Hypothetical Evaluations of Positive and Negative Alcohol Consequences in Adolescents across Various Levels of Drinking Experience

Abstract

Research supports the importance of the subjective evaluation of alcohol-related consequences, and theory suggests that these evaluations may depend on one’s prior experience. The goal of the present study was to understand how adolescents subjectively evaluate the potential negative and positive consequences of drinking and to test the hypothesis that evaluations differ as a function of personal experience with alcohol use and consequences. Participants were 697 adolescents (55% female) who completed online surveys assessing lifetime drinking experience and hypothetical evaluations of 13 negative and 9 positive consequences. Never having consumed a full drink of alcohol (vs having consumed a full drink, but not having negative consequences) was significantly associated with higher mean negative evaluations and lower mean positive evaluations. Those who had a full drink (vs. those who had not) rated close to half of the negative consequence items as significantly less bothersome, and all of the positive consequences as significantly more enjoyable. However, there was little evidence in this sample that evaluations differ between drinkers with and without experience with negative consequences. Overall, findings suggest that youth who have experience with simply consuming alcohol may place more value on the positive and less value on some of the negative consequences of drinking, which has the potential to impact decisions to continue to drink. Longitudinal research uncovering the direction of evaluation-experience effects and mechanisms other than consequence experience, are essential next steps.

Introduction

Alcohol misuse is a significant public health concern across a broad range of populations, and the majority of adult drinkers first experiment with alcohol during adolescence (Donovan, 2004; Jackson & Sartor, in press). Negative consequences associated with alcohol misuse among adolescents span a range of severity, and include but are not limited to neglecting academic obligations, engaging in risky behavior, having blackouts, and developing physiological dependence (Chassin, Colder, Hussong, & Sher, 2016; Miller, Naimi, Brewer, & Jones, 2007). Due to the public health concern presented by alcohol misuse, there is a focus in the research literature on understanding the phenomenology and early development of drinking. Such work is imperative to the refinement of existing targeted preventive interventions for those who have not yet progressed to alcohol use disorders and treatment protocols for those who have.

Several of the theories that drive research examining correlates of alcohol misuse focus on cognitive factors such as motives, attributions, and expectancies. Included among these is social cognitive theory (Bandura, 1986), which suggests that drinking behavior is learned both vicariously and through personal experience, and which places importance on cognitive interpretations of such learning experiences. As such, social cognitive theory supports the importance of the subjective evaluation of alcohol-related consequences and suggests that these evaluations may depend on one’s prior experience.

In general, subjective evaluations refer to the value one places on the consequences that may result from drinking, or how positive or negative those effects or consequences are deemed to be by an individual. These have been measured in different ways, with one key distinction being whether participants are asked to evaluate consequences they have actually experienced (actual evaluations; e.g., “How negative was your experience?”) or to evaluate consequences that may occur as a result of drinking (hypothetical evaluations; e.g., “How negative would this experience be?”). While the former may be of greater interest for samples of drinkers, the latter may be more relevant for youth who have yet to experience drinking or its consequences.

Though related, hypothetical evaluations differ from expectancies, which reflect the perceived likelihood of certain outcomes. Many researchers have suggested the unique importance of evaluations (e.g., Fromme, Stroot, & Kaplan, 1993; Neighbors, Walker, & Larimer, 2003), acknowledging that although expectancies for certain effects of alcohol are meaningful, expectancies may only result in drinking if the expected effects are deemed desirable (or may only result in avoiding drinking if those expected effects are deemed undesirable). Zamboanga, Schwartz, Ham, Hernandez Jarvis, and Olthuis (2009) demonstrated that while positive expectancy subscales of increased sociability and tension reduction and the negative expectancy subscale of risk and aggression were highly correlated with one another, and positive and negative hypothetical evaluation subscales were highly correlated with one other, correlations between expectancy and evaluation subscales were lower. Thus, expectancy outcomes and evaluations tap related but conceptually distinct constructs.

