take from book only
WRITE ONE LIKE THIS
Social learning theory is based on the idea that the same learning processes present in social structure, interaction, and situation produces both conforming and deviant behavior (ASJ, 2021). Ronald Akers expanded upon Edwin Sutherland’s original differential association theory and combined it with general behavioral learning principles. Social learning theory proposes that delinquent behavior is acquired, repeated, and changed by the same process as conforming behavior (ASJ 121). Akers’ theory also has a focus on four major concepts: differential association, definitions, differential reinforcement, and imitation.
Winfree, Sellers, and Clason (1993) explore the connection between social learning and adolescent deviance abstention with a focus on drug use. Although research has been previously conducted on why individuals resort to the use of drugs, little research has been done on the abstention from drugs and the cessation of their use. As a result, the researchers specifically focus on understanding the reasons for initiating, quitting, and avoiding drugs. The researchers propose that cessation and abstention are learned behavior and utilize the social learning variables to test their hypothesis (Winfree et al., 1993). Their goal was to distinguish among abstainers, current users and former users of illicit drugs within a sample of 1688 middle school and high school students in two nonurban communities. The results of their study indicated that the social learning variables clearly distinguish abstainers from current illicit drug users but were not as likely to distinguish between former users and current users or former users from abstainers (Winfree et al., 1993).
The study utilized a survey administered to middle school and high school students from two communities. The independent variables of focus were the personal-biographical characteristics and social learning variables. The social learning variables were divided into three subsets. The differential association concept was measured by questioning how many of the student’s best friends used marijuana, alcohol, or other drugs. The responses could be “(1) I don’t know and none, (2) less than half, (3) about one-half, and (4) more than one-half” (Winfree et al., 1993, p. 109). From this a composite score could be measured which represented the differential peer association. The concept of definitions was measured by personal approval, peer approval, and differential peer definitions. Differential reinforcement and imitation were also measured by asking questions about the reasons for initiating drug use and reasons for abstaining from drug use. The researchers hypothesized that the variables derived from social learning theory would distinguish best between abstainers and current users but not be as clearly defined between abstainers and former users or between current and former users. The results of the study confirmed their expectations. The social learning variables were accurate in predicting abstention and current drug use, which reflected continuous patterns of behavior. However, the cessation of drug use was found to result in a significant behavioral departure that cannot accurately be explained by social leaning variables independently. Social learning theory accounted for the difference in the learning environments of current users and abstainers from drug use. Abstainers were found to be exposed to nondrug using peers with antidrug orientations, whereas current users typically had drug-using friends who approved of drug use. The researchers proposed that the reason it was hard to distinguish between former and current users is due to the fact that they have similar social learning environments which include drug using peers, prodrug orientations, and reinforcers of drug using behavior.
When looking at the results of this study, it would be interesting to see how social learning was affected by the pandemic. For example, students were learning virtually and did not get to experience the normal socialization that goes along with typical learning situations. As a result of this, it would be expected that crime would decrease because individuals were not exposed to situations to learn deviant behavior.
References
Akers, R. L., Sellers, C. S., & Jennings, W. G. (2021). Criminological Theories: Introduction, Evaluation, and Application (8th ed.). Oxford University Press.
Winfree, L. T., Sellers, C. S., & Clason, D. L. (1993). Social Learning and Adolescent Deviance Abstention: Toward Understanding the Reasons for Initiating, Quitting, and Avoiding Drugs. Journal of Quantitative Criminology, 9(1), 101–125.
http://www.jstor.org/stable/23365886
Social learning theory has been used to refer to virtually any social behavioristic approach in social science, principally that of Albert Bandura and other psychologists (Bandura, 1977a, 1977b; Bandura and Walters, 1963; Miller and Dollard, 1941; Rotter, 1954; Patterson, 1975; Kunkel, 1975). As a general perspective emphasizing “reciprocal interaction between cognitive, behavioral and environmental determinants” (Bandura, 1977b:vii), variants of social learning can be found in a number of areas in psychology and sociology. Gerald Patterson and his colleagues at the
Oregon Social Learning Center
(OSLC) have a long history in the explanation, treatment, and prevention of delinquent and deviant behavior (Patterson et al., 1975; Patterson, 1995, 2002; Patterson and Chamberlain, 1994; Andrews, 1980; Andrews and Bonta, 2003; Akers and Jennings, 2009, 2019) from a social learning perspective.
In the field of criminology, however, social learning theory refers primarily to the theory of crime and deviance developed by Ronald L. Akers. Akers’s social learning theory was originally proposed in collaboration with Robert L. Burgess (Burgess and Akers, 1966a) as a behavioristic reformulation of Edwin H. Sutherland’s differential association theory of crime. It is a general theory that has been applied to a wide range of deviant and criminal behavior. It is one of the most frequently tested (Stitt and Giacopassi, 1992; Jennings and Akers, 2011; Sellers, Winfree, and Akers, 2012; Akers and Jennings, 2009, 2019) and is the theory of crime and delinquency most frequently endorsed by academic criminologists in recent surveys (Ellis and Walsh, 1999; Cooper et al., 2010).
SUTHERLAND’S DIFFERENTIAL Association Theory
The late Edwin H. Sutherland is widely recognized as the most important criminologist of the 20th century. Sutherland is known for pioneering sociological studies of professional theft (Sutherland, 1937) and white-collar crime (1940, 1949). He is best known for formulating a general sociological theory of crime and delinquency, the “differential association” theory.1
Sutherland was the author of a criminology textbook, Principles of Criminology, which was the leading text in the field for over 30 years. It was only in the pages of this text that he fully stated his theory, the final version of which was published in the 1947 edition. He proposed differential association theory as an explanation of individual criminal behavior and suggested that the theory was compatible with what he termed “differential social organization” as the cause of differences in group or societal crime rates. Sutherland gave only brief attention to differential social organization, however, and concentrated his efforts on fully explicating differential association theory, which is as follows (Sutherland, 1947:6–7):
1.Criminal behavior is learned.
2.Criminal behavior is learned in interaction with other persons in a process of communication.
3.The principal part of the learning of criminal behavior occurs within intimate personal groups.
4.When criminal behavior is learned, the learning includes (a) techniques of committing the crime, which are sometimes very complicated, sometimes very simple, and (b) the specific direction of motives, drives, rationalizations, and attitudes.
5.The specific direction of motives and drives is learned from
definitions
of the legal codes as favorable or unfavorable.
6.A person becomes delinquent because of an excess of definitions favorable to violation of law over definitions unfavorable to violation of law.
7.Differential associations may vary in frequency, duration, priority, and intensity.
8.The process of learning criminal behavior by association with criminal and anticriminal patterns involves all of the mechanisms that are involved in any other learning.
9.Although criminal behavior is an expression of general needs and values, it is not explained by those general needs and values, because noncriminal behavior is an expression of the same needs and values.
The first proposition is that criminal behavior is learned, and the terms learned and learning are included in other statements. Criminal behavior is learned in a process of symbolic interaction with others, mainly in primary or intimate groups. Although all nine statements constitute the theory, it is the sixth statement that Sutherland identified as the
principle of differential association
. This is the principle that a person commits criminal acts because he or she has learned definitions (rationalizations and attitudes) favorable to violation of law in “excess” of the definitions unfavorable to violation of law.
The theory explains criminal behavior by the exposure to others’ definitions favorable to criminal behavior, balanced against contact with conforming definitions. Although one expects that law-violating definitions are typically communicated by those who have violated the law, it is possible to learn law-abiding definitions from them, just as one can be exposed to deviant definitions from law-abiding people (Cressey, 1960). The seventh principle in the theory makes it clear that the process is not a simple matter of either criminal or noncriminal association but one that varies according to what are called the
modalities of association
. That is, if persons are exposed first (priority), more frequently, for a longer time (duration), and with greater intensity (importance) to law-violating definitions than to law-abiding definitions, then they are more likely to deviate from the law.
After Sutherland’s death, Donald R. Cressey revised Principles of Criminology from the 5th through the 10th editions (Sutherland and Cressey, 1978). Cressey became the major proponent of differential association, clarifying it and applying it to a number of different areas in criminology (see Akers and Matsueda, 1989; Matsueda, 1988) and on occasion included differential association as one of the concepts “related to general theories of social learning. . . . The content of learning, not the process itself, is considered the significant element determining whether one becomes a criminal or non-criminal” (Sutherland and Cressey, 1960:58; emphasis added). Through the final edition of the text (Sutherland et al., 1992) Sutherland’s theory was purposely left in its 1947 version without modification. Others had suggested partial modifications of the theory, but Akers’s social learning theory (first developed with Robert L. Burgess) is a thorough revision that integrates the sociological principles of the theory with principles of social behaviorism in psychology.
AKERS’S SOCIAL LEARNING THEORY
Development of the Theory
Sutherland asserted in the eighth statement of his theory that all the mechanisms of learning are involved in criminal behavior. However, beyond a brief comment that more is involved than direct imitation (Tarde, 1912), he did not explain what the mechanisms of learning are. These learning mechanisms were specified by Burgess and Akers (1966a) in their “differential association-reinforcement” theory of criminal behavior. Burgess and Akers produced a full reformulation that retained the principles of differential association, combining them with, and restating them in terms of, the learning principles of operant and respondent conditioning that had been developed by behavioral psychologists.2 Akers followed up his early work with Burgess to develop social learning theory, applying it to criminal, delinquent, and deviant behavior in general. He has modified the theory, provided a fully explicated presentation of its concepts, examined it in light of the critiques and research by others, and carried out his own research to test its central propositions (Akers, 1973, 1977, 1985, 1998; see also Akers and Jennings, 2009, 2019; Jennings and Akers, 2011; Sellers et al., 2012; Jennings and Henderson, 2014a, 2014b).
Social learning theory is not an alternative to differential association theory. Instead, it is a broader theory that retains all of the differential association processes in Sutherland’s theory (albeit clarified and somewhat modified) and integrates it with
differential reinforcement
and other principles of behavioral acquisition, continuation, and cessation (Akers, 1985:41). Thus research findings supportive of differential association also support the integrated theory. But social learning theory explains criminal and delinquent behavior more thoroughly than does the original differential association theory (see, e.g., Akers et al., 1979; Warr and Stafford, 1991).
Burgess and Akers (1966a) explicitly identified the learning mechanisms as those found in modern behavioral theory. They retained the concepts of differential association and definitions from Sutherland’s theory but conceptualized them in more behavioral terms and added concepts from behavioral learning theory. These concepts include differential reinforcement, whereby operant behavior (the voluntary actions of the individual) is conditioned or shaped by rewards and punishments. They also include
classical or “respondent” conditioning
(the conditioning of involuntary reflex behavior),
discriminative stimuli
(the environmental and internal stimuli that provide cues or signals for behavior),
schedules of reinforcement
(the rate and ratio in which rewards and punishments follow behavioral responses), and other principles of behavior modification.
Social learning theory retains a strong element of the
symbolic interactionism
found in the concepts of differential association and definitions from Sutherland’s theory (Akers, 1985:39–70). Symbolic interactionism is the theory that social interaction is mainly the exchange of meaning and symbols; individuals have the cognitive capacity to imagine themselves in the roles of others and incorporate this into their conceptions of themselves (Sandstrom, Martin, and Fine, 2003). This, and the explicit inclusion of such concepts as imitation, anticipated reinforcement, and
self-reinforcement
, makes social learning “soft behaviorism” (Akers, 1985:65). It assumes human agency and soft determinism (see Chapter 1). As a result, the theory is closer to cognitive/behavioral learning theories, such as Albert Bandura’s (1973, 1977b, 1986; Bandura and Walters, 1963), than to the radical or orthodox operant behaviorism of B. F. Skinner (1953, 1959), with which Burgess and Akers began.
