Delinquency and Violence as Affect-Control:
Reviving the Subcultural Approach in Criminology*


Will Kalkhoff
Kent State University


Work on delinquent and violent subcultures in criminology has focused on the relationship between holding values (attitudes, beliefs, etc.) in support of delinquency or violence and actual involvement in these activities. However, empirical research on this relationship is inconsistent. Heeding arguments from the sociology of culture and psychology, I suggest that these inconsistent findings result from conceptual oversimplification and inadequate measurement. I propose that an alternative approach to examining subcultural delinquency and violence grounded in affect control theory overcomes these problems. Specifically, I argue that an affect control theory approach to the study of delinquent and violent subcultures shows promise by providing 1) theoretical rigor rooted in the tradition of structural symbolic interactionism; 2) a well-honed measurement procedure founded upon several decades of psychometric research; 3) a developing methodology for identifying subcultural meaning systems; and, 4) a computer simulation program which can be used to generate hypotheses regarding behavior from survey input. Additionally, I demonstrate how the affect control theory approach outlined herein addresses key questions raised by Kornhauser’s (1978) classic critique of “cultural deviance theory.” Directions for future research are indicated.

Keywords: subculture, cultural deviance, violence, symbolic interaction, identity, affect

*An earlier version of this paper was presented at the annual meetings of the American Sociological Association, August, 1999. I would like to thank Karen Heimer, Stacy Decoster, and Lisa Troyer for comments on an earlier draft. Address all correspondence to Will Kalkhoff, Kent State University, Department of Sociology, PO Box 5190, Kent, OH, 44242-0001; Web site:

Work on delinquent and violent subcultures in criminology has focused primarily on the relationship between holding values and attitudes in support of delinquency or violence and actual involvement in these activities. Empirical research on this relationship, however, is inconsistent. The temptation appears to be increasing for criminologists to abandon subcultural approaches to delinquency and violence in favor of less “obviously” beleaguered structural theories. I propose that a more promising solution is to examine and attempt to remedy the flaws in existing subcultural approaches, rather than abandon these approaches altogether.

Two things might have limited criminologists’ understanding of subcultural processes as they relate to delinquency and violence. First, on a theoretical level, cultural sociologists advocate moving away from the view that personal values are the major link between (sub)culture and action, and toward more nuanced views of this relationship (see, e.g., Fine and Kleinman 1979; Swidler 1986; see also Gordon, Short, Cartwright, and Strodtbeck 1963). Second, on a methodological level, problems associated with measuring and assessing the link between attitudes and behaviors are well documented in psychology (see, e.g., Fishbein and Ajzen 1974, 1975; Ajzen 1988). Criminologists could benefit by applying findings and recommendations from these literatures to the development of subcultural models.

In this paper I use affect control theory (e.g., Heise 1979, 1985, 1986, 1987) from social psychology to develop an alternative approach to the study of subcultural delinquency and violence that overcomes the main limitations of previous approaches. Specifically, I seek to show that an approach to examining subcultural delinquency and violence grounded in affect control theory offers promise by providing 1) theoretical rigor rooted in the tradition of structural symbolic interactionism; 2) a well-honed (survey) measurement procedure founded upon several decades of psychometric research; 3) a developing methodology for identifying subcultural meaning systems; and, 4) a computer simulation program, INTERACT, which can be used to generate hypotheses regarding behavior from survey input. In short, I seek to provide theoretical strength and methodological rigor by articulating a view of delinquency and violence grounded in a formal interactionist model of affect control. Furthermore, as I argue below, an important benefit of this approach is that it provides solutions to problems raised by Kornhauser (1978; see also Hirschi 1969; Costello 1997) in her classic critique of “cultural deviance theory.” To date, no theory of subcultural delinquency (or violence) has provided satisfactory answers to the problems raised in the critique (Costello 1997; but see also Matsueda 1988, 1997; Akers 1996; Bernard and Snipes 1996). A fresh and convincing reply to the longstanding critique of “cultural deviance theory” would constitute a step forward in theorizing about subcultural processes in relation to delinquency and violence.

This paper proceeds as follows. First, I summarize in general terms both the theoretical and empirical problems that have complicated research on subcultural delinquency and violence. Second, I present affect control theory and briefly review some recent empirical studies that use the theory to investigate subcultural processes. Third, I show how affect control theory can be used to address the theoretical and empirical problems that I catalog in my discussion of subcultural work in criminology. Finally, in the paper’s conclusion, I consider the overall strengths and weaknesses of an affect control approach to investigating subcultures of delinquency and violence.

Theoretical And Empirical Problems IN The Subcultural Paradigm

Potentially the most serious criticism of subcultural theories in criminology concerns their “logical adequacy” (Costello 1997). The charge that the quintessential representative of this class of theories (i.e., differential association theory) may contain logical flaws can be traced back to Hirschi (1969). However, the most “seemingly devastating” (Matsueda 1988) version of this particular critique has come from Kornhauser (1978).

Kornhauser (1978) classifies the major theories of criminal or delinquent subcultures as either “pure cultural deviance theories” or “mixed models” of delinquency. Pure cultural deviance theories state that deviance is normative to the adherents of a delinquent subculture, that the root cause of delinquency lies in conflicting subcultures, and that the intervening cause of delinquency is socialization into a subcultural value system that condones as “right” what happens to be defined by the legal system as “wrong.”1 Other theoretical models of delinquency are not “pure” because they combine assumptions from cultural deviance theory and other analytical models of delinquency, such as control theory or strain theory. That is, mixed models of delinquency assume that individuals are “selected for delinquency on the basis of experienced strain or weak controls,” but that delinquency will not ensue “without the endorsement of a delinquent subculture” (Kornhauser 1978, p. 26).2

Kornhauser’s critique of pure cultural deviance theories and mixed models of delinquency concerns the logical status of their underlying assumptions. According to this critique, pure cultural deviance theories (namely differential association theory) are problematic in a logical sense because one can derive from them a set of incoherent assumptions that leave nothing left to explain. Arguably, mixed models of delinquency are even more dubious. Not only do they contain a set of untenable assumptions from which only absurdities can be derived, they are also internally inconsistent. Social disorganization theory (Shaw et al. 1929; Shaw and McKay 1942 1969) is the running exemplar of this problem in Kornhauser’s analysis. On the one hand, social disorganization theory holds that social disorganization produces weak institutional controls, which in turn loosen the constraints on individuals’ natural propensity to deviate. Most proximally, delinquency results from weak social bonds. On the other hand, Shaw and his associates emphasize, especially in their later work, that delinquent youths’ denial of the moral validity of the laws they violate explains their illegal behavior. In this sense, delinquency is assumed to require the endorsement of a delinquent subculture. This latter assumption, Kornhauser argues, is not necessary to make social disorganization theory work. In fact, it only confounds the logical structure of the theory, which can stand alone on the former assumption that disorganization produces weak institutional controls, and thereby frees people to deviate.

Kornhauser has been harshly criticized in return for her caricaturing of the theories (namely differential association theory) that she finds problematic (see Matsueda 1988; Akers 1996; Bernard and Snipes 1996). In short, Kornhauser’s adversaries argue that she has built a “straw man” of the subcultural approach, one replete with “misconceptions and oversimplifications” (Matsueda 1988, p. 293). Thus, the critique may actually stand on shaky ground.

In a recent volume of Theoretical Criminology, though, Costello has reframed the debate around four “crucial questions” (1997, p. 406) that cultural deviance theories – represented best by differential association theory, she argues – cannot adequately address. These are:

1) What is the content of deviant norms?

2) What is the origin of deviant norms and subcultures?

3) How can we identify a deviant subculture apart from its members’ behavior?

