Self-Cognitions, Risk Factors for Alcohol Problems, and Drinking in Preadolescent Urban Youth

Abstract

We examined relationships between self-structure and known precursors for alcohol problems in 9–12 year old primarily Black and Latino youth (N=79). Parental alcohol problems and being female predicted few positive and many negative self-cognitions and a future-oriented self-cognition related to alcohol (‘drinking possible self’). Nineteen percent of the sample reported ever drinking, but 40% of those with a ‘drinking possible self’ reported ever drinking. Compared never drinkers, youth who reported ever drinking had fewer self-cognitions. The self-structure may be an important mechanism through which parental alcohol problems and antisocial behavior lead to early alcohol use, and a viable target of interventions aimed at preventing early alcohol use.

Adolescent alcohol use is a significant public health problem, particularly among economically disadvantaged children and adolescents (Fothergill & Ensminger, 2006; Kost & Smyth, 2002; Poulton et al., 2002). Youth who start drinking before age 15 are four times more likely to develop alcohol dependence later in life than those who begin drinking at or after age 15 (Grant & Dawson, 1997). In spite of existing prevention efforts, however, underage drinking rates remain unacceptably high (National Institute on Alcohol Abuse and Alcoholism, 2008). This has led to renewed calls to explicate the determinants of early alcohol use and acceleration to alcohol problems that take adolescent development into consideration (U.S. Department of Health and Human Services, 2007).

Our previous research suggests that characteristics of the total collection of identities may be an important and modifiable determinant of early alcohol use and the acceleration to alcohol problems. While much of the existing self and identity literature is focused on the process of identity development, e.g., exploration and commitment as mechanisms of resolving identity crises (see Schwartz, 2001 for a review), our work is focused on the cognitive products of identity development, i.e., memory structures called self-schemas (Markus, 1977; Markus & Wurf, 1987)that influence information processing (Hoyle, 2006; Kendzierski, 1988; Markus, 1977; Stein & Corte, 2007) and behavioral regulation (Corte, 2007; Estabrooks & Courneya, 1997; Froming, Nasby, & McManus, 1998; Kendzierski & Whitaker, 1997; Stein, Roeser, & Markus, 1998; Stein & Corte, 2007). We found that young adults with early onset alcohol dependence were distinguished from social drinkers in that they had fewer positive and more negative self-schemas and a self-schema related to drinking (Corte & Stein, 2007). Because these young adults already had alcohol dependence, however, it is not clear whether this configuration of self-cognitions was a contributor to or a consequence of alcohol dependence. In a secondary analysis of data from adolescent participants in a high-risk family study for alcoholism, we found that in the subset of adolescents who reported drinking and getting drunk, having few positive and many negative self-schemas in early adolescence (12–14 years) distinguished earlier from later onset of drinking and getting drunk respectively (Corte & Stein, 2008). A self-schema related to drinking was not measured in this study.

The purpose of this study is to extend our previous findings to a younger and more ethnically diverse community-based sample. More specifically, we examined relationships among 1) the total collection of identities including the number of positive and negative self-schemas and a future-oriented conception of the self related to drinking, 2) known precursors for alcohol problems (antisocial behavior and parental alcohol problems), and 3) ever drinking in 9–12 year old primarily Black and Latino urban youth, with attention paid to potential gender differences in these relationships. Because identities are modifiable (Jacobs, Bleeker, & Constantino, 2003), they may be viable targets of interventions aimed at preventing early alcohol use and the acceleration to alcohol problems.

Schema Model of Self-Concept

From a social cognitive theory perspective, identities are cognitive products of development. An adolescent’s developmental experiences, which themselves are shaped by a variety of environmental, cultural, and biological factors, lead to the formation of cognitive memory structures in specific content domains that are referred to as self-schemas (Andersen & Chen, 2002; Campbell, Assanand, & DiPauls, 2000; Markus & Wurf, 1987; Sedikides & Skowronski, 1997) as well as future-oriented conceptions of the self called possible selves (Markus & Nurius, 1987; Oyserman & Fryberg, 2006).