The Central Concepts and Propositions of Social Learning Theory
The word learning should not be taken to mean that the theory is only about how novel criminal behavior is acquired. “Behavioral principles are not limited to learning but are fundamental principles of performance [that account for] . . . the acquisition, maintenance, and modification of human behavior” (Andrews and Bonta, 1998:150; see also Horney, 2006). Social learning theory offers an explanation of crime and deviance that embraces variables that operate both to motivate and control criminal behavior, both to promote and undermine conformity. (See the discussion of questions of criminal motivations and inhibitors in Chapter 6.) The probability of criminal or conforming behavior occurring is a function of the balance of these influences on behavior operative in one’s learning history, at a given time, or in a given situation:
The basic assumption in social learning theory is that the same learning process in a context of social structure, interaction, and situation, produces both conforming and deviant behavior. The difference lies in the direction . . . [of] the balance of influences on behavior.
The probability that persons will engage in criminal and deviant behavior is increased and the probability of their conforming to the norm is decreased when they differentially associate with others who commit criminal behavior and espouse definitions favorable to it, are relatively more exposed in-person or symbolically to salient criminal/deviant models, define it as desirable or justified in a situation discriminative for the behavior, and have received in the past and anticipate in the current or future situation relatively greater reward than punishment for the behavior. (Akers, 1998:50)
As these quotations show, although referring to all aspects of the learning process, Akers’s development of the theory has relied principally on four major concepts: differential association, definitions, differential reinforcement, and imitation (Akers et al., 1979; Akers, 1985, 1998).
Differential Association
Differential association has both behavioral-interactional and
normative dimension
s. The
interactional dimension
is the direct association and interaction with others who engage in certain kinds of behavior, as well as the indirect association and identification with more distant reference groups. The normative dimension is the different patterns of norms and values to which an individual is exposed through this association (see Clark, 1972).
The groups with which one is in differential association provide the major social contexts in which all the mechanisms of social learning operate. They not only expose one to definitions, but they also present one with models to imitate and with differential reinforcement (source, schedule, value, and amount) for criminal or conforming behavior. The most important of these groups are the primary ones of family and friends, although they may also be secondary and reference groups. Neighbors, churches, schoolteachers, physicians, the police and authority figures, and other individuals and groups in the community (as well as mass media and other more remote sources of attitudes and models) have varying degrees of effect on the individual’s propensity to commit criminal and delinquent behavior. With the large and growing social interaction taking place over cell phones, the Internet, and other voice, picture, and video technologies, these also include what Warr (2002) refers to as “virtual peer groups” that both overlap with and exist apart from groups with which one is in face-to-face interaction. The term social media that is so often applied to this phenomenon captures very well the facilitation of interaction that makes not only for virtual peer groups but other groups and even entire “communities.” In social learning terms, these would be different variations on “virtual differential association.” Those associations that occur earlier (priority), last longer and occupy more of one’s time (duration), take place most often (frequency), and involve others with whom one has the more important or closer relationship (intensity) will have the greater effect on behavior.
Definitions
Definitions are one’s own attitudes or meanings that one attaches to given behavior. That is, they are orientations, rationalizations, definitions of the situation, and other evaluative and moral attitudes that define the commission of an act as right or wrong, good or bad, desirable or undesirable, justified or unjustified. In social learning theory, one’s own definitions are distinct from (albeit shaped by) the normative dimension of differential association, which includes definitions shared with or held by others with whom one associates.
In social learning theory, these definitions are both general and specific. General definitions include religious, moral, and other conventional values and norms that are favorable to conforming behavior and unfavorable to committing any deviant or criminal acts. Specific definitions orient the person to particular acts or series of acts. Thus one may believe that it is morally wrong to steal and that laws against theft should be obeyed, but at the same time one may see little wrong with smoking marijuana and rationalize that it is all right to violate laws against drug possession.
The greater the extent to which one holds attitudes that disapprove of certain acts, the less one is likely to engage in them. Conventional beliefs are
negative definitions
of criminal behavior. Conversely, the more one’s own attitudes approve of a behavior, the greater the chances are that one will enact that behavior. Approving definitions favorable to the commission of criminal or deviant behavior are basically positive or neutralizing. Positive definitions are beliefs or attitudes that make the behavior morally desirable or wholly permissible. Neutralizing definitions favor the commission of a crime by justifying or excusing it. They view the act as something that is probably undesirable but, given the situation, is nonetheless all right, justified, excusable, necessary, or not really that bad. The concept of
neutralizing definitions
in social learning theory incorporates the notions of verbalizations, rationalizations, techniques of neutralization, accounts, disclaimers, and moral disengagement (Cardwell, Piquero, Jennings, Copes, Schubert, and Mulvey, 2015; Cressey, 1953; Sykes and Matza, 1957; Lyman and Scott, 1970; Hewitt and Stokes, 1975; Bandura, 1990; also see the discussion of neutralizations in Chapter 6). Neutralizing attitudes include such beliefs as “Everybody has a racket,” “I can’t help myself; I was born this way,” “I am not at fault,” “I am not responsible,” “I was drunk and didn’t know what I was doing,” “I just blew my top,” “They can afford it,” “He deserved it,” and other excuses and justifications for committing deviant acts and victimizing others. These definitions favorable and unfavorable to criminal and delinquent behavior are developed through imitation and differential reinforcement. Cognitively, they provide a mindset that makes one more willing to commit the act when the opportunity occurs. Behaviorally, they affect the commission of deviant or criminal behavior by acting as internal discriminative stimuli. Discriminative stimuli operate as cues or signals to the individual as to what responses are appropriate or expected in a given situation.
Some of the definitions favorable to deviance are so intensely held that they almost require one to violate the law. For instance, the radical ideologies of militant groups provide strong motivation for terrorist acts (Akers and Silverman, 2004; Akins and Winfree, 2017; Winfree and Akins, 2008). For the most part, however, definitions favorable to crime and delinquency do not directly motivate action in this sense. Rather, they are conventional beliefs so weakly held that they provide no restraint or are positive or neutralizing attitudes that facilitate law violation in the right set of circumstances. “The concept of definitions is not proposed as an either/or set of categories. Rather, it is a matter of the balance of definitions favorable and unfavorable along a continuum to which one may be more or less exposed by others and may personally internalize to a greater or lesser degree” (Akers and Jennings, 2009:107).
Differential Reinforcement
Differential reinforcement refers to the balance of anticipated or actual rewards and punishments that follow or are consequences of behavior. Whether individuals will refrain from or commit a crime at any given time (and whether they will continue or desist from doing so in the future) depends on the past, present, and anticipated future rewards and punishments for their actions. The probability that an act will be committed or repeated is increased by rewarding outcomes or reactions to it (e.g., obtaining approval, money, food, or pleasant feelings)—
positive reinforcement
. The likelihood that an action will be taken is also enhanced when it allows the person to avoid or escape aversive or unpleasant events—
negative reinforcement
. Punishment may also be direct (positive), in which painful or unpleasant consequences are attached to a behavior, or indirect (negative), in which a reward or pleasant consequence is removed. Just as there are modalities of association, there are
modalities of reinforcement
—amount, frequency, and probability. The greater the value or amount of reinforcement for the person’s behavior, the more frequently it is reinforced, and the higher the probability that it will be reinforced (as balanced against alternative behavior), the greater the likelihood that it will occur and be repeated. The reinforcement process does not operate in the social environment in a simple either/or fashion. Rather, it operates according to a matching function in which the occurrence of, and changes in, each of several different behaviors correlate with the probability and amount of, and changes in, the balance of reward and punishment attached to each behavior (Herrnstein, 1961; Hamblin, 1979; Conger and Simons, 1995).
Reinforcers and punishers can be nonsocial; an example is the direct physical effects of drugs and alcohol. Also, there may be a genetic or physiological basis for the tendency of some individuals (such as those prone to sensation-seeking) more than others to find certain forms of deviant behavior intrinsically rewarding (Brezina and Piquero, 2003; Fox, 2017; Higgins, Mahoney, and Ricketts, 2009; Wood et al., 1995). Social reinforcement, which is the peer, family, or other social context in which the actions take place, one’s learned moral attitudes, and other social variables, affects how much one experiences the intrinsic effects of substance use or committing certain acts as pleasurable and enjoyable or as frightening and unpleasant. “Individual differences in the propensity to derive intrinsic pleasure and reward from substance use appear to be related to social learning factors in ways predicted by the theory” (Brezina and Piquero, 2003:284). Thus the theory posits effects of non
social reinforcement
(see Higgins et al., 2011) but proposes that these interact with other social learning variables. The prediction is that most of the learning in criminal and deviant behavior is the result of social exchange in which the words, responses, presence, and behavior of other persons directly reinforce behavior, provide the setting for reinforcement (discriminative stimuli), or serve as the conduit through which other social rewards and punishers are delivered or made available.
The concept of social reinforcement (and punishment) goes beyond the direct reactions of others present while an act is committed. It also includes the whole range of actual and anticipated, tangible and intangible rewards valued in society or subgroups. Social rewards can be highly symbolic. Their reinforcing effects can come from their fulfilling ideological, religious, political, or other goals. Even those rewards that we consider to be very tangible, such as money and material possessions, gain their reinforcing value from the prestige and approval value they have in society. Nonsocial reinforcement, therefore, is more narrowly confined to unconditioned physiological and physical stimuli. In self-reinforcement the individual exercises self-control, reinforcing or punishing one’s own behavior by taking the role of others, even when alone.
Imitation
Imitation refers to the engagement in behavior after the observation of similar behavior in others. Whether the behavior modeled by others will be imitated is affected by the characteristics of the models, the behavior observed, and the observed consequences of the behavior (
vicarious reinforcement
; Bandura, 1977b). The observation of salient models in primary groups and in the media affects both prosocial and deviant behavior (Donnerstein and Linz, 1995). It is more important in the initial acquisition and performance of novel behavior than in the maintenance or cessation of behavioral patterns once established, but it continues to have some effect in maintaining behavior.
The Social Learning Process: Sequence and Feedback Effects
These social learning variables are all part of an underlying process that is operative in each individual’s learning history and in the immediate situation in which an opportunity for a crime occurs. Akers stresses that social learning is a process with reciprocal and feedback effects. The reciprocal effects are not seen as equal, however. Akers (1998) hypothesizes a typical temporal sequence or process by which persons come to the point of violating the law or engaging in other deviant acts.
This process is one in which the balance of learned definitions, imitation of criminal or deviant models, and the anticipated balance of reinforcement produces the initial delinquent or deviant act. The facilitative effects of these variables continue in the repetition of acts, although imitation becomes less important than it was in the first commission of the act. After initiation, the actual social and nonsocial reinforcers and punishers affect whether or not the acts will be repeated and at what level of frequency. Not only the behavior itself but also the definitions are affected by the consequences of the initial act. Whether a deviant act will be committed in a situation that presents the opportunity to do so depends on the learning history of the individual and the set of reinforcement contingencies in that situation:
The actual social sanctions and other effects of engaging in the behavior may be perceived differently, but to the extent that they are more rewarding than alternative behavior, then the deviant behavior will be repeated under similar circumstances. Progression into more frequent or sustained patterns of deviant behavior is promoted [to the extent] that reinforcement, exposure to deviant models, and definitions are not offset by negative formal and informal sanctions and definitions. (Akers, 1985:60)
The theory does not hypothesize that definitions favorable to law violation only precede and are unaffected by the initiation of criminal acts. Acts in violation of the law can occur in the absence of any thought given to right and wrong. Furthermore, definitions may be applied by the individual retroactively to excuse or justify an act already committed. To the extent that such excuses successfully mitigate others’ negative sanctions or one’s self-punishment, however, they become cues for the repetition of criminal acts. At that point they precede the future commission of the acts.
