Lau, Sigelman, Heldman, and Babbitt: The effects of negative political advertisements

Lau, Sigelman, Heldman, and Babbitt. 1999. The effects of negative political advertisements. APSR 93:851-75.

Research Question

Many studies of negative advertising have been done. They tend to address three questions: (1) whether negative ads "work," (2) whether people dislike them, and (3) whether negative ads lead to a disengagement with politics. Yet these studies are not cumulative. Thus, this paper performs a meta-analysis in an effort to aggregate the findings of dozens of studies to determine whether, on the whole, negative ads seem to matter.

Findings

No, negative ads don't matter. Although some studies find significant effects, these are balanced out by studies that do not. The authors control for a variety of factors that differentiate these studies: experiment vs survey, large- vs small-sample, actual vs fake ads, more recent vs older studies, subjectively "good" vs "bad" studies, etc. Even controlling for these factors, the literature does not produce a consistent result that negative ads have significant effects.

Method

See the article. In brief, meta-analysis works like this:

1. Sort the articles into a small number of independent variables (definitions of "negative ads")
2. Sort their dependent variables into a small number of outcomes being tested
3. Standardize the effects of each study, measuring statistical and substantive effects
4. Run some analysis and see whether the studies, on the whole, have produced consistent findings.