Self-Schemas

Self-schemas are complex and chronically accessible cognitive structures about the self that are stored in long-term memory (Lieberman, Jarcho, & Satpute, 2004; Macrae, Moran, Heatherton, Banfield, & Kelley, 2004). They consist of abstract knowledge about the self in a particular domain (e.g., athletic), memories of specific behavioral episodes in the domain (e.g., getting first basket while playing on basketball team), and action-based behavioral rules and strategies that enable efficient behavior and the ability to manipulate complex information (e.g., completing a series of complex moves on the basketball court (Markus & Wurf, 1987). Because self-schemas include both non-conscious rules for drawing inferences about the self as well as behavioral strategies and routines, they enhance processing of self-relevant information and serve as strong internal regulators of affect and behavior. Affect and behavior, however, depend in part on the valence of the individual self-schemas. Positive self-schemas generate positive affect and motivate behavior in the domain (Cross & Markus, 1994; Estabrooks & Courneya, 1997; Froming et al., 1998; Kendzierski, 1988; Kendzierski & Whitaker, 1997) whereas negative self-schemas generate negative affect and are associated with behavioral inhibition (Andersen & Cyranowski, 1994; Andersen & Cyranowski, 1995; Lips, 2004; Markus, Hamill, & Sentis, 1987). Having few positive and many negative self-schemas is a general cognitive vulnerability that has been shown to be associated with high levels of negative affect (Corte & Stein, 2007; Stein & Corte, 2007).

Possible selves

Possible selves are future-oriented conceptions of the self that one expects to be, wishes to be, or fears being in the future (Markus & Ruvolo, 1989; Oyserman, Bybee, Terry, & Hart-Johnson, 2004; Oyserman & James, in press; Pizzolato, 2006). Because they include vivid action-oriented images not only of an outcome but also of strategies required to achieve that outcome, possible selves play a powerful role in behavior (King & Raspin, 2004; King & Smith, 2004; Oyserman & James, in press). Because a major developmental task of adolescence is to construct the self one wants to become (Markus & Ruvolo, 1989; Oyserman et al., 2004), the content and salience of possible selves influences their behavioral choices. Of particular relevance for the proposed study is that possible selves have been shown to predict a cluster of risky behaviors (Freeman, Hennessy, & Marzullo, 2001; Stein et al., 1998) in youth and binge drinking in college students (Quinlan, Jaccard, & Blanton, 2006). Moreover, in a study of youth across the transition from 8^th to 9^th grades, Stein found that an 8^th grade ‘popular’ possible self predicted a cluster of risky behaviors in the 9^th grade, and that engaging in risky behaviors in the 8^th grade predicted formation of a deviant self-schema in the 9^th grade (Stein et al., 1998). These findings suggest that domain-specific possible selves play a critical role in the early enactment of risky behaviors, and that the risky behaviors feed back to the developing self-concept.

Overview of Present Study

In this exploratory study, we examined relationships among self-cognitions (number of positive and negative self-schemas and availability in memory of a ‘drinking possible self’), antisocial behavior, parental alcohol problems, gender, and ever drinking in 79 low-income, predominantly Black or Latino, preadolescent school-recruited youth who had not yet begun regular drinking.

METHODS

Participants

Participants were 9–12 year old youth (N = 79) recruited from 2 low-income public schools and 3 low-income summer youth programs in Chicago. Their average age was 10.7 (SD = 1.1) years. Fifty-four percent (n = 43) of the youth were female. Most were Black (53%; n = 42) or Latino (39%; n = 31); 5% (n = 4) were Multiracial and <3% (n = 2) were White. Eligibility criteria included: age 9–12, English speaking, attendance at one of five recruitment sites, parental permission, and cognitively able to provide assent and complete study measures.

Brief descriptions of the study were made in school classrooms, study descriptions were mailed to parents of 9–12 year old youth, and flyers were posted at the summer youth program centers. While youth had to be English speakers to participate, our parental permission forms were in Spanish for those youth who had Spanish-only speaking parents. The PI (CC) or research assistant met briefly (individually) with youth for whom we received signed parental permission forms in order to describe the study and obtain child assent. Of 82 youth who received parental permission, 79 provided assent, 2 children declined to participate and one child was cognitively unable to provide assent.

