We WILL NOT ever and CANNOT psychologically predict shootings? Read why.

We don’t have the stats for predicting these shootings nor will we. The issue is that “reasons” for such shootings are inconsistent at best. There is no consistent pattern of why people engage in these shootings. Consider this following list of reasons that have been cited. Then some of these potential reasons are simply guesses due to a postmortem analysis for those who have been killed or taken their own lives.

• vengeance
• white supremacy
• use of legal or illegal drugs
• alcohol
• instill violence (actually stated)
• bullying
• divorce
• loss of job
• isolation (stated)
• family problems (stated)
• upbringing
• violent video games
• mental disorders/delusions
• political point
• racial discrimination or in retaliation to it
• religious ideology or discrimination
• attempting to cause a race war (stated once)
• abortion rights (think planned parenthood examples)
• one girl said I don’t like Mondays (stated)
• hatred of group X (stated)

As you can see this list is seemingly random with no or rhyme or reason. Some even being the opposites of each other. I am stating that there is no pattern that will be predictive. Furthermore any good predictive model will distinguish (discriminate) between those who will become shooters and those who will NOT. Sure psychologists will come up with correlations at best yet with no predictive assurance. Graduate students with do some clever dissertations and find some odd missed variable that correlates and so on. After all you may yourself know individuals who may “fit” a “profile” of a shooter yet never enact a harming behavior. For example your acquaintance may now have a list of “variables” that correlate with individuals that have enacted these shootings. However, they never become a shooter. I personally have known many guys who match these variables that correlate with shootings. However they have not done anything. So the predictive model to have value must actually predict those who will NOT do it. Another way of thinking of it is this. It may be your weird neighbor who you think may flip out one day yet never does and then your quiet neighbor who does flip out and you never suspected it.

Of course people are going to attack guns as well. As I am sure you know I am not a fan of guns. It was swept under the rug that the shooting at Columbine over 20 years goes, the students initially tried pipe bombs. Those failed and they fell back on guns as a back up. Even, if we eliminated all the guns (an impossible task) people will come up with clever alternatives. Like pipe bombs, toxic ones, etc. who knows. There are too many clever people out there who want to do bad things.

We simply do not have enough shootings as morbid as that sounds for statistical analysis. We need to break them down by age groups, ideologies, religions, dozens and dozens of variables. Furthermore, then we need replication studies which means more shootings (morbid again) to provide supporting evidence that the model works.  Again, as I have stated in other posts you cannot have a one time study without replication to support your position. A one time study is merely that and may be coincidence hence we call it a pilot study. This is why we want dozens of studies that support the initial study. Yet that means more shooters.

Note to students: For those of you studying psychology we always have the external validity concern of the historical variable. Before, 911 if you were doing a study on say anxiety, depression or many types of research you were in trouble. You took a pre measure of anxiety/depression etc. before the atttack. Then the attack happens… any post measure thereafter is contaminated by the attack as one can imagine on those two stated studies and the data is useless for the main purpose of the study. The same can/may be true with shootings. Ones that happened decades ago are contaminated by a variety of historical variables that interrupt.