An Examination of the Importance of Strain in the Cannabis Gateway Effect

Abstract

Gateway theory has been the source of much debate in both the research literature and public policy. Support for gateway sequencing has been mixed, especially in research that has considered the role of criminological variables in the etiology of substance use. For example, limited prior research has observed as important in gateway sequencing the effects of severe stressors. Data from the National Longitudinal Study of Adolescent to Adult Health are utilized to test gateway theory and examine whether severe stressors affect the relationship between frequency of cannabis use and later use of other illicit drugs (OIDs). Findings suggest that while frequency of cannabis use does increase the likelihood of later use of OIDs, this relationship may be the result of the common cause of experiencing severe stress. Implications of the findings are discussed.

Keywords

gateway cannabis illicit drugs general strain theory Add Health

Introduction

Understanding Gateway Theory

The use of illicit substances from adolescence into adulthood has been the subject of much debate in the United States (Blanco et al., 2014; Murray, Morrison, Henquet, & Di Forti, 2007; Secades-Villa, García-Rodríguez, Jin, Wang, & Blanco, 2015). Various theoretical paradigms have been tested and used to explain the risk factors for continued use from adolescence into adulthood (de Wit, 1996; Fergusson, Horwood, Lynskey, & Madden, 2003; Jessor & Jessor, 1980; Kandel, 1975). One such explanation that has been a highly discussed topic among scholars and policy makers is the gateway hypothesis, which suggests that the sequential use of alcohol and tobacco, followed by cannabis, is associated with an increased likelihood of the subsequent use of illicit opioids, depressants, and stimulants (Degenhardt et al., 2010; Ellgren, Spano, & Hurd, 2007; Grau et al., 2007; Kandel, 1975, 1978, 1984; Kandel, Davies, Karus, & Yamaguchi, 1986; Kandel, Yamaguchi, & Chen, 1992; Van Ours, 2003; Yamaguchi & Kandel, 1984a, 1984b). Several studies have examined and found associations between cannabis and subsequent use of other illicit drugs (OIDs; Agrawal, Neale, Prescott, & Kendler, 2004; Fergusson, Boden, & Horwood, 2006; Fergusson & Horwood, 2000; Khan et al., 2013; O’Donnell & Clayton, 1982; Secades-Villa et al., 2015; Van Ours, 2003).

Empirical support notwithstanding, much of the findings for the gateway effect have been contested. Research critical of the gateway effect cites data limitations, lack of cultural considerations, and potential spuriousness (Golub & Johnson, 2002; Morral, McCaffrey, & Paddock, 2002). Research has adopted further utilization of criminological theory, cultural context, genetic traits, and social status (Golub & Johnson, 2002). Empirical consideration for licit and illicit drug sequencing continues, with scholars accounting for other theoretical and sociocultural factors (Choo, Roh, & Robinson, 2008; Degenhardt, Hall, & Lynskey, 2001; Fergusson & Horwood, 2000; Hall & Lynskey, 2005; Rebellon & Van Gundy, 2006; Secades-Villa et al., 2015; van Gundy & Rebellon, 2010).

The central assumption of the gateway hypothesis is the temporal sequence where the initiation of one substance is associated with an increased risk to use another substance (Kandel, 2002; Mayet, Legleye, Falissard, & Chau, 2012). Importantly, an adolescent who initiates the use of one substance, such as cannabis, will fall subject to this gateway sequence, as that individual will hold a higher risk of using other illicit substances. Recently, however, behavioral and mental health problems (Degenhardt et al., 2009) have shown to lead to later use. A study by Degenhardt et al. (2009) concluded that early drug initiation is not strongly predictive of later use, unless adolescents have mental health concerns putting them at greater risk of later OID use. Sociodemographics and environmental risk factors have also shown to have an impact with this hypothesis. A study by Fergusson and Horwood (1997) uncovered that initial cannabis use and later OID use can be explained through social disadvantage, adverse peer relations, and childhood adversity (Fergusson & Horwood, 1997, 2000; Yamaguchi & Kandel, 1984a, 1984b). Other factors that have been studied include, but are not limited to, exposure to opportunity through peer drug use (Wagner & Anthony, 2002), race and ethnicity, and economic measures (Secades-Villa et al., 2015). Some of these risk factors can be seen as stressors, which can be associated with deviant behaviors, such as drug initiation and later use. Furthermore, criminological theories such as strain as well as factors including social status, genetics, and cultural contexts have all shown the possibility to render away much of the gateway effect (Golub & Johnson, 2002; Measham & Shiner, 2009; Peele & Brodsky, 1997; Rebellon & Van Gundy, 2006; van Gundy & Rebellon, 2010).