Importantly, some of the alcohol-related effects assumed to be negative by researchers and clinicians are not necessarily rated as such by all individuals (e.g., blackouts, hangovers; Mallett, Bachrach, & Turrisi, 2008). In one study, Canadian adolescents (7^th–9^th grade) were asked to list the three or four most important things that would happen if they used alcohol, and to indicate for each outcome whether they would like it or not (hypothetical evaluations; Fulton, Krank, & Stewart, 2012). Similar to studies conducted among college students, some “negative” outcomes were deemed likeable by a subset of participants (e.g., impact on cognitive performance, 17%; intoxication, 40%). However, while studies of college students have examined subjective evaluations at the level of specific negative and positive consequence items (Barnett, Merrill, Kahler, & Colby, 2015; Merrill, Read, & Barnett, 2013), little work has been conducted to provide finer-grained data on the types of consequences that are viewed positively or negatively by adolescents.

There is variability among individuals in subjective evaluations and this variability has been shown to be related to drinking behavior. Cross-sectional studies reveal that hypothetical evaluations of alcohol outcomes contribute more significantly to alcohol use behavior than do expectancies among adolescents (Fromme & D'Amico, 2000). Moreover, these studies demonstrate partial support for mediational pathways from hypothetical positive evaluations (i.e., the importance of the positive effects of drinking) to drinking behavior and ultimately to alcohol problems among adolescents (D'Amico et al., 2002). In another cross-sectional study (Van Tyne, Zamboanga, Ham, Olthuis, & Pole, 2012), more positive hypothetical evaluations of consequences items were associated with higher levels of hazardous alcohol use in adolescents age 14–18, controlling for drinking motives, expectancy outcomes, and drinking game participation. In longitudinal research, Fulton et al. (2012) found that higher outcome liking scores (i.e., hypothetical evaluations) from the first year of the study were associated with both higher initial levels and an increased rate of change in drinking over three years. In sum, across different adolescent age groups, in both cross sectional and longitudinal work, hypothetical evaluations of consequences are related to drinking behavior, both within and across time.

Though, as noted, some work (e.g., Fulton et al., 2012) has examined whether evaluations are associated with later increases in drinking, theory suggests that the opposite direction of effect may also be relevant. That is, prior experience with drinking and/or consequences of drinking may be predictive of hypothetical evaluations of both negative and positive consequences among young adolescents. Social cognitive theory highlights the role of reinforcement from the environment in the prediction of both behavior and cognition. One possibility is that youth who have actually experienced the consequences of drinking may find the positive consequences to be particularly reinforcing and the negative consequences to be less bothersome than previously expected; this might lead to consequence evaluations that are different when reflecting upon actual experience as compared to considering consequences in the abstract. Similarly, self-perception theory (Bem, 1972) would suggest that those with experience simply consuming alcohol may infer positive attitudes toward drinking from their behavior. Thus, even in the absence of consequences, drinkers may rate the negative consequences as less bothersome and the positive consequences as more enjoyable. In young adult samples findings are mixed with respect to whether prior experience with alcohol use or consequences is associated with rating negative consequences as more (Merrill, Read, & Colder, 2013; White & Ray, 2013) or less bothersome or important (Barnett, Goldstein, Murphy, Colby, & Monti, 2006; Patrick & Maggs, 2011). This may be due to inconsistencies in the way in which evaluations are measured.

Importantly, aside from being conducted exclusively among young adults, research examining drinking experience as a predictor (rather than outcome) of evaluations (1) has relied only on measures of actual evaluations (i.e., ratings of consequences that were actually experienced by participants), and (2) has not considered evaluations of the positive consequences of drinking. A sample of youth characterized by a range of drinking risk level (non-drinkers, drinkers who have not had negative consequences, and drinkers who have experienced negative alcohol consequences) can help answer questions about whether experiences with alcohol use and/or consequences early in development are associated with evaluations of consequences. Understanding how youth evaluate the potential positive and negative effects of alcohol can aid in efforts to intervene upon cognitive targets early on, in order to minimize the development of alcohol problems. If we find that baseline evaluations differ depending on past experience with drinking, interventions that target attitudes toward or evaluations of negative and positive consequences of drinking among youth could use enhanced tailored approaches.

The Present Study

The present study investigates how adolescents with a range of drinking severity hypothetically evaluate the potential positive and negative consequences of drinking. Our first aim was to provide descriptive information on the extent to which different negative consequences are viewed as bothersome and the extent to which different positive consequences are viewed as enjoyable. Our second aim was to test whether the subjective evaluation of consequences is invariant between three drinking severity groups: (1) Never having had a full drink (ND) during the three years prior to the assessment of consequence evaluations, (2) having had a full drink with negative consequence(s) (FDC), and (3) having had a full drink with no negative consequence(s) (FDNC). We examined individual negative and positive consequences and aggregate positive and negative scales. Further, we controlled for the association between expectancies and evaluations when examining the impact of prior drinking. We hypothesized that negative consequences would be viewed as less bothersome, and positive consequences as more enjoyable, among teens in the FDC and/or FDNC groups compared to the ND group.

