Media_Effects

Paired spirals of influence page 1 10/27/05 Paired spirals of influence: A system dynamics model for understanding media use and effects Michael D. Slater School of Communication The Ohio State University Running head: Paired spirals of Influence The author thanks Lindsay Hoffman, The Ohio State University, for her insights and literature search efforts, and Kim Henry, University of Colorado for comments and suggestions regarding the manuscript. Paired spirals of influence page 2 Abstract The attitudinal or behavioral outcomes of media use studied by media effects researchers also can typically be expected to influence selection of and attention to media content. The consequence of this process of mutual influence over time can be conceptualized as a positive feedback loop, or mutually-reinforcing spirals. Implications of such a perspective include increased attention to more complex, longitudinal modeling of media selection and effects processes; impact of such spirals of influence in youth and adolescent development; analysis of social and psychological factors that control, dampen, or eventually extinguish the influence of such spirals. The influence of relatively subcultures or national cultures that are relatively open or closed with respect to communicative influences are also explored, with implications for social change, social conflict, or homeostasis. Paired spirals of influence page 3 Paired spirals of influence: A system dynamics model for understanding media use and effects Researchers in communication, psychology, sociology and allied social science disciplines have established strong empirical support for effects of media exposure on aggression (Anderson & Bushman, 2002), fears about the social world (Gerbner, Gross, Morgan, Signorielli & Shanahan, 2002), political behavior (Graber, 2002) and a wide variety of other outcomes (see Bryant & Zillman, 2002 for a review). Similarly, there is a rich body of literature documenting the reasons people select and use media channels, genre, and content (Kim & Rubin, 1997; Zillman & Bryant, 1985). Surprisingly, however, there has been limited systematic effort to synthesize the process of media selection and media effects into a more comprehensive model, though the potential value of such efforts have been previously noted (Rubin, 2002; Windahl, 1981). The present paper explores such a model. This model is intended in particular to provide a framework for studying media selectivity and effects processes as dynamic processes over time, adapting for this purpose concepts from general system theory and cybernetics (von Bertalanffy, 1968). Such a model should have provocative implications in particular with respect to understanding the influence of mediated communication on adolescent socialization and on development and maintenance of political, religious and lifestyle subcultures in contemporary societies. I therefore outline the basic propositions of the model, acknowledge the many earlier models that share at least implicitly some of the same propositions, describe distinctive and research questions that arise as a function of applying the model, and suggest some ways the model might be used to help understand important social phenomena. Paired spirals of influence page 4 Media Use as Endogenous As students of audience selectivity and uses and gratifications tell us, preferred media channels and genres of media content as well as the specific types of content or elements of content attended to are a function of a person’s age, gender, disposition, prior experience, mood, ideology, social influences, and social identity (Zillman & Bryant, 1985). Conversely, as noted above, there is ample evidence for effects of such content on beliefs, attitudes, and behavior—the components of social identity—even after controlling for these prior influences. From a theoretical perspective, the role of media use, then, is typically an intervening one. In statistical terms, media use should mediate or partially mediate (Baron & Kenny, 1986) the influences of individual difference variables such as age, gender, education, prior experience and interests on cognitive or behavioral outcomes such as health attitudes or behavior (citation withheld for blind review). Stated more formally, as a proposition and a testable hypothesis: Proposition 1. In most media effects contexts, the role of media can be modeled as endogenous and mediating or partially mediating the effect of other individual difference variables on the outcomes of interest. I hasten to note that this is not an assertion that every study of media effects must therefore analyze media use variables such as exposure or attention as mediators rather than independent variables. Sometimes the endogeneity of media use simply is not of theoretical or practical interest. Moreover, analyzing media use as endogenous can be clumsy when other mediators are the primary focus of research attention, potentially resulting in a four-step model. It is all the more complex when one seeks to parse out effects of the two major types of indicator of media use, media exposure and media Paired spirals of influence page 5 attention (e.g., citation withheld for blind review). However, proposition 1 is an assertion that the fullest and most accurate depiction of a media effects process can typically best be modeled by assessing both selectivity and effects within the same analysis. Media Selectivity and Effects as Structured, Dynamic Processes Mediation of the kind described above is for some phenomena an excessively simple formulation of the relationship between media selectivity and media effects processes. For example, Proposition 1 would argue for a model of media effects on adolescent aggression in which exogenous variables such as gender, age, substance use, and prior victimization predict both aggression and use of violent media, and the latter partially account for the relationship between the exogenous variables and aggression as well as, perhaps, providing an independent contribution of their own. However, it is also probable that media use is predicted by prior values on the outcome variable itself. In this example, an adolescent’s aggressiveness is likely to predict use of violent media. This pattern is commonplace enough that it may serve as an additional proposition: Proposition 2. In most media effects processes, the outcome influenced by media use also influences media use. The notion that media selectivity and media effects together form a