The Iraq Study Group's report states, "There is significant underreporting of the violence in Iraq" by the U.S. military. The too-high "standard for recording attacks acts as a filter to keep events out of reports and databases," as well as news stories, according to Editor & Publisher. Examples of unreported attacks include "a roadside bomb or a rocket or mortar attack that doesn't hurt U.S. personnel" and a "sectarian attack" by an unknown group. In addition, "a murder of an Iraqi is not necessarily counted as an attack." On one day in July 2006, only 93 violent acts were officially recorded, when some 1,100 actually occurred, according to the report. "Good policy is difficult to make when information is systematically collected in a way that minimizes its discrepancy with policy goals," it concludes. One recommendation is to "institute immediate changes in the collection of data about violence and the sources of violence in Iraq to provide a more accurate picture of events on the ground."