Method

Procedure

All procedures were approved by the university’s Institutional Review Board. Data come from a larger prospective study (N = 1023) investigating the development of adolescent drinking milestones (Jackson, Colby, Barnett, & Abar, 2015; Jackson et al., 2014). Participants were recruited through local middle schools, with information distributed in homeroom classes and also mailed to each student’s home address. Participants for whom we had parental consent attended an after-school orientation session where they were compensated $25 for completing a 45-minute computerized baseline survey. Over the next three years, six waves of web-survey data were collected on a semi-annual basis with the exception of Wave 6, which followed Wave 5 by one year. Surveys could be completed from any location and participants were compensated with $20 per survey.

The study was recently renewed to include additional waves of data collection. Eighty-two percent (n=839) of the original sample agreed to participate in the continuation of the study. Compared to those who declined, participants who agreed to participate were more likely to be younger (12.2 vs. 12.4, p<.01), non-Hispanic White (p<.01), and female (p<.01), but no difference was observed on lifetime consumption of a full drink of alcohol (p>.05). The subjective evaluation data for the current study are drawn from the seventh wave of data collection, when evaluations of positive and negative consequences were first assessed. Measures of drinking experience represent whether drinking behavior was reported at any time up to Wave 7.

Participants

Participants (n = 738, 88% response rate) completed measures of evaluations of positive and negative consequences; however 41 (5.56%) participants had missing data so were dropped through listwise deletion from the main analyses. This resulted in a final analytic sample of 697 youth (55% female, M age at Wave 7 = 15.56, SD = 1.01). Ten percent of the sample endorsed Hispanic ethnicity. Race was self-identified by 79.5% of participants as White, 6.2% Multiracial, 4.3% Black, 2.6% Asian, 1.5% American Indian/Alaskan, and 0.4% Hawaiian/Pacific Islander. Race was unknown for 5.6% of participants. At Wave 7, 17% were in the 12^th grade, 30% in the 11^th grade, 34% in the 10^th grade, and 19% in the 9^th grade. The sample was largely representative of the schools from which it was drawn, particularly with respect to gender and grade distribution, but with some evidence of greater racial diversity and less economic disadvantage in the current sample (Jackson et al., 2015).

Measures

Demographics

Participants reported age and gender at baseline.

Lifetime Drinking Experience

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 consumed a full drink of alcohol, not including alcoholic drinks consumed as part of a religious ceremony. They also reported whether they had experienced each of a list of 18 alcohol-related problems (White & Labouvie, 2000) that were a subset drawn from the full Rutgers Alcohol Problems Index (RAPI) (White & Labouvie, 1989)¹. From these measures, we categorized participants into the three groups (ND, FDC, FDNC) based on reports of drinking behavior at all points up to Wave 7.

Negative and Positive Evaluations

Evaluations were assessed for 13 negative consequence items (e.g., passing out, getting physically sick) adapted from the RAPI (White & Labouvie, 1989) and nine positive consequence items (e.g., having a good time, feeling more energetic) used in prior research among college students (Barnett et al., 2014). For each negative item, participants were asked “If you were to experience the following as a result of drinking alcohol, how bothersome would it be?” For each positive item, participants were asked “If you were to experience the following as a result of drinking alcohol, how enjoyable would it be?” Items were rated on a scale from 0 (not at all) to 3 (extremely). Items were averaged to create negative evaluation and positive evaluation scale scores; Cronbach’s alphas for both were .97. See Footnote ² for additional validity tests conducted in this sample.

Negative and Positive Expectancies

A set of items assessing nine positive expectancies (e.g., forget their problems, feel more friendly) and 13 negative expectancies (e.g., get into fights, act stupid) about the effects of alcohol were used (Schell, Martino, Ellickson, Collins, & McCaffrey, 2005) (one positive item, “look cool”, was dropped due to a low factor loading). The expectancy items cover affective, cognitive, and behavioral effects of alcohol use but did not map on directly to those assessed in the evaluation measures. Items were preceded by the following question: “How likely is it that the following things would happen to people your age if they had one or more drinks of alcohol?” Response options ranged from “very unlikely” (1) to “very likely” (4). Cronbach’s alpha for positive expectancies was .96, and .94 for negative expectancies.

