Discussion below will concern two issues of great significance to assessing the public health threat, if any, presented by marijuana use. The first issue is whether or not use of marijuana is a valid predictor of the use of more dangerous and more addictive drugs. The second issue is related, whether marijuana use necessarily an indication of psychological or behavioral problems. Both issues address the same question, does marijuana use provide the basis for successful predictions about an individual's behavior? What does marijuana use, alone, tell us about an individual? Advocates of marijuana prohibition have long argued that marijuana use may not appear harmful to an individual, but is a marker that the individual is ultimately headed for trouble as a result of their marijuana use alone.

If marijuana use is a valid predictor of harmful behavior, then, it is argued, marijuana use is a threat to public health regardless of the effect on the individual.

Social scientists are well aware of the pitfalls of addressing such questions; correlations do not establish causation. In an example widely used by radio commentator Rush Limbaugh, many criminals eat carrots, but eating carrots does not cause crime. Claims that marijuana presents a threat to public health by leading individuals to use and abuse other drugs can be criticized on both logical and empirical grounds. Let us examine the literature.

Gabriel Nahas summarizes the evidence supporting the theory connecting marijuana use to other drug use in the preface to 1990's 5th edition of Keep Off the Grass :

"It appears that the biochemical changes induced by marijuana in the brain results in a drug-seeking, drug taking behavior, which in many instances will lead the user to experiment with other pleasurable substances. The risk of progressing from marijuana to cocaine or heroin is now well documented.

"Marijuana users are sixty-six times more likely to use cocaine subsequently than subjects who have never consumed marijuana. This 1990 survey of PRIDE documents further the fact that marijuana is a "gateway" drug to more destructive dependency-producing drugs such as heroin or cocaine. Such a most significant risk of escalating from marijuana to cocaine was reported by Kandel and colleagues in 1975 (Science 190 (1975): 912). Clayton and Voss (U.S. Journal of Drug and Alcohol Dependence, Jan. 1982) confirmed Kandel's observation and reported that the risk for a marijuana user to progress to cocaine consumption is ten times greater than the risk of a heavy tobacco smoker to develop cancer of the lung. Dr. Herbert Kleber (Journal of Clinical Psychiatry, 1988, 48:3) reports that 75% of frequent users of marijuana have used cocaine at least once. It is a fact that the major epidemic of cocaine consumption besetting the country since the mid-eighties was preceded by the marijuana epidemic of the seventies."(7)

It widely accepted in the scientific community that a correlation does not establish causality.(8) Over reliance on correlations in the verification of hypotheses represent an error in reasoning well-known to statistical analysts in the social sciences -- the fallacy of affirming the consequent.(9)

The 1993 National Household Survey indicates that 79% of regular marijuana users do not use any other illicit drug.(10)

Denise Kandel, also a professor at Columbia University, and colleagues published a series of articles in the American Journal of Public Health in 1984 on patterns of drug use, updating the data published in 1975 with reports on the same cohort; the 1984 report studies the 1975 cohort, now at age 25. A summary of their findings:

"The sequence of involvement into drugs progresses from the use of at least one legal drug, alcohol and/or cigarettes, to marijuana, and from marijuana to other illicit drugs, and/or prescribed psychoactive drugs."(11)

The Office of Technology Assessment of the U.S. Congress, citing recent work by Clayton, Kandel, and other research, concluded:

"While study results vary somewhat, the sequence most often reported is that alcohol and cigarette use come first, followed by marijuana use and then by the use of other illicit substances."(12)

Neurobiological research has discredited the hypothesis that marijuana and other drugs target the same brain system (the mesocorticolimbic pathway), as will be discussed in section 7 below. Are there other explanations why some marijuana users also use more addictive drugs? One hypothesis is that:

"(T)he criminalization of marijuana may have caused some marijuana users to move on to other illicit substances through contact with the subculture of illicit users."(13)

Research is beginning to suggest that targeting marijuana as the key battleground for preventing drug abuse is a little like closing the doors after the horses have already left the barn. Marijuana use among adolescents appears to be a symptom of problem behavior, not a cause.

