Parental Drinking as Context for Parental Socialization of Adolescent Alcohol Use

Abstract

Introduction:

While alcohol-specific parenting practices have demonstrated unique effects on adolescent substance use, their efficacy in the context of parental drinking levels has not been studied. This study assessed the influence of three alcohol-specific parenting practices (rules, punishment, communication) on adolescent alcohol use, and the degree to which those associations varied by parents’ own drinking.

Methods:

We conducted logistic regression analyses among US adolescents (N = 1,023; 52% female; 12% Hispanic; 76% Caucasian, 5% Black, 8% mixed race, 11% other race/ethnicity; mean age at enrollment = 12.2 years) to examine the relationship between alcohol-specific parenting practices and the odds of ever having experienced two drinking milestones, having a full drink of alcohol and a heavy drinking episode, and whether parental drinking levels moderated those associations.

Results:

Strict rules for drinking, higher levels of cautionary communication messages, and punishment for drinking were associated with lower odds of alcohol use. Witnessing parent drinking increased the risk for both alcohol outcomes. Furthermore, parental drinking modified the influence of parental cautionary messages on alcohol use such that the effect was particularly salient for those youth who witnessed and whose parents reported higher levels of alcohol use.

Conclusions:

Family-based preventive interventions should include skills training in alcoholspecific parenting practices with emphasis on reducing parental alcohol use particularly when children are present.

INTRODUCTION

Adolescence is the developmental period in which initiation and escalation of alcohol use are most likely to occur (Windle et al., 2008). Despite ongoing prevention efforts, 64% of high school seniors have consumed alcohol, and 35% have done so in the past month (Johnston, O’Malley, Miech, Bachman & Schulenberg, 2016). The negative consequences associated with adolescent alcohol use make this risky behavior a public health priority and a concern of parents nationwide. Multiple parental socialization behaviors are known to have a substantial influence on adolescent alcohol use outcomes, however, the effects of these behaviors within the context of parental drinking in the home is not adequately understood. In the current study, we examine models of parental socialization that include the joint interaction of alcohol-specific parenting practices and parental drinking behaviors to elucidate which parenting practices are associated with initiation of adolescent alcohol use, and under which conditions of parental alcohol use those associations are most salient.

Parental socialization of alcohol

Parents play a critical role in the socialization of children, serving as their primary source of influence throughout childhood. Although peers become increasingly important during adolescence, parents continue to be instrumental in the socialization process throughout adolescent development (Wood, Read, Mitchell, & Brand, 2004). Parental socialization occurs via two processes. First, socialization occurs through general parenting behaviors, which reflect an overall parenting style and influence a range of adolescent outcomes including substance use, delinquency, and academic achievement. Research has long supported the conclusion that an authoritative parenting style, marked by high levels of support, and appropriate levels of behavioral control, is the most conducive to effective parenting (Baumrind, 1971; Steinberg, 2001). Second, socialization occurs through specific attempts to influence particular adolescent behaviors. One such targeted approach is alcohol-specific parenting, which reflects behaviors employed by parents to deter or prevent their child from consuming alcohol. These practices may include rules about drinking, punishment for drinking, and communication about drinking. Importantly, alcohol-specific socialization practices have been shown to be negatively associated with adolescent drinking, above and beyond the influence of general socialization behaviors (e.g. support and behavioral control; Handley & Chassin, 2013; van der Zwaluw et al., 2008). Thus, alcohol-specific parenting reflects a distinct component of the socialization process regarding alcohol. Given its association with adolescent alcohol use, and the modifiable nature of parenting behaviors, research on alcohol-specific parenting has implications for prevention and intervention strategies to reduce youth drinking.

A growing body of evidence supports longitudinal associations between alcohol-specific parenting practices and adolescent alcohol outcomes. We examine three alcohol-specific parenting practices relevant to family-based interventions. First, having strict rules about alcohol use has been repeatedly shown to have a significant protective effect on adolescent alcohol use (Jackson, Henriksen, & Dickinson, 1999; Mares, Lichtwarck-Aschoff, Burk, van der Vorst, & Engels, 2012; van der Vorst, Engels, Meeus, Deković, & Van Leeuwe, 2005; van der Vorst, Engels, Meeus, & Deković, 2006; van der Vorst, Engels, Deković, Meeus, & Vermulst, 2007; Janssen, Larsen, Peeters, Pronk, Vollebergh, & Wiers, 2014). That is, when parents actively set boundaries and limitations for drinking, their child is less likely to become involved in, and stay involved with alcohol. Second, parents may enforce punishments when rules are broken. Parental punishment tactics include grounding their child from activities or taking away privileges, among others. These actions may serve to reduce adolescent drinking by impelling teens to refrain from drinking because they do not want to incur the associated punishment. However, to date, no studies have independently examined punishment for drinking as an alcohol-specific parenting practice. Third, cautionary communication messages about alcohol are associated with reductions in adolescent drinking (Andrews, Hops, Ary, Tildesley, & Harris, 1993). The content of the communication message is important, as permissive messages about alcohol (e.g. drinking in moderation is acceptable, drinking is allowed in certain circumstances), are associated with higher levels of alcohol use and misuse (Jackson et al., 1999; Wood et al., 2004). Therefore, research supports policies recommending that parents talk to their children about alcohol using messages that relay the harms and consequences associated with alcohol (Jackson et al., 2016). Indeed, strict rule setting by parents in combination with high-quality communication has been associated with reduced levels of adolescent alcohol consumption (Koning et al., 2012).

