When a U.S. military operation results in the death or injury of civilians, how does that harm get reported? Who can make a report? And what does the investigation that follows—if any—look like?

To civilians living in Afghanistan, Iraq, Syria, Somalia, Yemen, and other countries where U.S. military operations take place, these questions take on critical importance when their communities, neighborhoods, and families suffer harm. Yet too often, those whose loved ones are killed, or who themselves are seriously injured, don’t know where to turn for answers. In many cases—for example, in a drone strike, or during an air campaign when many countries are carrying out airstrikes—they may not even know who is responsible. Rahat Gul, an Afghan man whose son was killed in a U.S. strike carried out in September on a gathering of pine nut farmers, told The National, “We’re sad and angry, but what can we do? We wouldn’t even know where and how to file an official complaint. We’re powerless.”

When investigations do occur, whether they’re triggered by civilians reporting attacks or internal military reports, we need to be able to rely on their conclusions. Without a thorough investigation, civilians harmed have no way to seek accountability, or claim redress.  The public can also fail to understand the full human costs of a military operation or a war. For example, in Operation Inherent Resolve, during the two first years of the campaign, official estimates limited the civilian death toll to 152 civilians, whereas Airwars, a civilian harm monitoring group, estimated a minimum of 1,500 civilian casualties—a 90 percent discrepancy.

The U.S. military has also repeatedly emphasized that careful investigations are necessary to maintain order within the institution, to learn lessons from mistakes, and to ensure that claims of civilian harm are properly understood and addressed.

So, if good investigations are so crucial to both the military and the public, why are they not happening more consistently?

“In Search of Answers,” a new in-depth report published by the Center for Civilians in Conflict and the Columbia Human Rights Institute, provides some answers to this critical question. Over a period of two years, our research team analyzed a total of 228 investigations into reported civilian harm in Afghanistan, Iraq, and Syria between 2002 and 2015. We conducted in-depth interviews with current and former military officers and U.S. government officials, as well as with civil society representatives with expertise on investigation procedures and responses to civilian harm. Among those interviewed were human rights experts from Afghanistan, Somalia, and Yemen who have worked extensively on the issue of civilian harm in these countries. Our interviews covered not only practice prior to 2015, but more recent operations which saw significant changes in how initial reports of civilian harm were evaluated, and whether or not they ultimately resulted in a full-blown military investigation.

We found that in a number of cases, the military pursued investigations rigorously, collecting information from witnesses and visiting the site of an attack. But overall, the military’s record over 18 years is inconsistent, and from the perspective of many involved, inadequate. In the incidents we reviewed, military investigators frequently did not show they had taken steps to acquire important evidence regarding an incident, nor did they explain why they had not taken these steps.

Many of the investigations relied exclusively on internal military records and military witnesses. While these sources tell critical parts of the story, they too often leave out other crucial facts, as both individuals with experience of the system, and public statements around investigations, told us.

For example, after the U.S.-led coalition airstrike struck a residence in the al-Jadida neighborhood of Mosul in Iraq, killing over 100 civilians, the U.S. military assembled a team of 12 investigators who spent several weeks looking into the incident. The investigation ultimately revealed that neither the U.S., nor its Iraqi partner forces, knew that dozens of civilians had sheltered in the building’s basement prior to the strike. An in-depth ground investigation told them what neither internal records nor U.S. military sources would have shown.

We found that the military had interviewed civilian witnesses—survivors, family members, eye witnesses to an attack—in only 21.5 percent of the incidents we reviewed, despite the fact that civilians often have both important factual details, and contextual information relevant to assessing an attack. And investigating officers had only visited the sites of harm to look for physical evidence and assess damage in only 16 percent of these cases. In some incidents, investigators made up for these deficits by asking interviewees to come to another location for interviews, asking community members or local partners to collect photographs and physical evidence, or by taking other measures to collect needed information. But we also noted that investigators rarely explained why they did not take such measures, and that their conclusions were still accepted by commanding officers and legal reviewers.

Most importantly to the civilians harmed by U.S. operations, investigations rarely resulted in any form of public disclosure. While incidents with large death tolls, like the airstrike in al-Jadida, or the 2015 bombing of the Médecins Sans Frontières hospital in Kunduz, Afghanistan, may result in a public briefing, those harmed in other incidents may receive no information on the results of an investigation, or updates on its progress. When the military does not publish investigation results, impacted countries and communities may believe the military failed to consider the harm they suffered, and will not hold itself to account. The good practices we noted in some investigations, for example when investigating officers overcame obstacles and sought out additional interviews or evidence, are difficult to replicate if they are not publicly shared. Meanwhile within the military, the way investigations are recorded leaves little opportunity for lessons learned to be transmitted across commands.

The Defense Department is currently working on a new, comprehensive policy to address the way it handles civilian harm, including investigations, reporting, and compensation. The new policy should work to create greater consistency and more widely applied standards across military investigations. It should also require investigating officers to document why they do not interview witnesses or visit sites of harm, and should ensure that investigation results are more regularly and consistently reported to the public. Without these changes, the military risks eroding public trust in how it investigates potential mistakes, and whether it is willing to improve. Meanwhile, until new policies are implemented, civilians impacted by U.S. operations—often with no way to contact those responsible—will continue to suffer terrible losses without any accounting for their harm.

Editor’s note: There will be a panel discussion about the report on Friday, Feb. 14 at 11 a.m. at the American Red Cross. The event is co-sponsored by Columbia Law School’s Human Rights Institute, the American Red Cross, the Leiber Society, and American Society of International Law.

Image: The damaged interior of the hospital in which the Medecins Sans Frontieres (MSF) operated is seen on October 13, 2015, following a U.S. air strike in the northern city of Kunduz. AFP via Getty Images