Analytic Plan

To address our primary aims, we ran two separate regression models to examine the influence of experience with alcohol on the total scale scores for (1) negative evaluations of negative consequences, and (2) positive evaluations of positive consequences. Both evaluation scales were normally distributed, and we did not identify any outliers at either the scale or item level. We created two dummy codes, one representing the contrast between the ND and FDNC groups, and one representing the contrast between the FDC and FDNC groups. Next, two separate MANCOVA models were used to examine the influence of experience with alcohol (using a three-group variable): one on the 13 negative consequence items, controlling for negative expectancies; and one on the nine positive consequence items, controlling for positive expectancies. Given omnibus effects in the MANCOVA models, univariate associations between drinking experience and each individual evaluation item score were examined, and Cohen’s d effect sizes were calculated.

Results

Descriptive Information

Sixty-five percent of our sample (n = 452) reported never having had a full drink of alcohol (ND group), 27% (n = 186) reported having had a full drink but no negative consequences (FDNC group), and the remaining 8% (n = 59) reported both a full drink and at least one lifetime negative alcohol consequence (FDC group). Among those who had consumed a full drink (FDC, FDNC), the mean number of lifetime negative consequences was 0.71 (SD = 1.81). The mean negative evaluation score was 2.26 (SD = 0.84) and the mean positive evaluation score was 1.43 (SD = 0.98). While negative and positive expectancies were highly positively correlated with one another (r = .46, p < .01), negative and positive evaluations were uncorrelated (r = .00, p = .99). Negative expectancies were positively correlated with negative evaluations (r = .46, p < .01) and positive expectancies were positively correlated with positive evaluations (r = .45, p < .01).

Substantive Analyses

Predicting scale scores

Multiple regression models predicting negative and positive evaluation scale scores from drinking experience group, controlling for gender, age, and negative or positive expectancies, respectively, are shown in Table 1. In the model predicting negative evaluation scale scores (F (5, 691) = 40.34, p < .001; Adjusted R² = .22), ND vs FDNC was significantly associated with higher negative evaluation scale scores. In the multiple regression model predicting positive evaluation scale scores (F (5, 691) = 53.11, p < .001; Adjusted R² = .27), ND vs FDNC was significantly associated with lower positive evaluation scale scores. The contrast between FDC and FDNC was not significant in either model.

Table 1.

Multiple Regression Model Predicting Mean Negative and Positive Evaluation Scale Scores

Predicting Mean Negative Evaluation Scores
Predictor	B	SE	Beta	p
ND vs FDNC	.16	.07	.09	.014
FDC vs FDNC	−.06	.11	−.02	.627
Negative expectancies	.51	.04	.45	.000
Age	.05	.03	.05	.124
Gender	−.08	.06	−.05	.162
Predicting Mean Positive Evaluation Scores
Predictor	B	SE	Beta	p
ND vs FDNC	−.51	.08	−.25	.000
FDC vs FDNC	.23	.13	.07	.065
Positive expectancies	.51	.04	.40	.000
Age	−.02	.04	−.01	.685
Gender	.03	.07	.01	.707

Note: ND = never full drink, FDC = full drink with at least one consequence, FDNC = full drink with no consequences

Predicting individual item scores

The MANCOVA model predicting the set of negative evaluation items (controlling negative expectancies), as well as the MANCOVA model predicting the set of positive evaluation items (controlling positive expectancies), revealed a significant omnibus effect of experience with alcohol (Table 2). Mean evaluation scores for each item across the full sample are presented in Tables 3 and 4, ordered from highest (most bothersome, enjoyable) to lowest. As seen in Table 3, at the univariate level, drinking experience group was associated with perceiving several of the negative consequences as less bothersome. Pairwise comparisons revealed that the biggest and most frequent differences involved the ND group evaluating consequences more negatively than the FDNC group (range Cohen’s d = .03 to .30; Table 3). Additionally, those in the FDC group rated a change in personality as significantly less bothersome than both the FDNC (d=.03) and FDC group (d=.07).

Table 2.