Parental drinking as context for alcohol-specific socialization

In addition to alcohol-specific parenting practices, parents also influence child outcomes directly by modeling behaviors that are then internalized and repeated by their children, as specified by social learning theory (Bandura, 1977). Evidence suggests that adolescents whose parents drink regularly are at increased risk for using alcohol themselves (Alati et al., 2014; Ary, Tildesley, Hops, & Andrews, 1993). Witnessing parents drink alcohol may lead adolescents to drink and/or adopt norms permissive of alcohol use (Duncan, Duncan, & Strycker, 2006; van der Vorst, Vermulst, Meeus, Deković, & Engels, 2009; White, Johnson, & Buyske, 2000). Parental drinking levels are also associated with the severity of adolescent drinking, as heavy episodic parental drinking has been linked to both earlier and heavier alcohol use among adolescents (Vermeulen-Smit et al., 2012).

Parental drinking behaviors also can be conceptualized as an important component of the overall parenting context. In their integrative model of parenting, Darling & Steinberg (1993) argue that specific parenting practices occur within an overall context of parental socialization. That is, the effects of a particular parenting practice will differ depending on the parenting context within which the practices are enacted. Thus, the effects of alcohol-specific parenting practices will depend on the broader drinking environment parents establish in the home. The effects of parenting practices to reduce or deter adolescent drinking might be enhanced by parental modeling of reasonable drinking behaviors, whereas problematic parental drinking might weaken any protective effects of the alcohol-specific parenting practices. To date, however, these interactive effects have not been empirically examined. Therefore, in our study, we examine the degree to which the association between each parenting practice and adolescent alcohol use is dependent upon parents’ own use of alcohol.

Current study

We sought to assess the associations of three parental alcohol-specific socialization behaviors (cautionary communication messages, punishment for drinking, and rules about drinking), on the odds of having ever experienced two milestones of adolescent drinking: ever having a full drink of alcohol and ever having a heavy drinking episode. We expected stronger socialization behaviors (i.e., more communication, higher levels of punishment, and strict rules about alcohol) to be negatively associated with adolescents having had a full drink and heavy drinking episode. Furthermore, we assessed the degree to which parental alcohol consumption moderated the effect of each socialization behavior on each alcohol outcome. We expected higher levels of parental alcohol use to weaken the positive effects of each socialization behavior on both outcomes. The purpose of the current study was to inform prevention efforts by identifying modifiable targets for intervention (i.e., alcohol-specific socialization behaviors), by pinpointing appropriate timing of interventions, and by delineating the role of parental alcohol use to ensure the most effective delivery of protective strategies to reduce adolescent alcohol use.

METHODS

Participants

Data for this study were drawn from an ongoing longitudinal study of contextual factors that influence the initiation and progression of alcohol use during adolescence. Adolescents (N = 1,023; 52% female; 12% Hispanic; 76% Caucasian, 5% Black, 8% mixed race, 11% other race/ethnicity; mean age at enrollment = 12.2 years) and their parents were recruited from six Rhode Island, USA secondary schools composed of Grades 6–8. As compared to data from the Rhode Island Department of Education (http://infoworks.ride.ri.gov), the sample was largely representative of the schools from which it was drawn with respect to gender and grade level, but with some evidence of greater racial diversity and less economic disadvantage than the school populations.

Procedures

Following a baseline assessment, participants completed semi-annual surveys for two years, and a three-year follow-up survey. The study design was altered at the point of re-funding and participants completed surveys at quarterly intervals thereafter. All surveys were web-based and could be completed from any location with Internet access, and participants were compensated $20 for the completion of each follow-up survey (for further detail on the study protocol, see Jackson et al., 2014; 2015).