Differential association with conforming and nonconforming others typically precedes the individual’s committing the acts. Families are included in the differential association process, and it is obvious that association, reinforcement of conforming or deviant behavior, deviant or conforming modeling, and exposure to definitions favorable or unfavorable to deviance occurs within the family prior to the onset of delinquency. On the other hand, it can never be true that the onset of delinquency initiates interaction in the family (except in the unlikely case of the late-stage adoption of a child who is already delinquent who is drawn to and chosen by deviant parents). Parents supervise and control their children’s selection of friends (Warr, 2005). However, behavioral similarity, including prior deviant behavior or tendencies, can play a role in the youth’s friendship selection and initiation of interaction in peer groups. Even so, associations with peers and others are most often formed initially around attractions, friendships, and circumstances, such as neighborhood, family, gender, race/ethnic, school, and other factors of similarity and proximity that have little to do directly with co-involvement in some deviant behavior (Weerman, 2011). In either case, after the associations have been established and the reinforcing or punishing consequences of the interaction and the deviant or conforming behavior are experienced, both the continuation of old and the seeking of new peer associations (over which one has any choice) will themselves be affected “in proportion to the relative rate of positive reinforcement generated during interaction with each of those peers” (Snyder, 2002:109):
Deviant peers reinforce each other for deviancy. Peers who maximize the child’s immediate payoffs get selected as friends. . . . The selection of deviant peers insures the maintenance of deviant behaviors as well as the development of new forms of deviancy. . . . The findings show that antisocial boys are mutually reinforcing for rule-breaking talk, and that this talk predicts both later delinquency and later substance use. (Patterson, 2002:12–13)
One may choose further interaction with others based, in part, on whether they too are involved in similar deviant or criminal behavior, assuming the relationships are mutually reinforcing. But the theory proposes and research supports 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 (see Osgood et al., 2013).
This recognition of sequence and reciprocal effects through social interaction is another clear indication that social learning is not, as some critics (Gottfredson and Hirschi, 1990) claim, simply a “cultural deviance” theory that assumes that violation of the rules of the larger society occurs only because individuals are conforming to the norms of deviant groups into which they have become completely socialized (see Akers, 1996; Sellers and Akers, 2006). One may learn deviant attitudes and behavior through exposure to street codes and deviant subcultures (Anderson, 1999; Baron, 2017; Benoit et al., 2003) or immersion in criminal and illegitimate activities as a career criminal (Steffensmeier and Ulmer, 2005). Many offenders have come into association with somewhat older criminal “mentors” who go beyond being co-offenders to provide direct tutelage in criminal attitudes and behavior and help the younger offenders develop and enhance their own criminal careers (Morselli, Tremblay, and McCarthy, 2006). Benoit et al. (2003) identified the balance of social learning variables, including what they refer to as “inverse imitation,” as the main mechanism by which the inner-city social structure, neighborhood, and family affect young men’s involvement in, and switching between, conforming and deviant norms and behavior such as drug use, trafficking, and violence (Benoit et al., 2003). However, the theory is not confined to delinquent, criminal, or deviant behavior acquired only through differential exposure to such deviant subcultures, environments, and individuals. It does not ignore, as Osgood and Anderson (2004:525) contend it does, “contextual and situational influences,” and it does not rest solely on “positing a causal role for peer culture that values delinquency.” Their finding that delinquency is related to unstructured or unsupervised adolescent peer interaction fits precisely into the notion of differential peer association. The theory proposes that conforming and delinquent, criminal, or deviant behavior is learned through differential exposure to conforming and deviant patterns. This includes incomplete or failed socialization in conventional norms and values as well as countervailing processes of reinforcement, imitation, and exposure to deviant definitions or attitudes. (See Akers and Jensen, 2006; Sellers and Akers, 2006; and Sellers et al., 2012, for a review of this and other common misinterpretations of social learning theory.)
Social Structure and Social Learning
Akers has proposed a social structure and social learning (SSSL) model in which social structural factors are hypothesized to have an indirect effect on an individual’s conduct. They affect the social learning variables of differential association, differential reinforcement, definitions, and imitation that, in turn, have a direct impact on an individual’s conduct. The social learning variables are proposed as the main ones in the process by which various aspects of the social structure influence individual behavior (see
Figure 5.1
):
The social structural variables are indicators of the primary distal macro-level and meso-level causes of crime, while the social learning variables reflect the primary proximate causes of criminal behavior that mediate the relationship between social structure and crime rates. Some structural variables are not related to crime and do not explain the crime rate because they do not have a crime-relevant effect on the social learning variables. (Akers, 1998:322)
As shown in Figure 5.1, Akers (1998) identified four dimensions of social structure that provide the contexts wherein the social learning variables operate:
Figure 5.1
SOURCE: Adapted from Akers, 1998:331.
(I)Differential Social Organization refers to the structural correlates of crime in the community or society that affect the rates of crime and delinquency, including age composition, population density, and other attributes that lean societies, communities, and other social systems “toward relatively high or relatively low crime rates” (Akers, 1998:332).
(II)Differential Location in the Social Structure refers to sociodemographic characteristics of individuals and social groups that indicate their niches within the larger social structure. Class, gender, race and ethnicity, marital status, and age locate the positions and standing of persons and their roles, groups, or social categories in the overall social structure.
(III)Theoretically Defined Structural Variables refer to anomie, class oppression, social disorganization, group conflict, patriarchy, and other concepts that have been used in one or more theories to identify criminogenic conditions of societies, communities, or groups (see Chapters 8–14).
(IV)Differential Social Location in Groups refers to individuals’ membership in and relation to primary, secondary, and reference groups such as the family, friendship and peer groups, leisure groups, colleagues, and work groups.
The differential social organization of society and community, as well as the differential location of persons in the social class, race, gender, religion, and other structures in society, provides the general learning contexts for individuals that increase or decrease the likelihood of their committing crime. The differential location in family, peer, school, church, and other groups provides the more immediate contexts that promote or discourage the criminal behavior of the individual. Differences in the societal or group rates of criminal behavior are a function of the extent to which their cultural traditions, norms, and social control systems provide socialization, learning environments, and immediate situations conducive to conformity or deviance. The structural conditions identified in macro-level theories can affect one’s exposure to criminal associations, models, definitions, and reinforcement to induce or retard criminal actions in individuals. It is possible, therefore, to integrate these structural theories with social learning. Although this has not yet been accomplished (see Chapter 15), the SSSL model is a step in that direction.
EMPIRICAL VALIDITY OF SOCIAL LEARNING THEORY
Research on Relationship of Criminal and Delinquent Behavior to Social Learning Variables
The great preponderance of research conducted on social learning theory has found relationships in the theoretically expected direction between social learning variables and a range of criminal, delinquent, and deviant behavior. Many studies have tested the theory using direct measures of one or more of the social learning variables, most often measuring differential association and definitions and less frequently imitation and differential reinforcement (see Brezina, 2009; Brauer and Tittle, 2012; Cochran et al., 2011; Punzo, 2016; Steele et al., 2011) and find that the theory’s hypotheses are upheld (see Winfree, Vigil-Backstrom, and Mays, 1994; Winfree, Mays, and Vigil-Backstrom, 1994; Mihalic and Elliott, 1997; Skinner and Fream, 1997; Batton and Ogle, 2003; Sellers, Cochran, and Winfree, 2003; Brezina and Piquero, 2003; Chappell and Piquero, 2004; McGloin et al., 2004; Osgood and Anderson, 2004; Triplett and Payne, 2004; Durkin, Wolfe, and Clark, 2005; Matsueda et al., 2006; Brauer, 2009; Fox, Nobles, and Akers, 2010; Dijkstra et al., 2010; Thomas, 2015; Watkins, 2016).3 The relationships between the social learning variables and delinquent, criminal, and deviant behavior found in the research are typically strong to moderate, and there has been very little negative evidence reported in the literature. The sequential and reciprocal effects of social learning variables and deviant behavior has also been supported by research on substance use (Krohn et al., 1985; Akers and Lee, 1996; Lee, Akers, and Borg, 2004) and on the use of physical aggression (Cochran, Maskaly, Jones, and Sellers, 2017). Most of the research on social learning theory has been done in the United States, but it is well supported by research in other societies as well (see Junger-Tas, 1992; Bruinsma, 1992; Hwang and Akers, 2003, 2006; Wang and Jensen, 2003; Aliverdinia and Pridemore, 2007; Antonaccio et al., 2010; Smangs, 2010; Scott-Parker et al., 2012; Meneses and Akers, 2011; Jennings et al., 2011; Miller et al., 2008, 2009; Lee, Moak, and Walker, 2016; Kabiri, Cochran, Stewart, Sharepour, Rahmati, and Shadmanfaat, 2018).
Pratt and Cullen’s (2000) meta-analysis covering the findings of many studies found support for the impact of two social learning variables (differential association and definitions) on offending, but the effects of those variables were not stronger than measures of self-control (Gottfredson and Hirschi, 1990). A later meta-analysis by Pratt et al. (2010) found that the four main social learning variables are related to crime and deviance in the expected direction but with differential association and definitions consistently having “quite strong” relationships and differential reinforcement and imitation variables having more modest effects. When social learning theory is tested against other theories using the same data collected from the same samples and analyzed in the same statistical models, it is usually found to have greater support than the theories with which it is being compared (e.g., see McGee, 1992; Benda, 1994; Burton et al., 1994; Rebellon, 2002; Preston, 2006; Neff and Waite, 2007; Aliverdinia and Pridemore, 2007; Smangs, 2010; Pauvels and Schils, 2016). When social learning variables are included in integrated or combined models that incorporate variables from different theories, the measures of social learning concepts have the strongest main and net effects (Elliott, Huizinga, and Ageton, 1985; Thornberry et al., 1994; Kaplan, 1996; Catalano et al., 1996; Huang et al., 2001; Jang, 2002; Cochran, Jones, Jones, and Sellers, 2016).
In addition to the consistently positive findings by other researchers, support for the theory comes from research conducted by Akers, his colleagues, and/or his students in which all of the key social learning variables of differential association, differential reinforcement, definitions, and imitation are measured (Akers, 1998, 2009). These include tests of social learning theory by itself and tests that directly compare its empirical validity with other theories. The findings in each of these studies demonstrate that the social learning variables, independently and in combination, are strongly related to the various forms of deviant, delinquent, and criminal behavior studied.
The combined effects of the social learning variables on adolescent alcohol and drug use and abuse are very strong. High amounts (31–68%) of the variance in these variables are accounted for by the social learning variables. Social bonding models account for about 15% and anomie/strain models account for less than 5% of the variance (see Akers et al., 1979; Krohn, Lanza-Kaduce, and Akers, 1984; Akers and Cochran, 1985).