Measures

Number of Positive and Negative Self-Schemas

Items endorsed as “Really True for Me” on the Harter Self-Perception Profile for Children (Harter, 1985) were used to identify self-schemas. The measure consists of 30 items that tap children’s perceptions of themselves in domains known to be important in youth—scholastic competence, social acceptance, athletic competence, physical appearance, and behavioral conduct (global self-worth items were not included in the computation of self-schemas). Each item is written as a statement describing two types of children—one with high competence (positive) and one with low competence (negative) in a specific content domain, e.g., “some kids have lots of friends,” BUT “other kids don’t have many friends.” For each item, children first decide which side of the statement best describes them (e.g., the positive or the negative pole) and then they decide if the statement is “really true for me” or “sort of true for me.” Because self-schemas reflect domains about which the person feels confident and certain, only items rated as “really true for me” were considered self-schemas. Valence of the self-schema (positive or negative) was determined based on whether the positive or negative pole of the item was endorsed. The total number of positive and negative self-schemas was computed separately by adding the total number of positive items and the total number of negative items endorsed as “really true for me” in domains the adolescent feels are important.

Although the measure does not account for perceived importance of the domains to the self, the Harter items reflect culturally salient and important domains for adolescents in Western cultures. Furthermore, a recent study of adolescents shows that taking into account the perceived importance of domains does not improve predictive power of specific domains to general self-worth (Shapka & Keating, 2005). As such, Roeser and colleagues (Roeser et al., 2008) argue that relying on importance ratings is too stringent a requirement for definition of identities in adolescent populations. We have used this method of determining the number of positive and negative self-schemas in a previous study of adolescents across two waves of data that were 3 years apart (Corte & Stein, 2008). In that study, we found that positive and negative self-schemas as measured by the Harter measure were correlated in expected directions with global self-esteem in early and mid-adolescence respectively: [Positive self-schemas: rs₍₂₆₄₎ = .68, .30; Negative self-schemas: rs₍₂₆₄₎ = −.36, −.33] and depressive symptoms in early and mid-adolescence respectively: [Positive self-schemas: rs₍₂₆₄₎ = −.16, −.09, Negative self-schemas: rs₍₂₆₄₎ = .13, .19] providing evidence of construct validity. Three year stability was satisfactory for the measure as an indicator of the number of positive, r₍₂₆₄₎ = .43 and negative self-schemas, r₍₂₆₄₎ = .35).

Drinking Possible Self

Availability in memory of a ‘drinking possible self’ was measured with a 5-point Likert scale item from a closed-ended questionnaire used to measure domains of the adolescent self (Stein et al., 1998). The item, “Drink Alcohol is Likely to Describe Me NEXT YEAR” (0=‘not at all’ to 4 ‘very much’) is embedded among other items, e.g., likelihood of being a good student, being popular, using drugs. Any affirmative response (1 = a little to 4 = very much) suggests availability in memory of a ‘drinking possible self,’ whereas a negative response (0 = not at all) suggests no ‘drinking possible self.’ Evidence of construct validity of the item as an indicator of a ‘drinking possible self’ comes from a longitudinal study of 8^th grade girls who reported some drinking (N=60). Availability of a ‘drinking possible self’ in the 8^th grade predicted alcohol misuse in the 9^th grade, even controlling for 8^th grade alcohol misuse, R²=.46, p<.001 (Corte & Stein, 2005). In addition, likelihood ratings have been successfully used to measure possible selves in a variety of domains, e.g., popularity, conventionality, deviancy (Stein et al., 1998) and academics (Kemmelmeier & Oyserman, 2001). Evidence of discriminant validity comes from other pilot work (Corte, 2007) we conducted in a sample of 39 predominantly minority youth aged 9–13 who completed the ‘drinking possible self’ measure and measures of depressive symptoms, oppositional defiant behavior, and alcohol expectancies. The ‘drinking possible self’ was not associated with depressive symptoms r₍₃₉₎ = .04, opposition/defiance r₍₃₉₎ = .16, or alcohol expectancies: positive social expectancy r₍₃₉₎ = .13, wild and crazy expectancy r₍₃₉₎ = .12, and sedated-impaired expectancy r₍₃₉₎ = −.08. The ‘drinking possible self’ was moderately and inversely associated with negative arousal expectancy, however, r₍₃₉₎ = −.29, p=.07, suggesting that youth who have a belief about themselves in terms of drinking too much in the future tend not to believe that they will experience negative feelings (e.g., feeling sad, mad, or scared) or behaviors (e.g., acting cocky, dangerous, rude, nasty, or mean, or hurting others) as a result of drinking. The same pattern of correlations between the ‘drinking possible self’ and alcohol expectancies was also found in a larger sample (N = 115) of college students using different measures of both ‘drinking possible self’ and alcohol expectancies (Hicks, 2007). These correlations support the notion that a ‘drinking possible self’ is distinct from alcohol expectancies. Further evidence of construct and discriminant validity comes from recent studies of college student samples that have shown that the ‘drinking possible self’ predicts alcohol use over and above attitudes, expectancies and intentions.