Data for this study are from the first quarterly survey (henceforth called Time 1, T1, average age=15.98 years), and two follow-up points, one a year later (Time 2, T2, average age=16.68 years), and one 1.5 years later (Time 3, T3, average age=17.17 years). These time points were selected based on when the constructs of interest were measured. Thus, the sample for this study consists of those participants (n=848; 82%) who agreed to participate in continued data collection (at the point of re-funding). As compared to those who did not agree to reparticipate, included study participants were more likely to be female (χ²(1) = 8.32, p < .01), non-Hispanic White (χ²(1) = 11.57, p < .01), and not receive a lunch subsidy, which we use as a proxy for family socioeconomic status (χ²(1) = 6.59, p < .05), but did not differ based on age.

Parents were mailed a paper-and-pencil survey to complete corresponding to T1 and T2 of the adolescent surveys. Data from the self-selected responding parent were used in this study, of which 87% were mothers or female guardians at T1 and at T2. All study procedures were approved by the University’s Institutional Review Board, adolescent assent and parental consent for study participation was required, and a Certificate of Confidentiality was obtained from NIAAA to protect participant confidentiality.

Parental socialization behaviors

Perceived parental communication and punishment for drinking.

Adolescents responded to eight questions assessing their parents’ anticipated response to finding out they were drinking alcohol. Parents responded to items identical in content but reworded to be from the parent perspective. The measure was originally developed to assess parent reactions to youth smoking and later adapted to assess smoking, drinking, and drug use (Chassin, Presson, Todd, Rose, & Sherman, 1998). Two subscales for this measure were originally identified by Chassin and colleagues (1998) reflecting communication about and punishment for substance use. We decided a priori to only use one of the three communication items, “If you were drinking and your parents found out, how likely is it that they would talk with you about the reasons you shouldn’t drink” as the measure for parental communication about drinking due to the fact that it was the only communication item that indicated a direction of the message content about drinking (i.e., negative/cautionary). Based on confirmatory factor analysis, four items including “Take away a privilege, like watching TV or using a cell phone”, “Take away something from you, like treats or allowance”, “scold/yell at you”, and “ground you” loaded on the punishment factor, and were averaged to create a measure for parental punishment for drinking (T1: α_adolescent = .95; α_parent = .87; T2: α_adolescent = .96; α_parent = .87). Response options for all items ranged from not at all likely (1) to very likely (5).

Parental rules about alcohol use.

Adapting a measure on rules for teenage drinking by Dishion, Nelson, and Kavanagh (2003), parents were asked “Which statement best describes the rules for your child’s drinking in your house?” The item was measured on a five-point scale ranging from “My child is not allowed to drink in my home under any circumstance” (1) to “I have not made any rules for my child drinking in my home” (5). Due to the skew in the data towards strict rules (82% of parents reported absolute strictness at Time 1), we created a binary measure for parental rules about alcohol such that 1 = not allowed to drink in the home under any circumstance, and 0 = any other response option. No parallel measure regarding rules for alcohol use was asked of adolescent participants.

Parental alcohol use

Parents reported on both their frequency of alcohol use (number of drinking days in past 30 days) and quantity of use (average number of drinks per drinking day in past 30 days). The two scores were multiplied to create a measure for parent-reported parental alcohol use. Adolescents were asked of all parents and caretakers in their home whether they had seen each person drink alcohol. Responses were: never (0), yes, but only very rarely (1), yes, occasionally (2), or yes, frequently (3). The maximum value reported by the participant across parents/caretakers was used as the measure for adolescent-reported parental alcohol use (witness).

Adolescent alcohol use

During orientation, a standard drink was defined as 12 oz of beer, 5 oz wine, or a shot (1.5 oz) of hard liquor. At each survey, participants indicated whether they had ever had a full drink of alcohol, not including alcoholic drinks consumed as part of a religious ceremony. From this measure, we categorized participants into two groups (yes or no) based on reports of drinking behavior at all time points up to Time 3, indicating ever having had a full drink of alcohol. Participants also indicated the maximum number of drinks consumed in a drinking episode at each survey. From this measure, we categorized participants into two groups (yes or no) based on whether they had consumed three or more drinks during a single drinking episode, indicating ever having had a heavy drinking episode, (as recommended by Donovan, 2009 as an index of heavy episodic drinking for our adolescent sample).

Covariates

Demographics.

Participants reported on sex, race, and Hispanic ethnicity at baseline. Age was calculated in quarter-year intervals to reflect the age of the participant at Time 1. Parents reported whether their child was eligible for free or reduced-price lunch, which was used as a binary indicator of family socioeconomic status.

General parenting (parental support and control).

