Bandura's (1986) social cognitive theory postulates two cognitive constructs as crucial in determining behaviour. "Expectancies" are defined as the individual's beliefs about the positive and negative consequences likely to result from performing a behaviour. "Self-efficacy beliefs" are defined as the individual's confidence in his or her ability to perform the behaviour. A subtype of self efficacy, "refusal self-efficacy", has been defined as confidence in one's ability to resist or refuse engagement in a specified addictive behaviour at will (Oei & Baldwin, 1994). Refusal self-efficacy has been adopted along with expectancies to provide a social cognitive explanation of addictive behaviours, and particularly substance abuse.
Baldwin and Oei's two-phase model of addictive behaviour (Oei & Baldwin, 1994) combines social cognitive theory with models of memory activation (eg. Tiffany, 1990), to postulate qualitatively different processes which occur in the initiation phase and the maintenance phase of addictive behaviour. The "initiation phase" of an addictive behaviour is seen as dominated by controlled processes and instrumental conditioning. The "maintenance phase" is seen as dominated by automatic processes and classical conditioning. An "automatic" spreading activation of nodes in memory provides a plausible explanation for features of habitual and dependent substance use, including the phenomena of compulsive use and craving, and the acknowledged low success rates of various treatment paradigms.
The current work explores the "initiation phase" of the two-phase model of addictive behaviour, with particular reference to drinking and smoking as the two most common addictive behaviours involving substance use. Exploratory factor analyses and factor invariance studies are used to derive two new instruments for use with adolescents, from instruments previously established in adult populations. These instruments are the Drinking Cognitions Profile-Adolescent Form, to measure adolescent drinking expectancies and drinking refusal self-efficacy, and the Smoking Cognitions Profile-Adolescent Form, to measure adolescent smoking expectancies and smoking refusal self-efficacy.
Theory suggests that it is important to be able to quantify the substance-related expectancies and refusal self-efficacy beliefs of adolescents, to enable screening for at-risk adolescents at an individual level, and to inform preventive education strategies at the public health level. A clear link has previously been demonstrated between early, regular, or heavy experimental substance use in childhood and adolescence, and problems with substance abuse and dependence moving into adulthood (e.g. Petraitis, Flay, & Miller, 1995). The two-phase model suggests that prevention efforts which specifically aim to modify expectancies and refusal self efficacy may help to prevent, delay, or curtail experimental substance use, and there is encouraging evidence from the literature that such intervention is possible and has positive effects (Darkes & Goldman, 1993; Kraus, Smith, & Ratner, 1994).
In the current work, the Drinking Cognitions Profile-Adolescent Form (DCP-A) was used to test a model representing the initiation phase of the two-phase model of drinking behaviour, in a sample of 800 adolescent boys and girls aged 12 to 18 years. Paths were tested among parental drinking behaviour, adolescents' positive drinking expectancies (PDE), adolescents' negative drinking expectancies (NDE), adolescents' drinking refusal self-efficacy (DRSE), and number of drinks consumed in the past four weeks. PDE were significantly predicted by parental drinking behaviour: however NDE and DRSE were not. No significant relationship was found between PDE and NDE, supporting the hypothesis that these are not two poles of a bipolar construct but instead should be considered independent predictors of drinking behaviour. PDE were negatively related to DRSE. Although an earlier discriminant analysis showed that DRSE contributed significantly to the discrimination between drinkers and non-drinkers, DRSE did not significantly predict number of drinks consumed in the past four weeks in the current model. However, both PDE and NDE significantly predicted number of drinks in the past four weeks.
An important tenet of the two-phase model is that an individual's expectancies and refusal self-efficacy are specific to specific substances. For comparison with drinking behaviour, therefore, the Smoking Cognitions Profile-Adolescent Form was used to test a model representing the initiation phase of the two-phase model of smoking behaviour, in a different sample of 1200 adolescent boys and girls aged 12 to 18 years. Paths were tested among parental smoking behaviour, adolescents' positive smoking expectancies (PSE) and negative smoking expectancies (NSE), smoking refusal self-efficacy (SRSE), and number of cigarettes in the past four weeks. PSE and NSE were significantly predicted by parental smoking in the full sample; however a separate analysis found that the NSE of adolescent smokers NSE were not related to parental smoking. PSE and NSE were found to be negatively correlated. PSE were negatively related to SRSE, and NSE were positively related to SRSE. Although an earlier discriminant analysis showed that PSE and NSE discriminated between smokers and non-smokers, neither type of smoking expectancy significantly predicted number of cigarettes consumed in the past four weeks in the current model. However, SRSE was a significant predictor, negatively related to number of cigarettes smoked in the past four weeks.
One interpretation of the findings is that, despite the samples covering the same age-range, those adolescents in the "drinking" sample who engaged in drinking behaviour were predominantly in the initiation phase of drinking, while those in the "smoking" sample who engaged in smoking were predominantly in the maintenance phase of smoking. This difference would explain the fact that some evidence was found for the utility of both expectancies and refusal self-efficacy in predicting drinking behaviour, but only refusal self-efficacy was a significant predictor of smoking behaviour.
The findings suggest that expectancies and refusal self-efficacy are important determinants of substance-using behaviour, but may operate differently in relation to drinking and smoking. Implications of the study for prevention programs are discussed.