The forthcoming article “The Variable Persuasiveness of Political Rhetoric” by Jack Blumenau and Benjamin E. Lauderdale is summarized by the authors below.
Which types of political rhetoric are most persuasive? Political arguments often feature common rhetorical elements, regardless of the specific policy at stake. Politicians can draw on endorsements from relevant authorities; take moral stances; articulate costs and benefits; impugn the motives of opposition actors; make comparisons with historical experiences; and so on. Learning whether some forms of rhetoric are generally more persuasive than others is important for speaking to contemporary normative theories of rhetoric, which suggest that some forms of political argument can threaten deliberative ideals. For instance, if voters consistently respond to arguments that are low in informational content but rich in bombast and élan, we might worry that the quality of deliberation has fallen.
However, evaluating the relative persuasiveness of particular types of rhetoric is difficult because the arguments that politicians make are highly multidimensional. Arguments might deploy common elements, but they also vary in many other ways that make certain strategies effective in some instances but not in others.
We propose a survey-experimental approach for measuring the relative persuasiveness of different rhetorical elements that addresses this challenge. In our experiment, we present pairs of arguments to survey respondents and ask them to assess which of the pair is most persuasive. We use 336 individual arguments, each of which uses one of 14 distinct rhetorical elements on each side of 12 policy issues in UK politics. By presenting many implementations of each element, we can draw inferences about the relative effectiveness of different rhetorical strategies averaged across different political issues, thereby minimizing the risk that our conclusions will be confounded by the idiosyncrasies of single implementations.
This design also recognizes that researchers seldom want to test the effects of particular treatment texts. Rather, they typically want to make broader claims about a latent treatment, of which a treatment text is just one implementation. Because our focus is not on the effects of specific texts, but rather the distribution of such effects across implementations, we use a multilevel-modeling approach which allows us to estimate this distribution using several implementations, each of which would be statistically underpowered if analyzed alone.
We find that there are modest average differences between different rhetorical element types. One of the strongest rhetorical elements is appeals to authority – arguments that seek support for an issue by reporting the view of an entity with relevant subject area expertise. By contrast, the weakest arguments are those that employ ad hominem attacks and those that rely on metaphor and imagery to win support for a policy stance. When compared to many other common forms of political rhetoric, arguments of these types are do have differentially persuasive effects in the eyes of the UK public, at least on average.
However, we find that the heterogeneity in the persuasiveness of specific implementations of rhetorical elements is much larger than these average differences. Appeals to authority are more persuasive than other rhetorical styles on average, but some appeals of this sort are still among the weakest arguments we test. Similarly, comparisons to other countries feature in the lists of the most and least persuasive arguments, depending on the specific implementation. This represents an important lesson for the interpretation of existing studies in political communication, many of which are based on experiments relating to single policy issues.
Ultimately, our goal is to understand which types of arguments induce voters to support or oppose policy proposals on different issues. However, persuasion is different from the self-reported judgements of argument persuasiveness that we elicit in our experiment. We therefore conduct a second, out-of-sample validation experiment and show that respondents’ evaluations of which arguments are more persuasive strongly predict the direction and magnitude of those arguments’ ability to persuade different respondents to change their stated attitudes. The validation demonstrates substantial persuasion effects on average, but we again observe large variation in these treatment effects across issues.
This central substantive finding of argument persuasiveness heterogeneity suggests why scholarship on persuasion has not clearly decomposed the sources of persuasive appeal into distinct rhetorical elements: it is very difficult to identify rhetorical strategies that are consistently more persuasive than others when considered across multiple policy issues. Our findings imply that the persuasiveness of different argument types is likely to be highly context-dependent, and that analysing the rhetorical structure of arguments allows us to predict persuasiveness only to a limited extent.
About the Authors: Jack Blumenau is an Associate Professor of Political Science at the University College of London and Benjamin E. Lauderdale is a Professor of Political Science and Department Head at the University College of London. Their research “The Variable Persuasiveness of Political Rhetoric” is now available in Early View and will appear in a forthcoming issue of the American Journal of Political Science.
Speak Your Mind