4) How easily do individuals learn deviant norms?

With respect to the content of deviant norms, Costello argues that cultural deviance theorists have vacillated on the issue of whether deviant norms “require…or simply allow deviant behavior” (p. 406; emphasis in original). According to Costello, both positions entail logical difficulties. Kornhauser (1978) has already addressed in detail the more extreme of these two positions, but even the “more moderate” (Costello 1997, p. 407) stance that cultural deviance theorists have come to adopt – i.e., that deviant norms allow deviant behavior – seems to produce logical problems. Perhaps the most serious difficulty along these lines is that cultural deviance theory may not be able to explain the “transmission of the subculture of acceptable behavior” (p. 408). That is, if the cultural deviance theorist argues that simply allowing certain behaviors to occur within its parameters transmits a delinquent subculture, then she or he is left unable to explain primary deviance. Additional assumptions from outside of the cultural deviance perspective (ones that are incompatible with it) must be invoked to explain why individuals engage in delinquent behavior in the first place.

With reference to the second of her questions, Costello argues that cultural deviance theorists cannot adequately explain the origin of deviant norms and subcultures, despite at least two motivated efforts to do so. First among these are the mixed approaches exemplified by Cohen (1955) and Cloward and Ohlin (1960), both of which combine elements of structural-strain and socialization/learning arguments to explain the formation of delinquent subcultures and the resultant illegal behavior of these groups’ members. However, according to Costello, “it is not logically sound to argue that the strained offender requires group approval of criminal behavior…The strain itself should provide sufficient motivation for crime…” (p. 413). In short, elaborated versions of cultural deviance theory are internally inconsistent, and for no good reason.

Cultural deviance theorists have also tried to explain the origins of deviant subcultures by arguing, “criminal behavior is self-interested behavior” (Costello 1997, p. 414). However, this explanation is not consistent with the view that criminal behavior reflects an individual’s acceptance and pursuit of alternative group (i.e., subcultural) values, attitudes, or norms. Does one owe allegiance strictly to oneself? Or does one owe allegiance to (or conform to) a set of values vis-à-vis group membership? Cultural deviance theorists cannot have it both ways, or so the argument goes.

The third problem for cultural deviance theory concerns the identification of deviant subcultures. According to Costello, “Sutherland et al. simply equate an escalation of delinquent behavior with increasing acceptance of values supporting such behavior” (p. 417; emphasis added). Though not mentioned by Costello, the most logically problematic result of this strategy for identifying delinquent subcultures is tautology. Since delinquent behavior is the thing to be explained, it obviously cannot be explained by itself. In more formal terms, delinquent behavior or rates of delinquent behavior cannot serve as both the explanandum as well as the explanans. Cultural deviance theories are not inherently disabled in this way, though. Their practitioners must simply avoid the convenient shortcut of assuming that patterns of differential behavior indicate highly correlated patterns of differential association or cognition. Rather, the latter elements must be measured independently and then examined for systematic variation with forms of behavior.

The final point of controversy in cultural deviance theories concerns the acquisition of deviant norms. In agreement with Kornhauser’s (1978) classic critique of cultural deviance theories, Costello argues that these theories portray individuals as “completely controlled by their culture” (p. 418). Thus, cultural deviance theories seem to leave us with the unreasonable assumption that once acquired, the content of a deviant subculture forces its carriers to enact delinquent behavior. In other words, one might argue that cultural deviance theories lack a sophisticated (and logically coherent) social psychological model for how deviant content is acquired and expressed.

In the same volume of Theoretical Criminology, Matsueda (1997) offers a reply to Costello’s reformulated critique of cultural deviance theory. In large, this paper is a restatement of his (1988) earlier arguments directed at Kornhauser (1978). The reason for this restatement is a consequence of Matsueda’s belief that Costello herself has “merely continued Kornhauser’s misinterpretation [of cultural deviance theory]” (1997, p. 429). The point of the present paper, however, is not to review each point and counter-point and then try to adjudge who seems to have won the day in this ongoing debate. If a resolution has not yet been achieved, it is unlikely that it will emerge anytime soon. My view, rather, is that Kornhauser’s (1978) critique has motivated more than enough unresolved controversy to warrant the exploration of alternative subculture of delinquency formulations that avoid the alleged pitfalls of “cultural deviance theory.” Toward this end, I will later use Costello’s questions as the orienting questions that such a formulation must answer.

Additional Theoretical Problems

Two important organizing pieces from the sociology of culture provide additional, though less specific grounds for critiquing the theoretical underpinnings of the subcultural paradigm in criminology. First, Swidler claims that the prevailing sociological model used to understand culture’s effects on action assumes that “culture shapes action by supplying ultimate ends or values toward which action is directed, thus making values the central causal element of culture” (1986, p. 273). While this model brandishes a good deal of intuitive plausibility – a fact which may explain its dominance in sociology and criminology more particularly – Swidler argues that a focus on people’s values is insufficient for understanding social action. The main problem is that the value-action model of culture simply ignores too much, and thus has little actual potential to explain what is distinctive and important about the behavior of individuals, groups, and societies. In relying on the oversimplified view that people’s values determine their behavior, students of culture “neglect other distinctively cultural phenomena which offer greater promise of explaining patterns of action…[including]… culturally-shaped skills, habits, and styles…” (p. 275; emphasis in original).

Similarly, Fine and Kleinman (1979) argue that a key problem in the sociological analysis of subculture is that the term is frequently equated with social objects such as values and norms. This preferred conceptualization, they argue, has negative consequences for both theory development and empirical research. On a theoretical level, Fine and Kleinman agree with Swidler that while values are indeed important “cultural elements,” an exclusive focus on them blatantly ignores “much of what is cultural about subculture” (p. 7; emphasis in original). Sociologists would have a much better grasp of the relationship between culture and action if they would fix the limits of their theoretical purview to include a fuller range of cultural elements.

Empirical Problems

The extent to which empirical research based on the reigning value-action model of culture can shed light on subcultural processes is highly questionable. This is because the model unrealistically assumes that subcultural members will be willing to share their attitudes with outsiders, such as interviewers (Fine and Kleineman 1979). This assumption is especially untenable when the members of a subculture hold values that conflict with those of the interviewer(s). Therefore, not only is it likely that information describing a group’s values provides the least effective theoretical path to understanding its behavior, in some cases the authenticity of this information may be highly disputable. This is one likely reason why studies of delinquent and violent subcultures have yielded such inconsistent results.

Another likely reason for the inconsistency of findings concerning the existence of subcultures of delinquency and violence is the fact that values, the choice operationalization of subcultural content among criminologists, can be regarded as a special type of “attitude object” (Bem 1970).3 Therefore, the empirical measurement of values is subject to all of the well-documented problems associated with attitude measurement in general, especially those problems concerning the link between attitudes and behaviors (Fishbein and Ajzen 1974, 1975; Ajzen 1988).

For the purposes of this paper, one especially relevant argument from the attitude literature in social psychology is that predictions from verbal (or written) response to actual behavior are greatly strengthened by making the question calling for the attitude response as specific as possible. Ajzen and Fishbein (1977; see also Ajzen 1982) demonstrate that when a measured attitude is general, and the behavior in question is very specific, there is very little correspondence between attitudes and actions. Ajzen and Fishbein examined 27 attitude-behavior studies conforming to this type of design. In all but one of the studies, attitudes failed to predict behavior. By contrast, attitudes were found to predict behavior in all 26 of a second set of studies in which the measured attitude was directly pertinent to the situation. Other studies have confirmed that specific, relevant attitudes do predict behavior well (Kim and Hunter 1993).

