So, What Difference Do Leaders Make? Candidates’ Images and the “Conditionality” of Leader Effects on Voting

Abstract

Do leaders make a difference? Are national political candidates and their public images influential in shaping voter choice and in determining election outcomes? Scholarly literature on “leader effects” tends to be discordant when assessing the impact of leader images on voting. This suggests that we should reframe the question in “conditional” terms, i.e. under what conditions do leaders make a difference? Indeed, we need to consider not only “how much”, but also “who” can make a difference along with “where”, “when”, “to whom”, and “how”. Having examined the findings of existing international research, this review article proposes an “epistemological tool kit” with which to examine the forms and limits of knowledge on the influence of leaders’ images on voting. In so doing, it considers the following “conditions of possibility” for leader effects: (1) structural constraints: the institutional, political, territorial and media contexts in which an election takes place; (2) political opportunities: the opinion climate, the economic situation, the presence of a systemic crisis, and the campaign environment; (3) individual moderators: the availability of leader‐oriented segments of voters; (4) image variables: the fundamental features of the candidates/leaders. Finally, the article examines the possible forms (types of effects and mediating processes) and impact (micro/macro assessment) of leader effects on voting.

Notes

1. The “conditionality” this review article refers to is not related to the statistical sense of this expression, but to its literal meaning: i.e. the “conditions of possibility”, or the “conditions under which” a phenomenon is possible.

2. The difference between plurality/majority and proportional (PR) electoral systems is not crucial with respect to the centrality of the national leaders, who remain most often formally absent from the ballot. More generally, the distinction between majority and proportional democracies is credited with a weak explanatory power in relation to the political process (Pasquino, 2004). Moreover, Curtice and Blais (2001) find no evidence to support the claim that leader effects are stronger where a majority electoral system is in place.

3. Local level elections, where the impact of national leaders is theoretically weaker and methodologically even more difficult to identify, are not taken into account in this analysis.

4. For example, in the Italy of the 1960s and 1970s, where political parties were the undisputed actors of political contests and citizens were often identified with them, the attention paid to political leaders was half as large as has been detected since 2000 (Barisione, 2007).

5. Here an analysis of where rejoins an analysis of when leaders matter.

6. The notion of “valence image” – that will be better specified in the section on the “types of effects” – offers a possible partial clarification in the comprehension of a theoretical construct such as a leader’s image (Barisione, 2009). In any case, scholars seem gradually to agree more about the central importance of politically relevant, performance‐related personal traits for candidate evaluation and vote choice (from Bean, 1993 to Ohr & Oscarsson, 2003).

7. The case of substituting a leader is more complex: with Sniderman et al. (1991) we see that it is only influential in the case of an incumbent. Nonetheless, this point will be developed in the indirect forms of the leader effect.

8. See, for example, the seven models presented in the October 2004 issue of PS: Political Science and Politics. On the 2000 presidential elections see Fiorina et al. (2003) and Bartels and Zaller (2001).

9. Similarly, Shaw (1999) speaks of “width of perception” and of “one‐sidedness” of campaign events as conditions of their impact.

10. While it may be difficult to detect the qualitative value of a campaign in operational terms, it should not be sacrificed for strictly quantitative indicators. In an adequate theoretical framework it is necessary to combine the analysis of the qualitative and quantitative properties of a campaign.

11. Such as in presidential campaigns, where even “knowledge gaps” between highly and poorly educated voters may be reduced (Holbrook, 2002).

12. The model of “one reason decision‐making” holds that the individual ceases to look for new evaluation criteria after having found one that allows him/her to sufficiently differentiate between the two options considered.

13. According to empirical studies conducted since the 1960s (Treneman & McQuail, 1961; Kinder, et al., 1980; Miller, et al., 1986; Hacker, 2004; Mondak & Huckfeld, 2006) voters assess traits such as competence, strength, reliability, integrity, and ability, but also warmth or “closeness to people”, as essential qualities for a political leader. Two main sets of traits (or image dimensions) seem to emerge as relatively more important: “energy” and “trustworthiness” (Caprara, et al., 2002; Barisione, 2009).

14. Bush Jr. fell below 50% in the summer surveys of 2004, but not in the overall last semester average.

15. The “rule of 50” seems to hold even if applied to a substitute of the outgoing president, like a kind of transfer of popularity between the leader and his party: Johnson and Bush would successfully seize on the popularity of Kennedy and Reagan, while Nixon would defeat the substitute of the unpopular Johnson. A single case seems to contradict this pattern: that of Gore, who lost despite the high approval ratings inherited from Clinton. Nonetheless, not only did Gore win the popular vote, but his inheritance was also quite problematic, given the significant (and unusual) split between Clinton’s popularity in terms of job approval (high) and favorability rating (less than 50%). According to Fiorina et al. (2003) it would have thus been sufficient to insert both of these variables in the equation concerning the 2000 vote to save the predictive model.

16. In relation to the literature on electoral forecasting, see for instance the 2004 special number of PS: Political Science & Politics (October), which presents seven different models, respectively conceived by: A. Abramowitz, J. E. Campbell, C. Wlezien and R. Erikson, T. Holbrook, M. Lewis‐Beck and C. Tien, B. Lockerbie, H. Norpoth. (Other references on this topic are: Rosenstone, 1983; Lewis‐Beck, 1985; Campbell & Garand, 2000; Bartels & Zaller, 2001; Lewis‐Beck, et al., 2004; Lewis‐Beck & Tien, 2005.)

17. In the literature on US electoral behaviour, there exist both research that focuses specifically on incumbency effect for Congressional elections (Gelman & King, 1990; Kulisheck & Mondak, 1996; Ansolabehere & Snyder, 2002), and prediction models about presidential elections, that take the incumbency factor into account, among others (from Brody & Sigelman, 1983 to Abramowitz, 2004).

18. Other possible survey‐based techniques attempting to capture the leadership factor in addition to party identifications, ideology and other political variables, are of a more qualitative nature. For instance, the simple open‐ended question “Who did you vote for?” can reveal, according to nature of the answer given (name of leader/party/ideology, etc.) the cognitively more accessible, if not the ultimately “true”, voting motivation. In addition, the apparently naïve question, included in each Itanes post‐election questionnaire, asking the interviewee what has been more important in her/his vote choice among such items as: the party, the leader, the local candidate, the coalition, the electoral platform, etc…’ performed unexpectedly well (Barisione, 2007).

19. The distinction between “macro” and “micro” impact is akin to that proposed by King (2002: 41) between “net” and “gross” leader effects.