Skip to main content

Is Marijuana Really a Gateway Drug?

Decorative Image

Cody Jorgensen

Cody Jorgensen earned a Ph.D. in Criminology from the University of Texas at Dallas. His main areas of interest include biosocial criminology, policing and forensics, and attitudes and perceptions in crime and justice. He is currently an assistant professor in the criminal justice department at Boise State University and has taught classes on policing, biosocial criminology, criminological theory, statistics, and law and social control. In his spare time he can be found on the trails riding dirt bikes and mountain bikes or on the streets riding motorcycles.

During the 2012 presidential election, then-candidate Mitt Romney told a wheelchair-bound man suffering from muscular dystrophy that he did not support medical marijuana. At the time Romney believed marijuana was a gateway drug that leads to harder substance use. Similarly during the 2016 Republican primary, several candidates including Chris Christie and Carly Fiorina argued against loosening federal regulations of marijuana because they believed that marijuana was a gateway drug. In the 80s, President Reagan quipped that marijuana may be the most dangerous drug because it was a gateway to harder drugs. This marijuana gateway hypothesis (MGH) remains a consistent justification for harsh policies controlling marijuana. However, it begs the question: Is marijuana really a gateway drug? The marijuana gateway hypothesis or the marijuana gateway effect has been invoked for the better part of eighty years to justify its prohibition and researchers have been investigating the accuracy of the claim for decades.


Critics of cannabis have given numerous justifications for prohibiting the plant including: the idea that marijuana causes harm to society and the user, the increased potency of marijuana, the addictive properties of cannabis, that decriminalizing marijuana sends a latent message that people should be using it, that drug use is on the rise, just to name a few. However, the most commonly cited justification for marijuana prohibition is the MGH. All of these justifications listed above have been debunked by empirical research and are more fiction than fact.

A 1994 report published by the Center on Addiction and Substance Abuse reported that marijuana users are 85 times more likely to use cocaine than non-users. These numbers have been heavily criticized by substance use researchers for using an illogical application of math. Photo credit: National Center on Addiction and Substance Abuse.

The MGH is a causal argument that suggests marijuana use causes hard drug use. For marijuana to really be a gateway drug three conditions must be met: sequencing, association, and causality. Sequencing means that there is a fixed relationship between two drugs and that the use of one drug (marijuana) is initiated before the use of another drug (e.g., cocaine). Association means that the initiation of first drug statistically and substantively increases the probability of using the second drug. Causation means that the use of the first drug actually causes the use of the second drug. For the MGH to be true the use of marijuana must precede the use of harder drugs like heroin or methamphetamine, that there is a statistical association between using marijuana (first) and using harder drugs (second), and that the marijuana use was the cause of the harder drug use. Another way to think about sequencing and causality is to consider people who ride motorcycles. People who ride motorcycles learned to ride bicycles earlier in their lives (during childhood). Yet, no one would entertain the thought that riding a pedal bicycle causes riding a motorcycle later on. The sequence is there, but causality is questionable at best.   If any of these criteria are not met then the gateway argument is not supported as alternative explanations may be invoked to explain the relationship between marijuana use and harder drug use. As with many areas of social science, the findings from studies investigating the MGH are mixed. Some studies find some support for the claim while others do not.

A commonly cited statistic showing support for the MGH comes from a 1994 report published by the Center on Addiction and Substance Abuse. The document reported that marijuana users are 85 times more likely to use cocaine than non-users. A 2002 report from the Office of National Drug Control Policy claimed that marijuana users were 8 times more likely to use cocaine and 15 times more likely to use heroin compared with people who do not use marijuana. These numbers have been heavily criticized by substance use researchers for using an illogical application of math; in order to get these statistics, the reports listed above simply divided the proportion of hard drug users who used marijuana prior to using hard drugs by the proportion of users who used a hard drug prior to using marijuana. Whenever you divide a large proportion (common event) by a small proportion (uncommon event) the quotient will be quite large. However, this method is not the appropriate method to calculate a probability estimate or make generalizable inferences. Seeing that marijuana users are 85 times more likely to use cocaine can be quite alarming and can make an individual believe in the dangerousness of using marijuana. But this statistic is in error due to its illogical application of math and should be taken with a grain of salt.


Monica Deza also found strong support for sequencing concluding that marijuana was a common “stepping stone” to harder drugs. However, the study also found that alcohol was the “stepping stone” to marijuana use.That said, there have been a number of peer reviewed studies published in scientific journals that have used proper methods and data analytics to investigate the MGH and found support for the argument. For example, Wayne Hall and Michael Lynskey suggested that in animal studies heavy cannabis use primes the brain to be pharmacologically receptive to other illicit drugs, thereby increasing the likelihood of harder drug use. However, alcohol and tobacco have similar priming effects. Additionally, David Fergusson and colleagues employed a sophisticated statistical method to test the MGH and found an association between heavy marijuana use and a marginal increased risk of subsequent illicit drug use. The study did find support for sequencing, but the authors concluded they could not establish causality. More recently, Monica Deza also found strong support for sequencing concluding that marijuana was a common “stepping stone” to harder drugs. However, the study also found that alcohol was the “stepping stone” to marijuana use. In a 2016 review of the literature, the National Institute on Drug Abuse concluded there is strong evidence for sequencing, i.e., that marijuana use precedes hard drug use. The report also concluded that sequencing goes farther back. That is, alcohol use precedes marijuana use and tobacco use precedes alcohol use. What is more, drinking caffeinated beverages precedes tobacco use. It can be seen that the sequencing to hard, illicit drug use begins much earlier in the substance use timeline. Yet, it is uncommon to hear the claim that caffeine is a gateway to tobacco, or that alcohol is a gateway to marijuana. The studies above do show strong support for sequencing and association, but these studies cannot support the previously mentioned causality requirement of the gateway model.