Adolescents completed the Network of Relationship Inventory (Furman & Buhrmester, 1985). Six items regarding support and companionship such as “How often do you turn to this person for support for personal problems?” and “When you are feeling down or upset, how often do you depend on this person to cheer things up?” were asked separately for their mother-figure (α = .87) and father-figure (α = .84). Scales for mothers and fathers were positively correlated (r(848) = .55, p < .001), so scores were averaged to represent an overall index of general parental support. Adolescents also completed the Parental Monitoring Questionnaire (PMQ; Kerr & Stattin, 2000), which operationalizes parental monitoring as parent knowledge of child activities. The PMQ is comprised of nine items such as “Do your parents know what you do during your free time?” and “Do your parents normally know where you go and what you do after school?” Items were measured on a five-point Likert scale ranging from no/never (0) to yes/always (5). One item, “In the last month, have there been nights when your parents had no idea where you were?” did not correlate well (r(848) < .15 with total), and was dropped. The remaining eight items were averaged to create a measure of general parental control (α = .93). Measures of general support and control were added to each model as covariates to ensure the effect of each alcohol-specific socialization behavior was above and beyond the effect of general parenting behaviors.

Data analysis

We first conducted descriptive analyses of univariate properties and correlations between key study variables. To test study hypotheses, logistic regressions were conducted in two steps. In the first step, we predicted likelihood of the alcohol behavior outcome by each of the main effects of the alcohol-specific parenting practices (for adolescent-report models: communication, punishment, witnessing parents drink; for parent-report models: communication, punishment, rules about drinking, parental alcohol use), and all covariates (socio-demographics, general parental support, general parental control¹). In the second step, interactions were calculated and entered in the model as the product of the parental alcohol use variable (for parent-reported models: parent alcohol use; for adolescent-reported models: witness) and each parenting practice. We used a likelihood ratio test to ascertain whether the set of interactions improved model fit, which was significant in all models at p <.05. We probed significant interactions, testing simple slopes associated with the effect of the parenting practice at varying levels of the moderating parental drinking variable (no, mean, and one standard deviation above the mean).

This sequence of regressions was conducted for each combination of alcohol behavior outcome (full drink or heavy drinking episode, both at Time 3), informant (parent or adolescent report), and timing of predictors (Time 1 or Time 2), resulting in eight total models. We include both adolescent- and parent-report of parenting variables as they have been shown to differentially predict adolescent substance use and may capture subtly different aspects of parental behavior (e.g. Latendresse et al., 2009). In all models including T2 parenting behaviors, the value of the behavior at T1 was also included in the model so that the coefficient for the T2 predictor reflects the change in that behavior from Time 1 to Time 2. Missing data was accommodated using full information maximum likelihood estimation. Descriptive analyses were performed using SAS (version 9.4, SAS Institute Inc.), and logistic regression analyses were performed using MPlus Version 7 (Muthen & Muthen, 2010).

RESULTS

Table 1 provides descriptive statistics of the sample, and associations among study variables. Overall, parental reports of communication about drinking and punishment were higher on average than adolescent-report of the same variables at both time points. Average levels of adolescent-report of each parenting practice decreased over time, though parent-report of behaviors remained relatively constant. All adolescent-report parenting practices (communication and punishment) were significantly negatively correlated with each alcohol outcome, though significance with parent-reported behaviors varied across time and outcome. Witnessing a parent drink at T1 and T2 was positively correlated with alcohol use.

Table 1:

Sample descriptive statistics and associations among study variables

Parent-reporting parenting behaviors
T1							T2
	Comm.	Punish	Rules	Parent alcohol use	Ever full drink (1=yes)	Ever heavy use (1=yes)	Comm.	Punish	Rules	Parent alcohol use	Ever full drink (1=yes)	Ever heavy use (1=yes)
Communication
Punishment	.40***						.32***
Rules (1=strict)	0.06	.28***					.18***	.25***
Parent alcohol use	−0.05	−0.10*	−.14***				−0.06	−0.04	−.15***
Ever full drink	0.02	−0.03	−.17***	0.05			−.11*	−.14**	−.26***	0.06
Ever heavy use	−0.01	−0.03	−.10*	0.05	.76**		−.08	−0.09	−.20***	−0.01	.76**
Range	1–5	1–5	0–1	0–165	0–1	0–1	2–5	1–5	0–1	0–182	0–1	0–1
Mean (SD) or %	4.83 (0.57)	4.25 (0.97)	82.25%	8.55 (17.40)	54.96%	41.37%	4.83 (0.48)	4.21 (0.99)	75.98%	8.19 (15.96)	54.96%	41.37%
Adolescent-reported parenting behaviors
T1							T2
	Comm.	Punish	Witness parent drink	Ever full drink (1=yes)	Ever heavy use (1=yes)		Comm.	Punish	Witness parent drink	Ever full drink (1=yes)		Ever heavy use (1=yes)
Communication
Punishment	.80***						.81***
Witness	0.02	−.01					0.07	0.07
Ever full drink	−.13***	−.14***	.09*				−0.16***	−.13***	.12**
Ever heavy use	−.17***	−.19***	0.04	.76***			−.20***	−.16***	.11**	.76**
Range	1–5	1–5	0–3	0–1	0–1		1–5	1–5	0–3	0–1		0–1
Mean (SD) or %	3.99 (1.44)	3.94 (1.39)	1.74 (0.99)	54.96%	41.37%		3.87 (1.43)	3.81 (1.43)	1.79 (0.99)	54.96%		41.37%
Correlations between parent- and adolescent-reported parenting behaviors
T1							T2
	Parent Comm.	Parent Punish	Adolescent Comm.	Adolescent Punish				Parent Comm	Parent Punish	Adolescent Comm.		Adolescent Punish
Parent Comm.							Parent Comm
Parent Punish	.40***						Parent Punish		.32***
Adolescent Comm.	.20***	.10*					Adolescent Comm.		.14**	.06
Adolescent Punish	.21***	.21***	.80***				Adolescent Punish		.09*	.15***		.81***