Similarly, adolescent cigarette smoking is highly correlated with the social learning variables. These variables also predict quite well the maintenance of smoking over a three-year period. They fare less well, however, when predicting which of the initially abstinent youngsters will begin smoking in that same period. The social learning variables do a slightly better job of predicting the onset of smoking over a five-year period. The sequencing and reciprocal effects of social learning variables and smoking behavior over the five-year period are as predicted by the theory (see Krohn et al., 1985; Spear and Akers, 1988; Akers and Lee, 1996).
Further support for social learning theory comes from studies among other age groups and on other behavior. The onset, frequency, and quantity of elderly drinking are highly correlated with social learning variables and the theory also successfully accounts for problem drinking among the elderly (see Akers et al., 1989; Akers and La Greca, 1991). The social learning variables explain self-perceived likelihood of using force to gain sexual contact or committing rape by college men (55% explained variance). They also account for the actual use of drugs or alcohol, nonphysical coercion, and physical force by males to obtain sex (20% explained variance). Social bonding, self-control, and relative deprivation (classic strain) models account for less than 10% of the variance in these behaviors (Boeringer, Shehan, and Akers, 1991). This support for social learning theory in comparison to other social psychological theories has been replicated in cross-cultural research on adolescent substance use in South Korea (Hwang and Akers, 2003, 2006; Kim, Akers, and Yun, 2013) and on drug use among university students in Latin America (Meneses and Akers, 2011), high school students in Puerto Rico (Miller et al., 2008), and professional athletes in Iran (Kibiri et al., 2018).
Research on Social Learning in the Family and Delinquency
There is abundant evidence to show the significant impact on criminal and deviant behavior of differential association in primary groups, especially family and peers:
Thus, the analysis confirmed social learning theory’s prediction, and the findings from existing literature, that family and friends were important in understanding what was reinforcing for a particular person. . . . The analysis also showed that this measure of reinforcement worked in the way expected by social learning theory. (Triplett and Payne, 2004:628)
Family structure (two-parent vs. single-parent or no-parent families) and the larger social context of “the economy, the polity, the church, and the neighborhood or community . . . necessarily influence the functioning and efficacy of families” (Simons, Simons, and Wallace, 2004:93–94). The family is a key primary group with which one is differentially associated; and the process of acquiring, persisting, or modifying conforming and deviant behavior in the family or family surrogate is a social learning process in which interaction in the family exposes the children to normative values, behavioral models and vicarious reinforcement, and differential reinforcement (Simons et al., 2004). The role of the family is usually as a conventional socializer against delinquency and crime. It provides anticriminal definitions, conforming models, and the reinforcement of conformity through parental discipline; it also promotes the development of self-control. But deviant behavior may be the outcome of internal family interaction (McCord, 1991b; Simons et al., 2004; Simons, Sutton, Simons, Gibbons, and Murry, 2016). It is directly affected by deviant parental models, ineffective and erratic parental supervision and discipline in the use of positive and negative sanctions, and the endorsement of values and attitudes favorable to deviance. Parents who remain very much in control of the children while also being highly supportive and sensitive to the children’s needs have a better chance of raising prosocial, conforming, well-adjusted children, whereas problems of misconduct and delinquency are more likely to be found among children from permissive, neglectful, or too highly controlling families (Simons et al., 2004; Simons et al., 2016). Patterson and his colleagues at the OSLC have shown that the operation of social learning mechanisms in parent–child interaction is a strong predictor of conforming and deviant behavior (Patterson, 1975, 1995; Reid, Patterson, and Snyder, 2002; Weisner, Capaldi, and Patterson, 2003). Ineffective disciplinary strategies by parents increase the chances that a child will learn behavior in the early years that is a precursor to his or her later delinquency. Children learn conforming responses when parents consistently make use of positive rewards for proper behavior and impose moderately negative consequences for misbehavior (Capaldi, Chamberlain, and Patterson, 1997; see also Ardelt and Day, 2002, for effects of parental discipline and supervision in early adolescence).
[The] relative rate of reinforcement for child coercion during family conflict bouts . . . also predicted the child’s rates of deviancy observed one week later. If we then added how frequent conflicts or training trials occurred, we could account for over 60% of the variance in individual differences in deviancy. (Patterson, 2002:12)
In some cases, parents directly train their children to commit deviant behavior (Adler and Adler, 1978). And, in general, parental deviance and criminality are predictive of the children’s future delinquency and crime (McCord, 1991a). Moreover, youngsters with delinquent siblings (especially same-sex older siblings) in the family are more likely to be delinquent, even when parental and other family characteristics are taken into account (Rowe and Gulley, 1992; Lauritsen, 1993; Rowe and Farrington, 1997; Ardelt and Day, 2002; Mathis and Mueller, 2015). Of course influence and effect on behavior in the parent–child relationship is mutual and not all one way; the behavior and reactions of children affect parents and custodial adults and the way they carry out socialization in the family (Reid et al., 2002; Simons et al., 2004). Research on intrafamily process in the etiology of sibling violence has found that
of the three theoretical perspectives, social learning theory [compared to feminist and conflict theories] garnered the strongest, most consistent empirical support. Social learning theory directs attention to the behavioral consequences of interaction patterns in families, emphasizing that children tend to adopt behaviors they learn from their parents. . . . We find that witnessing arguments between parents and being involved in verbal conflicts with parents and siblings is related to higher levels of sibling violence. (Hoffman, Kiecolt, and Edwards, 2005:1124)
Research on Peers and Group Contexts in Crime and Delinquency
Differential Peer Association
Among the range of actions and choices directly affected by families are the adolescent’s choices of friends and peer associations. Sibling effects in part are peer effects, and delinquent tendencies learned in the family may be exacerbated, whereas prosocial influences of the family may be counteracted, by differential peer association (Lauritsen, 1993; Simons et al., 1994; Simons et al., 2004; Simons et al., 2016). As one moves from childhood to adolescence, family influences diminish and peer groups play an increasingly prominent role in learning conforming and deviant behavior. Jones and Jones (2000) saw the influence of peer groups as embedded in two types of social networks involved in the “contagious nature of antisocial behavior.” One is a “cohesion” network in which there is direct communication in the network, and the other is a “structural equivalence” network that “consists of people who occupy the same niche in society. . . . Because the members of a teenage group also communicate directly with one another, a teenage group exemplifies both kinds of network and can function in both ways as a conduit of antisocial socialization” (Jones and Jones, 2000:33; emphasis added).
Other than one’s own prior deviant behavior, the best single predictor of the onset, continuance, or desistance of crime and delinquency is differential association with conforming or law-violating peers (Loeber and Dishion, 1987; Loeber and Stouthamer-Loeber, 1986; Liu, 2003; Snyder et al., 2005; Jennings et al., 2010). More frequent, longer term, and closer association with peers who do not support deviant behavior is strongly correlated with conformity, whereas greater association with peers who commit and approve of delinquency is predictive of one’s own delinquent behavior. It is in peer groups that the first availability and opportunity for delinquent acts are typically provided. Virtually every study that includes a peer association variable finds it to be significantly and usually most strongly related to delinquency, alcohol and drug use and abuse, adult crime, and other forms of deviant behavior (Warr, 2002; Pratt et al., 2010). There is a sizable body of research literature that shows the importance of differential associations and definitions in explaining crime and delinquency.4 The impact of differential peer association on delinquent behavior is among the most fully substantiated and replicated findings in criminology. Only the well-known relationships of crime rates to basic sociodemographic variables such as age and sex are as consistently reported in the literature both in research in the United States and in other countries (Antonaccio et al., 2010).
In fact, Mark Warr contended that there is no factor in criminal and delinquent behavior as powerful as peer associations:
No characteristic of individuals known to criminologists is a better predictor of criminal behavior than the number of delinquent friends an individual has. The strong correlation between delinquent behavior and delinquent friends has been documented in scores of studies from the 1950s up to the present day . . . using alternative kinds of criminological data (self-reports, official records, perceptual data) on participants and friends, alternative research designs, and data on a wide variety of criminal offenses. Few, if any, empirical regularities in criminology have been documented as often or over as long a period as the association between delinquency and delinquent friends. (Warr, 2002:40)
Warr (2002) documented these statements by reviewing the large body of research on the group character of, and the role of peers in, crime and delinquency in the United States and societies around the world. Peer influence is most noticeable and strongest in adolescence but is not restricted to the adolescent years. The vast majority of offenses by adolescents are committed in the company of others. These tend to be small groups and to be homogeneous by race, age, and gender. Warr (2002) proposes several mechanisms of peer influence consistent with differential reinforcement (e.g., “fear of ridicule” as negative reinforcement and “status” and “excitement” as positive reinforcement), imitation, and definitions (e.g., “group norms” and “dilution of moral responsibility”). Recent research has also demonstrated that delinquent peer effects on criminal involvement are observable for both virtual peer groups and face-to-face peer groups (Miller and Morris, 2016).
In longitudinal research on friendship networks, Haynie (2002) found strong support for the social learning principle of the balance or ratio of association with delinquent and nondelinquent friends. The higher the proportion of one’s friends who were delinquent, the greater the likelihood that he or she would become delinquent. Youth with only friends who had engaged in delinquency were twice as likely, and those with some delinquent friends were also much more likely, to commit delinquent acts than those youth who had only nondelinquent friends. This effect of the relative number of delinquent and nondelinquent friends remained strong even taking into account prior delinquency, time spent with peers, attachment to peers, and other characteristics of the network of friends. Peer influence is nearly always found in research, but the impact of that influence may vary; for instance, somewhat older peers may have greater influence than younger ones, and some may be more susceptible to peer influence than others (Harding, 2009; Miller, 2010). This peer influence is found not only on the overall frequency of one’s delinquency but also in the tendency for some to “specialize” in particular types of delinquent behavior such as violence, theft, or substance use (Thomas, 2015).
Delinquent Gangs
One special context of peer association is participation in delinquent gangs. Although delinquent gangs account for only a portion of all group-related delinquency (Warr, 2002), delinquent gangs and subcultures have received a great deal of attention in criminology for a long time (see Chapter 9). Research continues to find a significant correlation between delinquent peers and gang membership (Smith, Gomez Auyong, and Ferguson, 2019) and a strong influence of gang membership on serious delinquency. Curry, Decker, and Egley (2002) found that gang membership greatly increases involvement in both self-reported and official delinquency; even nonmembers who have only marginal association with gangs are more delinquent than youth who have no involvement at all with gang activities. Battin et al. (1998) found that, controlling for prior delinquency, adolescents with delinquent friends are more likely to engage in delinquent conduct and come before the juvenile court on delinquency charges, even if they are not part of a gang. But they are even more likely to do so if they and their friends are members of an identified delinquent gang. Whatever the frequency and seriousness of one’s previous delinquency, joining a gang promotes an even higher level of his or her delinquent involvement, in large part because “group processes and norms favorable to violence and other delinquency within gangs subsequently encourage and reinforce participation in violent and delinquent behavior” (Battin et al., 1998:108). In addition, Winfree et al. (Winfree, Vigil-Backstrom, et al., 1994; Winfree, Mays, et al., 1994) found that both gang membership itself and delinquency (gang-related as well as non-gang delinquency) are explained by social learning variables (attitudes, social reinforcers or punishers, and differential association). This is true even controlling for “personal-biographical characteristics, including ethnicity, gender, and place of residence” (Winfree, Vigil-Backstrom, et al., 1994:167). The processes specified in social learning theory are “nearly identical to those provided by qualitative gang research. Gang members reward certain behavior in their peers and punish others, employing goals and processes that are indistinguishable from those described by Akers” (Winfree, Vigil-Backstrom, et al., 1994:149).