MANCOVA Models Predicting Full Sets of Negative Evaluation Items and Positive Evaluation Items by Group

Predicting Set of Negative Evaluation Items
Predictor	Wilks' Lambda	F (dfa)	p
Group	.90	2.83	.000
Negative expectancies	.79	13.88	.000
Age	.97	1.57	.084
Gender	.96	1.93	.024
Predicting Set of Positive Evaluation Items
Predictor	Wilks' Lambda	F (dfb)	p
Group	.90	4.24	.000
Positive expectancies	.82	16.55	.000
Age	.99	.81	.612
Gender	.97	2.63	.005

Note:

^adf = 13, 679 for tests of expectancies, age and gender; df = 13, 1358 for group;

^bdf = 9, 683 for tests of expectancies, age and gender; df = 18, 1366 for group;

Group was coded 0 = never full drink (ND), 1 = full drink but never any negative consequence (FDNC), 2 = full drink and at least one negative con (FDC) (lifetime)

Table 3.

Item-level Negative Consequence Evaluation Scores for Full Sample and by Drinking Experience Status

	Full Sample M (SD)	ND (n = 452) EMM (SE)	FDNC (n = 186) EMM (SE)	FDC (n = 59) EMM (SE)	F	p	Cohen’s d for ND vs FDNC	Cohen’s d for ND vs FDC	Cohen’s d for FDNC vs FDC
Getting in a car with a teenage driver who had been drinking	2.50 (0.96)	2.49 (0.04)	2.51 (0.07)	2.57 (0.12)	.18	.839	−0.02	−0.09	−0.06
Driving after drinking alcohol	2.49 (0.97)	2.49 (0.04)	2.48 (0.07)	2.53 (0.12)	.08	.920	0.01	−0.05	−0.05
Getting in trouble for drinking alcohol (for example: with your parents, with the school)	2.40 (0.98)	2.43 (0.04)	2.35 (0.07)	2.38 (0.12)	.51	.602	0.09	0.06	−0.03
Getting physically sick (for example: vomit, stomach cramps)	2.37 (0.97)	2.43 (0.43)	2.29 (0.07)	2.22 (0.12)	2.21	.111	0.02	0.02	0.07
Having a fight, argument, or bad feelings with a friend	2.33 (0.96)	2.37 (0.41)	2.23 (0.06)	2.30 (0.11)	1.69	.185	0.02	0.01	−0.09
Causing shame or embarrassment to yourself or someone else	2.32 (0.98)	2.39 (0.04)	2.20 (0.07)	2.17 (0.12)	3.61	.028	0.22	0.26	0.03
Saying something you wish you hadn't	2.28 (0.98)	2.35 (0.04)a	2.13 (0.07)b	2.19 (0.12)	4.06	.018	0.25	0.19	−0.06
Passing out from drinking	2.27 (1.02)	2.31 (0.05)	2.22 (0.07)	2.15 (0.12)	.99	.372	0.09	0.15	0.07
Missing classes, being late for classes, or being unable to do your homework or study for a test	2.25 (1.02)	2.31 (0.05)	2.14 (0.07)	2.10 (0.12)	2.88	.057	0.16	0.20	0.04
Being unable remember some part of the day or night (saying or doing things you did not remember afterwards)	2.14 (1.03)	2.25 (0.04)a	2.00 (0.07)b	1.77 (0.12)b	9.70	.000	0.28	0.56	0.24
Neglecting your responsibilities	2.09 (1.04)	2.19 (0.46)a	1.94 (0.72)b	1.83 (0.13)b	6.40	.002	0.03	0.04	0.01
Having a headache after drinking	2.05 (1.03)	2.17 (0.05)a	1.86 (0.07)b	1.82 (0.13)b	8.43	.000	0.30	0.33	0.04
Noticing a change in your personality	1.93 (1.06)	2.07 (0.47)a	1.77 (0.73)b	1.39 (0.13)c	14.85	.000	0.03	0.07 0.07	0.04

Note: Items are listed in order of mean evaluation score (most to least negatively evaluated) in the full sample; p values in bold are less than .05;

ND = never full drink, FDC = full drink with at least one consequence, FDNC = full drink with no consequences; EMM = estimated marginal mean, evaluated at the following values: negative expectancies = 3.15, Age at time of baseline survey, in half-years = 12.40, Gender = .44; SE = standard error; Items rated on a scale from 0 (not at all bothersome) to 3 (extremely bothersome);

Means with differing subscripts within rows indicate significant pairwise comparisons, with a Bonferroni adjustment for multiple comparisons.

d = 0.2 is considered a 'small' effect size, 0.5 represents a 'medium' effect size and 0.8 a 'large' effect size.

