conditionality and leader effects in parliamentary elections”
The study of how party leaders, sometimes called prime ministerial candidates, influence both the individual voting decision and the inter-party distribution of the vote in parliamentary elections is an exciting and expanding subfield of mass political behaviour. It is also of relatively recent vintage. For a long time, the argument that party leaders enjoyed a discernibly independent, perhaps critical, electoral influence was greeted with scepticism. The conventional wisdom held that parties were the dominant force in this type of contest and their national leaders were barely distinguishable from them in the eyes of most voters. This was so for three reasons: (i) the institutional reality is that parliamentary elections were, and are, local as opposed to nationwide contests between parties; (ii) voters contextualize the complex world of parliamentary politics in terms of parties; and (iii) as a direct result, voters tend not to distinguish between parties and their leaders. Put differently, party leaders were dismissed as at best bit players in the larger parliamentary election drama (Butler and Stokes, 1969; Crewe and King, 1994; Kaase, 1994; King, 2002).
Tables and Figures:
Figure 1. Proportions of 'very strong' identifiers in Australia and Britain
Table 1. Baseline logit vote models with no leader effects.
Table 2. AIC goodness-of-fit statistics for leader efects models.
Table 3. Change in probabilityof voting for first-named party per standard deviation scale change.
Table 4. Change in probability of coalition vote per standard deviation scale chang: Australia 1998-2007.
This raises the question of the latter’s appropriateness as a heuristic for a valence model of vote choice. To be sure, the party leader ‘is a clearly identified, single individual who, if elected, will take ultimate responsibility for any hard choices that the government needs to make’ (Clarke et al., 2004: 9), but the admirable clarity and simplicity of the valence model become muddied when party evaluations necessarily enter into a full understanding of the parliamentary electoral calculus. ‘Parties typically have local, regional, and national organizations that do not always convey a single, unified political message. The platform that a party presents at election time is frequently the result of public, intra-party debate among competing personalities who have previously articulated different policy positions and even represented different intellectual traditions. In these circumstances, it is not always clear to voters what “the party” actually stands for’ (Clarke et al., 2004: 9). In other words, resort to the ‘leaders vs. other party model’ may specify the conditions under which leader effects are at their strongest, but this specification comes at the cost of seriously compromising the integrity of the so-called valence model of electoral choice.