Antisocial Behavior

The degree of antisocial behavior was measured using the Antisocial Behavior Checklist for Youth (ASB) (Zucker, & Fitzgerald, 1996). This 63-item questionnaire measures the frequency of a child’s participation in a variety of aggressive and antisocial activities. We omitted 13 items that were age-inappropriate for our sample (e.g., lied to your spouse, changed jobs more than 3 times in one year).The items on the ASB questionnaire include overt forms of aggression such as “cursed at your parents to their face” and more covert forms of aggression like “lying to your teacher” which may better reflect antisociality in girls. This measure is currently being used in adolescents from an ongoing study of familial risk for alcoholism (Zucker, Fitzgerald, Refior, Puttler, Pallas, & Ellis, 2000). Expected negative correlations with positive behavioral conduct were found in adolescent participants of the longitudinal family study of risk for alcoholism both in early adolescence, r₍₂₆₄₎ = −.31, and three years later in mid-adolescence, r₍₂₆₄₎ = −.42 providing evidence of construct validity. In the same sample of adolescents, 3-year stability of measure was r₍₂₆₄₎ = .62. Cronbach’s alpha was = .89.

Parental Alcohol Problems

Parental alcohol problems were assessed using the short-form of the Children of Alcoholics Screening Test (CAST-6) (Hodgins, MatickaTyndale, ElGuebaly, & West, 1993). This six-item (yes/no) questionnaire asks about worries and concerns about a parents’ alcohol use (e.g., “have you ever heard your parents fight when one of them was drunk?,” “did you ever feel like hiding or emptying one of your parent’s bottles of liquor?,” “did you ever wish that one of your parents would stop drinking?”). The CAST-6 is a widely used and well-validated screening tool to identify children who live with an alcoholic parent. A cutoff score of ≥3 positive responses has been shown to be a reliable indicator of parental alcohol problems (Hodgins et al., 1993). Significant correlations between child and parent reports of parental drinking in 3^rd–6^th graders have been shown (Smith, Miller, Kroll, Simmons, & Gallen, 1999). In addition, a one-year test-retest reliability of .78 has been documented in a sample of 6^th graders (Havey & Dodd, 1995). Cronbach’s alpha was = .86.

Alcohol Use

The Alcohol Use & Misuse Scale (Shope, Copeland, & Dielman, 1994) was used to measure ever drinking more than a few sips of alcohol. Because it may be difficult for children to understand the concept of what constitutes an alcoholic drink, separate questions were asked for beer, wine, and hard liquor or any other drink with alcohol in it (Shope et al., 1994). Because of the age of our sample, however, we did not expect many youth to be regular drinkers or to have alcohol misuse symptoms (trouble with parents, police or peers due to alcohol). For any youth who did report drinking, we distinguished “illicit” drinking from drinking with parental permission (e.g., as part of a religious ceremony or on a holiday) by asking questions about context. Participants were reminded that their answers were confidential before the measure was administered. Studies have shown that adolescent’s self-reports of substance use are valid when compared to biochemical indicators (Wills & Cleary, 1997) and even with the utilization of a bogus pipeline procedure in which students are faced with the prospect of independent biochemical verification of self-reports of substance use (Campanelli, Dielman, & Shope, 1987). Other studies also support test-retest reliability of self-reported alcohol use in adolescents over 30, 60, 90 days, and past year (Levy et al., 2004).