Table 4.

Item-level Positive Consequence Evaluation Scores for Full Sample and by Drinking Experience Status

	Full Sample M (SD)	Group 1: No full drink experience (n = 452) EMM (SE)	Group 2: Full drink experience without any cons (n = 186) EMM (SE)	Group 3: Full drink experience with cons (n = 59) EMM (SE)	F	p	Cohen’s d for Group 1 vs 2	Cohen’s d for Group 1 vs 3	Cohen’s d for Group 2 vs 3
Having a good time	1.55 (1.09)	1.33 (.05)a	1.91(.07)b	2.13 (0.13)b	32.20	.000	−0.56	−0.76	−0.23
Feeling less stressed or more relaxed	1.53 (1.09)	1.32 (.05)a	1.88 (.07)b	2.13 (0.13)b	31.81	.000	−0.54	−0.77	−0.26
Talking to someone you were attracted to	1.53 (1.09)	1.33 (.05)a	1.85(.07)b	2.05(0.13)b	25.97	.000	−0.50	−0.68	−0.21
Finding it easier to socialize	1.45 (1.06)	1.28(.05)a	1.72(.07)b	1.97(0.13)b	22.10	.000	−0.43	−0.65	−0.26
Being able to take your mind off problems	1.43 (1.10)	1.23 (.05)a	1.75(.07)b	1.99 (0.13)b	28.09	.000	−0.50	−0.72	−0.25
Feeling more self- confident and sure of yourself	1.42 (1.09)	1.23(.05)a	1.73(.07)b	1.94 (0.13)b	24.89	.000	−0.48	−0.67	−0.22
Feeling more energetic	1.38 (1.06)	1.18(.04)a	1.69(.07)b	1.92 (0.12)b	28.07	.000	−0.58	−0.86	−0.24
Expressing your thoughts or feelings to someone more easily	1.29 (1.06)	1.11(.05)a	1.56 (.07)b	1.86 (0.13)b	23.96	.000	−0.44	−0.71	−0.31
Feeling like you were part of the group	1.29 (1.06)	1.11 (.05)a	1.58 (.07)b	1.78(0.13)b	21.72	.000	−0.46	−0.64	−0.21

Note: Items are listed in order of mean evaluation score (most to least positively evaluated) in the full sample; p values in bold are less than .05;

EMM=estimated marginal mean, evaluated at the following values: positive expectancies = 2.80, Age at time of baseline survey, in half-years = 12.40, Gender = .44. Items rated on a scale from 0 (not at all enjoyable) to 3 (extremely enjoyable); SE = standard error.

Means with differing subscripts within rows indicate significant pairwise comparisons, with a Bonferroni adjustment for multiple comparisons.

d = 0.2 is considered a 'small' effect size, 0.5 represents a 'medium' effect size and 0.8 a 'large' effect size.

As seen in Table 4, at the univariate level, compared to the ND group, experience with a full drink (either FDC or FDNC) was associated with rating each and every one of the positive consequences as more enjoyable (range Cohen’s d = −.21 to −.77, Table 4). There were no differences between the FDC and FDNC groups. Because we observed a significant effect of gender in the MANCOVA models, Supplementary Tables 1 and 2 display item-level evaluation scores separately by gender.

Discussion

The present study provides descriptive information on the hypothetical subjective evaluations of a range of distinct positive and negative alcohol-related consequences in adolescents, as well as how these evaluations differ dependent on prior drinking history. In support of our hypotheses and consistent with social cognitive theory, we found that adolescents who had prior experience with alcohol generally rated negative consequences as less bothersome, with some specificity with respect to consequence items, and all positive consequences as more enjoyable. However, contrary to hypotheses, among those who had consumed a full drink, evaluations did not tend to differ between those who had experienced at least one negative consequence and those who had not.

