**Principal investigators:**
**Joel Middleton**
New York University
Email: joel.middleton@gmail.com
Homepage: http://www.joelmiddleton.com/
**Austin Hart**
American University
Email: ahart@american.edu
Homepage: https://sites.google.com/site/ahartau/
**Sample size:** 2085
**Field period:** 4/13/2011-10/17/2011
**Abstract:**
This study reevaluates the classic “media priming” hypothesis, which argues that, when news coverage raises an issue’s salience, voters align their overall evaluation of the president with their assessment of him on that issue. Experimental studies of media priming typically show greater correspondence between overall and issue evaluations among subjects exposed to issue-related news. The greater correspondence in the treatment group is identified as priming. However, this phenomenon is also consistent with another explanation. Precisely the opposite, the “projection” hypothesis argues that voters exposed to issue news align their opinion on the issue with their assessment of the president’s overall performance. Existing experimental studies cannot rule out this alternative explanation, so we conduct a survey experiment to evaluate the priming and projection hypotheses jointly. Despite recent evidence suggesting that projection is the true underlying effect, our findings support the priming hypothesis. This represents the first unconfounded evidence of media priming.
**Hypotheses:**
Priming: Voters align their overall approval with their issue approval when exposed to issue news.
Projection: Voters align their issue approval with their overall approval when exposed to issue news.
**Experimental Manipulations:**
Respondents read a news article about education or energy, or they read a control story.
**Key Dependent Variables:**
Overall presidential approval. Approval of the president on the issues: education, energy.
**Summary of Findings:**
Our results support the priming hypothesis. We find no evidence of projection in our study.
**References**
Presented at MPSA in 2012. Section: Public Opinion ([link to paper here][1])
[1]: http://tessexperiments.org/data/MiddletonHart.pdf