However, being able to obtain valid reports of respondents’ attitudes or values concerning specific, contextualized illegal acts is likely to be undermined by the intrusive nature of this kind of inquiry, especially when the reports solicited pertain to serious illegal acts such as violence. I question whether truthful and accurate reports could be obtained under these conditions. Thus, where criminologists might accommodate the considerations of social psychologists in their investigations of the link between delinquent values and delinquent behavior, it is likely that these advancements would be minimized or canceled out by the introduction of more invasive survey procedures.

All things considered, then, a desirable alternative perspective on subcultural delinquency and violence would not only answer to the longstanding critique of “cultural deviance theory,” but would also be able to avoid the conceptual oversimplification and complications of measurement that accompany the equating of “subculture” with individuals’ adoption of a distinctive set of values or attitudes. I propose that one solution to these problems lies in affect control theory.

Affect Control Theory

Affect control theory (hereafter ACT) offers a dynamic model of social action that focuses on how people’s attitudes toward identities, behaviors, social settings, and emotions (i.e., the key aspects of social interaction) inform the actions that individuals take toward one another. As an interactionist theory, ACT views social situations and the cultural context within which they occur as important determinants of behavior, including its conventional and deviant forms. Specifically, ACT’s mathematical model of attitudes gives formal rigor to the basic interactionist principles that people act toward things on the basis of the meanings that these things have for them (Shibutani 1961; Blumer 1969), and according to their definition of the situation (Goffman 1974). Another important interactionist feature of ACT is its emphasis on the notion that culturally shared meanings (for identities, behaviors, etc.) make interaction, communication, and society possible (Mead 1934).

Unlike many other symbolic interactionist theories, however, ACT conceptualizes meaning precisely (Smith-Lovin 1990). Based on the well-established psychometric research of Osgood and his colleagues (Osgood, Suci, and Tannenbaum 1957; Osgood, May, and Miron 1975), ACT suggests that interaction sequences can be modeled as if people think about the plethora of cultural identities, behaviors, settings, and emotions in terms of three universal (i.e., cross-culturally valid and reliable) dimensions of affective meaning: evaluation (goodness), potency (powerfulness), and activity (liveliness). People’s attitudes toward a given identity, behavior, setting, or emotion within a given culture are viewed as consisting of a particular EPA profile (i.e., a unique combination of quantitative values corresponding to evaluation, potency, and activity).

From affect control theory’s precise conception of meaning, we can obtain a comparably precise definition of culture, a task that has plagued sociologists and anthropologists. In ACT, culture, simply put, refers to “Shared meanings regarding people, processes, and non-human objects” (Heise 2002) Indeed, using a semantic differential measurement procedure, ACT research demonstrates a high level of intra cultural agreement about the affective meanings corresponding to a wide variety of identities, behaviors, settings, and emotions (Smith-Lovin 1990).

The usual or established meanings that identities, behaviors, settings, and emotions have for cultural members are referred to in ACT as fundamental sentiments. These are the meanings (as measured by evaluation, potency, and activity) that individuals are assumed to carry in memory and import into social situations. Because ACT assumes that meanings are shared within a culture, fundamental sentiments are discovered by sampling the culture’s members and averaging their EPA ratings for numerous identities, behaviors, settings, and emotions.

While fundamental sentiments are robust, the experience of actual social events (impression formation) produces situational meanings that may differ from our fundamental sentiments, sometimes in drastic ways. The situational meanings that arise from real social experience are referred to in ACT as transient sentiments, or transient impressions. More precisely, transient sentiments are “the affective meanings associated with the particular combination of identities, settings, behaviors, and emotions in the current situation” (Robinson, Smith-Lovin, and Tsoudis 1994). Compared to fundamental sentiments, situationally-invoked transient sentiments can make the identities and behaviors present in the setting appear better or worse (evaluation), stronger or weaker (power), or livelier or quieter (activity) than people learn to expect in a culture (Smith-Lovin 1990).

Deflection occurs when fundamental sentiments and transient sentiments are inconsistent. The greater the disagreement between fundamental and transient sentiments, the greater the deflection. In quantitative terms, deflection can be conceived of as “the sum of the squared differences between the fundamental sentiment and the transient impression across each of the three EPA dimensions” (Robinson et al. 1994, p. 177). Alternatively, using a graphical heuristic, “a deflection roughly corresponds to the distance between the fundamental sentiment and the transient impression [in three-dimensional space]” (Robinson et al. 1994, p. 177).

As its name implies, ACT assumes that individuals attempt to confirm or control established meanings or fundamental sentiments in social situations. More formally, the fundamental motivational principle of ACT is that people seek to construct social events that verify their fundamental sentiments about themselves and others – that is, people attempt to minimize deflections. Therefore, people view events that produce large deflections as improbable (Heise 1987). Research shows, in fact, that individuals’ perception of the likelihood of events is negatively related to the amount of deflection produced (Heise and MacKinnon 1987).

However, when events do happen to create transient sentiments that differ from fundamental sentiments, ACT assumes that people attempt to regain control of meanings in one of two ways. First, and most typically, individuals will respond to deflection by behaving in ways to generate new events that restore fundamental sentiments. Smith-Lovin illustrates this primary technique in the following example:

Suppose I attend a party and think that the Host is ignoring me, her Guest. Hosts and Guests normally are viewed as pleasant people, with the Host adjudged somewhat more powerful and lively than the Guest. When rated by female undergraduates on a scale ranging from –4 to +4, Host has an evaluation (goodness) of 1.5, a potency (powerfulness) of 1.2, and an activity (liveliness) of 0.5; Guest has an evaluation of 1.7, potency of 0.0, and activity of –0.2…The unpleasant and inappropriate act of Ignoring (with an evaluation of –1.8, potency of –0.2, and activity of 0.2) damages the reputation of both Host and Guest. A Host who has Ignored a Guest loses goodness and power, though not liveliness…the Host and Guest have both lost status as a result of the Ignoring. The theory [i.e., ACT] suggests that this event is unlikely, since it does not maintain and support the identities of the interactants…If the Guest points out the event as Ignoring and the Host accepts this account of what has transpired, the Host is likely to produce a restorative act: a behavior that, when processed, will move impressions back in line with fundamental meanings. Or she may expect the Guest to repair the situation with such an act. Possibilities for the Host include [acts such as] Appreciate, Soothe, and Apologize To; the Guest might Excuse or Caution the Host to restore her view of the occasion.4

(1990, pp. 241-2)

This example describes how individuals might respond to a situation that produces relatively modest deflections. In cases where deflections are relatively large or “incredible” (Heise 1987, p. 12), individuals must “redefine the situation” – i.e., recast identities or reinterpret behaviors – to establish an understanding of what has occurred in the situation. While the concept of an “incredible” deflection is not yet precisely defined in ACT, Smith-Lovin (1990) maintains that the event “grandfather rapes granddaughter” serves as an example of a social situation that produces massive deflection. In such cases – i.e., in cases where no subsequent action can effectively restore the dislocated interaction – people will redefine the situation. The result is underlying, cognitive change. Thus, evaluators of the event “grandfather rapes granddaughter” may label the grandfather a “rapist” or “child molester,” because both of these identities are bad enough and active enough to commit the act of rape (i.e., in terms of their affective ratings).

An attractive feature of ACT, then, is that it offers a unifying interactionist perspective. On the one hand, with its focus on fundamental sentiments and people’s tendency to seek their verification, ACT incorporates the ideas of role-playing and constraint associated with the “Iowa” school of symbolic interaction (Kuhn 1972a, 1972b). On the other hand, with its attention to transient impressions, disjunctions between commonly-held meanings and transient impressions, and people’s motivation to alleviate such disjunctions either behaviorally or cognitively, ACT concretely represents the notions of role-making and creativity associated with the “Chicago” school of symbolic interaction (Blumer 1969). MacKinnon (1994) includes a discussion of how the ACT model draws together the structural and more dynamic faces of symbolic interactionism (see, especially, pp. 84-6).