These findings suggest that, compared to having one or more non-gang delinquent friends, gang membership produces more frequent, intense, and enduring association with delinquent friends, exposure to delinquent models and definitions, and reinforcement for delinquent behavior. Similar conclusions have been made for the effects of gang membership on victimization as well (see Gibson et al., 2009, 2012; Gover et al., 2009). Moreover, Gagnon (2018) has recently extended social learning theory to explain victimization among gang and ex-gang members. She operationalized victimization as three separate indexes constructed by summing the number of yes responses to the specific forms of victimization within each index: (1) violent victimization—robbery, attacked with a weapon, attacked without a weapon, raped, stabbed, shot, shot at, or threatened with a weapon, during and after the duration of their membership; (2) property victimization—theft and vandalism, during and after the duration of their membership; and (3) gang-related victimization—carjacking, home invasion, drive-by shooting, and witness intimidation, during and after the duration of their membership. Social learning theory was measured by questions tapping into three of the four elements of the social learning process (i.e., definitions, differential reinforcement, and differential association). Gagnon’s analysis of the data collected from 2,500 inmates (approximately 300 of which claimed gang membership) in 14 jails in the state of Florida illustrated support for extending social learning theory to explain victimization among gang and non-gang members, net of the effects of demographic control variables including gender, race, and age.
Selection and Socialization in Peer Association
Some have argued that any research finding that shows a relationship between delinquency and differential peer association either in general or with regard to delinquent gangs shows nothing about peer influence on delinquency. Rather, the findings result from the fact that with whom one interacts is a consequence, not a cause, of one’s own delinquent behavior. Whatever relationship is found is dismissed as simply a case of “birds of a feather flock together.” Association with delinquent peers takes place only or mainly after peers have already independently established patterns of delinquent involvement; one first becomes delinquent and then seeks out other delinquent youths as friends. No deviance-relevant learning of any kind takes place in peer groups, and, therefore, association with delinquent friends has no effect on the onset or acceleration or on the continuation or cessation of delinquent behavior (Hirschi, 1969; Gottfredson and Hirschi, 1990; Sampson and Laub, 1993; Brauer, 2009). This issue is often presented as a
selection model
in which delinquent friendships are formed around similarity in behavior or character versus a
socialization model
in which delinquency is learned through peer association. This distinction is often made with the assumption that social learning theory strictly adheres only to the latter. This critique mischaracterizes the process spelled out in social learning theory. Although the theory’s focus is on the effects that association (and other variables) have on criminal and delinquent behavior, as has already been shown, reciprocal relationships between one’s own conduct and association with friends is clearly recognized in social learning theory. Therefore, the fact that delinquent behavior may precede the association with delinquent peers or that friendship selection in part may be based on behavioral homophily does not contradict the theory. “Social learning admits that birds of a feather do flock together, but it also admits that if the birds are humans, they also will influence one another’s behavior, in both conforming and deviant directions” (Akers, 1991:210). It would contradict the theory if research demonstrated that the onset of delinquency always or most often predates interaction with peers who have engaged in delinquent acts. It would not support the theory if the research evidence showed that whatever level of delinquent behavioral involvement preceded association with delinquent peers stayed the same or decreased rather than increased after the association with delinquent peers.
However, research has not produced such findings contradicting the theory. Instead, the findings favor the process proposed by social learning theory, which recognizes both direct and reciprocal effects and selection and socialization effects in peer interaction. That is, early social learning in the family affects both the child’s behavior prior to peer influences and (along with other factors) the choice of friends. The peer process in adolescent deviance then is that a youngster associates differentially with peers who are deviant or tolerant of deviance, learns definitions favorable to delinquent behavior, is exposed to deviant models and social reinforcement of delinquency, then initiates or increases involvement in that behavior, which then is expected to have some influence on further associations, with these associations having further effect on his or her behavior (Kandel, 1978; Andrews and Kandel, 1979; Krohn et al., 1985; Sellers and Winfree, 1990; Kandel and Davies, 1991; Warr, 1993; Esbensen and Huizinga, 1993; Thornberry et al., 1994; Menard and Elliott, 1994; Winfree, Vigil-Backstrom, et al., 1994; Akers and Lee, 1996; Esbensen and Deschenes, 1998; Battin et al., 1998; Gordon et al., 2004).
Kandel and Davies (1991:442) noted that “although assortative pairing plays a role, longitudinal research that we and others have carried out clearly documents the etiological importance of peers in the initiation and persistence of substance use.” Warr (1993) also referred to the considerable amount of research evidence showing that peer associations precede the development of deviant patterns (or increase the frequency and seriousness of deviant behavior once it has begun) more often than involvement in deviant behavior precedes associations with deviant peers. The reverse sequence also occurs, and Warr proposed that the process is “a more complex, sequential, reciprocal process: Adolescents are commonly introduced to delinquency by their friends and subsequently become more selective in their choices of friends. The ‘feathering’ and ’flocking’ . . . are not mutually exclusive and may instead be part of a unified process” (Warr, 1993:39). This is completely consistent with social learning theory:
A peer “socialization” process and a peer “selection” process in deviant behavior are not mutually exclusive, but are simply the social learning process at different times. Arguments that . . . any evidence of selective mechanisms in deviant interaction runs counter to social learning theory . . . are wrong. (Akers, 1998:56)
Of course the social learning theory of deviance is not focused on explaining what causes or guides the selection of friends. Thus research findings that show principally peer “socialization” effects with little “selection” effects would be more supportive of the theory, while research finding the opposite would not be supportive (Akers, 1998). Menard and Elliott (1994; Elliott and Menard, 1996) found some reciprocal effects but
in the typical sequence of initiation of delinquent bonding and illegal behavior, delinquent bonding (again, more specifically, association with delinquent friends) usually precedes illegal behavior for those individuals for whom one can ascertain the temporal order. (Menard and Elliott, 1994:174)
These results are strong enough to indicate that serious forms of delinquent behavior rarely, if ever, precede exposure to delinquent friends. Instead, in the vast majority of cases, exposure precedes index offending. (Elliott and Menard, 1996:43)
We found that exposure to delinquent peers preceded minor delinquent behavior in a majority of cases, and serious delinquency in nearly all cases where some order could be determined. . . . Having delinquent friends and being involved in delinquent behavior may influence one another, but the influence is not symmetric; the influence of exposure on delinquency begins earlier in the sequence, and remains stronger throughout the sequence, than the influence of delinquency on exposure. (Elliott and Menard, 1996:61–62)
Research tends to find both selection and socialization effects in peer associations (Kandel, 1996; Krohn et al., 1996; Matsueda and Anderson, 1998; Gordon et al., 2004; Osgood et al., 2013). However, the evidence is that delinquency is more probable for those in differential association with deviant peers, that whatever level of delinquent involvement already exists is increased through differential association with delinquent peers, and that “although many investigations offer evidence of reciprocal effects, no study yet has failed to show a significant effect of peers on current and/or subsequent delinquency” (Warr, 2002:42). Further, experimental studies have shown that an individual’s behavior can be changed by peer influence in a deviant direction in small groups formed by the experimenters (Gardner and Steinberg, 2005; Cohen and Prinstein, 2006). There was no choice of friends in these studies (and they probably underestimate the influence that would be found in naturally occurring groups of friends); therefore, selection could not have produced the similarity in behavior.
McGloin (2009) reported longitudinal data that similarity in peer behavior comes about because adolescents tend to change their own behavior to bring about (Scott-Parker et al., 2012) a greater “balance” between the level of their delinquent involvement and that of their best friends. Her analysis showed that
this tendency toward balance is not caused by selection. . . . [T]hese data provided no evidence that relative peer deviance predicted dimensions of friendship stability. This finding suggests that when confronted with delinquency incongruence with a best friend, a person is apt to change his or her behavior to be more like the friend rather than to end or change the status of the relationship. In this way, this investigation joins others in asserting that peers “matter” and offers a somewhat novel consideration of how and why they matter. (McGloin, 2009:467)
Monahan, Steinberg, and Cauffman (2009), using longitudinal data, examined socialization and selection effects at various stages of development: middle adolescence (ages 14–15), late adolescence (ages 16–20), and young adulthood (ages 21–22). They found that selection effects, in which antisocial behavior predicted peer delinquency, were restricted only to middle adolescence, but socialization effects, in which peer delinquency predicted later antisocial behavior, were in effect in both middle and late adolescence. By young adulthood, however, peer delinquency and antisocial behavior were unrelated, likely due to a concomitant increase in resistance to peer influence during this later period of development.
Dijkstra et al. (2010:207), in a study of weapons carrying among lower class Hispanic adolescents, concluded, “This reasoning favors an influence effect and goes against a direct-selection effect, which states that those who carry weapons flock together. The results of our study showed that peers indeed influenced the weapon carrying of adolescents; no support was found for the selection mechanism.” Similarly, Weerman (2011) found that friends’ behavior and changes in group membership had significant effects on one’s delinquency but that one’s delinquency had no significant net effects on selection of friends.
Recent research has begun to further investigate the “birds of a feather flock together” phenomenon, but from a “behavioral homophily” perspective (Boman, 2017; Boman and Mowen, 2018; McPherson, Smith-Lovin, and Cook, 2001). The behavioral homophily concept implies that individuals who engage in crime will tend to cluster together in peer groups, whereas individuals who do not engage in crime will tend to develop friendships and peer groups that are distinct from the peer groups formed by those engaged in crime (Boman and Mowen, 2018:30). Using data from pairs of friends, Boman and Mowen (2018:36) found that behavioral homophily was associated with less crime, or, in other words, “it appears that ‘birds of a feather tend to flock together’ more for normative behaviors than they do deviant behaviors.” Additional results clarified that the evidence illustrated that there were two distinct components of behavioral homophily (i.e., deviance homophily and non-deviance homophily), and that deviance homophily was positively associated with crime whereas non-deviance homophily was negatively associated with crime.
Perception and Behavior in Peer Association
Others have argued that the strong relationship between self-reported delinquency and peer associations is entirely a methodological artifact that results from measuring peer influence by relying on the individual respondents to report on how many or what proportion of their friends have been involved in delinquency. Such reports cannot be valid because they are filtered through the respondent’s perceptions of what friends are doing, and that is colored by their own behavior. They are strongly related because they are really measuring the same construct twice. It is the same underlying delinquent tendency, whether youngsters are asked about the delinquency of their friends or about their own delinquency. But research shows that the two are not the same and that the respondent’s reports of friends’ behavior are not simply a reflection of one’s own delinquent behavior (Agnew, 1991a; Warr, 1993; Thornberry et al., 1994; Elliott and Menard, 1996; Bartusch et al., 1997). Although the relationship is not as strong, the major impact of differential peer association on the adolescent’s own behavior continues to be found in research that measures the proportion of delinquent friends by asking the friends themselves about their delinquency, rather than relying on the adolescent’s report or perception of the friends’ behavior (Haynie, 2002; Weerman and Smeenk, 2005; Meldrum, Young, and Weerman, 2009; Thomas, 2015). Although both are reasonable indicators of differential peer association, the latter, perceptual measure is more relevant and accurate for testing peer influence on one’s own behavior, because “even if peer behavior is misperceived as more (or less) delinquent than it actually is, the peer influence will still come through that perception” (Akers, 1998:119). Recent research runs counter to this hypothesis and finds that one’s perception of peers’ behavior does not influence his or her delinquent behavior (see Young et al., 2014). However, other research (Thomas, 2015) finds that one’s behavior is significantly influenced by both the person’s perception of friends’ behavior and friends’ behavior as self-reported by friends.
