The media have recently printed many stories of violent acts which have been carried out by young people who were know users of cannabis (marijuana).  This has led to much discussion as to whether the use of cannabis (marijuana) has caused the violent acts.  The following comment shows that although correlation is not causal it is nevertheless sobering to see how often this occurs and clearly more research is needed.

 Studies showing an ‘Association between  Pot smoking and Violence

There are many correlation studies that find an “association” between two independent variables (e.g., pot smoking and violence). We cannot dismiss the importance of these associations but, at the same time, it would be unscientific to interpret causation from such an association. The latter requires far more than a linear association or correlation. Remember, correlation is NOT causation. There are statistical formulas for measuring correlation (e.g., Pearson product-moment correlation coefficient) and the closer to 1 that the measurement shows, the stronger is the relationship or association between the two variables. Measuring the correlation coefficient is a useful examination of data that intuitively may suggest a direct relationship. Scientists often use such studies to justify more elaborate and costly investigations.

In the 1950s, for example, British and American epidemiologists were concerned about the correlations they observed between cigarette smoking and lung cancer. As the data continued to show this association over time, it inspired additional research that, ultimately, some two decades later, resulted in finding that cigarette smoking “causes” lung cancer by causing changes and mutations at the cellular level.

 

Another example – and one from today – is the “association” or correlation reported between brain tumors and cellular telephone use. Hardly a month goes by without a report that claims one thing or another, usually the polar opposite of each other. Perhaps this mystery will be solved someday but for now, we must live with the uncertainty of the association. It’s much the same with pot and violence. For now we must acknowledge that there is an association, as the authors state in the article cited below. For us to make the leap to causation on the basis of these data, however, would be unscientific and incorrect. We may very well have that ability someday but for now we have to stick with what the data are telling us.

 

Let me give you one final example that I hope may explain the difference between correlation and causation. Tomorrow, I’m giving a presentation to a drug industry group in Washington, DC. In preparation, I obtained unpublished information from DEA on the medical distribution of oxycodone in the U.S. over an eight-year period. I then compared these data with hospital emergency department admissions for the same period in which the misuse or abuse of oxycodone was involved. You can see the results of this comparison below. The correlation coefficient when measured was 0.9796 – very close to 1, meaning a very strong linear association between the two variables (oxycodone and emergency department admissions). The probability of this relationship being a result of mere chance only was found to be one in ten-thousandths (p = 0.00001). This, I’m sure you would agree, is a very close relationship between these two variables. Would it be correct to assume causation? NO! The reason that this would be incorrect is quite simple.

The majority of oxycodone distributed for medical use in the U.S. is used properly and as directed to address a therapeutic need. We would be pushing the interpretation of these data into very troubled areas if we were to assume on the basis of the association we found that the use of oxycodone “causes” people to be admitted to hospital emergency departments for misuse or abuse of the drug. Nonetheless, as the graph shows quite clearly, there is a very close relationship between the two phenomena that warrants further research and attention.

Source:   John Coleman   Drugwatch International July 2013

Violent Behaviour as Related to Use of Marijuana and Other Drugs

Abstract

The relationship of the degree of use of each of ten types of illicit drugs with each of eight types of violent criminal offenses, is reported for an African-American, inner-city, low SES, young adult study sample (N = 612). Prospective data from the time of birth was available for the statistical analyses, to provide 51 control variables on factors other than substance use which might predict to later violent behavior.

Findings: Greater frequency of use of marijuana was found unexpectedly to be associated with greater likelihood to commit weapons offenses; and this association was not found for any of the other drugs, except for alcohol. Marijuana use was also found associated with commission of Attempted Homicide/Reckless Endangerment offenses. Cocaine/crack and marijuana were the only two types of drugs the frequency of use of which was found, for this sample, to be significantly related to the frequency of being involved in the selling of drugs. These findings may not apply to a middle-class African-American sample.

Source:  Journal of Addictive Diseases  Vol.20 Issue 1   2001

The Differential Disinhibition Effect of Marijuana Use on Violent Behaviour

Abstract

The following Kaplan/Damphouse hypothesis was tested and cross validated: The use of marijuana either predicts to or has a greater effect on increasing the degree of violent behavior for a group that is low on delinquent behavior, than it does for a group that scores high on these behaviors.

For the conventional, non-delinquent subgroup, a higher degree of significant relationship between degree of marijuana use and degree of violent behavior was found, compared to the degree of this type of relationship than was found for either cocaine/crack use, amphetamine use, or tranquilizer/sedative use. For example, for the commission of the offense of Attempted Homicide/Reckless Endangerment: for the conventional, non-delinquent group there was a highly significant relationship to the degree of marijuana use; but there was a non-significant relationship between this type of offense and the degree of use of each of the other types of drugs. Thus, this special disinhibition effect was found only for marijuana, and not for other drugs, regardless of whether they were stimulant types of drugs, or were sedative drugs.

Source: Journal of Addictive Diseases   Vol 22  Issue 3   2003