The ACT model can be represented formally by a set of impression formation equations (Smith-Lovin 1987a, 1987b) and their mathematical transformations (Heise 1987). The impression formation equations and their estimates were derived from empirical research using a sample of US undergraduate students. The transformations – which describe probable behavioral, emotional, and labeling (cognitive) responses to a social event – were derived from the empirically generated equations based on ACT’s fundamental motivational principle (i.e., that people seek to confirm fundamental meanings and minimize deflections).

To make the theory’s quantitative operationalization (Schneider and Heise 1995) accessible to social scientists, the mathematical equations underlying ACT have been assembled in a computer simulation program, INTERACT (Heise and Lewis 1988; Heise 1991). This computer program allows the investigator to simulate social interactions by inputing the social setting (e.g., a gas station) and the identities of the actors involved (e.g., a clerk and a thief). Additionally, the investigator may choose to modify the actor identities (e.g., a scared clerk and a mean thief). Once the elements of the social event are defined, INTERACT generates predictions concerning the likely behavioral, emotional, and labeling responses of the actors involved (in quantitative and/or qualitative, natural-language form). Although ACT has not yet been extensively tested, recent empirical studies provide general support for predictions derived from the theory (e.g., Wiggins and Heise 1987; Robinson and Smith-Lovin 1992; Robinson et al. 1994). Two ACT studies are directly relevant to the present concerns; hence, I review them in detail.

Act Research on Subcultures

ACT research on subcultures suggests modification of the theory’s assumption that fundamental associations are uniform within a given culture. First, in a study of two religious groups, Smith-Lovin and Douglas (1992) demonstrate how deviant identities in mainstream society can be transformed in subcultural contexts into positive, rewarding ones. One of the religious groups investigated for the study was a traditional Unitarian church, while the other group (the Metropolitan Community Church, or MCC) was identified as “an institution of gay subculture” (p. 226). The two groups were otherwise similar in terms of membership size and income distribution.

Using participant-observation techniques, Smith-Lovin and Douglas first compiled a list of 30 social identities and 30 behaviors significant to both religious groups. Representative members from each church then rated these stimuli with semantic differential scales measuring evaluation, potency, and activity. T-test analyses conducted on these data indicated extensive differences between the two groups’ ratings of gay and religious identities, but not for behaviors. The lack of rating differences for behaviors is consistent with Heise’s (1966) contention that most behaviors will not have different meanings across subcultures. It is also gives greater specificity to the claim that subcultures are only partially different from the larger, parent culture (Wolfgang and Ferracuti 1967).

In the next stage of their analysis, Smith-Lovin and Douglas entered each group’s mean EPA ratings of identities into INTERACT. As discussed above, the output from INTERACT is equivalent to predictions generated by ACT about what people who see themselves in a particular way (based on a particular set of EPA meanings) will do, feel, and think. The simulations conducted by Smith-Lovin and Douglas produced suggestive hypotheses. For the meanings associated with religious identities, INTERACT predicted deep, positive emotions for both groups. However, with respect to behavioral expectations, “the dramatic, flamboyant interactions generated by the MCC affective meanings contrast markedly with the staid role relationships among religious figures in the Unitarian church group” (p. 234).

The very different meanings associated with homosexual identities also generated provocative simulations. Using the MCC sentiments about homosexuals, INTERACT predicted positive interactions among gay individuals (e.g., Applaud, Play With, Court one another). By contrast, the Unitarian meanings led INTERACT to predict that homosexuals will experience negative, unhappy interactions with one another. It is also interesting to note Smith-Lovin and Douglas’ finding that “the most common term used by the group [MCC] to describe themselves – gay person – has a meaning that is strikingly positive and lively…[and has]…the least negative image among the Unitarians” (p. 235). This finding and the results of the simulations lend credence to the investigators’ belief that subcultures can enhance the identities of individuals who are stigmatized in the parent culture.5

To validate their simulation results, Smith-Lovin and Douglas created 60 descriptions of social events, each specifying an interaction between gay and religious identities. The behavior directed from one identity (actor) toward the other (object) in each event was selected from one of three sets of behavioral predictions: 1) those produced by the MCC sentiments, 2) those produced by the Unitarian sentiments, or 3) those produced through random selection of behaviors from the INTERACT corpus. Seven judges from the MCC church then rated all 60 of the events on a Likert-type scale. Based on the ACT assumption that the perceived likelihoods of events decline as more affective deflection is produced (Heise and MacKinnon 1987), Smith-Lovin and Douglas hypothesized that “the subjective likelihood of events generated by the MCC sentiments should be higher than events generated by the other church group’s sentiments” (p. 243). In fact, ANOVA produced a significant result for the 3-category grouping variable, and a follow-up test confirmed the investigators’ prediction of a significant difference between the MCC and Unitarian events. MCC raters saw the events generated from the MCC sentiments as more likely than events generated from the Unitarian sentiments. Smith-Lovin and Douglas conclude from their analyses that ACT shows promise as a “generative model of culture” (p. 243).

In a more recent article, Thomas and Heise (1995) demonstrate how the error variance in individuals’ EPA ratings of words can be “mined” in order to identify subcultural patterns of deviation in sentiment measurements. Previously, ACT researchers had treated rating variance as idiosyncratic deviance or random measurement error (Heise 1966). However, Thomas and Heise cite recent research that calls this assumption into question. Consistent with Wolfgang and Ferracuti’s (1967) arguments, these studies reveal that “members of subcultures give very different EPA ratings for stimuli that are centrally relevant to the subculture, even though their sentiments about other concepts are the same as in the larger culture” (Thomas and Heise 1995: 428; emphasis added). Smith-Lovin and Douglas’ (1992) study is one exposition of this phenomenon, as Thomas and Heise note. Other studies are cited which also show meaningful variation in the sentiment measurements of males and females, individuals in different psychiatric groups, and college-aged drug users and non-users.

Thomas and Heise (1995) shed more light on these kinds of subcultural phenomena with two studies, published together. In the first study, Thomas and Heise use previously collected EPA data (Heise and Thomas 1989). These data consist of 224 social identities, emotions, and combinations of identities and emotions rated by 326 respondents (162 males and 164 females) from an American university psychology subject pool. Analyzing 44 emotions and 64 identities from these data, Thomas and Heise note that about one-fifth of these concepts produced parallel coordinate displays where subgroup differences could be visually detected. However, k-means cluster analyses performed on individual concepts revealed much more fine-grained patterns of variation, even for concepts where differences could not be visually detected.

Suspecting that people may have distinctive sentiments for a variety of concepts, Thomas and Heise performed principal components analyses on the sets of emotion and identity concepts. These analyses consistently produced two meaningful components, leading to the conclusion that “individual variations in sentiment are somewhat predictable from one concept to another and that there are two main ways that individuals vary in sentiments [in these data]” (p. 430). With these procedures, Thomas and Heise demonstrate a new and promising way of both conceptualizing and identifying subcultural meaning systems within the ACT framework.

In their second study, Thomas and Heise used the results of the components analyses from Study 1 to select concepts (13 identities and 7 emotions) that maximally discriminated respondent subsets. These stimuli were presented to a sample (N=55) of undergraduate social psychology students who rated the concepts with semantic differential scales measuring evaluation, potency, and activity. Using principal components analysis, two components were extracted from these data, and the loadings were found to replicate the results from the first study.