^*p < .05

^**p < .01

^***p<.001

Predicting full drink.

Table 2 presents the complete set of models for predicting ever having a full drink. Having strict rules for drinking at T1 (OR = 0.41, p < .001, 95% CI [0.25, 0.66]) and T2 (OR = 0.37, p < .01, 95% CI [0.20, 0.68]) reduced the odds for ever having had a full drink. Parental report of punishment at T2 also reduced the odds of having a full drink (OR = 0.74, p < .05, 95% CI [0.57, 0.95]). Adolescent-report of parental communication about alcohol was negatively associated with having a full drink, but only at T2 (OR = 0.77, p < .05, 95% CI [0.62, 0.97]). Adolescent-report of parental punishment did not predict ever having a full drink, although their report of witnessing a parent drink alcohol was positively associated with ever having had a full drink at T1 (OR = 1.21, p < .05, 95% CI [1.03, 1.43]) and T2 (OR = 1.29, p < .05, 95% CI [1.04, 1.61]).

Table 2:

Logistic regression results for T3 full drink of alcohol predicted by T1 and T2 alcohol-specific parenting Parent-reported parenting behaviors

Parent-reported parenting behaviors		T1				T2a
		OR	p	95% CI		OR	p	95% CI
Step 1	Sex (1=male)	0.42	<.001	0.31	0.56	0.41	<.001	0.29	0.56
	Race/Ethnicity (1=non-Hispanic White)	1.08	0.35	0.92	1.26	1.05	0.53	0.90	1.23
	Age	1.64	<.001	1.40	1.90	1.63	<.001	1.38	1.92
	Lunch subsidy (1=yes)	0.77	0.16	0.53	1.11	0.89	0.54	0.60	1.31
	Support	0.88	0.22	0.72	1.08	0.90	0.31	0.72	1.11
	Behavioral control	0.64	<.001	0.53	0.78	0.63	<.001	0.51	0.77
	Communication	1.16	0.37	0.84	1.60	0.79	0.37	0.48	1.31
	Punishment	1.08	0.46	0.89	1.31	0.74	0.02	0.57	0.95
	Rules (1=strict)	0.41	<.001	0.25	0.66	0.37	<0.01	0.20	0.68
	Parent alcohol use	1.00	0.48	0.99	1.01	1.01	0.16	0.995	1.03
Step 2	Communication*Parent Alcohol Use	0.96	0.01	0.92	0.99	1.00	0.92	0.97	1.04
	Punishment*Parent Alcohol Use	1.01	0.08	0.99	1.02	1.02	0.11	0.996	1.04
	Rules*Parent Alcohol Use	1.00	0.95	0.98	1.02	1.01	0.77	0.97	1.04
Adolescent-reported parenting behaviors		T1				T2a
		OR	p	95% CI		OR	p	95% CI
Step 1	Sex (1=male)	0.41	<.001	0.31	0.56	0.41	<.001	0.30	0.55
	Race/Ethnicity (1=non-Hispanic White)	1.07	0.37	0.92	1.25	1.08	0.31	0.93	1.26
	Age	1.67	<.001	1.44	1.95	1.67	<.001	1.43	1.96
	Lunch subsidy (1=yes)	0.77	0.16	0.53	1.11	0.75	0.14	0.51	1.10
	Support	0.89	0.24	0.72	1.09	0.93	0.51	0.76	1.15
	Behavioral control	0.70	<0.01	0.57	0.86	0.70	<.001	0.57	0.86
	Communication	0.91	0.35	0.75	1.11	0.77	0.03	0.62	0.97
	Punishment	0.96	0.66	0.79	1.17	1.13	0.32	0.89	1.42
	Witness parent drinking	1.21	0.02	1.03	1.43	1.29	0.02	1.04	1.61
Step 2	Communication*Witness	1.23	0.045	1.01	1.51	1.19	0.16	0.94	1.50
	Punishment*Witness	0.88	0.23	0.71	1.09	0.81	0.09	0.63	1.03

^*Note: T2 models also control for the value of each alcohol-specific parenting behavior and parental alcohol use at T1.

