Understandably, suicide is a touchy subject under any circumstances. Many of us have lost someone to their own hand, or at least know someone who has been affected by such a tragedy. There is little to be said that can alleviate the aftershock of a suicide, but there inevitably comes the discussion of signs. Was it predictable? Preventable? Did it happen with little to no warning? It is unfortunate that our apparent awareness of these signs often comes after it’s too late.
A post-hoc (meaning “after it happens”) analysis of self-inflicted death is often part of the healing process. If we had direct contact with the person recently, we may feel the need to scrutinize those interactions. This can be a dangerous approach to dealing with a suicide if it is not put into the proper perspective.
Let me diverge for a moment. In scientific research, a post-hoc analysis is any form of statistic that is performed on the results of an intended mathematical analysis. If the post-hoc work is planned ahead of time then it is usually fine to proceed. However, if the math has been completed, and you then decide that a test of the results is further required, you risk introducing a bias into the analysis, thereby confounding and corrupting your results.
How does a post-hoc test hurt an analysis if it is not planned? Well, this is a little conceptual but stick with me and you’ll likely understand. If you have your results and then decide to do a test on those results, you may choose the test that best suits your desires for the intended outcome, and is most likely to produce that outcome given your result numbers. Experienced statisticians have insight on how different post-hoc tests interact with certain types of data. Also, doing an extra test without planning it will introduce an extra source of error (each test has a bit of error that must be reported) that was not accounted for in the initial design, thereby skewing the results and making the entire study unreliable.
Now, let’s look at the analysis of a suicide from a personal point of view. First, you must realize that a suicide, in any form, is an end result. I don’t mean “end” in the metaphysical sense, but rather in the finality of results sense. The suicide is the end result of the initial test (that of living). Rarely does someone plan ahead for the life-taking actions of a family member, friend, or associate, so we cannot assume the post hoc examination of suicide was planned ahead of time. This is especially true when the suicide is most shocking, though you could say that suicide is always shocking in some way.
So, using the statistical analogy, trying to find any suicidal signs in the memories of past interactions with the self-victim is like conducting a post hoc test without having planned to do so. The practice is scientifically invalid, and the information will almost surely be corrupted in some form. Of course this doesn’t make the task any easier, and the search for answers is a normal part of grieving, but putting the validity issue in perspective may at least save us from placing undue blame, especially upon ourselves.
Good vibes to those who need it most,
*This post is dedicated to the memory of Wade Belak, former NHL enforcer and all around nice guy. He was found dead in a Toronto hotel room on August 31, 2011. The cause of death has not been determined as of this writing, but suicide has been alluded to in the media. Wade’s mother recently informed the public that he was indeed suffering from depression. The revelation was shocking to say the least. No matter the cause, the loss of Wade Belak is being heavily felt throughout the NHL and its fans. I can only offer condolences and best wishes to his family and friends.
Photo by Jason Kuffer, available under a Creative Commons attribution license.