To shed light on the reasons why this pattern exists, Thomas and Heise computed and plotted component scores for respondents in their second study. Ten cases, representing the most extreme locations from each quadrant of the component-scores plot, were selected for in-depth, open-ended interviews. Nine of these individuals consented to and completed the interview.

The exploratory interviews revealed interesting uniformities among individuals sharing a distinctive pattern of sentiment deviations (see pp. 436-7). Thomas and Heise report that individuals who might be categorized as “socially isolated” had less intense sentiments about many concepts. These individuals also reported difficulties in “symbolic interaction” (e.g., possessing “nebulous opinions,” being “irresolute in discourse,” and “making jokes that aren’t funny”). By contrast, individuals who might be categorized as possessing broad and deep social relations (social capital) expressed strong sentiments, displayed “self assurance in interpersonal discourse,” and projected “intense, confident, and convergent [i.e., non-deviant] opinions.” These socially integrated individuals were also apparently less psychologically distressed than the individuals expressing deviant sentiments. Thomas and Heise attribute the verbosity and psychological well being of the socially integrated individuals to their greater ownership of and participation in mainstream culture. Given its capacity to foster and shed light on these conclusions, ACT shows further promise as a powerful tool for analyzing cultural and subcultural processes.

Reviving The Subcultural Approach IN Criminology

The main goal of this paper is to show how ACT contributes to criminology by providing solutions to the major limitations of existing subculture of delinquency and violence approaches. I have identified three such limitations: logical inadequacy, conceptual oversimplification, and measurement problems. In this section of the present paper, I tentatively propose an ACT answer to these limitations, focusing especially on Costello’s (1997) four questions concerning logical flaws in cultural deviance theory. To avoid redundancy, the issues of conceptual oversimplification and measurement problems are addressed within the context of these questions.

What is the Content of Deviant Norms?

According to Costello, cultural deviance theorists have over the years settled on the position that deviant norms allow (but do not require) deviant behavior. As mentioned above, the major problem with this position is that it allegedly cannot elucidate the reasons for an initial act of deviance. In my view, the essential problem here concerns the lack of an adequate theoretical model explaining how people’s attitudes are translated into social action. Among existing subculture of delinquency and violence explanations, the dominant model of this process is intra personal (i.e., a person’s values or attitudes cause her or his behavior), rather than inter personal. As Ball-Rokeach’s comments about her inability to detect a subculture of violence make clear, the pressing issue is one of conceptual oversimplification:

A plausible interpretation of these [insignificant] findings is that values and attitudes are relatively unrelated to violent behavior because violence is primarily interpersonal rather than intrapersonal…Thus, one ought not to expect a causal connection between attitudes and violent behavior when the values and attitudes of only one interacting party are taken into account. This interpretation suggests a line of research which regards violence as neither victim-precipitated nor assailant initiated…In other words, participation in violence is some interacting function of the values and attitudes of all parties involved. (1973, pp. 748-9)

To the extent that Ball-Rokeach is accurate, ACT provides a promising solution to the problem she raises. Indeed, a distinguishing feature of ACT is its rigorous focus on how people’s attitudes towards the key aspects of social interaction (i.e., identities, behaviors, emotional displays, and settings) relate to interpersonal behaviors. Consistent with a conflict view of crime and delinquency, recent research on ACT (see above) confirms that groups of individuals within a society are often working with partially distinguishable sets of fundamental sentiments (i.e., sets of sentiments that are centrally relevant to the groups’ uniqueness). However, unlike the dominant criminological approaches to subcultural delinquency and violence (e.g., Shaw and McKay 1942, 1969; Cohen 1955; Miller 1958; Cloward and Ohlin 1960; Wolfgang and Ferracuti 1967), ACT does not posit that delinquency results from the behavior of individuals who are driven to enact delinquent attitudes/sentiments across situations. Rather, what happens to be called “delinquency” stems from conflicting definitions of situations. From the orientation of some subgroups operating with a particular subset of fundamental sentiments, officially defined deviance is not always socially dislocating. Quite the contrary, for these groups, condemned acts may be confirmatory and expected in some circumstances, depending on the particular configuration of identities, behaviors, emotional displays, and features of the setting. These factors can be taken into account by ACT’s predictive machinery, which would seem to give the criminologist access to a more accurate portrayal of much crime and delinquency.

What is the Origin of Deviant Norms and Subcultures?

As discussed above, Costello (1997) has also critiqued cultural deviance theorists for being unable to account for the origins of subcultures. An important point in this argument is the claim that cultural deviance theorists “have not developed an explanation of why shared structural conditions can only influence behavior through their influence on values or “shared patterns of thinking”…” (p. 414). However, sociologists have recently come to emphasize a factor that affect control theorists (and interactionists more generally) might regard as the key structural determinant of shared patterns of meaning: social isolation. Indeed, ACT may give microlevel specificity to current efforts in urban sociology to explain the structural forces behind the effects of segregation and social isolation on the perpetuation of disadvantage and high crime rates among underclass Blacks (see, e.g., Massey and Denton 1993). According to the segregation argument, the “coincidence of racial isolation and high poverty” (Massey and Denton 1993, p. 170) has produced high levels of social disorganization in Black communities. In these severely disorganized communities, a “culture of segregation” has emerged that perpetuates social ills such as poverty and crime. Pointing to the severity of racial isolation in the US, Massey and Denton note, “many ghetto residents have come to speak a language that is increasingly remote from that spoken by American whites” (p. 162). Similarly, in their study of two religious groups, Smith-Lovin and Douglas cite Levine’s research on gay ghettos, which shows how socially isolated homosexuals have “created their own culture, developing their own language and social norms for interaction with other gays” (1992, p. 225). In addition, Smith-Lovin and Douglas discuss Warren’s ethnography of gays in Sun City, which serves as another example of how “Terminology that would seem inappropriate or mundane to an outsider is in fact rich in meaning for members of the gay community” (p. 225). ACT interprets these phenomena in terms of its interactionist roots. the meaning of things arises out of the social interaction one has with one’s fellows (Blumer 1969; Shibutani 1961). More specifically, ACT posits that language is the central symbolic system through which beliefs and their corresponding affective associations are “represented, accessed, processed, and communicated” (MacKinnon 1994, p. 74). Given that underclass Blacks and gays are socially isolated from mainstream society, it is not surprising that they have developed distinctive language patterns reflecting their situation of isolation. And given Smith-Lovin and Douglas’ (1992) findings, it would not be surprising to find that patterned deviation exists among inner-city Blacks’ sentiments, and that the “culture of segregation” comprising these unique sentiments is associated with delinquent or violent interpersonal behaviors. A key question that comes up now, though, is: why does social isolation not always lead to violence? In my view, attempting to answer this question requires attention to the macro-structural context of segregation.

Along these lines, current thinking by ACT researchers on “macro-structural representations” (Ridgeway and Smith-Lovin 1994) might be developed with the aid of criminologists and other sociologists to explain how the structural features of society create distinctive patterns of sentiments among social groups. To the extent that sentiment deviation is found to be related to delinquent or violent behavior, theoretical advancements in ACT specifying the linkages among structure, culture, and interaction would allow criminologists to explain why delinquent groups embrace different fundamental sentiments than other members of society.