Our descriptive examination of how adolescents evaluate hypothetical drinking consequences revealed that the three negative consequences rated as most bothersome on average were getting in a car with a teenage driver who had been drinking, driving after drinking alcohol, and getting in trouble for drinking alcohol. Research among college student drinkers also shows that driving after drinking is perceived to be one of the most negative consequences (Barnett et al., 2015; Merrill, Read, & Barnett, 2013), as compared to less negatively evaluated consequences such as blacking out or getting into physical fights. The least bothersome negative consequences were noticing a change in personality, having a headache after drinking, and neglecting responsibilities. However, evaluation of these items was highly dependent on experience with drinking, such that those with experience consuming a full drink (FDC and FDNC groups) viewed these negative consequences as less bothersome than non-drinkers. Also of note is the largest effect of experience was observed for alcohol-induced memory loss, where youth in the FDC group evaluated the item as significantly less bothersome than those in the ND group. That several negative consequences in this adolescent sample were viewed as less negative among those with drinking experience is consistent with research conducted with young adult samples that found drinking experience was linked to rating negative consequences as less bothersome or important (Barnett et al., 2006; Patrick & Maggs, 2011).

Regarding positive evaluations of positive consequences, those rated as most enjoyable included having a good time, talking to someone you were attracted to, and feeling less stressed/more relaxed. Interestingly, this mirrors the rank ordering observed among college students (Barnett et al., 2015), despite that here hypothetical evaluations were assessed, rather than actual evaluations of consequences one had experienced. Across the board, positive consequences of drinking were rated as more enjoyable by those with experience consuming a full drink as compared to those without, and this was true whether or not negative consequences were also reported. This may be due to the fact that positive effects of drinking are likely even at low levels of consumption; youth who have consumed alcohol have likely experienced these positive effects and therefore rate them more favorably. Of note, effect sizes for differences in evaluations between those in the non-drinking group and the other two groups were relatively larger for the positive consequences of drinking than for the negative consequences of drinking. This suggests either that experience with drinking may alter perceptions about positive outcomes more so than negative, or that positive evaluations of positive consequences may influence later decisions to drink to a greater extent than negative evaluations of negative consequences.

Those who have begun to drink have had more opportunities to directly experience both the negative and positive consequences of drinking. We suspected that one reason they may view the negative consequences as less bothersome than those without experience consuming a full drink is that they have experienced them and concluded that the consequences were not as bad as initially expected. Surprisingly though, with the exception of the consequence “noticed a change in your personality”, group differences in evaluations were evident regardless of whether or not experience with negative consequences was endorsed. It appeared to be consumption of a full drink, rather than experience of negative consequences, that mattered most with respect to hypothetical consequence evaluations.

Thus, it is important to consider mechanisms that may explain why experience with alcohol use (but not experience with consequences) is associated with evaluations. As noted above, one possibility is that having experienced positive consequences of drinking, which are even more likely to occur than negative consequences (Park, 2004), may lead youth to view the negatives as less bothersome in the broader context. Unfortunately, we did not measure lifetime positive consequences in this study in order to examine this possibility. Second, the mere exposure effect (Zajonc, 2004) may be at play, whereby even minimal experience with alcohol leads to increased preference for alcohol and hence more positive evaluations. Third, less negative evaluations among drinkers might also be explained by cognitive dissonance (Festinger, 1957): drinkers may report that the negatives are not so bothersome because they have engaged in drinking behavior that has the potential to lead to those negative consequences. Similarly, self-perception theory (Bem, 1972) supports that one’s attitudes may be derived from considering one’s prior behavior – knowing that one has consumed alcohol, one may infer a positive (or less negative) attitude to drinking-related consequences.

Additional mechanistic possibilities are consistent with social cognitive theory. First, a teen that has not directly experienced negative consequences of drinking may affiliate with peers who have. He or she may observe such peers downplaying the negatives or talking up the positives, or may assume those peers do not find the negatives to be bothersome, which in turn impacts his/her own attitudes. Similarly, social modeling may occur via parents. Adolescents who drink and evaluate consequences differently from non-drinkers may be those who have parents who drink at home. This is in line with research documenting different alcohol-related cognitions between children with and without alcoholic parents (Zucker, Kincaid, Fitzgerald, & Bingham, 1995). Alternatively, those youth in our sample who have early experience with alcohol may also be those youth who have a genetic propensity to display lower subjective responses to alcohol (e.g., Schuckit et al., 2009; Schuckit et al., 2008). In other words, youth who begin drinking earlier may be those same youth who will experience more positive and/or fewer negative effects of alcohol, and rate both more favorably.