Considering the literature at large, there are more studies that do not find support for the MGH than do. One reason for this lack of support is that very few marijuana users go on to try harder drugs yet alone use them on a regular basis. In short, given the population of marijuana users the probability of moving on to harder drugs is next to zero. For example, data from a 2001 study by the National Household Survey on Drug Abuse suggested that an estimated 83 million Americans had tried marijuana in their lifetime but have never tried heroin. Similarly, Mitch Earleywine found that for those individuals who had ever smoked cannabis in their lifetime the odds of them using cocaine within the last 30 days was 1 in 50. That number was 1 in 677 for heroin. These are incredibly low and non-substantive probabilities. Other studies have found that initiation to hard drug use was best explained by factors other than prior marijuana use such as genetic factors, drug abuse propensity, or access to illicit drug markets. In sum, the available empirical evidence regarding the MGH suggests that all 3 criteria in the gateway model are not met. A reasonable conclusion to draw from the literature so far is that the evidence for marijuana causing hard drug use is, at best, very weak.


The biggest limitation of the extant research testing the MGH is the lack of experimental methods. The gold standard in scientific research and the best way to establish causation between an independent variable and a dependent variable is the randomized controlled trial. With this method, study participants are randomly assigned to a treatment group (given the treatment) or a control group (not given the treatment). Any difference between these two groups on some outcome of interest will be due to the treatment.   However, this method is rarely applied in the social sciences due to ethical and practicality constraints. For example, to test the MGH under such experimental conditions researchers would have to randomly assign a group of high school students to smoke marijuana and randomly assign a control group of students to abstain. Then hard drug use rates measured at a later time for the treatment group would be compared to that of the control group. Any difference in hard drug use rates would be attributed to using marijuana. Doing so, though, is not possible because randomly assigning students to do drugs is highly unethical and illegal. The next best option, however, is to employ a quasi-experimental technique that approximates the randomized controlled trial known as propensity score matching. This statistical technique uses a matching algorithm to create a treatment group and a control group that can be compared within an existing dataset. Properly done, there will be no differences between the two groups for a variety of control variables. This allows the researcher to be confident that the treatment, and only the treatment, is driving the effect on some outcome of interest.

Recent research conducted by the author tested the MGH via propensity score matching using data from the National Longitudinal Survey of Adolescent to Adult Health, a nationally representative longitudinal dataset. In total, 18 tests of the MGH examining the effects of three different dosages of marijuana use on two different dosages of hard drug use were conducted. The technique employed meets a robust and rigorous standard of social science research. The first six tests examined the gateway effect of junior high and high school students’ marijuana use on harder drug use two years later on in their life. No relationship was found between marijuana use and harder drug use in this series of tests, a finding contrary to the MGH. The next six tests assessed the gateway effect of students’ marijuana use on harder drug use seven years later. In this series of tests, one test found a weak relationship between heavy marijuana use and light hard drug use later on in life. The other five tests did not find a relationship between marijuana use and hard drug use at any dosage. The last six tests examined the gateway effect of marijuana use when participants were between 14 and 20 years old on hard drug use five years later on in their life. Again, one test found a weak relationship between heavy marijuana use and light hard drug use later on. The results of this research suggest that both occasional and recreational marijuana use have no effect on hard drug use. Additionally, these results also suggest that heavy marijuana use only has a weak effect on light hard drug use (but not heavy drug use). What is more, the tests that did find a relationship between heavy marijuana use and light hard drug use were the least precise tests, allowing more room for error.


To put these findings in context, propensity score matching models were also employed to test the independent gateway effect of both tobacco and alcohol on hard drug use. The findings from these tests nearly mimicked the findings from the marijuana gateway tests. This means that both tobacco and alcohol independently have equivalent gateway effects on hard drug use as compared with marijuana. If the MGH is being invoked to justify its prohibition, then these tests suggest the same argument could be made to prohibit the use of tobacco and alcohol.

In sum, an objective reading of the literature examining the MGH and considering the findings from the research presented above leads to the conclusion that marijuana is not really a gateway drug. For the MGH to hold water, there should be more consistent and stronger associations to harder drug use. There is strong support for sequencing, however sequencing is not causal. Moreover, the most rigorous empirical evidence shows little to no causal gateway effects. As such, any public policy that prohibits the use of marijuana in an attempt to curb hard drug use is unlikely to succeed. Additionally, policymakers should consider the negative and unintended consequences of making a criminal out of an otherwise law abiding pot smoker, such as difficulty securing employment and an increased risk of committing future crimes. Former President Carter once said, “the penalty for drug possession should not be more harmful to an individual than the use of the drug itself.” This is certainly true for marijuana.