Steps in the direction of forging such micro-macro links in ACT have been taken by Ridgeway and Smith-Lovin who argue that positions in the social structure (i.e., identities) with varying degrees of evaluation, potency, and activity have their origin in “the control of material resources and meaningful ritual action” (1994, p. 221). While affective associations are learned during socialization, the particular affective meanings that a group comes to assign to cultural identities reflect their structural analogues. For example, the evaluation dimension of affective meaning mirrors an identity’s “positional resources” (i.e., status or esteem) within a social system. Potency reflects an identity’s power, or “control over physical force or valued resources.” Finally, activity relates to an identity’s “expressivity,” which reflects both that identity’s “technical activity” and “degree of [social] engagement” with other individuals in the social/institutional system. From the ACT perspective, then, one would expect isolated, inner-city ghetto Blacks to give very different responses to the questions: Who is worthy? Who is mighty? Who is connected? It follows from this that members of mainstream culture and members of inner-city ghetto cultures may attend to constructing and interpreting social scenes in very different ways. It is not surprising, then, that inner-city crime and violence would seem so “senseless” or “nihilistic” to outsiders. But how do we identify “outsiders” and “insiders”?

How Can We Identify a Deviant Subculture Apart From Its Members’ Behavior?

The third problem that Costello (1997) has attributed to cultural deviance theorists is their inability to identify a delinquent subculture without making reference to behavior. As I have argued, the most serious logical problem that results from this strategy is tautology: delinquent behavior cannot be explained by itself. However, in ACT, the concept of subculture is precisely delineated from individuals’ behavior. As I have interpreted two recent ACT investigations of subcultural phenomena, the term subculture is defined and operationalized as patterned deviation in affective sentiments. It is these acute deviations (detected with t-tests, k-means clustering, and principal components analysis) that the researcher comes to expect to be related to specific behaviors though INTERACT analysis. Nonetheless, simulation results generated from INTERACT must be validated with independently collected data before ACT’s explanatory and predictive ability can be assessed. This ability has been demonstrated in Smith-Lovin and Douglas’ (1992) as well as Thomas and Heise’s (1995) research on deviant subcultures.

A special benefit of ACT – contributing to its demonstrated ability to identify subcultures – is its measurement procedure. Previous individual-level studies of delinquent and violent subcultures, which have relied on direct questioning of individuals about general (delinquent or violent) attitudes, suffer from the “attitude-behavior problem” long known to psychology. The validity of these studies’ findings is disputable too, since members of subcultures may be motivated to give misleading answers to queries about their lifestyles and activities, especially in cases where subcultural boundaries between researchers and their subjects are clearly demarcated (Fine and Kleinman 1979).

ACT avoids the validity problem to some extent with its more innocuous measurement procedure. Respondents are asked to rate “concepts” on three abstract dimensions; they are not questioned directly about their attitudes, beliefs, or values. Furthermore, ACT evades the attitude-behavior problem with its sophisticated inter personal, rather than intra personal, conceptualization of the interrelationships among people’s affective associations about identities, behaviors, emotions, and settings. However, I stress that the ACT model of attitudes and interpersonal behavior is not necessarily better than its simpler counterpart. Indeed, the basic attitude-behavior model from psychology is more parsimonious and may work well for less serious forms of intrapersonal crime and delinquency where respondents attitudes about specific behaviors toward specific objects can be measured unobtrusively. However, where crime and delinquency are best captured as “some interacting function of the values and attitudes of all parties involved” (Ball-Rokeach 1973, p. 749-50), ACT is proposed as a better explanatory and predictive model. The question is one of “scope” or applicability, not one of unconditional superiority.

How Easily Do Individuals Learn Deviant Norms?

This question, as it stands, may be the most difficult for ACT to answer on its own terms. The most the theory says is that cultural meanings are acquired during the socialization process.6 However, Costello’s (1997) main concern with respect to the issue of the acquisition of deviant norms is that cultural deviance theories assume that once learned, the content of a deviant subculture forces its members to enact deviant behavior. Certainly, ACT does not fall prey to this problem. Again, the theory offers a more sophisticated, interpersonal model describing how deviant content is expressed in social situations. First, attitudes are conceived of in a multidimensional sense (EPA). Second, while ACT does incorporate the interactionist principle that human beings behave toward things on the basis of the meanings those things have for them, it does not assume a simple correspondence between a solitary object’s meaning and an actor’s behavior toward it. Rather, as should be clear by now, ACT conceives of social behavior as a contextualized phenomenon reflecting a complex interplay among the meanings associated with the plethora of aspects of social interaction. What is especially elegant about the theory is its unique ability to make sense of this complex process starting with a relatively parsimonious set of concepts, assumptions, and equations. Far from being overly deterministic, ACT does justice to the inherent creativity and nuances of social interaction without jettisoning the rigor of scientific explanation.


Criminologists face a decision. Despite its “remarkable persistence” (Matsueda 1997), the utility of the traditional subcultural approach to the study of delinquency and violence has been seriously questioned since its earliest formulations. One solution would be to abandon the approach altogether in favor of less “obviously” problematic structural theories. It does not seem that we have been able to achieve, at least within criminology, a perspective that makes sense of “the person in relation to other persons and groups as they function within the milieu of their specific participation in culture and society” (Hollingshead 1939, p. 822). Even by the publication of Ball-Rokeach’s test of the subculture of violence thesis, criminologists had not come to be able to make adequate theoretical sense of how delinquency, crime, and violence more particularly may be “some interacting function of the values and attitudes of all parties involved” (1973, p. 748-9). And twenty-nine years later, we are still found lamenting the fact that the “interactional level” of analysis has been grossly neglected in criminology (Short 1998).

In this paper, I have offered affect control theory (ACT) as one promising solution to this problem in criminology. First, with its rigorous and innocuous measurement procedure, ACT addresses important methodological problems faced by existing theories of delinquent and violent subcultures concerning measurement validity and the assessment of the complex link between attitudes and behavior. Second, with its specific and sophisticated interactionist portrayal of how people’s attitudes toward identities, behaviors, settings, and emotions are related to interpersonal behaviors, ACT defies the criticism of conceptual oversimplification and offers a model that is true to the likely (interpersonal) nature of much crime and delinquency. Finally, I have attempted to show how ACT speaks to Costello’s (1997) reframing of Kornhauser’s (1978) classic critique of “cultural deviance theory” around four main orienting questions that a theory of delinquent or violent subcultures must answer. Arguably, ACT can address the content, origins, procedures for identifying, and acquisition of deviant sentiments.

Of course, ACT is not without its weaknesses. First, ACT’s computer simulation program, INTERACT (used for deriving hypotheses from the theory), is prone to occasional prediction errors (Smith-Lovin 1987c). Though the ACT software has been enhanced with better equations since it first came to be used, INTERACT results will at times appear unreasonable or absurd to the “cultural expert.” However, efforts to further eradicate prediction failures are underway by ACT investigators.

A second weakness of ACT concerns the relative sparseness of the theory’s statements about linkages among macro-structural representations (structure), culture, and interaction-level dynamics. Yet, because criminology has come to emphasize recently the importance of understanding structure and culture as interacting causes of crime (Heimer and DeCoster 1999; Wilson 1991), the connection between these two forces will have to be made more explicit in ACT before the theory can provide a fully satisfying account of subcultural delinquency and crime. As discussed above, more attention to micro-macro relationships is needed to understand why, for example, social isolation does not always lead to violence.

One fruitful line of development in this area would borrow from Kathleen Carley’s (1985, 1986a, 1986b) excellent research on social structure and cognition to develop an ACT model that more clearly specifies how affective meaning is acquired and shared within social networks (Smith-Lovin 1987c). Such an elaboration of ACT using concepts from the social networks perspective would be particularly well-suited to questions posed by criminologists concerning differential opportunity and criminal networks (Cloward and Ohlin 1960), social capital and crime (Sampson and Laub 1993), and “criminal embeddedness” (Hagan 1993). And since ACT and the social networks perspective do not contradict one another logically, an integration of the two could be achieved without the incompatibility issues confronting existing “mixed models” (Kornhauser 1978) of crime and delinquency.

