The 'Emboldenment Effect'
By JAMES TARANTO

"Researchers at Harvard say that publicly voiced doubts about the U.S. occupation of Iraq have a measurable 'emboldenment effect' on insurgents there," United Press International reports:

Periods of intense news media coverage in the United States of criticism about the war, or of polling about public opinion on the conflict, are followed by a small but quantifiable increases in the number of attacks on civilians and U.S. forces in Iraq, according to a study by Radha Iyengar, a Robert Wood Johnson Scholar in health policy research at Harvard and Jonathan Monten of the Belfer Center at the university's Kennedy School of Government.
The increase in attacks is more pronounced in areas of Iraq that have better access to international news media, the authors conclude in a report titled "Is There an 'Emboldenment' Effect? Evidence from the Insurgency in Iraq." . . .
In Iraqi provinces that were broadly comparable in social and economic terms, attacks increased between 7 percent and 10 percent following what the researchers call "high-mention weeks," like the two just before the November 2006 election.

The study is here. One possible objection is the way the study counts "emboldening" statements (see pp. 9-10):

We construct an automated mentions count of potentially "emboldening" statements reported in major U.S. news outlets, which we define as the number of times top Bush administration officials--the President, Vice-President, Secretary of Defense, Secretary of State, Press Secretary, and the U.S. commander [in] Iraq--refer to statements or actions by other U.S. political figures that might encourage violent extremist groups in Iraq. This strategy provides an objective mechanism to classify what constitutes an emboldening or resolve-undermining statement or action.

This is probably the best the authors could do. There is no way to count "emboldening" statements directly; the definition invariably would be subjective and the count incomplete. But by relying on characterizations of Bush administration officials (and the commander in Iraq), they leave open the possibility that it is those characterizations, rather than the statements being characterized, that embolden terrorists in Iraq. In addition to "emboldening" statements, the authors also find a correlation between the release of major U.S. opinion polls about Iraq and violence there.