Study Sequence
Collins, 2002 AD or AT → M → HD
Graham et al., 1991 A or T → AT → AT D → AT DM Patrick et al., 2009 A or T → AT → M → HD
Maldonado-Molina & Lanza, 2010 A or T → ADorAT D → AT DM (alsoT M ) Lanza & Collins, 2002 A or T → AT orT M → AT DorAT M → AT DM Collins & Flaherty, 2002 A or T → AT → M
Kam & Collins, 2000 A or T → AT → M Tang & Collins, 2001 A or T → AT → M
(A “alcohol use”; T “tobacco use”; D “drunkenness”; M “marijuana use”; HD “hard drugs use”)
The results presented in the table above are taken from a “representative sample” of actual studies on the gateway theory. On the one hand they confirm the existence of a general sequence of stage development of drug use especially during adolescence. Indeed, although some sequences (especially concerning legal drugs) are not the same for all studies, a general trend can be identified where legal substances are entry drugs, followed by alcohol abuse, marijuana, and only after more dangerous illegal substances. On the other hand, although some of these differences could be led back to different measurement procedures and study designs, the sequence among legal substances and among harder drugs is still unclear and less understood (Kandel, 2002).
There is an ongoing debate about both the validity of the gateway hypothesis, and its contribution for the development and implementation of substance use prevention pro-grams. Knowing the sequence with which drug use develops can help policy makers to target populations at risk for specific gateway drugs, and thus prevent the use of sub-stances more advanced in the sequence. Recent efforts made in the USA to combat the diffusion of illicit drugs, especially in the youth population, have strongly relied on such a generalized interpretation of the gateway hypothesis. Many researchers doubt that the simple prevention of alcohol and tobacco use could lead to a significant reduction in illegal drug use (see Golub & Johnson, 2002). Such an application of the theory fails to take into account other important factors that might be at play in defining a specific stage sequential development in drug use. In fact, two major points should be considered.
The first problem was mentioned already, and is concerned with the validity of the se-quence proposed by the gateway theory. It should be kept in mind, especially when developing prevention strategies based on the gateway hypothesis, that if on the one hand a lot is known about the role of marijuana, on the other hand less is known about sequences among hard drugs, and it is still unclear the real sequence among legal sub-stances. More intercultural and cross-national studies might help to identify population specific sequences, but also support the idea that social and cultural factors, among oth-ers, are at play in defining time and place of specific sequential developments.
A second problem concerns the causality assumption that can be deducted from the two main propositions of the gateway theory. In fact, the presence of a given sequence, and the association among different stages could suggest that a substance in an early stage is the principal cause for the use of another substance at a more advanced stage. However,
2.6. Hypothesis
already Kandel and Jessor (2002) warned about taking causality into the gateway hy-pothesis. The existence of a progression does not directly imply a causal mechanism; the empirical results clearly suggests that being in an early stage is necessary to progress to a more advanced stage, but not all those who are in a particular stage do progress. The rates of progression through the sequence, in fact, can be easily represented by a pyramid, where the higher a substance is in the sequence, the less subjects use it. Furthermore, other empirical studies have also showed that reversal movements on the sequence are possible; only few move forward, many remain, and also many - especially in the more advanced stages - recede to a previous stage.
So far, no causal mechanism between drugs have been proven. There is instead strong support to the hypothesis that stage sequences, as found in the last thirty years, are strongly cultural determined. Other factors, in fact, concur to define a specific devel-opment within a specific culture. Drugs use, both legal and illegal, remain primarily a socialization process that is learned within specific socializing contexts, such as the fam-ily for what concern mainly legal substances, and the peer group and the school for the abuse of legal substances and illegal drugs. So it is highly probable that societal use and availability of particular substances might strongly influence their sequence.
Concluding, the gateway hypothesis offers a valid instrument to identify gateway drugs, especially in a developmental phase of life like adolescence. It has been proven, in fact, that the onset of a specific substance use is age specific. Knowing the existence of gateway substances might also help defining more accurate drug prevention programs.
However, it should be kept in mind that no causality processes are alone at work, and that a sequence that works for one society, might not work for another. A valid approach to the practical implementation of the gateway theory suggests that at first, the exact existence and form of the sequence should be found within the target population. Thereafter, external socio-psychological factors as well as influences of the setting (availability and social acceptance of a substance) should be included in the sequence to test for spurious effects. Only when an overall view over the influencing mechanisms is gained, prevention measures can target specific contextual factors related to specific gateway substances.
2.6 Hypothesis
There is a general consensus about some main results of the first longitudinal studies on marijuana use, concerned both with quantitative and qualitative development. This can help to formulate some hypothesis for the empirical analysis. From an epidemiological point of view, the following statements can be made.
Drug use, in particular cannabis use, generally starts during adolescence around the age of 14, and reach a peak in late adolescence/early adulthood, between the age of 18 and 21 (Kandel, 1980). By then, its rates of consumption (both frequencies and prevalence) decline to very low levels within the general population (Hser, 2002; Hser et al., 2007).
Keeping in mind that the time span covered in the data used for this empirical analysis covers only transitions from early into late adolescence (13-17), a similar general trend in the surveyed population is expected.
H1) By means of latent growth models I expect to find clues of a bell-shaped trajectory in the individual development of marijuana use across adolescence. This, in fact, should show an increasing frequency pattern that tends to stabilize at the end of adolescence (i.e., by the last measurement point, at the age of 17, I expect a still increasing trajectory or a stabilizing one).
Empirical research in DLC and drug use development have shown that more than a single developmental pattern is to be expected, also for what concern marijuana use. In fact, research on drug use careers suggests that multiple patterns are present at every stage of drug use developmental process (Hser et al., 2007). More precisely the literature
suggests that at least 4 trajectories are generally found for marijuana use (Brook et al., 2011).
H2) By means of growth mixture models I expect to estimated different developmental tra-jectories of drug use. They should resemble four different typologies of development across adolescence: non-users/low users, chronic users, late-onset users, and an adolescence-limited group.
Research on the gateway hypothesis on the development of drug use in adolescence has shown that a sequence exists that represents the stages at which different drugs are taken (Kandel, 1975; Kandel & Faust, 1975). Initiation with legal drugs is generally the first step; smoking cigarettes and drinking alcohol are equally considered entry substances.
In successive steps subjects experience alcohol abuse (such as binge drinking and drunk-enness), followed by marijuana use, and only after other illicit substances. For all these transitions a cumulative process is expected, for which the previously used substance is not dismissed, but rather used together.
H3) By means of latent transition analysis a cumulative model should be specified in which the following sequence is tested: use and abuse of licit substances precede marijuana use, which in turn is followed by hard drug use. I also expect that subjects will not skip stages but move always to the next in the sequence in a sequential fashion.
Some studies have also shown that not only forward progression through stages is possible, but also backward movements are possible (Tang et al., 2001; Collins & Flaherty, 2002). Subjects, in fact, especially in a period like adolescence where many changes are expected, could also temporally experiment with a substance and then give up its use. In this case one would step back from an advanced substance in the sequence to a previously used one.
H4) LTA model should also include the possibility of backward movements in the stage sequential process. I also expect that not all subjects will move forward, but that many would remain in a given stage across time, and some would also recede to a previous one.
These four hypothesis will be tested in the following chapters using empirical data taken from the longitudinal study Crime in the Modern City, a panel study on youth delinquent behavior conducted over the last decade in the German town of Duisburg (Boers et al., 2010). Before the empirical analysis using growth models and latent transi-tion analysis, two chapters will deal with the presentatransi-tion of the study design and some important descriptive statistics.