Procedures

Data collection occurred at the school or summer youth program. Each child met individually with the PI or Research Assistant in a quiet room. After child assent was obtained, all measures were administered in interview format. Data collection took approximately 30–40 minutes. Each child was compensated for their participation with a $10 gift card to a local merchant (e.g., bookstore, sporting goods store, movie theatre).

Data Analysis

Bivariate correlations were estimated to examine relationships between the number of positive and negative self-schemas, presence/absence of a ‘drinking possible self,’ and known precursors for alcohol problems -- antisociality (ASB-Youth sum score) and parental alcohol problems (CAST-6 sum score). Multiple regression models were fit to examine the relative predictive ability of parental alcohol problems, antisociality, and gender on the number of positive self-schemas and the number of negative self-schemas. We fit a logistic regression model with the same variables to predict presence/absence of a ‘drinking possible self.’ Finally, to determine whether self-cognitions differed for youth who reported ever drinking versus those who did not, t-tests and contingency table analyses were used. Additional correlations and contingency table analyses were completed on data from 28 youth who completed a one-year follow-up.

RESULTS

Descriptive Statistics

Across the entire sample, antisociality ranged from no antisocial behaviors to 24 antisocial behaviors (See Table 1). Similarly, there was a full range from 0 to 6 yes responses on the 6-item Children of Alcoholics Screening Test (CAST-6). The self-concepts of youth were characterized by many more positive than negative self-schemas. Consistent with a previous study of 8^th graders (Corte & Stein, 2005), approximately one in five youth showed evidence of a ‘drinking possible self’ available in memory. None of the youth were regular drinkers (at least one drink per month for six months), but nearly one in five reported that they had consumed more than a few sips of alcohol. Among those who reported ever drinking, the average age of a child at his/her first drink was 10.0 (SD=1.9) and 80% (12 of 15) did so with parental awareness, e.g., at a holiday or family celebration.

Table 1.

Descriptive statistics for all variables for the entire sample (N = 79).

	Mean (SD) or % (n)	Range
Antisociality (ASB-Youth score)	4.5 (5.1)	0–24
Parental Alcohol Problems (CAST-6 score)	1.6 (2.0)	0–6
#Positive Self-Schemas	13.1 (6.8)	0–29
#Negative Self-Schemas	5.2 (4.4)	0–16
‘Drinking Possible Self’ (% yes)	19% (n = 15)	--
Ever Drink? (% yes)	19% (n = 15)	--

Relationships between Self-Cognitions and Known Precursors for Alcohol Problems

Pearson correlation coefficients are shown in Table 2 for the three self-cognitions (number of positive self-schemas, number of negative self-schemas and availability in memory of a ‘drinking possible self’) and two known precursors for alcohol problems (parental alcohol problems and antisociality). Parental alcohol problems were significantly associated with all three self-cognitions; few positive self-schemas, many negative self-schemas and a ‘drinking possible self’ were associated with high Children of Alcoholics Screening Test scores. Antisociality was only associated with one self-cognition--having a ‘drinking possible self’ available in memory.

Table 2.