The Importance of Strain

Scholars have investigated various risk factors that can lead to illicit drug use, including genetic variation (Van den Bree et al., 1998), family environment (DiClemente et al., 2001), and community factors (Lillie-Blanton, Anthony, & Schuster, 1993).¹ In addition, stressors have been examined in association with drug misuse and abuse. One theoretical basis for the importance of stressors lies within the theoretical paradigm of strain. General strain theory, proposed by Robert Agnew (1992), is an individual-level, social-psychological explanation of deviance. General strain theory argues there are three types of strain: failure to achieve culturally valued goals, loss of positively valued stimuli, and the presentation of noxious stimuli (Agnew, 1992). Experiencing these types of strain in daily life can increase criminality. General strain theory proposes that strains are connected to criminal acts through negative affective states, including anger, depression, fear, and anxiety. These negative affective states, brought on through stressors in everyday life, lead to actions to correct or alleviate negative emotions and strains. Strains or stressors can cause individuals to participate in corrective action, such as illicit substance use to attenuate the negative emotions that can result from stress. These strains have consistently been shown to lead to deviance, including drug misuse and abuse (Agnew, 1992, 2001; Agnew & White, 1992; Boardman, Finch, Ellison, Williams, & Jackson, 2001; Carson, Sullivan, Cochran, & Lersch, 2008; Cloward & Ohlin, 1960; Cohen, 1955; Elliot, Huizinga, & Menard, 1989; Ford & McCutcheon, 2012; Hoffmann & Su, 1997; Merton, 1938; Neff & Waite, 2007; Schroeder & Ford, 2012; Schulenberg, Bachman, O’Malley, & Johnston, 1994; Slocum, 2010; Watts & McNulty, 2013).

In the context of the gateway hypothesis, strain, such as stressors related to school, family, and so on, can lead to corrective action or deviancy in the form of drug use. In this way, the link from cannabis initiation to later OID use is perhaps stronger for those experiencing strain as they seek to continue to correct for negative emotions. Those who do not experience these stressors will be less likely to further engage in corrective actions, such as future OID use. Importantly, there have been limited studies that test the impact of strain in conjunction with patterns of drug sequencing, particularly gateway theory (Peele & Brodsky, 1997; Rebellon & Van Gundy, 2006; van Gundy & Rebellon, 2010).

In the context of the gateway hypothesis, previous researchers have cited concerns with under sampling of illicit drug users² (Golub & Johnson, 2002). The present study utilizes data from the National Longitudinal Study of Adolescent to Adult Health (Add Health), which is a nationally representative sample of the United States. While prior researchers have utilized Add Health data to observe gateway effects, these previous studies have largely focused on genetic and developmental pathways, rather than the focus of the current study, strain (Cleveland & Wiebe, 2008; Lessem et al., 2006).

Specifically, we test one of the key components of the gateway hypothesis, that frequency of cannabis use is associated with an increased likelihood of later OID use, and whether strain effects this relationship using a representative sample of American adolescents. We hypothesize that there will be a positive relationship between cannabis and OIDs (Hypothesis 1), but that this relationship will be weakened when accounting for strain (Hypothesis 2).

Method

Sample

We draw on restricted-use data, which are only available contractually, from Waves I, II, and III of Add Health. Add Health is a panel study of a nationally representative sample of American adolescents who were first recruited during the 1994-1995 school year while they were in Grades 7 to 12 (Harris et al., 2003; Udry, 1998). Add Health utilized a multistage stratified sampling process to select 80 high schools and 52 middle and junior high schools for inclusion in the study. A total of 90,000 students completed in-school self-report surveys, and a subsample of this group was randomly chosen for the in-home component of Wave I of Add Health. A total of 20,745 adolescents and 17,700 of their primary caregivers participated in the Wave I in-home component (Harris et al., 2003). At the time of Wave I data collection, the average Add Health respondent was approximately 15 years old. Wave II data were collected approximately 1 year after Wave I. For Wave II, Add Health did not reinterview individuals who were in the 12th grade at Wave I, resulting in a Wave II sample of almost 15,000 individuals. Wave III interviews were conducted in 2001-2002 with every original Wave I respondent that could be successfully located at that time, with a total sample size of 15,170.

We eliminated from the analysis sample respondents who at Wave I reported that they had, at any point in their life, used, even if only once, any of the illicit drugs that are thought to be the outcome of the gateway effect of cannabis, such as cocaine, heroin, methamphetamine, and so on. This resulted in eliminating approximately 8% of Wave I respondents from the analysis.³ With this restriction and further limiting the analyses to respondents with complete data for the key independent and dependent variables, our analysis sample consists of 12,403 individuals. This analysis sample is nationally representative of adolescents who were not in Grade 12 and had not already used serious illicit drugs as of Wave I of Add Health data collection.⁴ Table 1 presents descriptive statistics for the full analysis sample.

Table 1.

Descriptive Statistics (n = 12,403).

Variable	Range	M (SE)
Dependent variable
Other illicit drug use, WII	0-1	0.04 (0.00)
Independent variable
Cannabis use, WI	0-100	0.71 (0.04)
Strain theory
Strain, WI	0-10	0.74 (0.01)
Controls
Sex (1 = male)	0-1	0.48 (0.00)
Age, WII	11-23	16.18 (0.02)
White	0-1	0.51 (0.00)
Hispanic	0-1	0.16 (0.00)
Black	0-1	0.22 (0.00)
Native American	0-1	0.02 (0.00)
Asian	0-1	0.07 (0.00)
Other	0-1	0.01 (0.00)
Parent’s education	0-1	0.28 (0.00)
Parental employment	0-1	0.91 (0.00)
Household income (in thousands of dollars)	0-999	46.12 (0.40)
Minor delinquency, WI	0-12	1.71 (0.02)
Substance-using peers, WI	0-9	2.14 (0.02)
Smoking, WI	0-1	0.15 (0.00)
Alcohol use, WI	0-1	0.51 (0.00)
Parental alcohol use	1-6	1.95 (0.01)
Low self-control	0-73	38.16 (0.06)

Note. Because these statistics are weighted and adjusted for survey design, standard errors are produced rather than standard deviations. WI = Wave I; WII = Wave II.

