A group of soldiers are silhouetted against the evening sky in Kandahar, Afghanistan, in January 2011. (Photo by Ensign Haraz Ghanbari/ Defense Department)

Five years ago, 21-year-old Marine Lance Cpl. Harry Lew kept falling asleep while on guard duty in Afghanistan, a major gaffe for any infantryman in combat. His sergeant told two other Marines in his unit that “peers correct peers,” and so Lew was punched, kicked and forced to do pushups, crunches and other exercises in the middle of the night while wearing body armor, according to a Marine Corps investigation of the incident. Soon after, Lew turned his gun on himself and ended his life.

Lew’s suicide jump-started a debate: What constitutes hazing in the military, and what should the Pentagon do to crack down on the practice?

Lew’s case generated significant interest in Washington in part because of his aunt: Rep. Judy Chu (D.-Calif.). She pressed successfully for an independent investigation by the Government Accountability Office, saying that the stories of her nephew and other victims of hazing — generally described as abusive behavior meant to correct a mistake or earn one’s way into a group — showed the military clearly needed to make improvements.

The GAO released the investigation’s findings this week, reporting that the services have no uniform way of tracking the practice and unclear definitions of what constitutes hazing in the first place. About 12 percent of rank-and-file service members surveyed by the GAO believe that hazing was going on in their unit, the organization said.

“We learned that despite having anti-hazing policies in place, these policies are unevenly implemented and done with little oversight,” Chu said. “In addition, the standards among branches can differ radically, with some not even having a system for collecting data on hazing. We cannot claim that any existing prevention and enforcement policies are adequate without understanding the full scope of the problem.”

[Male-on-male sexual assault in the military: Overlooked and hard to fix, investigation finds]

The GAO carried out the investigation from April 2015 through this month, and found that the Defense Department and Coast Guard have “limited visibility” over hazing incidents. For example, the Army, Navy and Marine Corps track data on hazing cases, but the data is “not complete and consistent due to varying tracking method that do not always include all reported incidents,” the report said.

“For example, until October 2015, the Army only tracked cases investigated by criminal investigators or military police, while the Navy required reports on substantiated hazing cases and the Marine Corps required reports on both substantiated and unsubstantiated cases,” it continued. “The Air Force and Coast Guard do not require the collection of hazing incident data, and instead have taken an ad hoc approach to compiling relevant information to respond to requests for such data.”

The GAO also held focus groups with Marines and sailors at Camp Pendleton and Naval Base Coronado in California, and found a mixture of attitudes and frustrations. Notably, mid-ranking enlisted troops known as noncommissioned officers “reinforced the suggestion that hazing definitions are not sufficiently clear” to determine what is hazing and what is not, the report said.

“The noncommissioned officers we met with generally agreed that the broad definition of hazing prevents them from effectively doing their jobs, including disciplining servicemembers, taking corrective action, or administering extra military instruction for fear of an allegation of hazing,” the GAO report said. “For example, non-commissioned officers during one site visit said that a servicemember need only say ‘hazing’ to prompt an investigation.”

When surveyed, however, more than a third of male Marines (14 of 39) and and nearly half of female Marines (eight of 17) said they had experienced hazing during their military career. About a quarter of male sailors (10 of 40) and female sailors (four of 15) reported the same, the GAO reported.

Clarence A. Johnson, the director of the Pentagon’s Office of Diversity Management, said in a response to the GAO’s preliminary findings including in the final report that the Defense Department’s new policy on hazing and bullying will help the service track cases in the services. It was signed Dec. 23, less than a week after GAO released the draft report to the Pentagon.

The GAO recommended that the services “provide additional clarification” to service members about what constitutes hazing, and Johnson concurred. The services have been directed to revise training or communicate in some other way “further guidance on hazing policies.”

Lew’s case ultimately resulted in the sergeant and a lance corporal being found not guilty of any crime in courts-martial. A second lance corporal was found guilty of assault, and sentenced to 30 days of confinement and reduction in rank to private first class.