TRUSTEES OF PRINCETON UNIVERSITY, THE
Media priming has been established as one of the most important ways in which the information environment of electoral campaigns affects public opinion. By changing the criteria citizens use to evaluate policies and politicians, the media conditionally affects mass preferences. Over the past two decades, dozens of studies have demonstrated media priming effects via a wide variety of media content. Recently, however, concerns have been raised about the measurement and identification strategies used to demonstrate media priming effects. A critical decision for researchers involves the timing of the measurement of the primed dimension relative to exposure to media messages. Two alternative strategies are used in the literature: pre-test versus post-test designs. Under the pre-test design the primed attitude is measured before the administration of stimuli, while under the post-test design the measurement is taken after the treatment is given. The importance of this design choice is highlighted in a recent debate about the effects of implicit racial cues, where some find no significant priming effects under the pre-test design, contradicting the empirical evidence previously obtained under the post-test design. The goal of the proposed research is to shed light on this general measurement issue as well as related substantive findings of media priming studies. The PIs demonstrate mathematically that neither pre-test nor post-test design alone is sufficient to identify media priming effects. Existing studies that employ the post-test design risk overestimating media priming effects, while recent studies employing the pre-test design may find no effects even when priming has occurred. To overcome this identification problem, this project combines the two designs in a creative way. This strategy produces more powerful inferences about the media priming effect and offers a new set of research methods that can be used to make stronger inferences about media priming. Experimental tests investigate the consequences of different measurement strategies and their implications for the substantive literature. The broader impacts of this project are twofold. First, the findings will refine the field's knowledge about the nature and extent of media priming effects in modern campaigns. In particular, results will bear on how, when, and why campaigns affect candidate evaluations and policy opinion. Second, the proposed methods will have a wide array of applications in other areas of political science and psychology where researchers face a similar dilemma about alternative measurement strategies and experimental designs.