Toll Booths on the Information Superhighway? Policy Metaphors in the Case of Net Neutrality

Abstract

Scholars have argued for centuries that metaphors are persuasive in politics, yet scant experimental research exists to validate these assertions. Two experiments about the issue of federally regulating the Internet were conducted to test whether metaphors confer a unique persuasive advantage relative to conventional messages. The results of these studies confirm that an apt metaphor can be a powerful tool of persuasion. Moreover, the evidence suggests that metaphor-induced persuasion works particularly well for politically unsophisticated citizens by increasing assessments of message quality. Ultimately, this research concerns how individuals make sense of politics and how policymakers can use what we know about human cognition to convey their platforms to the general public.

Keywords:

• metaphor
• persuasion
• policy
• net neutrality
• experiment

Acknowledgments

I would like to thank Howie Lavine, Milt Lodge, Stanley Feldman, Rick Lau, Chris Weber, Rune Slothuus, and Todd Belt for their helpful comments.

Notes

1. Lakoff and Johnson (1980; see also Gibbs, 1994, 1996) argue that all concepts are inherently metaphorical and embodied in our everyday experiences.

2. Note that a demand-side metaphor like drug use is an illness has different implications for how to address the problem in society; namely, drug users should treated like patients suffering from an illness

3. It may also be worth noting that metaphors can be used to obscure thinking, as Orwell (1947) argued in his essay “Politics and the English Language.”

4. The meta-analysis also included 12 unpublished works (largely from doctoral dissertations and master's theses), which would bring the total number of experiments on metaphor-induced persuasion to 29. However, given that these unpublished studies were not subjected to peer review, I opted to exclude them from my discussion.

5. For each study, Sopory and Dillard (2002) estimated effect sizes (r) based upon cell-to-cell comparisons calculated from the reported t or F statistics.

6. The bipartisan survey was conducted by the Glover Park Group and Public Opinion Strategies in September 2006.

7. The exact wording for these items is as follows (correct responses are indicated in italics; frequencies of correct answers are in parentheses): (a) Whose responsibility is it to determine if a law is constitutional or not? Supreme Court (79%); (b) Which party currently has the most elected members in the U.S. House of Representatives? Democratic Party (62%); (c) What job does Harry Reid currently hold? Senate Majority Leader (35%); (d) How much of a majority of both the House of Representatives and Senate are required to override a presidential veto? 2/3 (63%); (e) Which one of the parties is more conservative than the other at the national level? Republican Party (79%); (f) How many justices are there on the U.S. Supreme Court? 9 (53%); (g) What job does Condoleezza Rice currently hold? Secretary of State (82%); and (h) Which branch of government does the U.S. Constitution give the sole authority to declare war? Legislative Branch (44%).

8. Predicted values were calculated by holding all other variables constant at their mean values or reference categories.

9. The wording of the metaphor and literal conditions varied only slightly from Study 1, with the most notable difference being the addition of one line. For the metaphor-based message, the last line read: “I don't know about you, but I don't like the idea of having toll booths at every on-ramp on the information superhighway.” In contrast, the last line of the literal comparison read: “I don't know about you, but I don't like the idea of having special fees imposed on content providers on the Internet.”

10. Given that political sophistication is central to one of the hypotheses, I provide the summary statistics for this measure: KR-20 = 0.70, M = 0.60, SD = 0.29. The exact same items from Study 1, were used in this index, with the following frequencies of correct responses: (a) 79%, (b) 62%, (c) 35%, (d) 63%, (e) 79%, (f) 53%, (g) 82%, and (h) 44%.

11. The coefficient is negative in this case because the metaphor condition serves as the reference category.

12. For difference of means tests, all variables were recoded to range from 0 to 1 so that the results could be meaningfully interpreted.

13. Models in which policy attitudes were regressed on interactions of the experimental condition variable with computer expertise or interest were not statistically significant, regardless of whether I used the pooled data or considered each study individually.