Another potentially fruitful multilevel approach to understanding affective processes in violence and delinquency would develop ideas (not yet fully incorporated into the corpus of ACT) about social organization and “macroactions” (Heise 1996; Heise and Durig 1997). The term “macroactions” refers to social organization that emerges from the routinized, collaborative activities of interdependent actors. Macroactions materialize out of coalescing, chain-linked “event frames,” in which people/agents perform actions that align objects in a setting to generate products for a beneficiary. Production on an assembly line is an example of an event frame. Event frames are linked when the product of one event frame becomes the object to be manipulated in another. Purposeful activity occurring in a series of event frames creates a macroaction when a higher-level social system (e.g., an institution) is utilized as an instrument to transform an object into a product (e.g., when children become educated by public schools).

This perspective focuses on the cooperative relationship between individuals and environing social systems. Even so, it allows that power relations oftentimes are the motivation for directed activity (i.e., actors are induced to mobilize certain actions through exposure to a system of rewards and punishments). Therefore, when it comes to crime, we can conceive of relatively powerful actors at the level of the social system (e.g., the police) serving as “counter-instruments” in linked event frames. For example, a drug dealer (actor) uses potassium permanganate (an instrument) to purify a shipment of cocaine (object) and distributes (action) one-gram doses of it (product) to users (beneficiaries) in a disadvantaged neighborhood (setting). The beneficiaries in the first event frame become actors in a second, where the former product (i.e., a dose of cocaine) becomes a new object that is transformed into an experiential product: “getting high.” In either phase, the transformation of the object into a product may generate a “concert of social activity” (Heise 1996) involving the level of the social system (e.g., police intervention).

The key is to understand how and why the unfolding of certain event frames instigates disruptive macroactions. Luckily, criminology is rich with ethnographic and historical narratives about “organizational happenings” in crime-ridden milieus. Such data could be entered in Ethno, a computer program for qualitatively modeling likely structures among the micro- and macro elements of event frames, including dominance relations (for an illustration, see Heise and Durig 1997).7 The simulation program could be used to generate hypotheses about the synergy between criminal events and disruptive macroaction qua social control.

In spite of these hopeful prospects, a final problem with ACT is that more empirical tests of the theory are needed (Smith-Lovin 1987c). Certainly, the corroboration status of ACT has not been firmly established. After many years of theoretical refinement, ACT has only recently been subjected to rigorous empirical examination. As mentioned above, however, those tests of the theory that do exist provide general support its predictions. Therefore, while ACT would seem to have much to offer criminology, criminologists also have much to offer ACT insofar as their systematic application of the theory to crime and delinquency processes provides empirical feedback leading to its refinement and growth as a “general, generative interpretive sociology” (Smith-Lovin 1987c: 175).


Ajzen, Icek. 1982. “On Behaving in Accordance with One’s Attitudes.” In M.P. Zanna, E.T. Higgins, and C.P. Herman (eds.), Consistency in Social Behavior: The Ontario Symposium, vol. 2. Hillside, NJ: Erlbaum.

__________. 1988. Attitudes, Personality and Behavior. Chicago: Dorsey.

Ajzen, Icek, and Martin Fishbein. 1977. “Attitude-Behavior Relations: A Theoretical Analysis and Review of Empirical Research,” Psychological Bulletin 84:888-918.

Akers, Ronald L. 1996. “Is Differential Association/Social Learning Cultural Deviance Theory?” Criminology 34:229-47.

Ball-Rokeach, Sandra J. 1973. “Values and Violence: A Test of the Subculture of Violence Thesis,” American Sociological Review 38:736-49.

Bem, Daryl. 1970. Beliefs, Attitudes and Human Affairs. Monterey, CA: Brooks/Cole.

Bernard, Thomas J., and Jeffrey B. Snipes. 1996. “Theoretical Integration in Criminology.” In Michael Tonry (ed.), Crime and Justice. Chicago: University of Chicago Press.

Blumer, Herbert. 1969. Symoblic Interactionism: Perspective and Method. Englewood Cliffs, NJ: Prentice-Hall, Inc.

Carley, Kathleen. 1985. “An Approach for Relating Social Structure to Cognitive Structure,” Journal of Mathematical Sociology 12:1-26.

__________. 1986a. “Knowledge Acquisition and as a Social Phenomenon,” Instructional Science 14:381-438.

__________. 1986b. “Separating the Effects of Structure and Interaction,” Working paper. Department of Social and Decision Sciences, Carnegie-Mellon University. Pittsburg, PA.

Cloward, Richard A. 1959. “Illegitimate Means, Anomie, and Deviant Behavior,” American Sociological Review 44:588-608.

Cloward, Richard A., and Lloyd E. Ohlin. 1960. Delinquency and Opportunity. New York: Free Press.

Cohen, Albert K. 1955. Delinquent Boys. Glencoe, IL: Free Press.

Costello, Barbara. 1997. “On the Logical Adequacy of Cultural Deviance Theories,” Theoretical Criminology 1:403-28.

Fine, Gary Alan, and Sherryl Kleinman. 1979. “Rethinking Subculture: An Interactionist Analysis,” American Journal of Sociology 85:1-20.

Fishbein, Martin, and Icek Ajzen. 1974. “A Consideration of Beliefs and Their Role in Attitude Measurement.” In M. Fishbeing (ed.), Readings in Attitude Theory and Measurement. New York: Wiley.

__________. 1975. Belief, Attitude, Intention, and Behavior: An Introduction to Theory and Research. Reading, MA: Addison-Wesley.

Goffman, Erving. 1974. Frame Analysis: An Essay on the Organization of Experience. New York: Harper and Row.

Gordon, Robert A., James F. Short, Jr., Desmond Cartwright, and Fred L. Strodtbeck 1963. “Values and Gang Delinquency: A Study of Street-Corner Groups,” American Journal of Sociology 69:109-28.

Hagan, John. 1993. “The Social Embeddedness of Crime and Unemployment,” Criminology 31:465-91.

Heimer, Karen, and Stacy DeCoster. 1999. “The Gendering of Violent Delinquency,” Criminology 37:277-317

Heise, David R. 1966. “Social Status, Attitudes, and Word Connotations,” Sociological Inquiry 36:227-39.

__________. 1979. Understanding Events: Affect and the Construction of Social Action. New York: Cambridge University Press.

__________. 1985. “Affect Control Theory: Respecification, Estimation, and Tests of the Formal Model,” Journal of Mathematical Sociology 11:191-222.

__________. 1986. “Modeling Symbolic Interaction.” In S. Lindenberg, J. Coleman, and S. Nowak (eds.), Approaches to Social Theory. New York: Russell Sage Foundation.

__________. 1987. “Affect Control Theory: Concepts and Model,” Journal of Mathematical Sociology 13:1-33.

__________. 1991. Interact 2: A Computer Program for Studying Cultural Meanings and Social Interactions. Department of Sociology, Indiana University.

__________. 1996. “Social Order Through Macroactions: An Interactionist Approach.” Presentation at Panel on Micro-Macro Processes and Social Order, Ninth Annual Group Processes Conference, August, New York, NY.

__________. 2002. “Glossary of Terms Used in Affect Control Theory.” Retrieved July 19, 2002 from Indiana University (Bloomington) Affect Control Theory Web site:

Heise, David R., and Alex Durig. 1997. “A Frame for Organizational Actions and Macroactions.” Journal of Mathematical Sociology 22:95-123.

