Candidate Strategies for a Policy-only Turnout Model

Here we assess two alternative voting models concerning candidate strategies that place Hypotheses 1 and 2 (introduced in section 8.2) in perspective. Both the illustrative arguments from Chapter 7, section 7.3 and the results presented in prior spatial modeling studies (e.g., Riker and Ordeshook 1968; Hinich and Ordeshook 1970; Enelow and Hinich 1984; Erikson and Romero 1990; Anderson and Glomm 1992) suggest the following hypotheses for margin-maximizing candidates:

H3: For a policy-only turnout model – that is, one that omits measured nonpolicy variables – the two candidates' optimal policy positions converge along the policy scales.

H4: For a unified turnout model that omits abstention due to alienation – that is, one in which abstention is specified as being motivated solely by indifference – the two candidates' optimal policy positions converge along the policy scales.

Hypotheses 3 and 4 are important because if they are supported, these results will bolster our central theoretical claim that while the combination of voters' non-policy-related motivations and abstention due to alienation motivates candidate divergence (H1), neither alienation nor nonpolicy motivations alone motivates divergence (H3 – H4).

In contrast to the unified turnout model explored earlier, most spatial modeling studies omit measured nonpolicy variables. First, we evaluate Hypothesis 3, that such a voting model motivates candidates to present convergent policies – the standard result in the spatial modeling literature. In order to explore this hypothesis, we estimated the parameters of a policy-only turnout model for the 1988 ANES that included only respondents' policy evaluations as independent variables. The parameter estimates for this model are reported in Table A8.1.