Correlations between Self-Cognitions and Known Precursors for Alcohol Problems

	CAST-6 score	ASB-Youth score
# Positive Self-Schemas	−.38***	−.10
# Negative Self-Schemas	.37***	.01
‘Drinking Possible Self’ (0=no, 1=yes)	.25*	.28**

Note. CAST-6 score = Children of Alcoholics Screening Test sum score; ASB-Youth score= Antisocial Behavior Checklist for Youth score

^*p<.05;

^**p<.01;

^***p<.001

Known Precursors of Alcohol Problems as Predictors of Self-Cognitions

To examine the combined influence of known precursors for alcohol problems on the number of positive and negative self-schemas, we fit two regression models with parental alcohol problems (CAST-6 score), antisociality (ASB-Youth score) and gender as predictors. In the first analysis, high CAST-6 scores (β = −.43) and being female (β = −.28) predicted fewer positive self-schemas, R² = .22, F_(3,75)=6.9, p<.001. Antisociality was not a significant predictor of the number of positive self-schemas. The same predictors were used to predict the number of negative self-schemas. High CAST-6 scores (β = .44) and being female (β = .28) significantly predicted more negative self-schemas, R²= .22, F_(3,75)=6.9, p<.001. Antisociality was not a significant predictor of the number of negative self-schemas. Because our previous work demonstrated that having both few positive and many negative self-schemas was a cognitive liability for negative affect and risk behavior (Corte & Stein, 2007; Stein & Corte, 2008), we also fit a model with a composite score reflecting the overall valence of the self (number of positive self-schemas minus the number of negative self-schemas) as the dependent variable; high scores reflected a greater proportion of positive self-schemas, and presumably, a healthier self-concept. Once again, high CAST-6 scores (β = −.48) and being female (β = −.31) significantly predicted a lower proportion of positive self-schemas (lower composite score), R²= .27, F_(3,75)=9.0, p<.001. Antisociality was not a significant predictor of the composite self score.

Subsequently, we fit a logistic regression model to the data using the same variables to predict the ‘drinking possible self’ (0=no, 1=yes). Both known precursors of alcohol problems were significant predictors χ²(3,N=79)=9.0, p = .03. Each yes response on the Children of Alcoholics Screening Test increased the odds of having a ‘drinking possible self’ by 32% (OR=1.32, p=.05), and each point on the Antisocial Behavior Checklist increased the odds of having a ‘drinking possible self’ by 12% (OR=1.12, p=.05). Gender was not a significant predictor of the ‘drinking possible self.’

Differences in Self-Cognitions by Drinking Status

There were no differences in the number of positive self-schemas for those who reported ever drinking (M = 12.0, SD = 6.6) compared to those who reported never drinking (M = 13.4, SD = 6.9), t₍₇₇₎<1. There was a non-significant trend for those who reported ever drinking to have fewer negative self-schemas (M = 3.2, SD = 3.8) compared to those who reported never drinking (M = 5.6, SD = 4.5), t₍₇₇₎=1.92, p = .06. When looking at the total number of self-schemas (positive and negative combined) – an indicator of the degree of certainty about one’s self-definition -- those youth who reported ever drinking had significantly fewer overall self-schemas (M = 15.2, SD = 5.7) compared to those who reported never drinking (M = 19.0, SD = 5.2), t₍₇₇₎=2.46, p<.02. There were significant differences in whether or not drinking occurred depending upon the availability in memory of a ‘drinking possible self,’ χ²(1,N=79)= 5.3, p = .02. Although only 19% of the sample (15 of 79) reported ever drinking, among those youth who had a ‘drinking possible self’ available in memory, 40% reported ever drinking. Only 14% of those without a ‘drinking possible self’ (9 of 64) reported ever drinking.

Additional Analyses on Subset of Sample One Year Later

We were able to locate 28 youth from the original sample for a one-year follow-up. This subsample was comprised of 61% (n=17) boys and 39% (n=11) girls. Most were Latino (50%; n=14) or Black (43%; n=12). Mean age was 11.9 years. The same procedures and measures that were completed at baseline were completed at the one-year follow-up. There were no differences in any of the self-cognition variables in these 28 youth compared to the 79 youth in the original sample. In the follow-up sample, 25% (n=7) had a ‘drinking possible self’ available in memory and 36% (n=10) reported ever drinking (though only 3 of the 10 reported doing so outside parental awareness).