Measures

OID, Wave II

Our analysis focuses on predicting the use of serious illicit drugs at Wave II in the Add Health dataset. We focus on Wave II for the outcome variable because later waves of data in Add Health, Waves III and IV, were collected many years later and only cover a period of 12 months prior to those interviews in the questions concerning the use of illicit drugs. Any use since the prior interview that fell outside of those 12-month periods would not be reported in the survey. With this dependent variable, we pick up all use of illicit drugs since Wave I. The short follow-up period for measuring gateway effects is not necessarily ideal, but it is necessary when looking for these effects in the current data. The dependent variable is represented by one dichotomous measure that asked respondents if since the Wave I interview they had used, even if just once, any of the following drugs: cocaine (powder, freebase, or crack), lysergic acid diethylamide (LSD), phencyclidine (PCP), ecstasy, mushrooms, speed, ice, or heroin (1 = yes). As this measure is dichotomous, logistic regression techniques are utilized in the statistical models.⁵ Approximately 4% of respondents reported using these serious illicit drugs between Waves I and II (see Table 1).

Cannabis use, Wave I

Our independent variable cannabis use comes from Wave I. Respondents reported at Wave I the number of times in the 30 days prior to being interviewed they smoked cannabis, ranging from 0 to 100 or more times. As can be seen in Table 1, average levels of cannabis use in the month prior to Wave I interview are relatively low. In fact, almost 90% of respondents reported no cannabis use in the month prior to interview. Furthermore, another 7% or so of respondents reported using cannabis only 1 to 3 times during the month prior to interview. The low level of cannabis use in this sample is most likely due to our elimination from consideration respondents who reported serious illicit drug use prior to Wave I, and Add Health being a school-based sample.

Strain

To examine the robustness of the gateway effect of cannabis, we examine how strain affects this relationship. To do so, we constructed an index of strain at Wave I, with additional retrospective items from Wave III that refer to events that predate Wave I data collection. In all, 10 measures of strain are included. Two items from Wave I asked respondents if any of their friends or family had committed suicide in the past 12 months. Five items concerning recent experiences with violence at Wave I asked respondents if in the past 12 months they had witnessed someone getting shot or stabbed, had had a knife or gun pulled on them, had actually been cut/stabbed or shot, or been otherwise physically attacked. Three retrospective items were taken from the Wave III interviews that asked respondents about experiences of neglect and physical and sexual abuse at the hands of their caregivers before the start of the sixth grade. With neglect, respondents were specifically asked if before the sixth grade their parents or other adult caregivers had failed to take care of their basic needs, such as keeping them clean or providing food or clothing. Physical abuse was covered by asking respondents if their parents or other adult caregivers had ever slapped, hit, or kicked them. Sexual abuse was covered by asking if the respondent was ever touched by, forced to touch, or forced to have sex with a parent or adult caregiver. As these experiences come from before the sixth grade, and the youngest Add Health respondents at Wave I were seventh graders, these strain experiences predate Wave I data collection. All of the above measures fit the mold of severe strain in general strain theory, as they are very traumatic experiences, and are likely to be seen as unjust and/or undeserved, meaning they may be particularly problematic for coping behaviors like substance use (Agnew, 2001).

All of these strain measures are serious and fairly rare, and thus we coded each severe strain measure by whether a respondent reported that it had happened at least once (1 = yes), and summed these yes responses to create an index of strain. These same items have been used as objective measures of severe strain in previous research that utilized the Add Health dataset (Kaufman, 2009; Kort-Butler, 2010; Watts, 2015; Watts & McNulty, 2013). As can be seen in Table 1, the average Add Health respondent reported experiencing only about one of these severe strains.

Controls

We include a number of general controls in all analyses. These include age; sex (1 = male); race/ethnicity dummy variables for Hispanic, non-Hispanic Black, Native American, Asian, and Other (with non-Hispanic White as the reference category); and several measures of socioeconomic status, including respondents’ parent’s education (1 = 4-year degree or more), parental employment status at Wave 1 (1 = employed), and family income in thousands of dollars at Wave I.⁶