Research on Social Structure and Social Learning
A number of studies have also been done that provide evidence on the validity of the SSSL model. The correlations of sociodemographic variables such as age, sex, race, and class (which are indicators of the person’s differential location in the social structure) with various forms of deviant behavior have been found to be reduced when the social learning variables are taken into account. Differences in crime and deviant behavior across communities and even nations can be seen as the result of differential social organization, and the SSSL hypothesis is that such macro-level differences would be mediated by variables in the social learning process. That is what has been found for differences in levels of marijuana and alcohol use among adolescents in four types of communities and for differences in overall levels of drinking behavior among the elderly in four types of communities (see Akers, 1998) as well as cross-national variation in homicide rates (Jensen and Akers, 2003).
Another indicator of differential social organization is the level of social disorganization across neighborhoods and community areas. The theoretical expectation from SSSL is that the higher rates of violence, crime, and delinquency often found in socially disorganized neighborhoods, with low informal social control, poverty, and few social resources, occur primarily because adolescents residing in such neighborhoods are more apt to associate with delinquent peers, to be exposed to behavioral models of delinquent and criminal behavior, and to have their delinquent behavior differentially reinforced. Haynie, Silver, and Teasdale (2006:148, 162) found strong support for this hypothesis with regard to “violent peer networks” in socioeconomically disadvantaged urban neighborhoods, reporting that when peer associations were taken into account, “the effects of neighborhood disadvantage and immigrant concentration [were reduced] to non-significance.” Similar findings were reported by Gibson, Poles, and Akers (2010), whose research also showed that the effects of concentrated social and economic neighborhood disadvantage on high rates of violent delinquency were substantially mediated by differential peer association. Bellair, Roscigno, and Velez (2003:219) found that the effects of concentrations of low-wage service employment along with economic disadvantage, sociodemographic composition, and other structural community variables on adolescent violence “are mediated almost entirely” through the social learning process. Bellair and McNulty (2009) later reported that the higher levels of violence in disadvantaged neighborhoods involved the social learning process whereby living in those neighborhoods affected individuals’ gang participation and greater exposure to violent behavioral models, association with others who support procriminal definitions, and social reinforcement of violent behavior.
Lee, Akers, and Borg (2004) tested explanatory models containing all of the major social learning variables and the major dimensions of social structure specified in SSSL, finding that “variations in the behavioral and cognitive variables specified in the social learning process . . . mediate substantial, and in some instances virtually all, of the effects of gender [although mediation of gender was weaker, gender effects remained significant], socio-economic status, age, family structure, and community size on these forms of adolescent deviance” (Lee et al., 2004:29). Lanza-Kaduce and Capece (2003) examined variations in binge drinking among students in eight universities as a test of SSSL. Only two social learning variables were measured in the study (definitions and reinforcement), and while these variables did not substantially account for the variations across institutions, they did mediate other social structural factors of gender, race, fraternity/sorority membership, and the normative climates within universities; these “results support Akers’ mediation hypothesis” (Lanza-Kaduce and Capece, 2003:195). Holland-Davis (2006) tested a full SSSL model and found that the effects of the structural variables on adolescent substance use were substantially mediated through social learning. Jensen (2003) had earlier found that gender differences in both minor and serious delinquency were reduced to nonsignificance by inclusion of social learning variables in the models. However, Holland-Davis reported that although gender was somewhat mediated by the social learning variables, it retained significant net effects, suggesting that there may be “moderated mediation” relationships between social learning and social structure, at least with regard to gender. Similarly, Verrill (2008:139–140) suggests that the SSSL model may need to be modified in the direction of “moderation-mediation.” His research demonstrated that “social learning does relate to the social structural variables and their impact on delinquency,” but this relationship involves some moderation of social learning by social structure. Research on the SSSL model conducted in South Korea, which also tested the mediating effects of other social psychological variables taken from social bonding and self-control theories, found that the strongest mediating effects came from the social learning variables (Kim, Akers, and Yun 2013; Whaley et al., 2011).
The research results are favorable toward the main SSSL hypothesis that the variables specified in social learning theory are the principal social psychological (micro) variables that substantially mediate the relation of social structural (macro) variables to crime. However, there are findings that do not show this; some structural variations are mediated more than others, and there is some evidence of moderation in addition to mediation effects. (For information on the differences between mediation and moderation, see Baron and Kenny, 1986.) This seems to be found more with regard to gender as an indicator of location in the social structure. While research has found social learning mediation of gender effects in delinquent and criminal behavior, it is often not as substantial as proposed by the theory. “Main gender effects on crime seem to be more robust in the face of mediating social psychological variables (including social learning variables) than race, class, or age. For some kinds of deviance, non-trivial direct effects of gender may remain even after inserting learning or other social psychological variables into the model” (Akers, 2006:154). The question of how to account for gender differences in crime is, of course, one of the key issues to which feminist theory is addressed (see Chapter 13).
APPLICATIONS OF SOCIAL LEARNING THEORY IN PREVENTION AND TREATMENT PROGRAMS
If criminal and delinquent behavior is acquired and sustained through social learning processes in naturally occurring environments, then it should be possible to modify that behavior to the extent that one is able to manipulate those same processes or the environmental contingencies that impinge on them. This is the underlying assumption of prevention and treatment programs that have relied on the application of social learning principles. Reliance on one or more of the explanatory variables in social learning theory forms the implicit or explicit theoretical basis (often in combination with guidelines from other theories) for many types of group therapies and self-help programs; positive peer counseling programs; gang interventions; family and school programs; teenage drug, alcohol, and delinquency prevention and education programs; and other private and public programs. Behavior modification programs based on cognitive/behavioral learning principles, including both group and individually focused techniques for juveniles and adults, are operating in correctional, treatment, and community facilities and programs and in private practice (see Morris and Braukmann, 1987; Ellis and Sowers, 2001; Pearson et al., 2002; Andrews and Bonta, 2003; Hersen and Rosqvist, 2005; Akers, 2010). Some examples, both historical and contemporary, will serve to show how social learning principles have been put into practice.
Highfields and Essexfields
One of the first systematic uses of created prosocial peer groups to change delinquent behavior was in the Highfields project (Weeks, 1958). Highfields was an alternative treatment program for delinquent boys, who were allowed to go to school, work, and participate in other activities during the day and return to the Highfields residential facility at night. There was no education, vocational, or individual counseling program. The principal activity was regular participation in
Guided Group Interaction (GGI)
sessions. These were peer groups (guided by adult staff) in which common problems could be discussed in a group atmosphere that created and encouraged nondelinquent attitudes and behavior. At the end of the program, the boys had developed attitudes more favorable to obeying the law, and those who changed their attitudes the most were more likely to succeed and stay out of trouble later. However, the changes were not great and were more noticeable among the black than the white youth in the program. The Highfields boys did somewhat better than a comparison group of boys, who had been committed to the state reformatory school (even when adjustments for in-program failures were made), in avoiding reinstitutionalization, but again this primarily was due to the differences observed among the black youth (Weeks, 1958). A program called Essexfields later built on the Highfields experience using similar peer group sessions to affect changes in delinquent behavior and attitudes but in a nonresidential setting (often in a juvenile probation office). The Essexfields boys were not more successful in staying out of future trouble than those who were either given regular probation supervision or assigned to the residential program at Highfields. But they did do somewhat better than the comparison group of those who had been sent to the state reform school (Stephenson and Scarpitti, 1969).
The Pinehills Experiment
LaMar T. Empey’s delinquency treatment experiment in Provo, Utah (Empey and Erickson, 1972), offered a semiresidential alternative to regular juvenile probation and to incarceration in a state training school for delinquent boys. Empey instituted a peer-group program with officially adjudicated delinquents in a group home facility known as Pinehills. He drew on differential association theory and applied techniques similar to the GGI of Highfields. With guidance from adult counselors, a prosocial, antidelinquent peer culture was cultivated. The Pinehills boys had the responsibility and authority to form groups, orient new boys coming in, set standards and rules for behavior, determine punishment for rule violation, and decide when a boy was ready to be released from Pinehills. The boys gained recognition and status in the group for conforming rather than antisocial behavior. The intent was to use this peer-group interaction to foster the boys’ definitions favorable to conforming behavior and peer influence to motivate them toward conformity and away from delinquency.
The Pinehills project was designed to have court-adjudicated delinquent boys randomly assigned to the Pinehills facility, to the secure-custody state training school, or to a regular juvenile probation caseload in the community. However, the experimental design did not last because the juvenile court judge began purposely assigning boys to Pinehills once he learned of the strong positive effects demonstrated by the program. The boys who had participated in the peer groups at Pinehills were very much less likely to be repeat offenders (27% recidivism among all the boys initially assigned and 16% recidivism among those completing the program) than those boys randomly committed to the state institution (60% recidivism) six months after release. Even four years later, recidivism of the Pinehills boys was still less than half that of a comparison group released from the state training school.
The Pinehills recidivism rate was also much better than the recidivism rate (about 45%) that had prevailed among delinquents placed on regular juvenile probation supervision prior to the Pinehills project. However, during the time the project was in operation, the boys assigned to the regular probation group experienced about the same level of recidivism as the Pinehills boys. This dramatic improvement over the preproject recidivism may have resulted from the tendency of the regular probation officers, who knew their caseloads were being compared with Pinehills’s, to reduce revocations and to intensify their efforts with the boys under their supervision. The Pinehills project ended prematurely due to opposition from county officials and received no additional public funding when its private funds ran out (Lundman, 1993).
The Teaching Family Model
Later community-based residential programs moved beyond the almost exclusive reliance on the peer group of the Highfields and Pinehills projects to create more of a family environment. They also explicitly applied the principle of differential reinforcement to modify behavior. The best examples are Achievement Place (in the 1960s) and its successor, the Teaching Family model (since the 1970s). The Teaching Family group homes became “perhaps the most systematic, and certainly the most long-lived and widely disseminated, application of the behavioral approach with juvenile offenders” (Braukmann and Wolf, 1987:135). The Teaching Family model involves a married couple (“teaching parents”) and six to eight delinquent or “at-risk” youth living together as a family. A token economy is in effect in which the youth can earn reward points by proper behavior or have points taken away for improper behavior in the home or at school. The parents are responsible for teaching social, academic, and prevocational skills and maintaining mutually reinforcing relationships with the adolescents. But the youths in the home also operate a peer-oriented self-government system. Thus, in addition to the shaping of behavior by the teaching parents, the Teaching Family model promotes conforming behavior through exposure to a prosocial peer group.
Studies have shown this model works quite well to maintain good behavior and inhibit misconduct and delinquent behavior in the school and community while living with the teaching family. However, outcome evaluations in many states have found little difference in subsequent delinquent behavior between those who had been in the Teaching Family group homes and youth in control groups who had participated in other types of programs. The delinquency-inhibiting effects on the adolescents’ behavior did not survive their release from the behavior modification system of the Teaching Family group homes back to their previous family and neighborhood environments (Braukmann and Wolf, 1987). Questions have been raised about this conclusion; the methods used in the evaluation research have been criticized, and at least one analysis has found some evidence of lower recidivism one year after release from Teaching Family treatment (Kingsley, 2006).
