although conversations among non-professionals and even law-makers bandy-about such a phrase as if there is.
Because there is no standard, every time the phrase 'severe mental illness' is used it really needs to be defined.
Even so, getting a correct list isn't the only concern when looking at severe mental illness "by the numbers". The numbers are almost always collected from subpopulations that are not infrequently made up of dx's in special settings such as prisons and mental hospitals. Those settings DO influence the reporting and thereby the descriptive statistics on occurrence of violent events. They also limit the associations of occurrence of violence to the group analyzed and discussion of the numbers beyond their context should be considered critically.
Swanson in an often referenced book chapter on violence and mental illness from 1994 defines severe mental illnesses as: Schizophrenia and schizophreniform disorder; major depression, mania, bipolar; obsessive compulsive disorder; panic disorders and phobias.
Such a list may be unsettling to various readers. No one wants to be considered severely mentally ill, that labeling comes with severe social penalties. But Swanson's list can probably be criticized for both its inclusions and exclusions.
One of the things happening in American society this spring, that we must keep in mind, is that Americans are more than ever frightened by mental illness. Much of the professed concern about the reduced state of care for mental illness in this nation is built upon that fear. Americans want to be safe from what Wayne LaPierre calls the lunatics among us.
This creates a strange social/political climate referred to in a handful of other articles in this group about advocating doing the right thing (funding mental health care) for the wrong reason (because society fears the mentally ill).
I think the MJ article in the op is situated within that mixed sentiment.