There were two significant interactions between parenting practices and parent drinking. The interaction between parent-reported communication at T1 and parental alcohol use (see Figure 1) was significant (OR = 0.96, p < .05, 95% CI: 0.92, 0.99), such that with increasing levels of parental communication, the probability of a having a full drink increased for those who did not witness their parents drink alcohol, and decreased for those who witnessed high levels of parental drinking. The simple slope (i.e. the effect of the predictor on the outcome at a certain level of the moderator) for no parental drinking was significantly different from zero (b = .39, p < .05), marginally significant for high parental drinking (b = −.81, p = .06), and non-significant for mean levels of witnessing a parent drink (b = −.01, p = .96). The interaction between adolescent-reported communication at T1 and witnessing their parent drink was also significant (OR = 1.23, p < .05, 95% CI: 1.01, 1.51), and displayed a similar pattern of effects (See Figure 2), though no simple slopes were significantly different from zero.

Figure 1:

Interaction between T1 parent-reported communication and parental drinking on probability of having a full drink.

Figure 2:

Interaction between T1 adolescent-reported communication and parental drinking on probability of having a full drink.

Predicting heavy drinking episode.

Table 3 presents the complete set of models for predicting ever having a heavy drinking episode. No T1 parenting practices reported by either informant predicted having a heavy drinking episode. At T2, having strict rules about alcohol reduced the odds of ever having had a heavy drinking episode (OR = 0.33, p < .001, 95% CI [0.18, 0.61]). Adolescent report of parental communication at T2 was negatively associated (OR = 0.76, p < .05, 95% CI [0.61, 0.94]), and witnessing the parent drinking alcohol at T2 was positively associated (OR = 1.43, p < .01, 95% CI [1.13, 1.81]) with heavy drinking.

Table 3:

Logistic regression results for T3 heavy alcohol use predicted by T1 and T2 alcohol-specific parenting

Parent-reported parenting behaviors		T1				T2
		OR	p	95% CI		OR	p	95% CI
Step 1	Sex (1=male)	0.49	<.001	0.36	0.67	0.51	<.001	0.37	0.70
	Race/Ethnicity (1=non-Hispanic White)	1.11	0.20	0.95	1.29	1.08	0.32	0.93	1.27
	Age	1.73	<.001	1.48	2.03	1.70	<.001	1.44	2.01
	Lunch subsidy (1=yes)	0.80	0.23	0.56	1.16	0.88	0.52	0.59	1.31
	Support	0.72	<.01	0.58	0.89	0.73	<.01	0.58	0.91
	Behavioral control	0.59	<.001	0.49	0.72	0.58	<.001	0.48	0.71
	Communication	1.08	0.63	0.78	1.51	0.92	0.70	0.58	1.45
	Punishment	1.08	0.44	0.89	1.32	0.82	0.14	0.64	1.07
	Rules (1=strict)	0.68	0.10	0.43	1.08	0.33	<.001	0.18	0.61
	Parent alcohol use	1.01	0.26	0.996	1.02	1.00	0.54	0.98	1.01
Step 2	Communication*Parent Alcohol Use	0.94	<.01	0.91	0.98	1.00	0.96	0.97	1.04
	Punishment*Parent Alcohol Use	1.01	0.28	0.995	1.02	1.02	0.16	0.99	1.04
	Rules*Parent Alcohol Use	0.99	0.47	0.97	1.02	1.00	0.80	0.96	1.03
Adolescent-reported parenting behaviors		T1				T2
		OR	p	95% CI		OR	p	95% CI
Step 1	Sex (1=male)	0.49	<.001	0.36	0.67	0.48	<.001	0.35	0.66
	Race/Ethnicity (1=non-Hispanic White)	1.10	0.23	0.94	1.28	1.11	0.18	0.95	1.30
	Age	1.72	<.001	1.47	2.01	1.72	<.001	1.46	2.02
	Lunch subsidy (1=yes)	0.78	0.21	0.53	1.15	0.76	0.18	0.51	1.13
	Support	0.71	<.01	0.57	0.89	0.75	0.01	0.60	0.94
	Behavioral control	0.66	<.001	0.54	0.81	0.65	<.001	0.53	0.80
	Communication	1.00	0.96	0.83	1.20	0.76	0.02	0.61	0.94
	Punishment	0.85	0.09	0.70	1.02	1.10	0.41	0.87	1.40
	Witness parent drinking	1.09	0.32	0.92	1.29	1.43	<.01	1.13	1.81
Step 2	Communication*Witness	1.12	0.25	0.92	1.36	1.12	0.37	0.88	1.42
	Punishment*Witness	0.87	0.18	0.71	1.07	0.87	0.28	0.68	1.12

^*Note: T2 models also control for the value of each alcohol-specific parenting behavior and parental alcohol use at T1.