Oregon Social Learning Center
Gerald R. Patterson and his colleagues at the OSLC (Patterson, 1975; Dishion, Patterson, and Kavanagh, 1992; Patterson and Chamberlain, 1994; Snyder and Patterson, 1995; Dishion, McCord, and Poulin, 1999; Reid et al., 2002; Snyder et al., 2005; Forgatch et al., 2009; Forgatch and Kjøbli, 2016) have long been conducting research on social learning theory and applying it in programs with families and peer groups. Their “coercive family” model proposes that the child’s behavior and interaction with others are “learned in the family, and under more extreme conditions carries over to a child’s interactions with others outside the family, including peers and teachers” (Dishion et al., 1992:254–255):
Poor parent discipline practices increase the likelihood of child coercive responses, and high rates of child coercion impede parents’ attempts to provide evenhanded, consistent, and effective discipline. It is in this sense that parental limit setting for behaviors such as lying, stealing, or fighting often fail in the quagmire of the child’s arguments, excuses, and counteraccusations. (Dishion et al., 1992:258)
The OSLC Adolescent Transition Program involves targeting family management skills in parent-focused and parent/teen groups as compared to teen-focused groups and self-directed change. In the parent groups, several sessions are held with a therapist to help parents develop monitoring, discipline, problem-solving, and other effective socialization and disciplinary skills through instruction, discussion, and role playing. “Social learning parent training is a step-wise, skill-based approach for developing effective parenting skills and strategies for maintaining change” (Dishion et al., 1992:263). The teen-focused group and individual sessions are run for at-risk youth in the preteen and early adolescent years (ages 10–14). The goals of these sessions are to help develop and improve communication skills, self-control, prosocial attitudes, and prosocial peer associations. Evaluation research has shown improvements in parenting skills and reductions in antisocial behavior among the youth in the program (Dishion et al., 1992). However, intervention groups that include older delinquents can increase, rather than decrease, deviant behavior of the youths (Dishion et al., 1999).
Similar cognitive and behavioral principles are applied in another successful OSLC program, the Multidimensional Treatment Foster Care (MTFC), designed to change the behavior of youth who have been involved in chronic, high frequency, and serious delinquencies adjudicated by the juvenile court (Chamberlain, Fisher, and Moore, 2002; Eddy and Chamberlain, 2000). The adolescents in the treatment program are assigned to foster care families (one or two per family) based on the court finding that, although the offenders are eligible for community rather than institutional placement, return of the youth to their own families is not feasible because their parents are no longer able to control them. Competent foster parents were recruited and trained to use “behavior management methods . . . to implement and maintain a flexible and individualized behavior plan for each youth within the context of a three-level point system that made youth privileges contingent on compliance with program rules and general progress” (Eddy and Chamberlain, 2000:858). This training equips the foster parents “to notice and reinforce youngsters” for proper behavior in a positive way. In addition, each youth takes part in weekly sessions with behavioral therapists “focused on skill building in such areas as problem solving, social perspective taking, and nonaggressive methods of self-expression” (Chamberlain et al., 2002:205–206). The foster home placement, school attendance and homework, therapy sessions, and other aspects of the intervention are coordinated and supervised by case managers. After the youth complete the program they are returned to their biological or stepparents who also receive support and training in parenting skills.
Outcome evaluation has shown that the delinquents in the MTFC treatment group had lower rates of both official and self-reported delinquency than comparable youth placed in other community treatment programs. In a random-assignment study, boys in the MTFC were found to have significantly fewer arrests and self-reported delinquencies than boys assigned to group care homes one year after release from the programs and continued to have fewer arrests for up to three years later (Chamberlain et al., 2002). Research also shows that the treatment effects were mediated by family and peer association variables. That is, the reduction in and desistance from further delinquent behavior were achieved because the program “caused levels of family management skills [parental supervision, discipline, and positive reinforcement] to increase and deviant peer associations to decrease” (Eddy and Chamberlain, 2000:858). Positive differences in MTFC outcomes between treatment and control groups continued to be found when the participants had reached early adulthood (Rhoades et al., 2014).
Linking the Interests of Families and Teachers (LIFT) is an OSLC delinquency-prevention project utilizing social learning principles. This program is designed to affect troublesome childhood behavior that is often a precursor to delinquent and violent behavior in adolescence. Although high-risk areas of the community are targeted, the approach is a “universal strategy” that provides services to all first and fifth graders (as the transitional grades into elementary and middle schools, respectively), their parents, and teachers in those areas rather than attempting to identify individual youth at risk and singling them out for preventive intervention. LIFT is focused on modifying the interaction of the children with their social environment at home, with peers, and in school. Theoretically, “the driving force in a child’s conduct problems are the reinforcement processes that occur in his or her day-to-day relationships . . . [A]ntisocial behavior develops as an interaction between the child’s antisocial behavior and the reactions of those with whom he or she interacts on a daily basis” (Reid and Eddy, 2002:222). With the cooperation of the schools and voluntary involvement of the children and families, “the three major components of the LIFT are (a) classroom-based child social and problem skills training, (b) playground-based behavior modification, and (c) group-delivered parent training” (Eddy, Reid, and Fetrow, 2000:165).
The classroom interventions involve 30-minute sessions twice a week for 10 weeks with lectures, group discussion, and practice of listening, cooperation, problem-solving, and other skills promoting prosocial behavior with peers (as well as study skills for the fifth graders). The approach on the playground following the classroom sessions is to have the children participate in a “good behavior” game with their behavior observed by project monitors. In this game, the children are rewarded by the monitors for good behavior (e.g., prosocial behavior, sharing, cooperation) and for refraining from aggressiveness or other antisocial acts toward peers. The reward points earned by accumulation of armbands, minus points deducted for observed displays of antisocial behavior, are later exchanged for group rewards. The parent training part of the program attempts to aid parents (through six weekly 1½-hour evening sessions held in school rooms with groups of 15 parents) to develop or strengthen “consistent and effective positive reinforcement, discipline, and monitoring skills” (Reid and Eddy, 2002:224). These sessions include video presentations and role playing with reading and practice at home. Group leaders monitor progress in the families weekly, and a voice messaging system is installed in the classrooms for daily communication between parents and teachers.
There were some modest to strong effects of the program almost immediately on reductions in physical aggression on the playground and improved behavior in the classrooms. Also, in the longer term, by the time the fifth graders were in middle school and the first years of high school, the LIFT participants had experienced significantly fewer police arrests, lower levels of self-reported drug use, and fewer associations with deviant peers (as observed by the teachers) than did the youth in the control group.
Other OSLC programs, such as the early intervention parental management training program for single parents and their at-risk elementary schoolchildren involving parental training and efforts to limit association with deviant peers, have also shown some success. A nine-year follow-up showed reduced delinquency as indicated by teachers’ reports and by arrests in the treatment group compared to the control group (Forgatch et al., 2009).
Andrews’ Experiments and Model of Treatment and Prevention
Donald R. Cressey (1955) proposed that both juvenile and adult treatment and reformation groups, whether naturally occurring or deliberately organized as an intervention technique, are based on the process of differential association. Cressey proposed that the effects of such a process come from what he termed
retroflexive reformation
. By agreeing to participate in a group to help reform others, the criminal comes to accept the prosocial purposes and values of the group while becoming alienated from more procriminal values and thereby affecting his or her own balance of definitions unfavorable to criminal behavior more than he or she affects others’ definitions.
Retroflexive reformation may have been operating along with other social learning processes in a pioneering prison and probation program designed and evaluated by D. A. Andrews (1980). Andrews applied three major principles in a program to change attitudes (definitions) and behavior of convicted adult offenders: (a) contingency (differential association and reinforcement in groups led by prosocial volunteers), (b) quality (modalities) of interpersonal relationships in those groups, and (c) self-management (a program of self-monitoring and self-reinforcement by the individual inmates). Manipulation of the associations and reinforcers produced significant changes toward less adherence to criminal definitions and greater acceptance of noncriminal definitions. Similarly, the more skilled the outside volunteers who worked with the groups were in creating cohesive groups (based on the principle of intensity of primary group relationships), the more likely the inmates were to develop anticriminal attitudes. Adult probationers in the program developed greater respect for the law and were less likely to repeat criminal offenses. Prisoners assigned to self-management and prosocial-attitude groups also changed their own attitudes in a law-abiding direction and had lower recidivism rates.
Andrews has continued to emphasize the relationship and
contingency principle
s in a more fully developed PIC–R (personal–interpersonal and community-reinforcement) model applied to treatment and prevention of criminal and deviant behavior. In this model, the
relationship principle
refers to the process whereby “interpersonal influence by way of antecedent and consequent processes is greatest in situations characterized by open, warm, and enthusiastic communication and by mutual respect and liking.” The contingency principle refers to “the procriminal/anticriminal content of the messages communicated or by the procriminal/anticriminal nature of the behavior patterns that are modeled, rehearsed and subject to reinforcement and punishment contingencies” (Andrews and Bonta, 1998:273).
Meta-Analyses of Cognitive-Behavioral Programs
Andrews and his associates have carried out their own research and conducted a meta-analysis of the literature to gauge the effectiveness of behaviorally oriented models such as his as well as various other cognitive-behavioral treatment and prevention strategies. They have found that social behavioral strategies utilizing the family and peers are the most effective, especially when targeted specifically to pro- or anticriminal attitudes and behavior rather than general psychological states such as self-esteem. The “cognitive behavioral and social learning approaches” are more effective than “nondirective relationship-oriented counseling or psychodynamic insight-oriented counseling” and “behavioral treatments had a substantially greater average effect [mean effect size of 0.25] on recidivism than did non-behavioral treatments [mean effect size of 0.04]” (Andrews and Bonta, 1998:262–263, 267–268, 282–290). Lipsey, Landenberger, and Chapman (2007) also found that on average cognitive-behavioral programs for criminal offenders resulted in a 25% reduction in recidivism. Other recent meta-analyses confirm the greater effectiveness of cognitive-behavioral drug abuse treatment programs in correctional facilities, which are consistent with “social learning theory [that] is broader than behavioral reinforcement theory because it includes as variables cognitions, verbalization, and social modeling to explain (and to change) behavior patterns” (Pearson et al., 2002:480; see also Lipsey and Landenberger, 2005; Gendreau, Smith, and French, 2006):
The core criteria of successful programs in developmental prevention are similar to those in offender treatment. For example, such programs have a sound theoretical basis in social learning theory, follow a cognitive-behavioral approach, are well structured and address multiple risk and protective factors. (Losel, 2007:4)
Research on correctional programs supplies strong and consistent support for theories—such as differential association/social learning theory—that link offending to antisocial associations and to the internalization of antisocial values. A consistent finding across meta-analyses, including cross-cultural studies, is that “cognitive-behavioral” programs tend to achieve higher reductions in recidivism than other treatment modalities. (Cullen et al., 2003:353)
Gang Resistance Education and Training
Gang Resistance Education and Training (G.R.E.A.T.) is a prevention program for students with the goals of promoting antigang attitudes, reducing involvement in gangs, and increasing positive relations with law enforcement, with the expectation that achieving these goals among the students would prevent delinquent behavior in the community. The program consists mainly of in-classroom sessions, including lessons on conflict resolution, resisting peer pressure, and the negative effect of gangs on quality of life in the community. It is entirely a cognitive-based program delivered in the classroom primarily by uniformed police officers without any direct intervention in gang activity or any components for specific behavioral change. Cross-sectional and four-year quasi-experimental evaluations of G.R.E.A.T. found that youths in the program developed more positive attitudes toward police and more negative attitudes toward gangs, lower levels of victimization, less risk-taking behavior, and more association with peers in prosocial activities. However, the research also showed that gang membership was not prevented by participation in the program and that self-reported delinquent behavior was not lower among the youths taking part in the G.R.E.A.T. sessions than in control groups of youths not participating in the program (Esbensen et al., 2001).
