalization of politics. On the contrary, personalization is most prevalent among older voters and those who strongly identify with a party, which is the opposite of the expected outcome. Can personalization be a solution to difficult client choices? Adopting the role of client on election day implies freedom of choice, but also that the voter has sufficient time and information to make wise choices. In that regard, two related issues have been identified in the foregoing discussion. One is that the voter perceive that there is more than one alternative to choose from; the other is that the voter may not perceive any significant difference between the available alterntives. It is possible to study the first of those two issues iindirectly, by determining whether or not voters near the centre of the ideological spectrum, where several parties have positioned themselves, exhibit a greater tendency than other voters to personalize. In that case, voters who locate themselves near the centre of the left-right political should be more incllned to personalize than are voters near the extremities of the spectrum. For this hypothesis, therefore, it does not matter if a voter is on the political left or right; the results shown in Figure 2 are only concerned with distance from the ideological midpoint.
Mean Mean Figure 2. Personalization among voters at various distances from the ideological midpoint. (Median values) 6 5 4 3 2 1 0 Ideological extremity Increased distance to midpoint Ideological midpoint Note: Definition of the ideological centre was based on the question: It is sometimes said that political opinions can be placed on a left-right spectrum. Where would you place yourself on such a spectrum? Respondents could position themselves along an 11-point scale where 0 meant Far to the left, 10 was Far to the right, and 5 was the midpoint. Nothing in the results indicates that personification is used by centrist voters to help them perceive differences between the parties. A possible explanation is provided by Figure 3. Figure 3. Personification among voters with different ideological orientations. (Mean values) 7 6 5 4 3 2 1 0 Far to the left 1 2 3 4 Centre 6 7 8 9 Far to Left-right spectrum the right The degree of personification increases from the left to the right of the spectrum. That trend is evident among those who voted for the Social Democratic Party, which is near the political centre.
But the trend toward increasing personalization is strongest among those who located themselves ideologically to the right and voted for parties of the centre-right coalition. Among voters who considered that they had more than one alternative from which to choose, it may be assumed that personal factors were decisive. Table 5 provides an answer to the question of whether or not those who could consider more than one party as their choice displayed a greater tendency to personalize. Table 5. Significance of similar parties for personalization (Median values) Similar parties Degree of personalization No. of respondents Only one party 4.22 366 Two or more parties 4.35 1001 While there is a difference in the direction of the observed trend, closer inspection indicates that it is not significant. That result is repeated when similar parties within the same bloc are studied. Accordingly, the conclusion is that there is no empirical support for the similar-party hypothesis, either. The extent to which voters personalize because the available alternatives are similar has also been investigated in connection with two other questions: (a) To what extent, in your opinion, do the parliamentary parties differ with regard to substantive issues? (b) To what extent, in your opinion, do the centre-right coalition and the red-green coalition differ with regard to substantive issues? The results are shown in Table 6. Table 6. Personalization in relation to perceived differences. (Mean values) Parties differ: Blocs differ: Personalization No. of respondents Personalization No. of respondents Very much 3.69 88 4.13 155 Quite a lot 4.42 661 4.42 769 Not very much 4.27 610 4.18 437 Not at all 4.73 26 4.27 26 The results do not support either hypothesis i.e. that voters who perceive little or no difference between (a) the parties or (b) the political blocs are more inclined to personalize. Personalization as an expression of responsible citizenship Is the inclination to personalize politics stronger among voters who regard an election as decisive for the nation s future? Respondents to Survey 2010 were asked for their opinion regarding which of the two political blocs would be better for Sweden s economy and publicly financed general welfare system. They could answer on a scale from 0 to 10, where each of those extremities indicated that the corresponding bloc would definitely be better. The midpoint on the scale indicated no difference.
The hypothesis is that personalization should be most prevalent among voters who felt that the election presented a choice between blocs, which was decisive for the nation s economy and publicly financed general welfare system. If that hypothesis is correct, the responses would form a U-shaped pattern. To test the hypothesis, two variables were introduced. They indicate distance from the midpoint, i.e. the point at which the two opposing blocs meet (see Figure 2 for a similar construction). In addition to the mean values we are using Pearson s r, which varies between -1.00 (perfect negative linear relationship) and +1.00 (perfect positive linear relationship). Values near 0 indicate absence of linear relationship, while higher values indicate strong relationships. If a strong sense of citizen responsibility is associated with a high degree of personalization, the results should demonstrate a positive (+) relationship. The results are shown in Table 7. Table 7. Personalization and responsible citizenship (Mean values and Pearson s r) Sense of responsibility for: Degree of responsibility Economy Public general welfare system 5 (greatest responsibility) 4.83 4.26 4 4.49 4.60 3 3.91 4.26 2 4.16 4.21 1 3.16 4.43 0 (least responsibility) 3.58 4.13 Number of respondents 1355 1359 Pearson s r +.20* +.02 Notes: Statistical significance of p<0.10 according to the following: p<0.01: ***, p<0.05: **, p<0.10: * Degree of citizen responsibility for the economy was based on the question: Regardless of which parties form the government, which of the two political blocs in your opinion would be better for the Swedish economy during the next 3-4 years? Degree of citizen responsibility for the publicly financed general welfare system was based on the question: Regardless of which parties form the government, which of the existing blocs would in your opinion be better for the general welfare services financed by state, county administrative boards and municipalities during the next 3-4 years? Answers to both questions were give on an 11-point scale from 0 ( Definitely the centre-right coalition ) to 10 ( Definitely the red-green coalition ), with the midpoint at 5 ( There is no difference ). Answers to both questions were coded as follows: 0 and 10: code 5, 1 and 9: 4, 2 and 8: 3, 3 and 7: 2, 4 and 6: 1, 1 and 5: 0. The only statistically significant relationship to support the hypothesis is that between personalization and citizen responsibility for the economy. There may be good reason to consider that relationship more closely. According to the hypothesis, personalization should be most prevalent among voters who regard the choice between the two blocs as decisive for the nation s economy and general welfare system. If so, the original formulation of the question would have yielded answers that formed a U-shaped pattern. That is the case with regard to the national economy, but not to the general welfare system. For the latter, the relationship is almost perfectly linear: Voters who perceived the centre-right coalition as guarantor of the general welfare system during the following 3-4 years were also most likely to personalize politics.
