ProPublica is a Pulitzer Prize-winning investigative newsroom. Sign up for The Big Story newsletter to receive stories like this one in your inbox.

This story also appeared in ProPublica

Each year, hundreds of Army soldiers face trial in military courts for offenses that range from murder to failure to report for duty.

The military justice system largely operates separately from the civilian legal process and is unknown to many Americans. Under the system, commanders, who are not required to be trained lawyers, wield significant influence and can detain soldiers while they are awaiting trial through a process known as pretrial confinement.

An investigation by ProPublica and The Texas Tribune found that soldiers accused of sexual assault are less than half as likely to be placed in pretrial confinement than those accused of offenses like drug use and distribution. Soldiers accused of other more minor offenses such as disobeying an officer or damaging nonmilitary property also have higher rates of pretrial confinement than those accused of sexual assault.

That gap held up even when narrowing in on certain types of sexual assault cases that we suspected might be more likely to result in pretrial confinement.

When we limited the analysis to charges involving the most violent sexual offenses, the pretrial confinement rate remained well below that of drug offenses. (Sexual assault charges include a range of offenses, from rape to nonconsensual contact of body parts like the inner thigh.)

Our analysis also focused on charges, not the outcome of the trial, since the decision to put a soldier in pretrial confinement is made before a case is adjudicated and because guilt is not a requirement for pretrial confinement. But even for soldiers who were eventually found guilty of at least one sexual assault offense, the rate of pretrial confinement remained well below the rate for drug cases, with or without a drug offense conviction.

Here’s how we did this

ProPublica and the Tribune examined nearly 8,400 cases over a decade that went to the Army’s general court-martial and special court-martial, which are sometimes likened to felony and misdemeanor courts, respectively, in the civilian system.

A case was considered to have pretrial confinement if the soldier was held or given credit for at least one day in confinement.

We analyzed the rate of pretrial confinement for the 25 offenses most frequently referred to courts-martial, including sexual assault, drug crimes, physical assault, larceny and being AWOL.

ProPublica and the Tribune spoke with military law experts and individuals familiar with the way Army records are maintained to inform our analysis and review our findings. Experts provided guidance on the accuracy of data fields, informed our approach to standardize charges and helped to vet our findings.

Over time, the military has changed how it charges some offenses. For our analysis, older charges were standardized to the most recent edition of the Manual for Courts-Martial, which guides trial proceedings. We did this by looking at the description of the offense and articles the case was charged under in the Uniform Code of Military Justice at the time an alleged offense took place. (Our analysis focuses on cases that went to courts-martial in the past decade, but a handful of those cases used charges from older editions of the manual.)

For example, the military’s criminal laws used to group certain consensual and nonconsensual sexual acts under the same article when charging soldiers. We aligned these rape and sexual assault charges with the recent edition of the manual and did not count charges related to consensual acts as assault.

We analyzed sexual assault offenses by looking at cases with charges of either sexual assault of an adult or of a child.

Cases with a murder or sexual assault charge were grouped only with their most serious offense. Put another way, if a soldier’s case included both murder and sexual assault charges, our analysis of pretrial confinement grouped it with murder charges, not with sexual assault offenses. And cases that included sexual assault offenses and more minor charges were grouped only with sexual assault.

If a case did not include one of those more serious offenses but had multiple charges, we analyzed pretrial confinement for each charge. For example, if a case included larceny and drug use, it was grouped with the larceny offense and the drug offense cases. This was necessary because beyond murder, homicide and sexual assault, there is not a clear way to rank the roughly 600 offenses by severity.

We also looked at pretrial confinement by trial location. For about 1% of cases, we corrected the trial location because it was entered imprecisely.

We found that the rate of pretrial confinement varied greatly by location. For sexual assault cases, it ranged from just under 4% at Fort Sill in Oklahoma to about 19% at Fort Leonard Wood in Missouri. We looked at U.S. installations that tried or arraigned at least 20 sexual assault cases and 20 cases of other types, excluding murder.

At 20 of the 29 installations, sexual assault cases had a lower rate of pretrial confinement than other cases that were tried or arraigned.

We will continue to investigate military justice

The database is expansive, with trial records from five continents and cases going back to the 1980s. But it has noteworthy limitations.

Military regulations require commanders to consider if lesser restrictions (like requiring regular check-ins) are insufficient before placing a soldier in pretrial confinement. For this story, we could not assess whether individuals received or followed restrictions before confinement because this information was not reliably recorded in the database. Of cases that had pretrial confinement, only 7% noted pretrial restrictions. Our review of case documents revealed instances where soldiers were subject to pretrial restrictions that were not recorded in the database.

We also know that some types of cases are not captured in the database, including cases where a soldier withdrew from military service ahead of arraignment or was punished outside of court. According to the most recent report to Congress on military justice, the Army used nonjudicial punishment in more than 25,000 cases in the past fiscal year. During that time, only about 700 people were arraigned in the branch’s highest trial courts.

We know other offenses are never reported at all. A 2020 Department of Defense report on sexual assault in the military estimates that for every service member who reports a sexual assault, at least two more do not.

ProPublica and the Tribune will continue to investigate the military justice system, which regulates the conduct of more than 1.3 million active-duty service members.

Evidence from trials, nonjudicial punishment and administrative actions are essential to our investigations. Your story is important to us.