Limitations, Strengths and Future Directions

One limitation to this study is that although we assessed drinking behavior across several waves of data, evaluations were assessed only at the final time point, and we only examined a single direction of effect – whether prior drinking impacts current subjective evaluations. As other research has shown, the opposite direction of effect is also of relevance - evaluating negative outcomes of drinking less negatively may be associated with subsequent increases in drinking (Fulton et al., 2012). Thus, repeated measures of evaluations and analytic methods such as cross-lagged panel models are necessary to examine this reciprocal process. Second, we categorized certain outcomes as “negative” and “positive” a priori, asking students to rate how bothersome and enjoyable, respectively, they thought those consequences would be if experienced. However, prior research highlights that some negative consequences are perceived positively by drinkers (Fulton et al., 2012; Mallett et al., 2008). Future research should assess evaluations in a more unbiased manner. For example, evaluations (at least of negative outcomes) could be assessed along the full spectrum of bothersome-enjoyable. Alternatively, it may be beneficial to ask participants to rate whether they believe each consequence is positive or negative, before responding with respect to degree along the positive or negative spectrum.

Third, we chose to define drinking experience as consumption of a full drink, with or without at least one negative consequence. However, different results may emerge given definitions that require lower thresholds such as consumption of a sip of alcohol, or higher thresholds such as prior experience of multiple consequences. A more comprehensive list of potential consequences such as the full 23-item RAPI, may have yielded a larger group of participants who had experienced consequences and perhaps greater power to detect group differences, although it is unlikely that this is the case given our light drinking sample. Additionally, given that consequences are more likely to occur in the context of heavy episodic drinking (HED), it may also be useful to examine evaluations of consequences based on experience with HED in a heavier drinking sample, bearing in mind that the operationalization of HED will vary with respect to both gender and age of the adolescents under study (Donovan, 2009).

Moreover, evaluations of positive consequences may differ based on experience with positive consequences, which were not measured in this study. We also did not collect data on whether participants had experienced each of the unique negative consequences they were asked to evaluate in this study, and generally, endorsement of consequences was low in our sample (Jackson et al., in press). It is possible that evaluations of specific negative and positive outcomes are linked to prior experience with those same outcomes. Finally, while a strength of our study was statistical controls for alcohol expectancies when examining the link between drinking experience and consequence evaluations, expectancies and evaluations were not assessed for the same alcohol-related outcomes.

Despite these limitations, this study had several notable strengths. This is one of the first studies to present descriptive data on how adolescents evaluate a range of unique positive and negative consequences of drinking. Further, we had a large sample size of youth represented by both experience and lack of experience with alcohol use and consequences. Of note, 35% of our overall sample, and 59% of those in 12^th grade endorsed lifetime consumption of a full drink of alcohol, consistent with rates of drinking observed in other studies (Colder et al., in press; Johnston, O’Malley, Miech, Bachman, & Schulenberg, 2016). Finally, though consequence evaluations were measured at a single time point, we were able to make use of longitudinal data by examining whether alcohol use was reported over several earlier repeated measures as a predictor of these evaluations.

Conclusion and Implications

In sum, there is much variability in evaluations of consequences, across both individuals and consequence types. Descriptive information on which negative and positive consequences are perceived as most bothersome and enjoyable in this age group can help determine optimal intervention content. It is important to acknowledge that evaluations may differ depending on past experience with drinking, and a “one size fits all” approach to intervention may not be optimal. Whereas some negative consequences are viewed quite negatively by the majority of participants, others with more variability leave more room for modification.

Generally, heterogeneity in the causes of who experiments with drinking (e.g., Reyna & Farley, 2006) may be an obstacle in the development of personalized interventions. Identifying youth characterized by risky patterns of perceived consequences of drinking may provide a framework to identify those at risk for the emergence of heavier and problematic forms of drinking, and to test targeted intervention efforts on those youth. Such types of targeted interventions have been previously developed to prevent drinking amongst youth who display elevated levels of aggression (Lochman & Wells, 2004). Youth may be intervened upon by shaping alternative, non-drug sources of positive reinforcement, and/or motivational enhancement interventions aimed at moving the youth towards goals that may be incompatible with drinking. Our findings suggest that another potential avenue for underage drinking prevention programs is to more directly target the evaluations of negative and positive consequences of drinking. Though interventions have been developed to target the cognitive variable of normative perceptions of drinking behavior (D'Amico & Edelen, 2007; Ellickson, McCaffrey, Ghosh-Dastidar, & Longshore, 2003), none have been developed to target subjective evaluations of consequences. Future research is needed to examine whether evaluations of alcohol consequences can in fact be modified, among both those who have and have not started drinking, and whether such modifications translate into positive outcomes on future drinking behavior.