That result suggests a potentially interesting path of inquiry. It appears that personalization is related to voters sense of their responsibilities as citizens. In one case, however, that seems primarily to be a phenomenon that is associated with the centre-right coalition. Forces driving voters personalization of politics Now, for the final test of the foregoing hypotheses. Thus far, the only hypothesis which has been supported by the survey results is the one relating to responsible citizenship, and then only in part. The question is whether or not the same results occur when the various hypotheses are controlled against each other, which is accomplished by means of linear regression analysis. Information about the strength of a relationship can be gained by analysing the effects of the various factors by means of regression analysis. The higher the value, the stronger the relationship. The values also convey information about the direction of the relationship. In this case, a positive relationship (+) indicates that the analysed factor is associated with increased personalization, while a negative relationship ( ) indicates a tendency to decrease. It is only when a relationship contradicts established criteria of statistical significance that it is possible to draw definite conclusions about the relationship. The foregoing analysis has yielded clues as to which other factors may be involved when hypotheses have not been supported. One such clue suggests that it would be appropriate to investigate whether personalization is a phenomenon that is primarily associated with the centre-right coalition. Another appropriate question is whether or not the observed differences among age groups are a consequence of older voters lower education levels compared with younger voters. Table 8 presents the results of the regression analysis. The hypotheses are tested primarily on the basis of statistically significant effects, and secondarily on the basis of the regression curve value (positive or negative).
Table 8. Tests of hypotheses: factors related to personalization (Linear regression, unstandardized regression coefficients) Model 1 Model 2 Explanatory factor b b Constant +2.64 *** +2.67 *** Uncertain; decided late +0.04 +0.05 Unaffiliated 1; younger voter +0.01 * +0.01 * Control: education level 0.08 ** 0.07 * Unaffiliated 2; weak party identification 0.23 * 0.18 Unaffiliated 3; little or no interest in politics +0.05 +0.01 Difficult decision 1; near ideological centre 0.04 0.02 Control: places self on ideological right +0.16 *** Difficult decision 2; more than one party a possible choice +0.12 +0.12 Control: Voted for a party in centre-right coalition +1.69 *** +0.75 *** Difficult decision 3; sees no difference between parties 0.01 0.04 Difficult decision 4; sees no difference between blocs +0.22 +0.19 Citizen responsibility 1; national economy +0.09 * +0.07 Citizen responsibility 2; general welfare system +0.03 +0.04 Control: red-green coalition seen as guarantor of general welfare system 0.07 ** Explanatory power: Adj-R 2 0.125 0.145 Number of respondents 1176 1176 Note: Statistical significance coded as follows: p<0,01: ***, p<0,05: **, p<0,10: **. These results essentially confirm the preceding analysis: No relationship is found between personalization and uncertainty (insignificant), non-affiliation (insignificant or wrong sign), or difficult decision (insignificant). The results also indicate that the less educated are more inclined to personalize than are the better educated, but that the age differences remain. The effects of party identification and citizen responsibility are close to being statistically significant. However, personalization is most prevalent among politically active citizens. Further, it is evident that personalization is more prevalent among centre-right voters: Those who voted for the centre-right coalition, who placed themselves on the ideological right and who regarded the centre-right
coalition as a guarantor of the publicly financed general welfare system were more inclined to personalize than were other voters. PERSONALIZATION AMONG ENGAGED CITIZENS This study has found very little to support a number of hypotheses derived from theories which suggest that personalization of politics is a sign of dysfunctional democracy. It has not been possible to demonstrate that voters who personalize politics are more unaffiliated, uninformed or indecisive. Voters who personalize describe themselves as engaged citizens who choose the alternative that they regard as best for the nation, especially if they are ideologically to the right. The only hypothesis which is supported is one that relates personalization to positive democratic value active citizenship. Among voters on the political right, it is the strongly affiliated and those who regard the election as a choice of presidents who personalize. Among voters on the left, on the other hand, parties tend to be more important than party leaders. How to interpret the apparent fact that personalization is widespread only among centre-right voters? One possibility is that it may be related to a more favourable attitude in those parties to the election of individual candidates. To acknowledge to oneself and others that one has been influenced by personal factors is thereby more acceptable to right-oriented than to left-oriented voters. That could, of course, also be a consideration which strongly affects the process of voting, in accordance with a conviction that the election of individual candidates is relevant. However, there is perhaps another explanation, one that is strongly related to the circumstances surrounding the Swedish national election of 2010 namely, enthusiasm for the centre-right alternative. Prime Minister Fredrik Reinfeldt enjoyed exceptionally high popularity, both in absolute numbers and in relation to his challenger, Social Democrat Mona Sahlin. The election presented a clear choice of alternative governments and their principal representatives. It is therefore difficult to generalize from the 2010 election. That the survey results indicate a clear presidential effect among voters may have more to do with Reinfeldt s personal popularity than with a more general phenomenon that applies to all national elections. To oversimplify somewhat, personalization in the 2010 election depended to a great extent on enthusiasm among centre-right voters for their incumbent leader, in combination with the perception that he represented the best guarantee for the nation s well-being, in an election that was preceded by a worldwide financial crisis. Accordingly, it would be interesting to conduct the same sort of study of an election in which the left bloc is led by an individual with similarly high popularity figures, in order to determine if the observed effect applies even in such a case despite the left s more sceptical attitude toward individual-focused elections. How, then, to interpret the results of this study within a democratic perspective? The results indicate that personalization is not necessarily associated with political non-affiliation or uncertainty and, in fact, that it is even compatible with such a positive democratic value as active citizenship.
However, the possibility cannot be excluded that the observed relationship was the result of a favourable trend. It remains to be seen whether the positive relationship between personalization and active citizenship applies even during periods when parties are experiencing political difficulties.
REFERENCES Aarts, Kees, André Blais & Hermann Schmitt Oxford (red) (2011) Political Leaders and Democratic Elections. Oxford: Oxford University Press Ahlbäck Öberg, Shirin, Jörgen Hermansson & Lena Wängnerud (2007) Exit riksdagen. Malmö: Liber. Asp, Kent (1986). Mäktiga massmedier. Studier i politisk opinionsbildning. Stockholm: Akademilitteratur. Barrling Hermansson, Katarina (2004) Partikulturer. Kollektiva självbilder och normer i Sveriges riksdag. Uppsala: Uppsala universitatis upsaliensis. Bauman, Zygmunt (2006) Liquid Times: Living in an Age of Uncertainty. Cambridge: Polity Press. Berggren, Niklas m.fl.. "The looks of a winner: Beauty and electoral success". Journal of Public Economics, Volume 94, Issues 1-2, February 2010, Pages 8-15. Brors, Henrik (2010): Reinfeldt slår Sahlin på samtliga punkter. Dagens Nyheter 2010-06-16. Internet: http://www.dn.se/nyheter/valet-2010/reinfeldt-slar-sahlin-pa-samtliga-punkter (hämtad 2011-04-12). Campbell m.fl. 1960 (gemensam) Dahl, Robert (2002) Demokratin och dess antagonister. Stockholm: Ordfront förlag. Downs, Anthony. (1957). An Economic Theory of Democracy. New York: Harper and Row. Giddens, Anthony (1990) The Consequences of Modernity. Cambridge: Polity. Hermansson, Jörgen m.fl..(2010). Regereingsmakten i Sverige. Ett experiment i parlamentarism 1917 2009. Stockholm: SNS Förlag. Hernes, Gudmund (1978) Det mediavridde samfunn. I Hernes, Gudmund (red): Forhandlingsøkonomi og blandningsadministrasjon. Osl o: Universitetsförlaget. Holmberg, Sören & Oskarsson, Henrik (2004) Regeringsskifte : väljarna och valet 2006. Stockholm: Norstedts. Karvonen, Lauri. 2010. The personalisation of politics. A study of parliamentary democracies. Colchester: ECPR Press. Manin, Bernard. 1997. Principles of Representative Governmenty. Cambridge: Cambridge University Press. Miller & Shanks 1991. Miller & Shanks 1996.
Oskarsson, Henrik & Sören Holmberg (2008). Pettit, Philip (1997) Rebublicanism: A Theory of Freedom and Government. Oxford: Oxford University Press. Ruin, Olof (2002) Sveriges statsminister och EU. Stockholm Hjalmarson och Högberg Bokförlag. SCB (2010): Partisympatiundersökningen maj 2010. Internet: http://www.scb.se/pages/pressrelease 295430.aspx (hämtat 2011-04-12). Strömbäck, Jesper (2000): Makt och medier. Om samspelet mellan medborgarna, medierna och de politiska makthavarna. Lund: Studentlitteratur. Strömbäck, Jesper (2008) På nyhetsmediernas agendor - en studie av hot och risker i det svenska nyhetsur valet i morgonpress, kvällspress och TV. Studier i Politisk Kommunikation nr 18. - - - - - -