We examined cross-sectional (at one-year follow-up) and longitudinal bivariate relationships among self-cognitions, known precursors for alcohol problems and ever drinking. At the one-year follow-up, the ‘drinking possible self’ remained positively associated with parental alcohol problems, r₍₂₈₎ =.33, p=.08 and was much more strongly associated with antisocial behavior, r₍₂₈₎ =.59, p=.001. In addition, there continued to be significant differences in whether or not drinking occurred depending upon the availability in memory of a ‘drinking possible self,’ χ²(1,N=79) =5.2, p = .02. Although 36% of the sample (10 of 28) reported ever drinking at the one-year follow-up, among those youth who had a ‘drinking possible self’ available in memory, 71% (5 of 7) reported ever drinking. Only 24% of those without a ‘drinking possible self’ (5 of 21) reported ever drinking.

In longitudinal analyses, only one marginally significant difference was found. Although 25% (7 of 28) of the youth had a ‘drinking possible self’ at the one-year follow-up, 44% (4 of 9) of those who reported ever drinking at baseline had a ‘drinking possible self’ one year later and only 16% of nondrinkers at baseline had a ‘drinking possible self’ one year later, χ²(1,N=28) =2.7, p = .10. Positive and negative self-schemas remained associated with known precursors but not with drinking.

Stability coefficients over the one-year time frame were high for the Children of Alcoholics Screening Test sum score (r = .67), Antisocial Behavior Checklist-Youth sum score (r = .75), number of positive self-schemas (r = .66), number of negative self-schemas (r = .69) and ever drinking (Phi = .77). The stability coefficient for the ‘drinking possible self,’ an emerging and therefore less well developed cognition, was lower (Phi = .24). Of the 28 youth, 20 (71%) were consistent in their reports of a ‘drinking possible self’ (yes-yes or no-no) over the one year time period, 4 (14% of the sample) did not report a ‘drinking possible self’ at baseline and did report one at one year follow-up, and only 4 (14% of the sample) who reported a ‘drinking possible self’ at baseline did not report one at follow-up.

DISCUSSION

In this study, we examined relationships among the total array of self-cognitions (number of positive and negative self-schemas), including availability in memory of a future-oriented conception of the self related to drinking (‘drinking possible self’), and two known precursors for alcohol problems (antisocial behavior and parental alcohol problems) in preadolescent youth prior to the onset of regular alcohol use. This study was a first attempt to determine whether self-cognitions, which are powerful proximal predictors of behavior, are related to known precursors for alcohol problems and prior to regular alcohol use.

From our theoretical perspective, self-cognitions may be an important proximal risk factor that mediates the effects of parental alcohol problems and antisocial behavior on early alcohol use. As such, we anticipated significant relationships between self-cognitions and both parental alcohol problems and antisocial behavior. Parental alcohol problems were modestly but significantly associated with few positive self-schemas, many negative self-schemas and availability in memory of a ‘drinking possible self.’ We theorize that the chaos of an alcoholic home may provide insufficient opportunities for exploration of the environment necessary to develop positive self-schemas (Eccles, Barber, 1999). In addition, the modeling of alcohol in the home may be conducive to the formation of a ‘drinking possible self.’ Zucker and colleagues (Zucker, Kincaid, Fitzgerald, & Bingham, 1995) found that 3–6 year old children of alcoholic fathers correctly identified more alcoholic beverages and attributed more alcoholic beverage use to male adults than children of nonalcoholics, suggesting the availability of a rudimentary alcohol-related cognition in children of preschool age. Although this alcohol-related cognition was not a self-cognition, it is certainly plausible that over time, children of alcoholics may develop a future-oriented cognition about themselves in relation to alcohol, i.e., a ‘drinking possible self.’ Antisocial behavior, was moderately positively associated with availability in memory of a ‘drinking possible self,’ but was not associated with positive or negative self-schemas. Because youth with conduct problems and antisocial behavior are likely to have a family history of alcohol problems (American Psychiatric Association (4th Ed.), 2000; Clark, Cornelius, Kirisci, & Tarter, 2005), and are at high risk for developing alcohol problems themselves (American Psychiatric Association (4th Ed.), 2000; Clark et al., 2005) they may have a ‘possible self’ related to alcohol articulated in memory at a very young age.

