This undated photo from a MySpace page that appeared to belong to Chris Harper Mercer shows him holding a rifle. Authorities identified Mercer as the gunman who went on a deadly shooting rampage at Umpqua Community College in Roseburg, Ore., on Thursday. (MySpace via AP)

After the recent mass shooting at Umpqua Community College in Oregon, many people, including President Obama, explicitly tied the shooting to issues of mental health. Indeed, this frame has become a staple of the National Rifle Association and others who champion gun ownership rights.

For those concerned with the stigmatization of mental health in America, the connection between mass shootings and mental health is troubling. Existing research suggests that people with mental health problems are often stigmatized because we associate antisocial behaviors with mental illness. Previous research has shown that media coverage is one key source of these stigmatizing attitudes.

So what happens to news coverage of mental health after well-publicized acts of violence? Our research shows that it becomes much less favorable.

We gathered every New York Times article between 1985 and 2014 that discussed mental health issues (about 30,000 articles). This gives us a sense of any trends in the amount of coverage.

Then, using an automated content analysis program and dictionary that classifies words as positive or negative, we measured the tone of these articles. For example, words like “hysterical,” “dangerous” or “fear” would suggest less favorable coverage of mental health issues, while words such as “calm” or “improved” would suggest more positive coverage.

The figure below shows that the number of mental health articles published in the New York Times has increased sharply over the past 30 years.

The figure below shows that the tone of mental health coverage in the New York Times has improved only modestly since the early 1990s. At its lowest point, New York Times coverage featured about 11 more negative words than positive words for the average article discussing mental health.

At its highest point, New York Times coverage still featured about seven more negative words than positive. In fact, of the 360 months in our dataset, in only three was the tone of coverage positive on average.

To put these numbers in context, a recent study that applied this same method found that even during the height of the Great Recession, news articles about the economy used only about four more negative words than positive for an article of the same average length.

In other words, the most negative coverage of the economy — even in the peak of a deep recession — is almost twice as positive as the most positive coverage of mental health issues.

But it gets worse. We used a statistical technique to estimate how what we refer to as “rampage violence” affects mental health coverage. By rampage violence, we mean attacks on multiple parties, even on whole institutions, such as schools or communities. The Oregon shooting is one example.

We found that the typical act of rampage violence (an average of nine fatalities) is linked to coverage of mental health that is even more negative than usual. We also found that an act of rampage violence is linked to an increase in the number of articles that discuss mental illness.

In other words, not only does coverage of mental health become more negative, it also becomes more prominent. This combination has the potential to exacerbate the stigma associated with mental illness.

There’s no easy solution here. Although the stigmatization of mental health is troubling, many perpetrators of rampage violence appear to have suffered from mental illness.

At the same time, however, research shows that “only a small proportion of the violence in our society can be attributed to persons who are mentally ill.” For this reason, even in the aftermath of tragedies like the one in Oregon, we need to be careful about the words we use if we care at all about the perceptions of those with mental illness.

Denver McNeney is a PhD candidate at the Center for the Study of Democratic Citizenship at McGill University. Stephanie Parent is a researcher at Reichert and Associates.