"[In one study] delinquency and youthful sexual activity tended to precede the use of marijuana and hard liquor. . . The early use of so-called gateway drugs, such as beer and cigarettes, may contribute to later problem behaviors, while the later use of marijuana, hard liquor, and other illicit substances may be more the result of extended participation in problem behaviors."(14)

Scientific theories are scientific because they are the basis for consistently successful predictions. Gravity is a prediction that when apples leave a tree branch, they fall. Gravity is a scientific theory because things fall, largely without exception. The Office of Technology Assessment does not say that the gateway drug theory is unscientific, but they reach a similar conclusion.

"Because many individuals who use substances do not go on to substance abuse, and because one level does not guarantee use at a higher level, these stages are descriptive but not predictive."(15)

Yamaguchi and Kandel elaborate in their conclusion:

"The existence of sequential stages of progression, however, does not necessarily imply causal linkages among different drugs since the observed sequences could simply reflect the association of each class of drugs with different ages of initiation and/or individual attributes rather than the specific effect of the use of one class of drug on the use of another. Furthermore, it is important to keep in mind that although a clear development sequence in drug involvement has been identified, use of a drug at a particular stage does not invariably lead to the use of other drugs higher up in the sequence. Many youths stop at a particular stage and do not progress further."(16)

In a related article, Yamaguchi and Kandel conclude that "the probability that individuals who never use marijuana will initiate the use of other illicit drugs is very low."(17) They conclude that teenagers are especially at risk to drug abuse if they begin to use marijuana. The dangers have less to do with marijuana itself than with use of marijuana by adolescents.

Generally the risk of initiation to the use of any drug rapidly decreases about age 20, although with marijuana there appears to be a residual initiation rate at about age 23 - 24.(18) A maturation trend is also apparent in marijuana use, with use stabilizing at age 19 and amount of use decreasing after age 20 even as use continues strongly (50)% of males and 33% of females in the cohort) into the mid twenties.

Yamaguchi and Kandel's findings deserve close consideration, and must not be removed from their original context. The authors specifically argue that their findings are relevant to public policy considerations regarding prevention and educational programs for adolescents.(19)

This 1984 article by Yamaguchi and Kandel has served as the basis for the 'gateway theory' that asserts that prior use of marijuana is necessary to progress to other drugs. Their specific policy recommendation, though, involves reduction of the use of legal drugs by school aged youth.

"The findings in this paper suggests potentially important policy implications for the development of preventive and educational efforts, namely that prevention of early involvement in legal drugs would reduce the use of marijuana, and that prevention of early marijuana use would reduce involvement in other illicit drugs. . . It is important to remember that age effects determine, in part, the sequential patterns observed between the licit drugs and marijuana. . ."(20) (emphasis added)

The focus of attention is on early use, and this is one of the most overlooked aspects of Kandel's gateway theory. It is early use itself, in high school for example, that presents a risk of further drug exposure, not later use of marijuana by adults.

"The probabilities of initiating other illicit drug are very much reduced if marijuana is initiated at age 20 or over. . ."(21)

Both current and former users of drugs share the same risk of progression to other drug use, and for this reason Kandel advocates making prevention of early use the fundamental priority of public policy.

In 1992 Kandel and Yamaguchi published another follow-up study on the same cohort, now followed from age 15 to age 35.(22) They reiterate the importance of early use of drugs to later use of dangerous drugs.

"Two characteristics of an individual's drug history are especially important predictors of progression from a lower to a higher stage of drug use: age of initiation into use of a drug class and extent of use of the drug."(23)

Here is how they update their description of the stages:

"While the patterns described during adolescence hold for the transitional period into adulthood, the use of prescribed psychoactive drugs, first identified in the mid-twenties for the cohort, had been confirmed as a further step in the sequence by the mid-thirties. The last stage, involving prescribed drugs, however, is the least strong in the sequence: it accounts for two-thirds of those who have ever used these drugs compared with 80% for stages involving legal and illegal drugs. The potential existence of problem drinking as an intervening stage between the use of marijuana and other illicit drugs (Donovan and Jessor, 1983)(24) could not be investigated with the data available in this study."(25)

The authors repeat that their gateway theory of stages does not mean that drug users "invariably" use stronger drugs; many youths stop at a particular stage. They elaborate on their 1984 comments:

"The notion of stages in drug behavior does not imply that these stages are either obligatory or universal such that all adolescents must progress through each in turn, as has been proposed by Piaget or Kohlberg for stages in moral development. The use of a particular drug does not invariably lead to the use of other drugs higher up in the sequence. The model is not meant to be a variant of the controversial "stepping stone" theory of drug addiction in which the use of marijuana was assumed to lead inexorably to the use of other illicit "hard" drugs, especially heroin. Rather the phases in drug behavior are facilatative. Entry into a particular stage is a common and perhaps even a necessary although not a sufficient prerequisite for entry into the next higher stage . . . [I]t is because all drug users come from a common pool of young people, who all started using the same drug classes, that they share certain factors in common."(26) (emphasis added.)

