Fixing May's Law? An Empirical Assessment of a New Measurement Approach

Is May's Law empirically valid? In accordance with May’s Law, empirical research shows that a political party's so-called “non-leaders” (e.g. voters) typically hold more moderate policy-preferences than its “sub-leaders” (e.g. active party members). However, when it comes to higher party ranks, May's idea of a curvilinear intraparty opinion structure is not well supported. In fact, a party's “top-leaders” (e.g. elected officials) rarely hold more moderate policy-views than the respective sub-leaders. On average, top-leaders even take the most radical positions within a party (Narud/Skare 1999; Van Holsteyn et al. 2017). I argue that this divergence is partly predicated on a basic mistake in the conceptualisation, and ultimately operationalisation, of May's Law. More precisely, May's key argument for the moderation by the top-leaders is flawed. Since the top-ranking positions are typically electively assigned, top-leaders have an incentive to maximise votes by moving closer to the political stands of the Centrist-minded non-leaders. This is a valid reason for strategically modified political behaviour. However, the idea does not help to understand individuals' political views. Hence, we expect the hypothesised curvilinear opinion pattern to be more prevailing when the actual issue positions of a party are contrasted with the preferences of the sub-leaders and the non-leaders. Using data from the German Party Membership Studies of 1998, 2009 and 2017, our analysis is based on the most comprehensive data available for investigating May's Law. It enables us to avoid common problems, such as insufficient observations in a given echelon or heterogeneous survey dates and items for varying strata (e.g., Van Holsteyn et al. 2017). In line with previous research, our analyses indicate that May's Law should by no means be considered a law. However, when the operationalisation is aligned with May's argument by using party positions instead of top leaders' policy attitudes, the results change significantly in favour of a curvilinear intraparty structure, making it one of the dominant patterns observed.