Doing what Simon says? Estimating the underlying causal structures of delinquent associations, attitudes, and serious theft.

The misbehavior of companions has been firmly established as one of the most important determinants of juvenile delinquency and adolescent substance use (e.g., Elliott, Huizinga, & Ageton, 1985; Johnson, Marcos, & Bahr, 1987). It is not clear, however, exactly how the relationship between juveniles and their peers fosters delinquency. The most well-known theory of peer influence and delinquency is social learning theory (Akers, 1985, 1994). Within its general theoretical framework, at least three key social processes -- socialization, social selection, and rationalization -- are described as underlying the connection between peer influence and delinquent conduct. Social learning and differential association theorists generally contend that juveniles learn the necessary attitudes and/or techniques to engage in wrongdoing through delinquent associations. It is their association with already delinquent peers that begins the socialization process of misconduct in previously conforming kids. Conversely, the process of social selection entails adolescents that are already delinquent and who seek out friends most like themselves. Delinquent peers only increase each other's misconduct to the extent that already formed delinquent attitudes and practices are reinforced by the nature of group activity. Finally, the rationalization process dictates that delinquent attitudes are shaped by delinquent conduct: Delinquents form their attitudes after engaging in misconduct.

Research testing of each of these models has been abundant. For the most part, studies have examined recursive models: It has been assumed that either delinquent associations foster delinquency or that delinquency promotes the selection of delinquent peers. Recent developments in quantitative criminology, however, challenge conventional wisdom, suggesting that the direction of causal influence between delinquent associations and delinquency is uncertain at best (see Greenberg, 1985, for a general discussion of the problem). For instance, although delinquency researchers frequently assume that delinquent associations affect delinquency, the actual relationship could indicate that: (a) Delinquency affects delinquent associations, (b) delinquent associations and delinquency are reciprocally related, or (c) a third variable affects both delinquent associations and delinquency. This is important because if the causal structure underlying the relationship between delinquent associations and delinquency is reciprocal rather than recursive, as previously assumed, then results from recursive studies are inadequate to describe the actual social processes in which delinquent behavior is embedded.

In this article, we argue that delinquency is as likely to affect delinquent associations and attitudes as associations and attitudes are to affect delinquency. To date, only a few longitudinal studies have examined the reciprocal nature of peer influence and delinquency (Agnew, 1991a; Burkett & Warren, 1987; Thornberry, Lizotte, Krohn, Farnworth, & Jang, 1994). These studies find overall support for the idea that delinquent associations, delinquent attitudes, and delinquent behavior are intricately related factors. The direction of influence, however, remains disputed. For instance, Thornberry et al. (1994) find that delinquent peer associations influence both delinquent beliefs and behavior. In turn, these influence the selection of peers. Burkett and Warren (1987) find that beliefs influence peer selection, which in turn shapes behavior. Their study shows that religiosity affects adolescent marijuana use, but only indirectly, through its impact on peer associations. The relationship between beliefs and behavior, then, is mediated by peer associations. Burkett and Warren (1987) contend that the primary influence of peer associations on marijuana use is through peer pressure (p. 122).

In keeping with the idea that unidirectional studies are inadequate for understanding the intricate relationships leading to crime, our research examines the reciprocal influences of delinquent associations, delinquent attitudes, and serious theft. We use panel data from the National Youth Survey (NYS) to estimate the direct and indirect paths between these three factors to test the relative contributions of differential association, situational group pressure, social selection, and rationalization processes in explaining peer influence and serious theft.

THEORETICAL BACKGROUND

Social learning theory is among the most prominent theories of delinquency today. Akers's (1985, 1994) social learning theory is a behavioristic reformulation of Sutherland's (1947; Sutherland & Cressey, 1970) differential association theory. He proposes a more complex differential association process than Sutherland, which includes imitation and reinforcement as well as the acquisition of definitions. Perhaps more important, the theory describes the typical sequence of events and feedbacks of deviant behavior, associations, and definitions. According to Akers (1985, 1994), deviant associations and definitions precede initial deviant acts, but once these acts have been performed, the associational patterns may themselves, in turn, be altered. Akers (1994), however, goes' on to point out that the reciprocal effects are not seen as equal, but that "The sequence of events, in which deviant associations precede the onset of delinquent behavior will occur more frequently than the sequence of events in which the onset of delinquency precedes the beginning of deviant associations" (p. 100). Recent research has identified four major social processes that specify and describe the feedback effects of the differential association variables within the general framework of social learning theory (Reed & Rountree, 1997).

DIFFERENTIAL ASSOCIATION

In the field of criminology, Aker's (1985, 1994) social learning theory is rooted in the writings of Sutherland (1947; Sutherland & Cressey, 1970). Sutherland presented differential association as a developmental or historical theory that explains the link between past experiences and criminality. The theory posits that delinquency is a product of norms, values, attitudes, rationalizations, motives, and drives (conceptualized as "definitions of law") learned in intimate, face-to-face interaction and communication with significant others. Most contemporary researchers simplify the concept of definitions by measuring only delinquent attitudes. Whereas Akers places primary importance on both family and peer groups, Sutherland and Cressey assign special importance to peer influences, and most research has followed by concentrating on the role of delinquent peers in the production of delinquency.

Sutherland and Cressey (1970) presented nine propositional statements from which the theory's causal structure can be implied. The key mechanism in becoming delinquent is that one associates differentially with a variety of social circles. Some of these circles may define delinquency as favorable, whereas others define it as unfavorable. They go on to state that "A person becomes delinquent because of an excess of defenitions favorable to violation of law over definitions unfavorable to violation of law" (p. 77). Exposure to criminogenic and conventional forces promotes the internalization of defenitions both favorable and unfavorable to law violation; a person exposed to an excess of crime-favorable defenitions is likely to engage in criminal behavior. Furthermore, people are exposed to associations in varying frequency, duration, priority, and intensity. Thus, the theory assumes that as the frequency, duration, priority, and intensity of delinquent associations increase relative to conventional associations, the probability of crime increases relative to conventional behavior. The differential association model (also called the socialization model) specifies a causal order in which delinquent associations affect delinquent behavior indirectly through its effect on delinquent attitudes. Although Akers and his associates (1985; Akers, Krohn, Lanza-Kaduce, & Radosevich, 1979; Krohn, Skinner, Massey, & Akers, 1983) elaborate the process to include imitation and reinforcement, the crux of Sutherland's framework lies in the notion of exposure to definitions (or attitudes).

SITUATIONAL GROUP PRESSURE

Some researchers interpret the causal structure implied by Sutherland's statements differently. Short (1957, 1958) and Reiss and Rhodes (1964) assume only direct effects from delinquent associations to criminal behavior. Contrary to differential association theory, most of the impact of a friend's delinquent behavior on personal delinquent behavior bypasses attitudes. Adolescents commit delinquency simply because their friends do. Just how the direct effect is supposed to occur is rarely stated, although some delinquency researchers speculate about the processes underlying such a finding. Briar and Piliavin (1965) and Short and Strodtbeck (1965) suggest that this effect reflects situational group pressure. Briar and Piliavin (1965) argue that the delinquent peer group may be a source of "situationally induced motives" and that delinquent peers may provide the impetus to deviate before one has come to accept crime-favorable definitions and quite often in spite of commitments to conventional normative standards. Adolescents whose friends …