Because of the correlation between illicit drug use and other forms of delinquency in adolescence, we also control for minor delinquency at Wave I, measured as how often respondents said they painted graffiti on other’s property, lied to their parents about their whereabouts, ran away from home, or acted unruly in public in the past 12 months. We additionally control for the presence of substance-using peers at Wave I with a measure that combines three questions that asked how many of the respondent’s three closest friends smoke cigarettes, drink alcohol, and smoke cannabis. We further control for parental use of alcohol, because parental substance use correlates highly with substance use among children (Chassin, Pillow, Curran, Molina, & Barrera, 1993; Chassin, Rogosch, & Barrera, 1991). During the Wave I in-home parental interview, the respondent’s parent was asked how often they drink alcohol, with possible answers including “never,” “once a month or less,” “2 or 3 days a month,” “once or twice a week,” “3 to 5 days a week,” or “nearly every day.” Because of its consistent correlation with various forms of delinquency during adolescence, we also control for levels of self-control (Beaver, DeLisi, Mears, & Stewart, 2009; Gottfredson & Hirschi, 1990; Pratt & Cullen, 2000). This index is a slightly altered version of one developed by Beaver et al. (2009) for use in the Add Health survey, and includes 18 items that taken together measure a respondent’s temper, self-centeredness, attention span, and use of rational decision making. This composite measure was created by summing these 18 items, with higher scores denoting lower levels of self-control (α = .67).

Last, because gateway theory focuses on tobacco and alcohol in addition to cannabis, we control for lifetime tobacco and alcohol use at Wave I. For tobacco, we utilize one item that asked respondents if they had ever smoked regularly, defined by smoking at least one cigarette every day for 30 days (1 = yes). For alcohol, we utilize one item that asked respondents if they had had a drink of beer, wine, or liquor, more than just a sip, a couple of times, or more in their life (1 = yes).⁷

Statistical Analyses

The dependent variable is dichotomous, and therefore the results consist of reporting odds ratios from logistic regression models.⁸ We utilize the appropriate weight, cluster, and strata variables in all analyses to account for the complex survey design found in Add Health. Because of the complex Add Health survey design, all of the tables include standard errors rather than standard deviations. These models allow us to assess whether levels of cannabis use at Wave I significantly affect the likelihood of use of OIDs at Wave II (Model 1, Hypothesis 1), and whether strain affects this relationship (Model 2, Hypothesis 2). Tests using variance inflation factors (VIFs) show that multicollinearity is not an issue in any of the presented equations.⁹

Results

Table 2 presents the results of logistic regression models where the Wave II OID measure is regressed on cannabis use at Wave I, strain, and the control variables. Model 1 includes the Wave I cannabis use measure and the controls. As can be seen in Model 1 of Table 2, frequency of use of cannabis at Wave I does significantly increase the odds of having tried OIDs between Waves I and II, supporting the gateway hypothesis. However, this effect is rather small. Based on the log odds, each event of smoking cannabis only increases the odds of trying OIDs between waves by about 1.55%. So while frequency of cannabis use at Wave I does exert a statistically significant effect on the likelihood of using OIDs at Wave II, the size of this effect is marginal. Among the controls, those who identify as Blacks are significantly less likely than non-Hispanic Whites to have used OIDs at Wave II, while higher household income, more delinquency, more affiliating with substance-using peers, having smoked and drank in one’s lifetime, having a parent who drinks, and lower self-control all significantly increase the odds of OID use at Wave II.

Table 2.

Other Illicit Drug Use WII Regressed on Cannabis Use WI, Strain, and Controls (n = 12,403).

Variable	Model 1	Model 2
Variable	OR (SE)	OR (SE)
Independent variable
Cannabis use, WI	1.015 (.01)*	1.015 (.01)
Strain theory
Strain, WI		1.143 (.06)*
Controls
Sex (1 = male)	1.130 (.14)	1.053 (.14)
Age, WII	0.949 (.04)	0.948 (.04)
Hispanic	0.744 (.17)	0.700 (.16)
Black	0.181 (.05)**	0.170 (.05)**
Native American	0.885 (.41)	0.866 (.39)
Asian	0.496 (.22)	0.481 (.22)
Other	0.699 (.32)	0.655 (.31)
Parent’s education	0.875 (.13)	0.890 (.13)
Parental employment	1.274 (.31)	1.287 (.32)
Household income	1.003 (.00)*	1.003 (.00)*
Minor delinquency, WI	1.118 (.04)**	1.096 (.04)*
Substance-using peers, WI	1.211 (.04)**	1.199 (.04)**
Smoking, WI	1.388 (.19)*	1.374 (.19)*
Alcohol use, WI	1.834 (.34)**	1.858 (.34)**
Parental alcohol use	1.147 (.07)*	1.150 (.07)*
Low self-control	1.043 (.01)**	1.042 (.01)**
McKelvey and Zavoina’s R²	.763	.765

Note. Non-Hispanic White is the reference category for all race/ethnic groups. This table includes ORs (linearized SEs) from logistic regression models. WII = Wave II; WI = Wave I; ORs = odds ratios.

p < .05. **p < .01.

Model 1 in Table 2 establishes marginal support for the gateway hypothesis concerning cannabis. In Model 2, we introduce the strain measure to assess whether this measure affects the cannabis–OIDs relationship that has been previously observed. Looking at Model 2 in Table 2, we see that strain at Wave I exerts the significant effect of increasing the odds of OID use at Wave II. Turning back to cannabis use, we see that this effect is no longer statistically significant when accounting for strain. Among the controls, all the same variables that were significant in Model 1 retain similar significance and strength in Model 2.

The results from Table 2 suggest that the link between cannabis use and the later use of OIDs may in part be the spurious outcome of experiencing strain during childhood and adolescence. However, for this to be the case, strain must significantly increase cannabis use. Table 3 presents a test of this possibility. Table 3 presents the results of a negative binomial regression model with cannabis use at Wave I regressed on strain and the controls. Negative binomial regression is utilized as the most appropriate statistical option when looking at the cannabis use variable because it is a highly skewed count measure that violates the assumption of normality required for traditional ordinary least squares regression (Gardner, Mulvey, & Shaw, 1995). As can be seen in Table 3, even with the controls in place, strain does indeed have a significant association with Wave I cannabis use.

Table 3.

Cannabis Use W1 Regressed on Strain and Controls (n = 12,403).

Variable	Model 1
Variable	Coefficient (SE)
Strain theory
Strain, WI	0.16 (.06)*
Controls
Sex (1 = male)	0.63 (.11)**
Age, WI	0.10 (.04)*
Hispanic	0.78 (.24)**
Black	1.21 (.23)**
Native American	−0.26 (.32)
Asian	0.15 (.29)
Other	0.07 (.75)
Parent’s education	−0.36 (.15)*
Parental employment	−0.05 (.27)
Household income	0.00 (.00)
Minor delinquency, WI	0.15 (.04)**
Substance-using peers, WI	0.52 (.03)**
Smoking, WI	0.46 (.22)*
Alcohol use, WI	0.97 (.17)**
Parental alcohol use	0.17 (.06)**
Low self-control	0.01 (.01)
Constant	−7.07 (.84)**

Note. Non-Hispanic White is the reference category for all race/ethnic groups. This table includes unstandardized coefficients (linearized SEs) from NB regression models. WI = Wave I; NB = negative binomial.

p < .05. **p < .01.

Combining the results from Tables 2 and 3 suggests that, in the current analytic sample at least, the cannabis–OIDs gateway relationship may simply be a classic case of spuriousness. When not considering strain, it looks like cannabis use is associated with an increase in the likelihood of using OIDs. However, these two variables are correlated only because of their joint relationship with experiencing strain during childhood and adolescence, which is significantly associated with both the frequency of cannabis use and the likelihood of using OIDs. Thus, strains may serve as a factor that increases the likelihood of using any illicit drug, with the wider availability of cannabis making it the most likely substance to be used first in the average individual’s drug use sequencing.

Discussion

While many studies to date have examined gateway theory, there have been methodological and data limitations, and few have tested the theory in conjunction with the role that strains play in the etiology of substance use (Peele & Brodsky, 1997; Rebellon & Van Gundy, 2006; van Gundy & Rebellon, 2010). The current study fills a gap in the literature by examining gateway theory in a nationally representative, longitudinal sample of American adolescents to understand the role of strain in gateway sequencing.

Like some of the literature to date, our findings only weakly support the cannabis–OID gateway sequence. Without accounting for strain, cannabis use at Wave I does exert a positive, statistically significant, but marginal, effect of increasing the odds of OID use between Waves I and II. Importantly, however, this marginal effect is reduced when accounting for strain. Strain increases the odds of using OIDs between Waves I and II, and subsequent analysis shows that strain also increases the use of cannabis. While our findings may seem to run counter to some previous studies that have found support for a gateway effect while controlling for criminological variables (Rebellon & Van Gundy, 2006), we must be clear that even though our variable of strain was based on general strain theory, this test is not a full test of general strain theory as we do not include strains more common to adolescents in our analysis. What should be taken from this study is that severe strains play an important role in drug use.

While these findings may be important, the limitations of the current study should be noted, which concern some data limitations. First, our OIDs outcome measure at Wave II is a single dichotomous indicator as to whether an individual used any of a number of drugs between Waves I and II. This measure is not ideal for a couple of reasons. One reason is that this means we are not able to focus on the effect of cannabis use on the frequency of use of OIDs, which could be as important as the difference between use versus nonuse. Another reason this measure is not ideal is because of its inclusivity, with many different substances put together, failing to distinguish between their various taxonomies and the differences between them. Any of the substances included in this variable are worthy of examining on their own to measure the gateway effect of cannabis. Due to limitations in the Add Health data, we cannot test for these differences in the current study.

Our second key limitation also relates to the design of Add Health, namely, that Wave I data were not collected until most respondents were around 15 years old. This means that for many individuals in the dataset, the chance to observe the gateway sequence had passed because they were already serious illicit drug users. This perhaps makes the current study a fairly conservative test of gateway sequencing. Another limitation to note is that all of the data utilized in this study were collected through self-report. Issues in the reliability of self-report data have been noted in the past, in particular as it relates to reporting on sensitive topics, many of which are present in the current study (Stone et al., 1999). Despite the overall value of the current study, this important limitation must be clearly noted. Last, it should again be pointed out that based on decisions made by the Add Health research team, the final analytic sample that includes measures from Waves I, II, and III is considerably smaller than the size of the Wave I in-home sample.

These limitations notwithstanding, we believe the current study makes a contribution to the literature on gateway theory, particularly the finding that in our analysis sample, the cannabis-OID gateway sequence appears to be the spurious outcome of strain, with strain serving a factor that shapes all forms of illicit substance use. Given the current U.S. political climate and differences of opinion in the scientific community concerning gateway theory, it is important to continue this line of research as we attempt to better understand the link between adolescent illicit drug use and subsequent use later in life. Specifically, we need to further understand the role of demographics and other sociocultural factors in the context of gateway theory to better identify populations who are more at risk for illicit drug use and work to identify other factors that increase illicit drug use of all kinds.

Footnotes

Authors’ Note

Both authors contributed equally to this article.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

References

Agnew

(1992). Foundation for a general strain theory of crime and delinquency. Criminology, 30, 47-88.

Agnew

(2001). Building on the foundation of general strain theory: Specifying the types of strain most likely to lead to crime and delinquency. Journal of Research in Crime & Delinquency, 38, 319-361.

Agnew

White

H. R.

(1992). An empirical test of general strain theory. Criminology, 30, 475-500.

Agrawal

Neale

M. C.

Prescott

C. A.

Kendler

K. S.

(2004) A twin study of early cannabis use and subsequent use and abuse/dependence of other illicit drugs. Psychological Medicine, 34, 1227-1237.

Alexander

B. K.

(2008). The globalisation of addiction: A study in poverty of the spirit. New York, NY: Oxford University Press.

Beaver

K. M.

DeLisi

Mears

D. P.

Stewart

(2009). Low self-control and contact with the criminal justice system in a nationally representative sample of males. Justice Quarterly, 26, 695-715.

Blanco

Rafful

Wall

M. M.

Ridenour

T. A.

Wang

Kendler

K. S.

(2014). Towards a comprehensive developmental model of cannabis use disorders. Addiction, 109, 284-294.

Boardman

J. D.

Finch

B. K.

Ellison

C. G.

Williams

D. R.

Jackson

J. S.

(2001). Neighborhood disadvantage, stress, and drug use among adults. Journal of Health and Social Behavior, 42, 151-165.

Carson

D. C.

Sullivan

C. J.

Cochran

J. K.

Lersch

K. M.

(2008). General strain theory and the relationship between early victimization and drug use. Deviant Behavior, 30, 54-88.

10.

Chassin

Pillow

D. R.

Curran

P. J.

Molina

B. S.

Barrera

Jr. (1993). Relation of parental alcoholism to early adolescent substance use: A test of three mediating mechanisms . Journal of Abnormal Psychology, 102(1), 3-19.

11.

Chassin

Rogosch

Barrera

(1991). Substance use and symptomatology among adolescent children of alcoholics. Journal of Abnormal Psychology, 100, 449-463.

12.

Chen

Chantala

(2014). Guidelines for analyzing Add Health data. Chapel Hill: Carolina Population Center, University of North Carolina.

13.

Choo

Roh

Robinson

(2008). Assessing the “Gateway Hypothesis” among middle and high school students in Tennessee. Journal of Drug Issues, 38, 467-492.

14.

Cleveland

H. H.

Wiebe

R. P.

(2008). Understanding the association between adolescent marijuana use and later serious drug use: Gateway effect or developmental trajectory? Development and Psychopathology, 20, 615-632.

15.

Cloward

R. A.

Ohlin

L. E.

(1960). Delinquency and opportunity. Glencoe, NJ: Free Press.

16.

Cohen

A. K.

(1955). Delinquent boys: The culture of the gang. Glencoe, IL: Free Press.

17.

Corr

M. L.

(2014). Young people’s early offending: The context of strained leisure careers. Young, 22, 113-133.

18.

Degenhardt

Chiu

W. T.

Conway

Dierker

Glantz

Kalaydjian

. . . Kessler

R. C.

(2009). Does the “gateway” matter? Associations between the order of drug use initiation and the development of drug dependence in the National Comorbidity Study Replication. Psychological Medicine, 39, 157-167.

19.

Degenhardt

Dierker

Chiu

W. T.

Medina-Mora

M. E.

Neumark

Sampson

. . . Kessler

R. C.

(2010). Evaluating the drug use “gateway” theory using cross-national data: Consistency and associations of the order of initiation of drug use among participants in the WHO World Mental Health Surveys. Drug and Alcohol Dependence, 108, 84-97.

20.

Degenhardt

Hall

Lynskey

(2001). The relationship between cannabis use and other substance use in the general population. Drug and Alcohol Dependence, 64, 319-327.

21.

de Wit

. (1996). Priming effects with drugs and other reinforcers. Experimental and Clinical Psychopharmacology, 4, 5-10.

22.

DiClemente

R. J.

Wingood

G. M.

Crosby

Sionean

Cobb

B. K.

Harrington

. . . Oh

M. K.

(2001). Parental monitoring: Association with adolescents’ risk behaviors. Pediatrics, 107(6), 1363-1368.

23.

Ellgren

Spano

S. M.

Hurd

Y. L.

(2007). Adolescent cannabis exposure alters opiate intake and opioid limbic neuronal populations in adult rats. Neuropsychopharmacology, 32, 607-615.

24.

Elliott

D. S.

Huizinga

Menard

(1989). The etiology of delinquency and ADM problems. In Multiple Problem Youth (pp. 137-168). New York, NY: Springer.

25.

Fergusson

D. M.

Boden

J. M.

Horwood

L. J.

(2006). Cannabis use and other illicit drug use: Testing the cannabis gateway hypothesis. Addiction, 101, 556-569.

26.

Fergusson

D. M.

Horwood

L. J.

(1997). Early onset cannabis use and psychosocial adjustment in young adults. Addiction, 92, 279-296.

27.

Fergusson

D. M.

Horwood

L. J.

(2000). Does cannabis use encourage other forms of illicit drug use? Addiction, 95, 505-520.

28.

Fergusson

D. M.

Horwood

L. J.

Lynskey

M. T.

Madden

P. A.

(2003). Early reactions to cannabis predict later dependence. Archives of General Psychiatry, 60, 1033-1039.

29.

Ford

J. A.

McCutcheon

(2012). The misuse of Ambien among adolescents: Prevalence and correlates in a national sample. Addictive Behaviors, 37, 1389-1394.

30.

Gardner

Mulvey

E. P.

Shaw

E. C.

(1995). Regression analyses of counts and rates: Poisson, overdispersed Poisson, and negative binomial models. Psychological Bulletin, 118, 392-404.

31.

Golub

Johnson

B. D.

(2002). The misuse of the “Gateway Theory” in US policy on drug abuse control: A secondary analysis of the muddled deduction. International Journal of Drug Policy, 13, 5-19.

32.

Gottfredson

M. R.

Hirschi

(1990). A general theory of crime. Stanford, CA: Stanford University Press.

33.

Grau

L. E.

Dasgupta

Harvey

A. P.

Irwin

Givens

Heimer

(2007). Illicit use of opioids: Is OxyContin® a “gateway drug”? American Journal on Addictions, 16, 166-173.

34.

Hall

W. D.

Lynskey

(2005). Is cannabis a gateway drug? Testing hypotheses about the relationship between cannabis use and the use of other illicit drugs. Drug and Alcohol Review, 24, 39-48.

35.

Harris

K. M.

Florey

Tabor

Bearman

P. S.

Jones

Udry

J. R.

(2003). The National Longitudinal Study of Adolescent Health: Research design. Retrieved from http://www.cpc.unc.edu/projects/addhealth/design

36.

Hoffmann

J. P.

S. S.

(1997). The conditional effects of stress on delinquency and drug use: A strain theory assessment of sex differences. Journal of Research in Crime & Delinquency, 34, 46-78.

37.

Jessor

(1980). A social-psychological framework for studying drug use. In Lettieri

D. J.

Sayers

Pearson

H. W.

(Eds.), Theories on drug abuse: Selected contemporary perspectives (pp. 102-109). Washington, DC: U.S. Government Printing Office.

38.

Kandel

D. B.

(1975). Stages in adolescent involvement in drug use. Science, 190, 912-914.

39.

Kandel

D. B.

(1978). Convergence in prospective longitudinal surveys of drug use in normal populations. In Kandel

D. B.

(Ed.), Longitudinal research on drug use: Empirical findings and methodological issues (pp. 3-38). Washington, DC: Hemisphere.

40.

Kandel

D. B.

(1984). Marijuana users in young adulthood. Archives of General Psychiatry, 41, 200-209.

41.

Kandel

D. B.

(Ed.). (2002). Stages and pathways of drug involvement: Examining the gateway hypothesis. Cambridge, UK: Cambridge University Press.

42.

Kandel

D. B.

Davies

Karus

Yamaguchi

(1986). The consequences in young adulthood of adolescent drug involvement: An overview. Archives of General Psychiatry, 43, 746-754.

43.

Kandel

D. B.

Yamaguchi

Chen

(1992). Stages of progression in drug involvement from adolescence to adulthood: Further evidence for the gateway theory. Journal of Studies on Alcohol and Drugs, 53, 447-457.

44.

Kaufman

J. M.

(2009). Gendered responses to serious strain: The argument for a general strain theory of deviance. JQ: Justice Quarterly, 26, 410-444.

45.

Khan

S. S.

Secades-Villa

Okuda

Wang

Pérez-Fuentes

Kerridge

B. T.

. . . Blanco

(2013). Gender differences in cannabis use disorders: Results from the National Epidemiologic Survey of Alcohol and Related Conditions. Drug and Alcohol Dependence, 130, 101-108.

46.

Kort-Butler

L. A.

(2010). Experienced and vicarious victimization: Do social support and self-esteem prevent delinquent responses? Journal of Criminal Justice, 38, 496-505.

47.

Lessem

J. M.

Hopfer

C. J.

Haberstick

B. C.

Timberlake

Ehringer

M. A.

Smolen

. . . Hewitt

J. K.

(2006). Relationship between adolescent marijuana use and young adult illicit drug use. Behavior Genetics, 36, 498-506.

48.

Lillie-Blanton

Anthony

J. C.

Schuster

C. R.

(1993). Probing the meaning of racial/ethnic group comparisons in crack cocaine smoking. Jama, 269(8), 993-997.

49.