Heise, David R., and Elsa Lewis. 1988. Introduction to INTERACT: Documentation for Programs INTERACT and ATTITUDE. Raleigh, NC: National Collegiate Software Clearinghouse.

Heise, David R., and Neil J. MacKinnon. 1987. “Affective Bases of Likelihood Judgments,” Journal of Mathematical Sociology 13:133-51.

Heise, David R., and Lisa Thomas. 1989. “Predicting Impressions Created by Combinations of Emotion and Social Identity,” Social Psychology Quarterly 52:141-48.

Hirschi, Travis. 1969. Causes of Delinquency. University of California Press.

Hollingshead, A.B. 1939. “Behavior Systems as a Field for Research,” American Sociological Review 4:816-22.

Kim, M-S, and Hunter J.E. 1993. “Attitude-Behavior Relations: A Meta-Analysis of Attitudinal Relevance and Topic,” Journal of Communication 43:101-42.

Kornhauser, Ruth. 1978. Social Sources of Delinquency. University of Chicago Press.

Kuhn, Manford. 1972a. “Major Trends in Symbolic Interaction Theory in the Past Twenty-Five Years.” Pp. 57-76 in Symbolic Interactionism: A Reader in Social Psychology, 2 nd ed., edited by J.G. Manis and B.N. Meltzer. Boston: Allyn and Bacon.

Kuhn, Manford. 1972b. “The Reference Group Reconsidered.” Pp. 171-84 in Symbolic Interactionism: A Reader in Social Psychology, 2 nd ed., edited by J.G. Manis and B.N. Meltzer. Boston: Allyn and Bacon.

MacKinnon, Neil J. 1994. “Symbolic Interaction as Affect Control.” Albany, NY: State University of New York Press.

Massey, Douglas S., and Nancy Denton. 1993. American Apartheid. Cambridge: Harvard University Press.

Matsueda, Ross. 1988. “The Current State of Differential Association Theory,” Crime and Delinquency 34:277-306.

__________. 1997. “Cultural Deviance Theory: The Remarkable Persistence of a Flawed Term,” Theoretical Criminology 1:429-52.

Mead, George Herbert. 1934. Mind, Self and Society. Chicago: University of Chicago Press.

Miller, Walter B. 1958. “Lower Class Culture as a Generating Milieu of Gang Delinquency,” Journal of Social Issues 14:5-19.

Osgood, C.E., W.H. May, and M.S. Miron. 1975. Cross-Cultural Universals of Affective Meaning. Urbana: University of Illinois Press.

Osgood, C.E., G.J. Suci, and P.H. Tannenbaum. 1957. The Measurement of Meaning. Urbana: University of Illinois Press.

Ridgeway, Cecilia, and Lynn Smith-Lovin. 1994. “Structure, Culture, and Interaction: Comparing Two Generative Theories,” Advances in Group Processes 11:213-239.

Robinson, Dawn T., and Lynn Smith-Lovin. 1992. “Selective Interaction as a Strategy for Identity Negotiation,” Social Psychology Quarterly 55:12-28.

Robinson, Dawn T., Lynn Smith-Lovin, and Olga Tsoudis. 1994. “Heinous Crime or Unfortunate Accident? The Effects of Remorse on Responses to Mock Criminal Confessions,” Social Forces 73:175-90.

Sampson, Robert J., and John H. Laub. 1993. Crime in the Making: Pathways and Turning Points Through Life. Harvard University Press.

Schuman, Howard. 1995. “Attitudes, Beliefs, and Behavior.” In Karen S. Cook, Gary Alan Fine, and James S. House (eds.), Social Psychology. Needham Heights, MA: Allyn and Bacon.

Schneider, Andreas, and David R. Heise. 1995. “Simulating Symbolic Interactionism.” Journal of Mathematical Sociology 20:271-87.

Shaw, Clifford R., and Henry D. McKay. 1942. Juvenile Delinquency and Urban Areas. Chicago: University of Chicago Press.

__________. 1969. Juvenile Delinquency and Urban Areas. Revised Edition. Chicago: University of Chicago Press.

Shaw, Clifford R., Frederick Zorbaugh, Henry D. McKay, and Leonard S. Cottrell. 1929. Delinquency Areas. Chicago: University of Chicago Press.

Shibutani, Tomatsu. 1961. Society and Personality: An Interactionist Approach to Social Psychology. Englewood Cliffs, NJ: Prentice-Hall, Inc.

Short, James F., Jr.. 1998. “The Level of Explanation Problem Revisited: The American Society of Criminology 1997 Presidential Address.” Criminology 36:3-36.

Smith-Lovin, Lynn. 1987a. “Impressions of Events,” Journal of Mathematical Sociology 13:35-70.

__________. 1987b. “The Affective Control of Events within Settings,” Journal of Mathematical Sociology 13:71-101.

__________. 1987c. “Affect Control Theory: An Assessment,” Journal of Mathematical Sociology 13:171-92.

__________. 1990. “Emotion as the Confirmation and Disconfirmation of Identity: An Affect Control Model.” In Kemper, T.D. (ed.), Research Agendas in the Sociology of Emotion. Albany: State University of New York Press.

Smith-Lovin, Lynn, and William Douglas. 1992. “An Affect Control Analysis of Two Religious Groups.” In V. Gecas and D. Franks (eds.), Social Perspectives on Emotions, vol. 1. CT: JAI Press.

Sutherland, Edwin H.. 1939. Principles of Criminology. 3 rd ed. Philadelphia: Lippincott.

Sutherland, Edwin H., Donald R. Cressey, and David F. Luckenbill. 1992. Principles of Criminology. 11 th ed. General Hall.

Swidler, Anne. 1986. “Culture in Action: Symbols and Strategies,” American Sociological Review 51:273-286.

Thomas, Lisa, and David R. Heise. 1995. “Mining Error Variance and Hitting Pay-Dirt: Discovering Systematic Variation in Social Sentiments,” The Sociological Quarterly 36:425-39.

Wiggins, James A., and David R. Heise. 1987. “Expectations, Intentions, and Behavior: Some Tests of Affect Control Theory,” Journal of Mathematical Sociology 13:153-69.

Wilson, William Julius. 1991. “Studying Inner-City Social Dislocations: The Challenge of Public Agenda Research,” American Sociological Review 56:1-14.

Wolfgang, Marvin E., and Franco Ferracuti. 1967. The Subculture of Violence. Tavistock.


1. Walter Miller’s (1958) subcultural theory of “focal concerns,” differential association theory (Sutherland 1939; Sutherland, Cressey, and Luckenbill 1992), and Wolfgang and Ferracuti’s (1967) classic “subculture of violence thesis” can be considered among the class of pure cultural deviance theories.

2. Social disorganization theory (Shaw et al. 1929; Shaw and McKay 1942, 1969), status-frustration theory (Cohen 1955), and differential opportunity theory (Cloward 1959; Cloward and Ohlin 1960) can be considered among the class of mixed models of delinquency.

3. Indeed, social psychology informs us that cognitive elements as seemingly diverse as values, attitudes, and norms can all be thought of as “conceptual cousins” (Schuman 1995).

4. These predictions were derived from an ACT computer simulation program, INTERACT (discussed below). Online and downloadable versions of the program, other related software, and useful ACT readings are available at David Heise’s ACT Web site:

5. This theoretical position is similar to Cohen’s (1955), who argues that delinquent subcultures provide a means for denigrated working-class boys to cope with their “problems of adjustment.”

6. Yet, one could import more thorough interactionist work on socialization (see, e.g., Shibutani 1961, especially pp. 471-501) to specify how meanings are learned in (sub)cultural contexts.

7. Ethno, including documentation and references, is freely available on the Web at the following URL:

© Electronic Journal of Sociology