Mobilizer, demobilizer–or artefact? Measuring the influence of research designs on the direct effect of negative advertising on voter turnout in U.S. election campaigns

Abstract

The effects of negative campaign advertising on turnout in U.S. elections vary greatly. Some researchers have suspected that differences in research methods are partly responsible for the conflicting results; however, meta-analyses have only been able to show such effects to a very limited extent. We propose a new approach by using the categorical information of which effect a study found as the dependent variable. Based on all U.S. studies available (29 studies with 133 effects), we examine the influence of research design on the reported direct effects of attack ads on turnout. Our results suggest that some research design choices do indeed impact the results: Null effects are significantly more likely with small samples, mobilizing effects are more likely when contrast ads are considered, and demobilizing effects are more likely when individual-level data are used.