The policy recommendation advanced by Kandel and her colleagues is that the optimum public health policy is to delay the age of onset of use of drugs. Once again they conclude that:

"Early intervention would not only prevent participation in behaviors, such as smoking cigarettes, that have been detrimental consequences of their own but would also probably deter and/or reduce progression to other forms of drug use."(27)

A 1993 paper by Kandel and Yamaguchi on the sequence from beer to crack illustrates the importance of delaying onset of first use. In the sample for this study, adolescents who use cocaine or crack began use of gateway drugs (alcohol, tobacco, or marijuana) two years earlier than those adolescents who do not use cocaine or crack, despite their use of gateway drugs. Age of first use at a lower stage is a strong predictor of further progression.(28)

In 1995 Chen and Kandel published "The Natural History of Drug Use from Adolescence to the Mid-Thirties in a General Population Sample" in the American Journal of Public Health.(29) Summarizing their previous findings, the conclusion of their abstract reads simply that "cigarettes are the most persistent of any drug used. Drug-focuses intervention must target adolescents and young adults."(30)

The natural history of marijuana use is that it begins in adolescence after the use of cigarettes and/or alcohol; consumption quantities and frequency of use increase to age 19, and then stabilizes until age 23-24, after which marijuana use declines. Like alcohol, there is an apparent maturation trend with the highest use of marijuana between ages 19 and 22. The daily consumption of marijuana drops as individuals reach their thirties.

An interesting aspect of these findings is that they lend support to an earlier hypothesis of Johnston and colleagues. In 1984, O'Malley, Bachman and Johnston published a paper on "Period, Age and Cohort Effects on Substance Use Among American Youth, 1976-82."(31) The purpose of this study was to develop a hypothesis with a close fit to the longitudinal data of the Monitoring the Future Project. One theory developed is that "transitions" mediate drug use as adolescents grow older. This lends credence to the notion that teens take drugs because they are available and interesting, and that most of them grow out of it.

"Age effects also encompass a very broad range of possible underlying causes. For example, many behaviors are age-regulated (voting, driving, marrying, purchasing alcohol), and thus age per se ought to have an effect on such behaviors. There are also consistent biological changes associated with aging, as well as some important transitions in social roles and environments. The latter are especially important for the age span under study here, with all the major transitions that occur between adolescence and adulthood. One important social transition is graduation from high school, and we see a clear effect of this transition on the frequency of cigarette use. Other important transitions which typically occur during the years after high school graduation include becoming a full-time worker, moving out of the parental home (possibly to go to college, to enter military service, or to marry and set up an independent living situation), and becoming a parent. Some of these transitions would be expected to lead to a decrease in drug use."(32)

Throughout the discussions conducted by Kandel, Johnston, and their colleagues social rather than biological factors dominate the analysis. The connection between marijuana use and other drug use is a function of set and setting, not a function of the pharmacological properties of the drug. As Brady pointed out in section 5, if you want to influence decisions about the use of some drugs, circumstances can be arranged. The paramount issue is whether or not the pharmacological properties of the drug outweigh the influence of set and setting. This is certainly not the case with marijuana, and this will be discussed in more detail in section 7. The addictive qualities of nicotine, by contrast, are a factor that outweighs or at least rivals the influence of set and setting -- note the persistence of cigarette use found in Chen and Kandel's 1995 report.

It is important to distinguish between the policy recommendations of Kandel's team about teenage marijuana use and the implications their work has for the ramifications of adult marijuana use.

It would be a horrible misrepresentation of their findings to argue that Kandel and Yamaguchi support the use of criminal sanctions for adult marijuana use because their findings indicate that adult marijuana users are likely to move on to the use of cocaine and/or heroin.