We found that girls and children who report more parental alcohol problems have both fewer positive and more negative self-cognitions. This is important given our previous studies with older adolescents (Corte & Stein, 2008), college students (Stein & Corte, 2008), and young adults (Corte & Stein, 2007; Stein & Corte, 2007) that have shown that such a self-concept configuration is associated with high levels of negative affect. In the presence of other factors, such as a ‘drinking possible self,’ the turn to alcohol may be one way of escaping the discomfort associated with high levels of affect. We also found that both of the known precursors for alcohol problems were associated with increased odds of having a ‘drinking possible self’ available in memory. As noted earlier, modeling of heavy alcohol use by parents may contribute to the development of rudimentary cognitions about oneself vis a vis drinking. The fact that antisocial behavior predicted the ‘drinking possible self’ is not surprising given that alcohol use in youth is one expression of antisocial behavior. A child with high levels of antisocial behavior may consider drinking to be consistent with the way they see themselves in the future.

While the majority of the sample had not yet begun drinking (and most had never had even a sip of alcohol), we took an exploratory look at differences in self-cognitions and known precursors for alcohol problems for those who had reported ever drinking versus those who had not (the majority of whom did so with parental awareness). Although there were no statistically significant differences in the number of positive or negative self-schemas for youth who reported ever drinking compared to youth who did not, those youth who reported ever drinking had fewer overall self-schemas compared to youth who never drank, suggesting that drinkers were less clear or certain about their sense of self-definition. Nineteen percent of the sample reported ever drinking, however, among those youth who had a ‘drinking possible self’ available in memory (approximately 1 in 5 youth), a much larger proportion (40%) reported ever drinking. In the one year follow up sample, 36% reported ever drinking, but among those youth who had a ‘drinking possible self’available in memory, a much larger proportion (71%) reported ever drinking. Longitudinal relationships in the follow up sample (n=28) suggest that ever drinking at baseline is numerically though not statistically associated with a ‘drinking possible self’ one year later. Because both drinking and availability in memory of a ‘drinking possible self’ are low frequency descriptors, this may be clinically significant. Given that the majority of those who reported ever drinking said that they did so with parental awareness, this pattern of findings raises interesting questions about whether any alcohol use, i.e., with or without parental awareness, may contribute to the formation of a conception of the self related to drinking.

The findings of this study should be interpreted in light of the following limitations. First, the sample size of 79 was small which limited our ability to examine interactions between known precursors for alcohol problems and self-cognitions as predictors of drinking. Second, although we were able to locate 28 youth for a one-year follow-up, this sample size severely limited power to detect effects of self-cognitions on drinking. Third, the ‘drinking possible self’ measure was a single item measure that was dichotomized to reflect presence or absence of a ‘drinking possible self.’ An earlier study, however, using the identical item as an indicator of the ‘drinking possible self’ showed that availability in memory of a ‘drinking possible self’ predicted alcohol misuse one year later in adolescent girls, controlling for the effects of alcohol misuse at baseline (Corte & Stein, 2005). Fourth, given that all variables were measured with self-report, it is possible that social desirability could be an alternative explanation of the results. We do not believe, however, this is a likely explanation of the findings given the wide range of scores for two of the variables that might call for a socially desirable response—antisocial behaviors and worries or concerns about a parent’s alcohol use. The other two variables that might call for a socially desirable response—’likely to drink too much in the future’ and ‘ever drink’ were endorsed by 19% of the sample, which is consistent with previous results in other samples of youth in this age range.

To our knowledge this was the first theoretically based study to examine relationships between self-structure and known precursors for alcohol problems in preadolescents prior to regular alcohol use. This study raises interesting questions about the potential role of a ‘drinking possible self’ in early alcohol use, as well as potential factors that contribute to development of a ‘drinking possible self.’ Larger longitudinal studies are needed to further explicate these relationships. Because the self is modifiable (Jacobs et al., 2003), it may be a viable target of interventions aimed at preventing early drinking onset.