An aerial view of the Pentagon. (Charles Dharapak/AP)

On Aug. 25, President Donald Trump issued a memorandum directing the military to ban transgender individuals from service. Among other defenses, some supporters of the ban have argued that transgender people are especially prone to high-risk behaviors — particularly, suicidality — that make them ill-suited for life in the military.

But these arguments — and more important, the statistics they’re based on — suffer from four important limitations.

1. Different studies measure ‘suicidality’ very differently

First, “suicidality” is a broad term that includes more specific concepts, such as suicidal ideation, planning, physical gestures, attempts and completed suicide. Studies on suicidal behavior in different groups often use data on different outcomes. That makes it harder to compare “suicidality” across different groups.

For example, studies of transgender individuals often measure suicidality by asking about suicide attempts. The Williams Institute’s 2014 study, which is widely relied upon, estimates that transgender and gender nonconforming individuals have a national rate of attempted suicide of about 40 percent. But that study’s analysts note that it’s quite difficult to collect data on completed suicides by transgender individuals, in part because gender identification is frequently unavailable. Still, a 2013 study of LGBT youth suggests that there is less evidence that transgender individuals complete suicides at higher rates than other groups.

By contrast, many studies of suicide in the military focus on completed suicides. Scholars note that reliable data on suicide attempts within the military are not available. Though the military has tried to improve observation and reporting of suicide attempts, a 2015 report states that military’s system for documenting suicide attempts is encumbered by uneven reporting across individual bases, clinics and providers.

2. Different studies collect data very differently

Second, different methods of data collection can make it hard to compare different groups. Research on transgender individuals often gathers its data by spreading the word through informal networks and asking volunteers to fill out surveys, called a “convenience sample.” The Williams Institute study mentioned above notes that “since the NTDS [National Transgender Discrimination Survey] utilized convenience sampling, it is unclear how representative the respondents are of the overall U.S. transgender/gender nonconforming adult population.” As a result, it’s difficult to know whether we can generalize from that study.

Researchers from the Williams Institute report also note that the question wording used in these surveys can inflate the numbers of suicide attempts, as individual survey respondents sometimes interpret questions in different ways. When presented with a simple “yes” or “no” question, individuals who say they have indeed attempted suicide may include experiences like thinking about and planning to commit suicide, not just actual physical attempts.

By contrast, the military has the ability to systematically collect data on completed suicides by active duty personnel — but not on attempted suicide, which is likely underreported. A 2009 article by the American Psychological Association explicitly notes that many military personnel see mental health treatment as professionally risky, and fail to seek help — which would enable health-care providers to record reported suicide attempts. There is also evidence that the military may classify data in ways that undercount even completed suicides.

President Trump's tweeted transgender military ban on July 26 drew immediate criticism from both Democrats and Republicans, who were caught unaware by the decision. (Jenny Starrs/The Washington Post)

3. We’re comparing the wrong groups

These arguments often compare groups that aren’t comparable. Studies do report that transgender individuals have high aggregate rates of depression, anxiety and suicidality — at least compared to the general population. A more relevant comparison is to the U.S. military, but this comparison requires caution.

The military is made up of individuals who aren’t representative of the general population. In particular, military suicide rates have been higher than that of the general population. A 2012 study reports that suicide rates within the U.S. Army have increased steadily since 2001. Another study published in 2015 found a significantly lower risk of suicide attempts for Army officers than for enlisted personnel. This study also found that the risk of suicide attempts decreases as a soldier serves more time in the Army. Suicide rates further vary by service branch and occupational specialty.

Using the entire military as a reference group masks the fact that some parts of the military have rates of suicidality that are considerably higher than the general population, or even the military as a whole.

But even comparing the military population with the transgender population at large is a problem. We know that the military population is different from the U.S. population as a whole; we can’t assume that transgender people who enter the military are representative of the broader transgender population.

The ideal comparison would be between transgender and cisgender individuals serving in the military. But there is little reliable data comparing rates of suicidality between transgender and cisgender servicemembers.

The study closest to this ideal uses data collected by the Veterans Health Administration (VHA). That finds that veterans who receive treatment from the VHA and who identify as transgender do report higher rates of suicide-related behaviors than cisgender veterans. But even this study is limited.

First, the outcome of interest, “suicide-related behaviors,” includes several types of behaviors — but excludes some key results that would better enable us to compare rates directly with other populations, like actual suicide deaths.

Second, this study only includes data on 8 million veterans receiving care through the VHA. That’s roughly one-third of the total estimated veteran population. There may be systematic differences between the types of veterans treated through the VHA and those who get their health care elsewhere.

Last, the VHA study focuses on veterans, not active duty soldiers. The evidence shows that suicide rates are driven by a range of personal, situational, social, and economic factors — giving us good reasons to expect different rates in active duty and veteran communities.

4. The military already screens for suicidal behavior; it doesn’t need overbroad proxies like group identity

Suicidal behavior is not an innate trait to any one group, but the result of several causal factors. Not all individuals in a high-risk group are equally at risk, and the military already screens directly for mental health risks in its recruits and personnel. For example, research suggests that screening soldiers’ mental health status before deployments can reduce troubling mental health incidents in theater. Such procedures treat mental health risks directly, improving unit safety without excluding entire groups.

Consider, for instance, the fact that white males have higher suicide rates than most other demographic groups — but the U.S. military is largely white and male.

Banning any demographic group from military service would rule out many qualified individuals. Medical and psychological screening procedures that specifically target mental health risks would appear more effective at enabling the military to recruit and maintain an effective force.

Michael Flynn is an assistant professor of political science at Kansas State University specializing in U.S. foreign policy.

Read more on this issue from the Monkey Cage: