Parental Influences on Adolescent Marijuana Use and the Baby Boom Generation

CHAPTER 6: PARENTAL MARIJUANA USE AND OTHER PREDICTORS OF MARIJUANA USE AMONG CHILDREN

6.1 Introduction

Three analyses were implemented to examine the nature of parental influences on their children's marijuana use. The first analysis investigated the effect of parental membership in the baby boom generation on offspring marijuana use. Two additional multivariate analyses investigated the influence of parental marijuana use on the drug use of their children, controlling for other covariates, and identified other significant predictors of child marijuana use. One set of analyses examined the effects of parent and child characteristics in logistic regression models. The final set of analyses estimated structural equation models to specify the direct and indirect paths of selected parental and child variables on child marijuana use.

6.2 Parental Exposure to the Marijuana Epidemic

A basic question underlying this research was to what extent did parental membership in the baby boom generation account for the sharp increase in the prevalence of marijuana use observed among adolescents in the 1990s. To answer this question, we developed a typology of differential parental exposure to the marijuana epidemic that took into account exposure to different incidence and prevalence rates in late adolescence. As described in Chapter 2, we identified five historical periods of the marijuana epidemic and nine different types of exposure among the parental cohorts. These exposure types classified parental birth cohorts according to exposure to these five periods, i.e., whether parents spent their adolescent years in historical periods prior to or after the marijuana epidemic or in periods characterized by different combinations of low or high incidence and prevalence rates. Because of the very small sample size in exposure type #9, groups #8 and #9 were combined. Offspring marijuana use was examined as a function of parental membership in cohort types and parental drug use.

Since rates of marijuana use vary with age, descriptive data about the ages and rates of parental and children use in the eight cohorts types are presented in Table 6.1. Parent age ranged from a mean of 47.3 years to a mean of 28.5 years across the cohorts, and child age ranged from a mean of 15.8 to a mean of 12.8 years. Parental lifetime prevalence of use was very low in the oldest cohorts, increased sharply in the second oldest cohorts, who reached adolescence prior to the marijuana epidemic and in a period of low incidence; prevalence rates increased gradually in successive cohorts, peaked in the cohorts that experienced both high incidence and high prevalence, and declined gradually thereafter. By contrast, the rates of parental last year use increased gradually in the first six cohorts and stabilized in the two youngest cohorts. The rates of lifetime and last year used among the children decreased across the eight parental cohorts, in part because the children were younger.

We examined the four associations between parental lifetime and last year use with child lifetime and last year use. The association between parent last year use and child lifetime use was examined to assess whether parents who continued to use marijuana in their thirties, at a time when the majority of users have stopped using, would have a greater impact on their children than parents who were no longer using marijuana. Unadjusted odds ratios and odds ratios adjusted for parent and child age are presented for the four combinations of parent and child lifetime and last year marijuana use in Table 6.2. The odds of child lifetime and last year use were more likely to be statistically significant for parental lifetime than last year use, and for older than younger cohorts. The odds for the aggregated baby boom cohorts (#2-#7) were not significantly different from those for the pre-baby boom generation cohorts (#1). We also examined the differences for all the pair-wise comparisons among the cohorts. There were variations among cohorts within the baby boom generation. In three of the four combinations of parent and child use the odds for the four oldest cohorts (#1-#4), including the pre-baby boom cohort, were higher and significantly different from those in the younger cohorts (#5, #6 or #7). The exception was parental lifetime use on child last year use. Cohorts could be classified into two groups: cohorts in the pre-baby boom and early periods of the baby boom generation, showing higher levels of parental influence; cohorts in the late phase of the baby boom and post baby boom, showing lower levels of influence. The difference between the two groups of aggregated cohorts was significant for parent last year use on child lifetime use. Unexpectedly, the odds were significantly lower among parents who experienced high incidence (Group #5) or a combination of high incidence and high prevalence in their adolescence (Group #6) than among preceding cohorts. We expected parents who spent their adolescence in the period of greatest exposure to marijuana use incidence and prevalence to have the strongest influence on their children. Differential censoring in children's opportunities to initiate marijuana use across parental birth cohorts could partially explain these results. Parental birth cohorts who experienced both high incidence and prevalence (1960-62) were among the youngest in the sample ( age =32.5 years) and therefore had younger children ( age = 13.6 years) than earlier parental birth cohorts (child age range from 14.2 to 15.8 years). The youngest adolescents were even less likely than the older ones to have gone through the entire period of risk for initiation of marijuana use.

In a final test of the baby boom hypothesis, we examined the effect of the interactions between birth cohort and parent lifetime (Table 6.3) and last year (Table 6.4) marijuana use on the child marijuana use. This provided a more definitive test of whether, given a specific type of parental marijuana use history, parents from different birth cohorts would have differential influence on their children. The interaction term was significant only for parental last year use on child lifetime use (p<.03) (Table 6.4). This significant interaction effect reflected the cohort specific patterns described above, in which the highest similarity between parent and child was observed for the three oldest birth cohorts (1946-1956).

A major hypothesis of the study was not confirmed. Parental membership in the baby boom generation did not appear to account for the differential rates of children's marijuana use, even though parents born between the years 1946 and 1964 used marijuana at higher rates than parents born before 1946 or born after 1963. Association in marijuana use between parents and children did not vary according to parental membership in the baby boom cohorts, who experienced different periods of low and high marijuana use prevalence or incidence in their youth. There was no effect of parental birth cohort with control for ages of child and parent and for the relationship of marijuana use in the cohorts. [Footnote #1]. Furthermore, rates of use among children of parents born between 1946 and 1964 were lower than for parents born before 1946; and rates of use among children of parents born between 1946 and 1964 were similar to those of parents born after 1964. The lack of effect of parental membership in baby boom cohorts on children's marijuana use is illustrated in Figures 6.1 and 6.2. Figure 6.1 shows the distribution of birth cohorts among parents of children aged 12-17 in NHSDA surveys from 1979 to 1996 by the level of marijuana use incidence parents experienced in their youth. Parents who experienced high incidence of marijuana use constituted an increasingly larger proportion of parents in the years 1991-1996. However, as shown in Figure 6.2 marijuana use among parents and adolescents diverged and were mirror images of each other in that same period. Lifetime marijuana use rates among parents of youths and young adults approximately doubled from 1979 to 1994, reflecting the increasing dominance of the baby boom cohorts among parents. However, most of this increase occurred during the 1980's, a period in which youth and young adult drug use rates were declining. During the period of rapid increase in youth marijuana use (1992 to 1995), the percent of parents who were baby boomers or who had ever used marijuana did not change enough to have been a major factor in the youth increase. [Footnote #1- Similar correlations between parental and child marijuana use, when rates of parental use vary across birth cohorts, may not be reflected in parallel variations in children's rates of use, since other factors than parental use affect the child's use. The regressions that we estimated controlled for a limited number of relevant factors and provided adjusted estimates of parental effects.

We investigated which factors in addition to parental marijuana use accounted for child marijuana use.

Source: SAMHSA, Office of Applied Studies, National Household Survey on Drug Abuse

6.3 Predictors of Marijuana Use

Child's lifetime and last year marijuana use were regressed on selected predictors. A limited number of potential predictors were available in the NHSDA and their availability varied across surveys. Due to this variability in the availability of measures, models were estimated on three different samples of aggregated surveys to analyze the largest sample for each combination of predictors. The analyses based on the more restricted samples allowed the inclusion of several individual characteristics found in prior research to predict the use of marijuana by young people (for review, see Hawkins et al., 1992). We chose not to include an interaction term between parental cohort and marijuana use in the multiple regression models including other covariates because of the reduced sample sizes in these analyses.

(1) The 1979-1996 comprehensive aggregate sample, consisted of all the surveys and included the most restricted number of covariates. Only parent and child sociodemographic characteristics (i.e., family intactness, ethnicity, gender, education), parent lifetime, past year and extensiveness of marijuana use, use of other substances (i.e., cigarettes, alcohol, cocaine), and child drop out status were specified in the models.

(2) The 1991-1994A aggregate sample added parent and child attitudinal and personal characteristics, including parent and child delinquency, perceived risk of occasional marijuana use, and two sociodemographic characteristics, household income and population density.

(3) The 1994B-1996 aggregate sample included, in addition to the variables listed in (1) above, individual characteristics not included in (2): parent depression and anxiety, child behavioral and emotional problems, as well as two sociodemographic variables included in (2) above: household income and population density. Parent and child delinquency were not available in 1994B and 1996, and perceived risk of marijuana use was not available in 1995 and could not be included in the 1994B-1996 aggregate models.

Each model was estimated five times for each sample to assess the impact of five different measures of parental marijuana use on offspring use. The measures distinguished currency and extensiveness of use: marijuana use lifetime, last year use versus former use, frequency of lifetime use, frequency of past year use, and frequency of past month use. There were too few cases of past month use, especially frequent past month use, to provide stable estimates. These models are presented in Tables A.6.10 and A.6.11. To the extent possible, the definitions of the three other drug variables included in each model (cigarettes, alcohol, cocaine) were the same as for marijuana. However, this strategy could not be implemented with respect to frequency of lifetime and last year use for alcohol and cigarettes, either because this information had never been obtained or had been obtained only in selected years. For these two drugs, the former past year user variable was included in the lifetime frequency of marijuana use model, and the past month frequency of use variable was included in the past year frequency of marijuana model. Because the frequency of past year cocaine use was very low, the former/past year use classification was included instead in the frequency of past year marijuana model.

Since the effects of covariates other than drug use were very similar across models for child lifetime and last year use, only one set of results for these variables predicting lifetime use from lifetime parental marijuana are presented and discussed in this chapter (Table 6.5). Exceptions to the general pattern are mentioned. When patterns for lifetime and last year child marijuana use are the same, we usually refer to child marijuana use. Results for the four parental drug use variables included in each set of models are presented for child use lifetime (Table 6.6) and last year (Table 6.7) to facilitate comparison of coefficients based on different measures of drug use. The full models for each parental drug use parameterization and child lifetime and last year use are presented in Tables A.6.2-A.6.11.

One important point needs to be noted before examining the effects of specific covariates. The overall effect of parental marijuana use on the child increased when parent and child sociodemographic covariates and child dropout status were included in the models. The adjusted odds ratio of parental lifetime use on child lifetime use was 2.75 compared with an univariate odds ratio of 1.61 (Panel A, Table 6.6). The comparable odds ratios of parental last year use on child last year use were 2.55 and 1.71 respectively (Panel A, Table 6.7). However, compared with Panel A, the adjusted odds ratios declined somewhat when personal parent and child characteristics were controlled for, in addition to the sociodemographic variables.

Sociodemographic characteristics. Of the parent and child sociodemographic characteristics that were examined, many were associated, although weakly, with child lifetime (Panel A, Tables 6.5; A.6.2, A.6.4, A.6.6, A.6.8, A.6.10) and last year (Panel A, Tables A.6.3, A.6.5, A.6.7, A.6.9, A.6.11) marijuana use. Several of these relationships had been noted in the earlier discussion of descriptive data on the epidemiology of drug use in the dyads (Chapter 4). The strong univariate effect of parental birth cohort shows striking reductions with control for sociodemographic characteristics (Page 1, Tables 6.5; A.6.2-A.6.11). Adolescent males were more likely than females to use marijuana (Page 2, Tables 6.5; A.6.2-A.6.11), and whites were more likely than African-Americans to use marijuana only in their lifetime (Page 1, Tables 6.5; A.6.2, A.6.4, A.6.6, A.6.8, A.6.10). Marijuana use increased gradually with age, peaked between ages 19-21, and declined thereafter. Children from more recent birth cohorts, those born between 1970 and 1984, were less likely to use marijuana than those born between 1962-1964 (Page 2, Tables 6.5; A.6.2-A.6.11). Sociodemographic and structural characteristics of the family affected child marijuana use. Compared with children in intact families, children in widowed families were more likely to use marijuana lifetime and last year; while those in divorced families were more likely to use in the last year only (Page 1, Tables 6.5; A.6.2-A.6.11). Children in mother-child dyads were more likely to use marijuana than those in father-child dyads. Higher levels of parental education were associated with higher rates of child last year and to a lesser extent lifetime marijuana use. Living in the Western region of the United States was associated with the highest rates of adolescent marijuana use, living in the Southern region with the lowest. Household income had no impact on child marijuana use (Page 1, Tables 6.5; A.6.2-A.6.11).

Child personal characteristics. Child personal characteristics, including behavioral and emotional problems, delinquency and attitudes toward marijuana use predicted child marijuana use (Panels B and C, Table 6.5; Tables A.6.2-A.6.9). The coefficients were generally higher for child last year than lifetime use, probably because these predictors were measured within the last year. The most significant predictor was attitude regarding the risks involved in using marijuana. Children who perceived little or no physical risk associated with occasional marijuana use were nine times as likely to use marijuana in their lifetime and 12 times as likely to use in the last year compared with children who perceived that occasional marijuana use posed great risk (Panel B, Tables 6.5; A.6.3). Because of the cross-sectional nature of the data, the causal relationship between use and attitudes cannot be specified. Whether attitudes precede use, or whether behavior leads to more favorable drug-related attitudes remains to be determined.

Child delinquency (Panel B, Tables 6.5; A.6.2-A.6.11) and behavioral problems (Panel C, Tables 6.5; A.6.2-A.6.11) were significantly associated with marijuana use. For the continuous delinquency variable the AOR for last year use was 1.6, and for the dichotomous behavior problems the AOR was 4.1. Child emotional problems (i.e., withdrawal, anxiety and somatic complaints) were associated with lifetime marijuana use but not for last year use (Panel C, Tables 6.5; A.6.2, A.6.4, A.6.8, A.6.10).

Although very few adolescents in the sample had dropped out of high school (5.3%), those who did were almost three times as likely to use marijuana as non drop-outs (adjusted AORs=2.2-3.3) (Tables 6.5; A.6.2-A.6.11).

Parent Characteristics. In contrast to the predictive strength of child personality characteristics, parent personality characteristics, including major depression, anxiety problems and delinquency in the past year, did not predict child marijuana use. Parental marijuana attitudes predicted only child lifetime marijuana use in the univariate models. Without control for other factors, children's lifetime marijuana use was slightly higher (UOR= 1.5, p<.05) when parents perceived that occasional marijuana use posed little or no physical risk than when parents perceived great risk (Panel B, Tables 6.5; A.6.2, A.6.4, A.6.6, A.6.8).

Parental use of marijuana and other drugs. Unique effects of parental use of marijuana and use of other drugs on child lifetime and last year marijuana use were present (Tables 6.6, 6.7; A.6.2-A.6.11). Parental marijuana use, irrespective of the time frame of the measure, was significantly and uniquely associated with child marijuana use, controlling for other covariates, including parental use of other drugs. Parents who had used marijuana in their lifetime but were not current (last year) users and those who reported using marijuana within the last year had children who used marijuana at similar rates. Although the odds of child lifetime marijuana use were slightly higher (AOR=3.7) when parents had used marijuana more than 200 days in the past year compared with parents who had used 200 or fewer days (AOR=2.7) (Panel A, Table 6.6; A.6.8), the difference was not statistically significant (See Table 6.7 for child last year use.)

Parental use of other drugs significantly predicted child marijuana use. Parental use of cigarettes, alcohol, and cocaine, as well marijuana, each uniquely predicted child lifetime and last year marijuana use (Tables 6.6, 6.7; A.6.2-A.6.11). As we had observed for parental marijuana use, with rare exceptions, former and past year use of these three other drugs had similar associations with child marijuana use. Without control for personal characteristics, current parental drinkers had a stronger effect on the child marijuana use than former drinkers. Former and past use of cigarettes and cocaine were equally associated with child marijuana use. Extensiveness of drug consumption in the last year was not associated with increased rates of child marijuana use. A puzzling association was observed between frequency of parental smoking and child smoking (Panel C, Tables 6.6, 6.7; A.6.8-A.6.11): the lowest odds were observed among the heaviest smokers, who smoked more than 35 cigarettes a day. The low prevalence of parent last year cocaine use prevented the estimation of the effect of last year frequency of cocaine use. Percentages of children using marijuana as a function of parental use of cigarettes, alcohol and cocaine are presented in Table A.6.1.

 6.4 The Role of Attitudes Toward Marijuana

Given the importance of adolescent marijuana attitudes as a correlate of marijuana use, analyses were undertaken to explore the effects and determinants of these attitudes, especially in relationship to parental marijuana use and attitudes. These analyses were restricted to the 1991-1994A aggregated surveys, where data about child attitudes as well as delinquency were available.

Structural equation models were estimated to specify the direct and indirect paths of influence of selected parental and child variables on child marijuana use. Two structural models were estimated. The first model examined the direct impact of parental marijuana use, parental marijuana attitude and child marijuana attitude on child marijuana use as well as the impact of parental attitude on child attitude. Child attitude and use were endogenous variables (Figure 6.3). The second model added exogenous variables for parental use (cigarette, alcohol, cocaine), two indicators of child deviance (delinquency and being a school drop out), and age (Figure 6.4). With the exception of age, school drop out and child delinquency, which were represented by manifest indicators, the remaining constructs were represented by latent variables. The paths for child age are not shown. (The correlation matrix is presented in Table A.6.12)

Figure 6.3. Effects of Parent Marijuana Use and Attitude and Child Marijuana Attitude on Child Marijuana Use¹ (Standardized Coefficients, NHSDA 1991-1994a, N=4,957)

¹ Child age is included in the model; paths are not shown.

Source: SAMHSA, Office of Applied Studies, National Household Survey on Drug Abuse.

The simpler structural model (Figure 6.3) elucidates the direct and indirect effects of parental marijuana use and attitudes on child marijuana use and attitudes. Parental effects on the child are present across the same domain, from parental behavior to child behavior, or from parental attitude to child attitude. There are significant direct effects of parental marijuana use on child marijuana use, and of parental marijuana attitudes on child marijuana attitudes, but no direct cross-over effects of parental marijuana behavior on child marijuana attitudes and of parent marijuana attitudes on child marijuana behavior. As was noted in the logistic regression models, there is a very strong and significant direct path from child marijuana attitudes to child marijuana use. This effect is by far the strongest path in the model. There is a significant indirect effect (.05, p<.001) of parent marijuana attitude on child marijuana use mediated through child marijuana attitude, but no direct effect. As a further exploration of the consequences of membership in the baby boom generation, we examined parental marijuana attitudes in each of the eight types of birth cohorts. We expected parents in the baby boom generation to have more favorable attitudes toward marijuana than other cohorts. Perceived risks associated with occasional and regular use were examined (Table 6.8). There was a decline in perceived risk from the oldest to the youngest cohorts, with similar rates for five of the intermediate cohorts. This downward trend may be related to the decreasing age of cohort members rather than to type of exposure to the marijuana epidemic.

The preeminence of child marijuana attitude was retained in the more comprehensive model (Figure 6.4). Attitude was the strongest predictor in the model and was five times as strong as parental use. The next most important predictor was child delinquency. Its association with marijuana use was four times as strong as the association between adolescent and parental use. In addition to its direct effect on use, delinquency also had an indirect effect through the child attitude (.08, p<.001), for a total effect of (.33, p<.001). The same pattern was observed with respect to being a school drop out, although the size of the path was smaller. Of the four parental drug use factors, marijuana and cigarette use had the same effect. The effect of cocaine was slightly lower. There was no direct effect of alcohol use on the child marijuana use, but an indirect effect through its impact on the child marijuana related attitude. Surprisingly, this was the only one of the four substances to impact on the child attitude, after controlling for the use of other drugs.

The most striking results presented in this chapter pertain to the importance of children's marijuana related attitudes on their marijuana use and the importance of parental use of drugs other than marijuana in addition to parental marijuana use.

To assess the substantive implications of the results, predicted changes in rates of adolescent marijuana use were estimated from assumed changes in parental behaviors and attitudes and youth attitudes. Coefficients obtained in Figure 6.4 were used to calculate these estimates.

Figure 6.4. Predictors of Child Marijuana Use¹ (Standardized Coefficients, NHSDA 1991-1994a, N=4,957)

¹ Child age is included in the model; paths are not shown. Correlations among exogenous variables not shown.
Source: SAMHSA, Office of Applied Studies, National Household Survey on Drug Abuse.

Effect of parental marijuana use on child marijuana use

For a one-unit decrease in parental last year marijuana use, e.g., from 1-2 days a year to not using at all, 7 out of 100 youths would reduce their marijuana use by one level, e.g., from using marijuana 6 times a month to using marijuana 3 times a month.

Effect of parental marijuana attitude on child marijuana use

For a one-unit decrease in favorableness of parental attitudes toward marijuana use, e.g., from moderate to great perceived risk of occasional use, 4 out of 100 youths would reduce their marijuana use by one level, e.g., from using marijuana 6 times a month to using marijuana 3 times a month.

Effect of parental marijuana attitude on child marijuana attitude

For a one-unit decrease in favorableness of parental attitudes toward marijuana use, e.g., from moderate to great perceived risk of occasional use, 13 out of 100 youths would increase their perceived harmfulness of marijuana use by one unit, e.g., an increase from moderate to great perceived risk of occasional use.

Effect of child marijuana attitude on child marijuana use

For a one-unit decrease in favorableness of youth attitudes toward marijuana use, e.g., from moderate to great perceived risk of occasional use, 36 out of 100 youths would reduce their marijuana use by one level, e.g., from using marijuana 6 times a month to using marijuana 3 times a month.

As we noted earlier, in the absence of longitudinal data, the causal relationship between young people's marijuana attitudes and use cannot be determined. However, at any point in time, the association of use with attitude is the strongest of any other factor that was examined.

Table 6.1. Parent Age, Child Age, Parent and Child Lifetime and Last Year Marijuana Use by Parental Exposure to the Marijuana Epidemic¹ (NHSDA 1979-1996)

				Parent	Parent	Child	Child
				Lifetime	Last Year	Lifetime	Last Year
		Parent Age	Child Age	Marijuana Use	Marijuana Use	Marijuana Use	Marijuana Use
	N	(years)	(years)	%	%	%	%
Total N	9,463	41.5	15.1	32.8	5.3	18.0	13.2
Parental Birth Cohort
1. Pre-epidemic (before 1946)	2,119	47.3	15.8	14.1	3.0	26.0	18.8
2. Pre-epidemic/low incidence (1946-1948)	1,066	43.3	15.4	34.6	4.7	19.0	13.3
3. Low incidence (1949-1953)	1,951	40.8	15.2	38.4	5.3	16.3	12.0
4. Low incidence/high incidence (1954-1956)	1,235	37.6	14.5	44.9	6.3	9.3	6.9
5. High incidence (1957-1959)	1,379	34.9	14.2	48.9	8.2	12.6	9.7
6. High incidence/high prevalence (1960-1962)	1,165	32.5	13.6	52.8	10.4	8.8	7.7
7. High prevalence (1963-1964)	366	30.9	13.5	47.8	9.4	7.9	5.0
8. Post epidemic (after 1964)	182	28.5	12.8	44.9	10.0	7.5	6.5

¹ In 1979, 1982, 1988 and 1990, children aged 12-17 were selected. In all other years, children aged 12-25 were selected.

Note: Birth cohort groupings reflect parental exposure to the marijuana epidemic at ages 15-18.

Source: SAMHSA, Office of Applied Studies, National Household Survey on Drug Abuse.

Table 6.2. Association in Marijuana Use Between Parents and Children Aged 12-25¹, by Parental Birth Cohorts and Exposure to the Marijuana Epidemic, Unadjusted and Adjusted² Odds Ratios³ (NHSDA 1979-1996)

								Parent Marijuana Use
		Lifetime								Last Year
		Child Lifetime				Child Last Year				Child Lifetime					Child Last Year
	N	OR		AOR		OR		AOR		OR		AOR			OR		AOR
Total	9,463	1.6	***	2.2	***	1.7	***	2.2	***	1.7	***	2.5	***		1.9	***	2.5	***
Parental Birth Cohorts: Single Groups
1. Pre epidemic (before 1946)	2,119	2.4	***	2.8	***ab	2.3	***	2.4	***	2.5	**	3.0	**ac		2.6	**	2.6	**a
2. Pre epidemic/low incidence (1946-1948)	1,066	2.0	**	2.9	***ab	2.6	***	3.1	***	1.7		1.9	ac		1.5		1.6	a
3. Low incidence (1949-1953)	1,951	2.8	***	3.5	***ab	2.9	***	3.4	***	2.4	*	2.9	**a		3.0	**	3.7	***ac
4. Low incidence/high incidence (1954-1956)	1,235	4.1	***	5.1	***a	3.3	***	3.7	***	3.4	*	4.9	***		4.2	**	6.2	***ac
5. High incidence (1957-1959)	1,379	2.0	**	2.2	**b	2.3	**	2.4	**	1.8		1.6	bc		1.6		1.5	b
6. High incidence/high prevalence (1960-1962)	1,165	2.1	**	2.1	**b	1.9	*	1.9	*	1.2		1.2	b		1.2		1.3	b
7. High prevalence (1963-1964)	366	2.4		2.2	ab	1.2		1.1		.4		.3	b		.6		.5	b
8. Post epidemic (after 1964)	182	1.5		2.0	ab	2.2		3.0	*	6.9	*	10.	**a		9.0	**	17.5	**c
Parental Birth Cohorts: Dichotomy
1. Cohorts 1-4 (before 1957)	6,371	1.8	***	2.3	***	1.9	***	2.2	***	2.1	***	2.9	***		2.3	***	2.9	***
2. Cohorts 5-8 (after 1956)	3,092	2.0	***	2.2	***	2.1	***	2.1	***	1.5		1.5	b		1.5		1.5
Parental Birth Cohorts: Trichotomy
1. Pre Baby Boomer (before 1946)	2,119	2.4	***	2.8	***	2.3	***	2.4	***	2.5	**	3.0	***		2.6	**	2.6	***a
2. Baby Boomer (1946-1964)	7,162	2.3	***	3.1	***	2.4	***	3.0	***	1.9	***	2.4	***		2.1	***	2.5	***a
3. Post Baby Boomer (after 1964)	182	1.5		2.0		2.2		3.0	*	6.9	*	10.	**		9.0	**	17.5	**b

¹ In 1979, 1982 and 1990, children aged 12-17 were selected. In all other years, children aged 12-25 were selected.

² Adjusted for parent and child age.

³ Weighted estimates with SUDAAN PROC LOGISTIC, unweighted N's.

^a-c Odds ratios with different superscripts are significantly different from each other, Wald test (p<.05).

*p<.05; **p<.01; ***p<.001, T-test.

Note: Birth cohort groupings reflect parental exposure to the marijuana epidemic at ages 15-18.

Source: SAMHSA, Office of Applied Studies, National Household Survey on Drug Abuse.

Table 6.3. Logistic Regressions of Child Lifetime and Last Year Marijuana Use by Parent Lifetime Marijuana Use, Parent and Child Age at Survey, and Parental Exposure to the Marijuana Epidemic^1,2 (NHSDA 1979-1996)

	Child Lifetime Marijuana Use								Child Last Year Marijuana Use
	Model 1 (Unadjusted)				Model 2 (Adjusted)				Model 1 (Unadjusted)				Model 2 (Adjusted)
Predictors	OR		(95% CI)		AOR3		(95% CI)		OR		(95% CI)		AOR3		(95% CI)
Parent lifetime marijuana use	1.62	***	(1.36-1.92)		2.98	***	(1.99-4.49)		1.71	***	(1.42-2.06)		2.51	***	(1.65-3.82)
Parent age at survey (in years)	1.05	***	(1.04-1.06)		.95	***	( .93- .97)		1.04	***	(1.03-1.05)		.96	***	( .94- .98)
Child age at survey (in years)	1.36	***	(1.31-1.41)		1.41	***	(1.35-1.47)		1.24	***	(1.20-1.27)		1.26	***	(1.22-1.31)
Cohort 2 (1946-1948) (versus Cohort 1)	.67	***	( .51- .87)		.48	***	( .33- .70)		.66	*	( .48- .92)		.45	***	( .29- .68)
Cohort 3 (1949-1953)	.55	***	( .43- .71)		.30	***	( .19- .47)		.59	***	( .44- .78)		.34	***	( .22- .54)
Cohort 4 (1954-1956)	.29	***	( .22- .39)		.13	***	( .08- .22)		.32	***	( .23- .45)		.17	***	( .09- .31)
Cohort 5 (1957-1959)	.41	***	( .31- .54)		.30	***	( .19- .48)		.46	***	( .34- .63)		.31	***	( .18- .54)
Cohort 6 (1960-1962)	.27	***	( .20- .37)		.21	***	( .13- .36)		.36	***	( .26- .49)		.27	***	( .15- .49)
Cohort 7 (1963-1964)	.25	***	( .14- .42)		.17	***	( .07- .40)		.23	***	( .13- .41)		.22	**	( .09- .56)
Cohort 8 (after 1964)	.23	***	( .12- .46)		.24	***	( .11- .52)		.30	**	( .14- .64)		.22	***	( .09- .52)
Parent lifetime marijuana use X Cohort 2					2.61		(1.39-4.85)						3.00		(1.51-5.99)
Parent lifetime marijuana use X Cohort 3					3.39		(1.95-5.81)						3.10		(1.75-5.47)
Parent lifetime marijuana use X Cohort 4					4.53		(2.23-9.21)						3.39		(1.54-7.54)
Parent lifetime marijuana use X Cohort 5					2.03		(1.11-3.78)						2.23		(1.13-4.44)
Parent lifetime marijuana use X Cohort 6					1.99		(1.07-3.78)						1.86		( .95-3.60)
Parent lifetime marijuana use X Cohort 7					2.36		( .77-7.17)						1.19		( .35-4.10)
Parent lifetime marijuana use X Cohort 8					1.75		( .41-7.10)						2.44		( .60-9.87)

¹ In 1979, 1982 and 1990, children aged 12-17 were selected. In all other years, children aged 12-25 were selected.

² Weighted estimates with SUDAAN PROC LOGISTIC, unweighted N's.

³ The OR coefficients listed for the interactions are the total effects for each birth cohort, including the main effects. The ORs are calculated as the exponentiated

sum of the main effect of parental lifetime marijuana use and the interaction effect with each parental birth cohort. The interaction terms are not significant.

*p<.05; **p<.01; ***p<.001, T-test.

Source: SAMHSA, Office of Applied Studies, National Household Survey on Drug Abuse.

Table 6.4. Logistic Regressions of Child Lifetime and Last Year Marijuana Use by Parent Last Year Marijuana Use, Parent and Child Age at Survey, and Parental Exposure to the Marijuana Epidemic^1,2 (NHSDA 1979-1996)

	Child Lifetime Marijuana Use								Child Last Year Marijuana Use
	Model 1 (Unadjusted)				Model 2 (Adjusted)				Model 1 (Unadjusted)				Model 2 (Adjusted)
Predictors	OR		(95% CI)		AOR3		(95% CI)		OR		(95% CI)		AOR3		(95% CI)
Parent last year marijuana use	1.74	***	(1.27-2.38)		3.42	***	(1.68-6.93)		1.93	***	(1.38-2.71)		3.04	***	(1.50-6.18)
Parent age at survey (in years)	1.05	***	(1.04-1.06)		.95	***	( .93- .97)		1.04	***	(1.03-1.05)		.96	**	( .94- .99)
Child age at survey (in years)	1.36	***	(1.31-1.41)		1.39	***	(1.33-1.45)		1.24	***	(1.20-1.27)		1.25	***	(1.21-1.30)
Cohort 2 (1946-1948) (versus Cohort 1)	.67	***	( .51- .87)		.60	***	( .44- .82)		.66	*	( .48- .92)		.62	*	( .43- .90)
Cohort 3 (1949-1953)	.55	***	( .43- .71)		.45	***	( .33- .61)