Mayet

Legleye

Falissard

Chau

(2012). Cannabis use stages as predictors of subsequent initiation with other illicit drugs among French adolescents: Use of a multi-state model. Addictive Behaviors, 37, 160-166.

50.

Measham

Shiner

(2009). The legacy of “normalisation”: The role of classical and contemporary criminological theory in understanding young people’s drug use. International Journal of Drug Policy, 20, 502-508.

51.

Menard

(2000). Coefficients of determination for multiple logistic regression analysis. The American Statistician, 54, 17-24.

52.

Merton

R. K.

(1938). Social structure and anomie. American Sociological Review, 3, 672-682.

53.

Mittlböck

Schemper

(1996). Explained variation in logistic regression. Statistics in Medicine, 15, 1987-1997.

54.

Morral

A. R.

McCaffrey

D. F.

Paddock

S. M.

(2002). Reassessing the marijuana gateway effect. Addiction, 97, 1493-1504.

55.

Murray

R. M.

Morrison

P. D.

Henquet

Di Forti

(2007). Cannabis, the mind and society: The hash realities. Nature Reviews Neuroscience, 8, 885-895.

56.

Neff

J. L.

Waite

D. E.

(2007). Male versus female substance abuse patterns among incarcerated juvenile offenders: Comparing strain and social learning variables. Justice Quarterly, 24, 106-132.

57.

O’Donnell

J. A.

Clayton

R. R.

(1982). The stepping-stone hypothesis—Marijuana, heroin, and causality. Chemical Dependencies, 4, 229-241.

58.

Peele

Brodsky

(1997). Gateway to nowhere: How alcohol came to be scapegoated for drug abuse. Addiction Research & Theory, 5, 419-425.

59.

Pratt

T. C.

Cullen

F. T.

(2000). The empirical status of Gottfredson and Hirschi’s general theory of crime: A meta-analysis. Criminology, 38, 931-964.

60.

Rebellon

C. J.

Van Gundy

(2006). Can social psychological delinquency theory explain the link between marijuana and other illicit drug use? A longitudinal analysis of the gateway hypothesis. Journal of Drug Issues, 36, 515-539.

61.

Schroeder

R. D.

Ford

J. A.

(2012). Prescription drug misuse a test of three competing criminological theories. Journal of Drug Issues, 42, 4-27.

62.

Schulenberg

Bachman

J. G.

O’Malley

P. M.

Johnston

L. D.

(1994). High school educational success and subsequent substance use: A panel analysis following adolescents into young adulthood. Journal of Health and Social Behavior, 35, 45-62.

63.

Secades-Villa

García-Rodríguez

Jin

C. J.

Wang

Blanco

(2015). Probability and predictors of the cannabis gateway effect: A national study. International Journal of Drug Policy, 26, 135-142.

64.

Slocum

L. A.

(2010). General strain theory and the development of stressors and substance use over time: An empirical examination. Journal of Criminal Justice, 38, 1100-1112.

65.

Stone

A. A.

Bachrach

C. A.

Jobe

J. B.

Kurtzman

H. S.

Cain

V. S.

(Eds.). (1999). The science of self-report: Implications for research and practice. Mahwah, NJ: Lawrence Erlbaum Associates, Inc.

66.

Substance Abuse and Mental Health Services Administration. (2014). Results from the 2013 National Survey on Drug Use and Health: Summary of national findings (NSDUH Series H-48, HHS Publication No. (SMA) 14-4863). Rockville, MD: Author.

67.

Udry

J. R.

(1998). The National Longitudinal Study of Adolescent Health (Add Health). Chapel Hill: Carolina Population Center, University of North Carolina at Chapel Hill.

68.

Van den Bree

M. B.

Johnson

E. O.

Neale

M. C.

Pickens

R. W

. (1998). Genetic and environmental influences on drug use and abuse/dependence in male and female twins. Drug and alcohol dependence, 52(3), 231-241.

69.

van Gundy

Rebellon

C. J

. (2010). A life-course perspective on the “Gateway Hypothesis.” Journal of Health and Social Behavior, 51, 244-259.

70.

Van Ours

J. C

. (2003). Is cannabis a stepping-stone for cocaine? Journal of Health Economics, 22, 539-554.

71.

Wagner

F. A.

Anthony

J. C.

(2002). Into the world of illegal drug use: Exposure opportunity and other mechanisms linking the use of alcohol, tobacco, marijuana, and cocaine. American Journal of Epidemiology, 155, 918-925.

72.

Watts

S. J.

(2015). 5-HTTLPR, suicidal behavior by others, depression, and criminal behavior during adolescence. Journal of Adolescent Research, 30, 800-820.

73.

Watts

S. J.

McNulty

T. L.

(2013). Childhood abuse and criminal behavior: Testing a General Strain Theory model. Journal of Interpersonal Violence, 28, 3023-3040.

74.

Yamaguchi

Kandel

D. B.

(1984a). Patterns of drug use from adolescence to young adulthood: II. Sequences of progression. American Journal of Public Health, 74, 668-672.

75.

Yamaguchi

Kandel

D. B.

(1984b). Patterns of drug use from adolescence to young adulthood: III. Predictors of progression. American Journal of Public Health, 74, 673-681.