These disappointing results led to a revision of G.R.E.A.T. based on the advice of consultants with expertise in gangs and school-based prevention programs. There are now four “components” of the program: elementary school, middle school, summer school, and family. The school components focus on teaching the students about gangs, violence, and drugs and attempt to foster prosocial attitudes and skills as well as “refusal skills” and other techniques of avoiding joining gangs and other deviant peer groups. In the family component, the officers teach good parental/family skills; effective parental discipline and monitoring; intrafamily communication; prosocial role modeling; and control of access to television, movies, video games, and the Internet. The revised program also has added some other “trained and certified” instructors from federal and local criminal justice agencies.
The revised program has been subjected to a well-designed and rigorous short-term and long-term (four years follow-up) evaluation with findings of more positive outcomes than was true for the original program. The differences were not large, but youth who had been exposed to the program, compared to control groups, had significantly lower chances of joining a gang. However, no significant differences were found in overall differential association with prosocial and delinquent peers or in the extent to which the participants held prosocial attitudes. Preventive effects on delinquent behavior were weak, but the findings were encouraging enough for G.R.E.A.T. to be designated by federal agencies as a “promising” prevention program for gang membership and delinquent behavior (Esbensen et al., 2013).
It appears that the designers of G.R.E.A.T. recognize that differential peer association has effects on delinquency and in particular assume that lowering frequency and intensity of association with deviant peers by reducing gang involvement will prevent delinquency that would be consistent, in part, with social learning theory. Also, the program attempts to undergird social skills and prosocial attitudes that could counteract gang-favorable and delinquency-favorable attitudes and that would appear to echo the concept of definitions in social learning theory. However, both the original and the revised program descriptions are vague about the theoretical assumptions on which they are based. Although there are references to the Seattle Social Development Model and to Life Skills Training (Esbensen et al., 2013), neither social learning nor any other criminological theory is specifically delineated as providing the theoretical basis for the program activities. Some of it seems to be fashioned on social bonding theory (see Chapter 6), and there are aspects of G.R.E.A.T. that do not follow from, or even run counter to, what one would expect from social learning theory. What social learning principles, for instance, would be the basis for projecting that a police-led classroom curriculum (even after revision in the program to involve teachers more and to move away from a strictly didactic approach in the classroom) and family sessions would have the desired outcomes? Network analyses of gang membership and relationships suggests that (consistent with social learning theory) a potentially more effective approach, perhaps in addition to or in conjunction with, classroom instruction would involve targeting direct intervention with individual gang members, especially those who occupy central “cut points” (sole connecting links between different individuals and groups) in the gang networks (McGloin, 2005; see also McGloin et al., 2004).
Positive Behavioral Interventions and Supports (PBIS) Program
Recent program evaluations have documented the effectiveness of the school-wide Positive Behavioral Interventions and Supports (PBIS) program for reducing school suspensions and improving perceptions of school safety (Bradshaw, Koth, Thornton, and Leaf, 2009; Bradshaw, Pas, Goldweber, Rosenberg, and Leaf, 2012; Bradshaw, Waasdorp, and Leaf, 2012; Waasdorp, Bradshaw, and Leaf, 2012). The PBIS program is a universal, school-based program focused on establishing a social culture where students expect appropriate behavior from their peers and appropriate behavior is reinforced and inappropriate behavior is punished, principles consistent with social learning theory. The implications are that this cultural change will lead to a safer school environment and less school-related delinquency. The program is targeted toward elementary school populations, and involves seven elements summarized by Bradshaw and colleagues (2008:463):
A PBIS team is formed within the school to provide leadership regarding school-wide implementation of PBIS. The team, which is comprised of six to eight staff and an administrator attends annual training events, establishes an action plan for implementing PBIS, develops materials to support program implementation, trains other staff members, and meets monthly to discuss school-wide behavior management. A behavioral support “coach” provides on-site technical assistance regarding PBIS. The coach is typically a school psychologist or guidance counselor who has prior experience working with PBIS and conducting functional behavioral assessments. The school team establishes three to five positively stated school-wide behavioral expectations regarding student behavior (e.g., “be respectful, responsible, and ready to learn”) that are posted in all classrooms and nonclassroom settings, and are known by all students and staff. Plans are developed by the school staff for defining and teaching the school-wide behavioral expectations on a regular basis. A school-wide system is developed to reward students who exhibit expected positive behaviors. School staff establish and use a school-wide system for reinforcement that includes a tangible reinforcer and is used consistently by all school staff across all settings. Staff and administrators create an agreed upon system for responding to behavioral violations that includes definitions about what constitutes a classroom-managed versus an office-managed discipline problem. Students across all classrooms receive consistent consequences for disciplinary infractions. A formal system is developed to collect, analyze, and use disciplinary data (e.g., office discipline referrals, suspensions) to make decisions regarding program implementation.
Horner and colleagues (2009) conducted an early
randomized controlled trial
, a research design deemed to provide the most compelling scientific evidence for evaluating the effectiveness of a program, on the PBIS program. In this randomized controlled trial, 33 schools from two states (Illinois and Hawaii) were randomly assigned to receive the PBIS program and 30 schools were assigned as “waitlist control” schools. Their evaluation demonstrated that the PBIS program had a noticeably large effect on improving perceptions of school safety (d = -0.86). In a subsequent randomized controlled trial, Bradshaw et al. (2010) documented a small to medium effect (d = 0.27) for the PBIS program on school suspensions when comparing schools randomly assigned to receive the PBIS intervention versus those that did not receive the PBIS program.
Other Prevention Programs
Alcohol and drug education/prevention programs that employ various “social influence and skills” strategies and techniques reflect, to some extent, social learning assumptions. They are designed to teach about the influences of peers, media, and family that can affect drug-using and abstaining decisions and how to deal with and resist these influences.
Participation in such programs appears to have modest effects on reducing or preventing drug behavior. The most effective are peer programs that focus on learning prosocial and life skills while discouraging drug-favorable attitudes (Tobler, 1986; Akers, 1992; Botvin et al., 1995). Although they can claim some success, it should be noted that even the best programs have not had large effects. It has not been firmly established that they will produce long-term reductions in the prevalence of substance use, not only among white, middle-class youth but also among lower class black and Hispanic youth.
In their review of scientifically sound studies of crime prevention, Sherman et al. (1998) list a number of family and school programs under the category of “what works.” They conclude that the most effective programs in the prevention of delinquency, drug use, and conduct problems are those designed for high-risk youths that rely on cognitive-behavioral approaches. These include clarifying and communicating norms; reinforcing positive behavior; and training in social competency, life skills, thinking skills, and self-control using behavior modification techniques.
Community and school-based programs directed toward families, children, and youth often draw on principles from both learning and bonding theories (see Chapter 6) as well as from general strain theory (see Chapter 9). These range from Head Start and other generalized children-oriented programs to early-intervention programs specifically geared to prevent misconduct and delinquency (Greenwood, 1998). They attempt to reduce the “risk factors” and promote “protective factors” in the family, school, and community by providing social, economic, and learning resources to schoolchildren and families, often in minority, disadvantaged, and high-risk neighborhoods (Jessor, 1996; Zigler, Tausig, and Black, 1996). The most notable program that melds learning and bonding theories is the Social Development Model (SDM; Hawkins et al., 1992; Hawkins et al., 1999). The SDM is an early-intervention, long-term project that has demonstrated some success in preventing adolescent delinquency and misconduct. The SDM and research evaluating it are described in the next chapter (Chapter 6).
SUMMARY
Akers’s social learning theory combines Sutherland’s original differential association theory of criminal behavior with general behavioral learning principles. The theory proposes that criminal and delinquent behavior is acquired, repeated, and changed by the same process as conforming behavior. While referring to all parts of the learning process, Akers’s social learning theory in criminology has focused on the four major concepts of differential association, definitions, differential reinforcement, and imitation. That process will more likely produce behavior that violates social and legal norms than conforming behavior when persons differentially associate with those who expose them to deviant patterns, when the deviant behavior is differentially reinforced over conforming behavior, when individuals are more exposed to deviant than conforming models, and when their own definitions favorably dispose them to commit deviant acts.
This social learning explanation of crime and delinquency has been strongly supported by the research evidence. Research conducted over many years, including that by Akers and associates, has consistently found that social learning is empirically supported as an explanation of individual differences in delinquent and criminal behavior. Although there is not yet a large number of studies directly testing the SSSL model, the evidence thus far provides some support of, but also some reservations about, the main hypothesis that social learning processes substantially mediate the effects of social structural variables on criminal and deviant behavior. The behavioral, cognitive, and social interactional principles of social learning theory have been applied in a range of prevention and treatment programs, mainly with adolescent populations but also with adult offenders. The outcomes of these programs often have been disappointing, but there is also some evidence of success, and generally such programs are more effective than alternative approaches. Finally, promising results have recently been documented with school-based interventions such as the Positive Behavioral Interventions and Supports (PBIS) program.
KEY WORDS
principle of differential association
definitions
modalities of association
differential reinforcement
classical or “respondent” conditioning
discriminative stimuli
schedules of reinforcement
symbolic interactionism
interactional dimension
normative dimension
general definitions
specific definitions
negative definitions
positive definitions
neutralizing definitions
discriminative stimuli
positive reinforcement
negative reinforcement
modalities of reinforcement
social reinforcement
nonsocial reinforcement
self-reinforcement
vicarious reinforcement
selection model
socialization model
Guided Group Interaction (GGI)
retroflexive reformation
relationship principle
contingency principle
randomized controlled trial
NOTES
1.For detailed accounts of Sutherland’s career and the way in which he developed this theory, see Cohen, Lindesmith, and Schuessler (1956), Gaylord and Galliher (1988), and Sutherland (1973). For further study of the development and revisions of differential association theory, see Chapter 2 of Akers (1998).
2.For classic statements of behavioristic “operant conditioning” principles of learning, see Skinner (1953, 1959). See also the full statement of behavioral learning theory in Burgess and Akers (1966b). Prior to the full revision of differential association by Burgess and Akers, C. Ray Jeffery (1965) proposed replacing all of Sutherland’s theory with a single statement of operant conditioning, essentially rejecting the theory. Burgess and Akers criticized Jeffery for doing this and retained all the major features of Sutherland’s theory in their revision.
3.See also the in-depth review of the research literature on social learning theory in Akers and Jensen (2006).
4.There is a very long list of supportive research on these variables both as part of the original differential association theory and the reformulated social learning theory going back to the pioneering studies by Short (1957) and continuing to the present time. For a review of this research, see Akers and Jensen (2003, 2006; see also Pratt, Travis, Cullen, et al., 2010). See also the critiques of social learning theory and the SSSL model and a reply to them in the symposium edited by Robert Agnew in the November 1999 issue of Theoretical Criminology (Sampson, 1999; Morash, 1999
Order an Essay Now & Get These Features For Free:
Turnitin Report
Formatting
Title Page
Citation
Outline