There was one significant interaction between parent-reported communication at T1 and parent drinking (OR = 0.94, p < 0.01, 95% CI: 0.91, 0.98). The pattern of results (see Figure 3) was similar to results found for predicting a full drink of alcohol, though only the simple slope for high levels of parental drinking was statistically significant (b = −1.16, p < .01). The simple slope for no parental drinking was marginally significant (b = .38, p = .06), while the simple slope for mean levels of parental drinking was non-significant (b = −.13, p = .51).

Figure 3:

Interaction between T1 parent-reported communication and parental drinking on probability of having a heavy drinking episode.

DISCUSSION

This study sought to expand our understanding of parental socialization behaviors on adolescent alcohol use in two ways. First, the study examined the associations of three alcoholspecific parenting practices (rules about drinking, punishment for drinking, and communication about drinking) with two adolescent drinking milestones (ever having a full drink of alcohol and heavy drinking episode). All analyses adjusted for general parenting behaviors to indicate unique effects of alcohol-specific parenting practices on adolescent alcohol use. Second, we assessed the contribution of parental drinking as an integral factor within the overall parenting context in which alcohol-specific socialization behaviors are enacted. We expected increased rules about drinking, punishment for drinking, and cautionary communication about drinking to be protective of both drinking behaviors, and that those associations would be moderated by parental drinking. We found protective influences of parental communication and rules for drinking across multiple outcomes and time points, whereas punishment was only associated with having a full drink in one model. Additionally, the influence of parent-reported parental communication at T1 on both alcohol outcomes, and adolescent-reported parental communication at T1 on having a full drink was dependent on levels of parent alcohol use. We discuss the findings regarding the main effects and interactions and their implications in detail below.

Results of this study corroborate previous evidence regarding the importance of setting rules about drinking to reduce adolescent alcohol use (Mares et al, 2012; van der Vorst et al., 2006; van der Vorst et al., 2007; Koning et al., 2010). Having strict rules about drinking consistently decreased the odds of alcohol involvement. Establishing clear boundaries for alcohol use when youth are young, and maintaining those parameters as youth age is an effective familybased strategy for preventing adolescent alcohol use. Prevention programs should provide exemplar models of rules about drinking.

Punishment for drinking was only protective of adolescent drinking in the parent-reported model at T2. In this case, maintaining punishment for drinking as an adolescent gets older reduced the odds of reporting having had a full drink of alcohol. This may reflect youth not wanting to miss out on social events (for example, if he/she were grounded) given the importance of peers and social connections in adolescence (Brown & Larson, 2009). However, punishment for drinking was not protective of adolescent alcohol use in any other multiple regression model. It may be that the mere act of setting punishments for drinking is only the first step, and that following through consistently to enforce the rules and enact punishments for drinking is needed to limit youth drinking. It is also important to note that the regression models in this study investigated the unique effect of setting punishments in the context of other parenting behaviors (i.e., with other parenting behaviors included in the model), and that adolescent-reported punishments were negatively correlated with ever having a full drink and drinking heavily. Given the dearth of research regarding punishment for drinking, future research should investigate the extent to which enforcement of rules and punishment impacts adolescent drinking, and to include the influence of parent drinking on those associations.

Finally, adolescents’ perception of cautionary communication messages about alcohol reduced the odds of having a full drink and a heavy drinking episode, though this effect only held for T2 accounts of parental communication. Given that levels of T1 communication were included in the models of T2 behaviors, this indicates that change in parental communication over time was associated with heavy drinking. This demonstrates that protective effects of parental communication may only apply when messages are consistently reinforced over time. The salience of negative communication messages about alcohol may increase over time not only through repetition of the message, but also in delivering those negative messages as adolescents get older and alcohol use is more prevalent.

Parent reports of their own drinking behavior did not have any direct effect on adolescent drinking in multiple regression models, a finding that contradicts previous studies (van der Vorst et al., 2009; White et al., 2000). However, we found a significant interaction of parental report of their own drinking behavior and their use of cautionary communication messages on both having had a full drink and heavy drinking episode. For adolescents whose parents reported high levels of drinking themselves, increasing levels of parental communication resulted in lower probabilities for having a full drink and heavy drinking episode. Stated differently, in circumstances in which adolescents are already at higher risk for drinking (i.e., their parents are not regularly communicating about the hazards and consequences of drinking), risk for drinking is even higher when their parents are drinking at elevated levels. However, it may also be that when heavier drinking parents who communicate openly about the negative consequences they experience with drinking they mitigate the risk their own drinking exerts on their children. Collectively, these interactions suggest that parental drinking does contribute to the general parenting context, which extends previous work that suggests parenting style, including parental demandingness and responsiveness, provides a framework within which behaviors specific to alcohol use are enacted (Ennett et al, 2016).

