Abstract
The conventional wisdom in the literature about political advertising effects—e.g., going negative risks backlash, stick to issues your party owns—has been derived from studies of general elections. Much less attention has been paid to primary elections, in which a partisan audience may be receptive to attacks on the opposing party and may judge most issues to be handled better by their own party. This experiment (N = 223) sets out to investigate the roles of tone (positive versus comparative), target (none, primary opponent, or general election opponent), and issue ownership (party-owned issue or unowned issue) in responses to political advertising during primary versus general elections. As predicted, partisans in primary election conditions had lower ad and sponsoring candidate evaluations for comparative ads attacking a primary opponent than for positive ads or comparative ads attacking the eventual general election opponent, but there were no differences between the latter two. Independents in the general election conditions responded more positively to positive ads than comparative ads. Issue ownership had no main effects.
Notes
Note. Eta-squared and univariate Fs shown only for significant results. Univariate results not shown where multivariate Fs were not significant. Degrees of freedom for the three-way interaction were (6, 394) for the omnibus and (2, 199) for the univariate results. All others had degrees of freedom of (3, 197) for the omnibus results and (1, 199) for the univariate analyses.
∧ p < .10; *p < .05; **p < .01; ***p < .001.
Note. Cell values are means; those in parentheses are standard deviations. In the final column, values in parentheses are eta-squared. F values are for planned contrasts between comparative ads targeting same-party opponents versus the other two types of ads. Means in the same row with different subscripts are significantly different at p < .05 in post hoc Scheffe tests.
∧ p < .05, one-tailed; *p < .05.
Note. Cell values are means; those in parentheses are standard deviations. In the final column, values in parentheses are eta-squared.
∧ p < .10, one-tailed.
This experiment was approved by the university's Institutional Review Board.
The factual basis for the belief that primary divisiveness dooms a party is contested in the literature, with some concluding that divisiveness resulting from primary negativity can hurt the party's candidate in the general election (e.g., Djupe and Peterson Citation2002) and others finding that divisive primaries have little bearing on general election outcomes (e.g., Atkeson Citation1998).
The Pew survey (Pew Research Center 2005) asked which party could do a better job on nine issues, listed here in order from most Democratic-owned to most Republican-owned: health care (51% to 23% Democrats), environment, energy problems, Social Security, education, economy, handling disasters, Iraq, and “dealing with the terrorist threat at home” (45% to 34% Republicans).