Individual Submission Summary
Share...

Direct link:

Don’t Shoot the Messenger

Sat, August 31, 2:00 to 3:30pm, Hilton, Jay

Abstract

The media is often criticized for the way it reports survey findings, particularly for election “horserace” polling. The assumption is the media frames polling results with more certainty than what is warranted given the statistical limitations of survey data. Yet, the media’s tendency to overstate the certainty of horserace polling is partly a product of how survey research outlets present their findings to the media. Although outlets report their margin of error, they often fail to provide an appropriate interpretation of how the margin of error should be used to interpret their survey findings and often completely ignore additional uncertainty introduced by large percentages of undecided voters. This paper analyzes polling reports issued from survey research outlets in the 2018 midterms to demonstrate that media coverage of election horserace polling largely reflects how survey research outlets present their survey findings, with too much certainty.

Author