Predicting problem behaviors with multiple expectancies: expanding expectancy-value theory

Adolescence, Fall, 2004 by Ashley Borders, Mitchell Earleywine, Stanley J. Huey

According to expectancy-value theory, individuals choose behaviors based on the outcomes they expect and the values they ascribe to those expected outcomes. Expectancies, or anticipations of likely consequences for a given action, result from individuals' learning history and then become the basis for future behavioral choices (Del Boca et al., 2002). Empirical support for this theory includes studies of such diverse behaviors as aggression (Perry, Perry, & Rasmussen, 1986; Slaby & Guerra, 1988), alcohol consumption (Del Boca et al., 2002), and academic performance (see Wigfield & Eccles, 2000). Across .the board, expectations of greater reinforcement relate to more of the target behavior.

Nevertheless, research on expectancy-value theory is often limited by a focus on a single behavior and its accompanying expectancies. In studying aggression, for example, researchers generally examine only aggression behaviors and no alternatives to aggression. However, decisions about behavior are rarely this simple. Instead, individuals usually choose from several alternative behaviors in any given situation and therefore compare various expectancies. For instance, aggression expectancies likely compete with expectancies for more prosocial behaviors (e.g., talking with the provoker or getting outside help) in determining behavior choice. Cognitive theories of behavioral choice rarely address this problem of multiple behaviors.

Applied behavior analysis suggests the matching law, which does account for competing behaviors. In his original formulation of the matching law, Herrnstein (1961, 1970) suggested that any given behavioral decision arises from reinforcement for that behavior as well as reinforcement for alternative behaviors. As reinforcement for alternative behaviors increases, performance of the target behavior decreases. Likewise, environments that do not provide alternative reinforcement promote an increase in the target behavior. Although initial studies of this phenomenon focused on animals (for reviews, see Davison & McCarthy, 1988; de Villiers, 1977), researchers now successfully apply the matching law to understanding human alcohol consumption (Vuchinich, 1995) and students' behavior in special education settings (Martens & Houk, 1989; Neef et al., 1992).

The matching law may provide an important contribution to expectancy-value theory (and other cognitive models of behavior choice). Specifically, individuals may make a behavioral decision based on anticipated reinforcement for that behavior as well as anticipated reinforcement for competing, alternative behaviors. In a previous investigation of multiple expectancies, Levy and Earleywine (2003) examined the relations between alcohol use expectancies, studying expectancies, and drinking problems in college students. For students with high alcohol use expectancies, high studying expectancies were associated with fewer perceived drinking problems than were low studying expectancies. In other words, positive studying expectancies attenuate the effect of positive alcohol use expectancies and may buffer against drinking problems. By contrast, students with low alcohol use expectancies did not report drinking problems, regardless of competing expectancies. Besides providing a more complex understanding of drinking behavior, discovering the influence of alternative expectancies also suggests new interventions for drinking problems.

In this study, we examined multiple expectancies to better understand the problem behavior of high school students. Instead of focusing on drinking behavior, which is typically studied with college populations, we examined general acting-out behavior. Most problem behavior measures for children and adolescents include externalizing (e.g., hitting, throwing things, yelling) and hyperactivity (e.g., trouble sitting still, interrupting) problems. High scores on these scales generally suggest conduct problems, school misbehavior, and poor academic functioning (Achenbach, 1999). In addition, both student self-reports and adult reports correlate with more objective measures of delinquent behavior (Hinshaw et al., 1995; Vazsonyi et al., 1999). We therefore defined problem behaviors as adolescents' self-reports of their externalizing and hyperactive behaviors.

We predicted that problem behaviors would be related to positive expectancies about misbehaving. Adolescents who expect some benefit (e.g., being popular, exercising power over peers, avoiding unpleasant academic demands) from acting out are more likely to engage in problem behavior. Allen and colleagues (1990) found that high school students' expectancies about delinquent behavior correlate with self-reported levels of delinquency, hard drug use, and unprotected sexual activity. Moreover, aggressive adolescents report positive aggression expectancies more often than their nonaggressive peers (Slaby & Guerra, 1988). Specifically, aggressive adolescents expect that aggression will increase self-esteem and help avoid a negative image with peers. In work with younger children, Perry and colleagues (1986) found that aggressive children expect aggression to yield tangible rewards and end aversive behaviors by the victim. Furthermore, myriad research on college student drinking behavior suggests that drinkers report higher expectations of relaxation and increased sociability (Del Boca et al., 2002). Finally, behavioral researchers suggest that tangible rewards, peer attention and approval, and avoidance of aversive tasks or people are common reinforcers for classroom misbehavior (Witt, Elliott, & Gresham, 1988). We therefore expected that adolescents' problem behavior would relate to positive expectancies about the consequences of misbehaving.