Although no direct effects of parent reported drinking were identified, adolescents’ report of witnessing their parents drink was positively associated with having both a full drink and heavy drinking episode, findings that corroborate previous research that suggest witnessing parents drink is risk-inducing for adolescent alcohol use (van der Vorst et al., 2009; White et al., 2000). Furthermore, witnessing parents drink also modified the effect of perceived communication of negative messages about alcohol. The synergistic effects of parental communication of negative messages about alcohol and parental drinking suggest a need to jointly address these behaviors in family-based prevention programs. Collectively, these results substantiate the importance of alcohol-specific parenting practices in family-based preventive interventions. This line of inquiry can be expanded to other alcohol-specific parenting practices. For example, when parents exhibit permissive attitudes about youth alcohol use, their children are more likely to initiate and use alcohol regularly (Koning, 2010). Examining these alcoholspecific socialization practices in the context of parental drinking is a worthy venture for future research.

It is important to note that we found differing results for parental communication based on the informant: results were only significant for adolescents’ perception of cautionary communication messages from their parents. Studies have often assessed parenting behaviors from the perspective of both the parent and adolescent, which represent two distinct constructs. Parental reports reflect the behavior parents believe they have exhibited, whereas adolescent reports reflect the awareness and perception of the behavior by the child. Importantly, studies have demonstrated only weak to moderate positive associations between parent and child report of well-studied socialization behaviors such as parental monitoring and knowledge (Abar, Jackson, Colby, & Barnett, 2015; De Los Reyes, Goodman, Kliewer, & Reid-Quinones, 2010; Lippold, Greenberg, & Feinberg, 2011; Pettit, Laird, Dodge, Bates, & Criss, 2001; Reynolds, MacPherson, Matusiewicz, Schreiber, & Lejuez, 2011), indicating important discrepancies between reporters. Largely, adolescent report of socialization behaviors demonstrates stronger predictive utility for delinquent youth outcomes, including alcohol use (Abar et al., 2015; Cottrell et al., 2003; Kerr & Stattin, 2000; Latendresse et al., 2009; Reynolds et al., 2011). Our results substantiate this finding, replicating a weak correlation between informant ratings on parenting practices [T1 communication: r(580) = .17, p < .001; T2 communication: r(457) = .15, p < .001; T1 punishment: r(581) = .20, p < .001; T2 punishment: r(457) = .16, p < .001]. This indicates a need to assess adolescents’ perception of parental socialization behaviors in addition to parents’ report of their own behavior.

Although the present study has several strengths, such as its longitudinal design and multiple-informant data, the study should be viewed in the context of several limitations. First, our sample is drawn from a subset of secondary schools in a single geographic region and thus is not representative of the entire United States. Second, our measure of parental communication was limited to a single item to capture cautionary message content. Third, parent-reported alcohol only accounted for the responding parent, which may not reflect the overall context of drinking in the home. Fourth, the items regarding communication and punishment reflect perceptions of how a parent would react to their child’s drinking – that is, anticipated response as opposed to actual response. Future research would benefit from empirically testing anticipated versus actual reports of parental behavior. Finally, while the outcomes of this study reflect two important milestones along the adolescent drinking continuum (having a full drink and heavy drinking episode), the directionality of the effects cannot be completely ascertained as the drinking outcome may have occurred prior to report of the parental socialization behavior, and studies have shown that parents alter their behaviors in response to youths’ drinking (Glatz, Stattin, & Kerr, 2012). Furthermore, data from this study reflect a community sample of adolescents that drink at relatively low rates. Future studies should replicate these findings using a sample of more alcohol-involved youth.

Despite the limitations, the current study advances our understanding of parental socialization of adolescent alcohol use. To further inform prevention efforts, future studies need to investigate if adding a brief module on alcohol-specific parenting practices to parent training interventions for substance use (for a review, see Kuntsche & Kuntsche, 2016). For example, in their family-based program, Jackson and colleagues (2016) recommend a module focused on engaging parents in rule setting against youth alcohol use, and monitoring opportunities for the child to deviate from such rules. If effective, the brief and relatively simple nature of coaching parents to communicate cautionary messages and rules about alcohol can lead to interventions that can be delivered to families in multiple settings such as health care clinics (Cunningham et al., 2015). Furthermore, this study emphasizes the need to encourage parents to limit their own drinking, especially when their children are present.