reciprocal, mutually-influencing process is hardly new. For example, Klapper’s classic review of media effect research argued that most often media content served to reinforce existing beliefs (Klapper, 1960). In building this argument, he cites a community study (Berelson, Lazarsfeld & McPhee, 1954) with results not only suggesting that political campaign content primarily reinforced existing beliefs, but that increased exposure to the campaign led to selective exposure to campaign content consistent with their beliefs. Paired spirals of influence page 6 Another classic review of television effects suggest that actions influenced by television may lead to additional information seeking from television regarding those or alternative behaviors (Comstock, Chaffee, Katzman, McCombs & Roberts, 1978). For a more recent example, it has been argued that the sexual content in media may increase interest in sexuality that stimulates further interest in sexual content as well as possible increases in sexual activity among adolescents (Brown, 2000). Acknowledging the reciprocal nature of selectivity and media effect is an important starting point. However, the concept of reciprocity is misleading. Causal relationships, as we will discuss in more detail below, do not go “back and forth” as the term “reciprocal” implies. They move forward in time, and should be conceptualized as such. Therefore, it is a relatively straightforward matter to specify the minimum assumptions of such reciprocal relationship as it must occur dynamically, over time. Exposure at baseline must lead to an effect at time 2 that influences exposure at time 3; simultaneously, the status of the effect or outcome at baseline should lead to exposure at time 2 leading to the effect at time 3: Proposition 3. The assertion of reciprocal relationships between media use and effects of such media use in its simplest case implies a three-step cross-lagged process (see Figure 1). INSERT FIGURE 1 HERE Calling a causal association a reciprocal relationship, with at minimum the features illustrated in Figure 1, raises far many more questions than it answers. What are the questions that arise even from this simplest case' As is detailed below, these questions include issues of time lag, differences in lag for different paths, relative strength of paths, identifying those social or psychological forces that may restrain or Paired spirals of influence page 7 limit the mutual influence process, and exploring the social and cultural implications of these patterns of media selectivity and influence. Paired Spirals of Influence It should be apparent from Figure 1 that if some type of media use tends to influence some belief or behavior (e.g., support for strong controls over illegal immigration or aggressiveness), and that belief or behavior tends in turn to increase that type of media use (e.g., listening to conservative radio talk shows or seeking violent media content), the process should be mutually reinforcing over time. Persons entering into this process should tend towards continued or increased use of the media content in question, and to maintenance or strengthening of the attitude or behavior in question. Whether one takes media use or selectivity effects as a starting point, then, reciprocal effects over time can be conceptualized or visualized as a spiral of ongoing influence. The term spiral also has another advantage: the connotation of a situation escalating or spiraling out of control does allude to the potentially cumulative effects of a process of mutual reinforcement (see the discussion of positive feedback loops, below). The weakness of the spiral image is that it illustrates only prospective relationships. As Figure 1 suggests, one can conceptualize and analyze these relationships as two paired, complementary spirals, one beginning with the outcome predicting media use and one with media use predicting the outcome. This permits portrayal of concurrent—or virtually concurrent—influences as well as prospective ones. Spiral processes of media and other communication influence are most familiar from spiral of silence theory (Noelle-Neumann, 1974). This theory will be addressed later from the perspective of the present approach, once that approach has been more fully explained. Another theoretical analysis, however, also deserves acknowledgement Paired spirals of influence page 8 here: a paper by Price and Allen (1990) in which they propose the argument that the principal value of spiral of silence theory is not its specifics but the linkage it makes between an individual’s communication choices and behavior (in this case, small group dynamics), and the influence of mass media. They argue that understanding such influences is central to understanding public opinion processes (Price & Allen, 1990). In the present paper, Price and Allen’s argument is extended and generalized with respect to media effects on beliefs and behavior in general, and focuses in particular on selectivity of media use and other communicative choices as a key element of the process. Public opinion may be considered a particularly complex and important case for this more general model. Two recent lines of work have exemplified the paired spiral of influence approach to understanding media effects. Recent work on violent media content and adolescent aggressiveness found that, as predicted, use of violent media content was prospectively predictive of adolescent aggressiveness and aggressiveness prospectively predicted use of violent media content in a four-wave panel data set. Moreover, when a stricter model was imposed looking only at effects of fluctuations from a youth’s own mean over-time trend in violent media content use or aggression, greater than an individual’s typical violent media content use was prospectively predictive of greater aggressiveness; the opposite path was true concurrently though it didn’t reach significance prospectively (Slater, Henry, Swaim & Anderson, 2003). These authors referred to the pattern of mutual reinforcement they found as a “downward spiral model.” Another study of a two-wave national panel data set found that that motivation to obtain political information was prospectively predictive of information processing and information processing predicted knowledge. However, such relationships were also Paired spirals of influence page 9 found in the other direction (Eveland, Shah & Kwak, 2003). Eveland and his colleagues noted the over-time implications that would be likely as this pattern repeated over time, resulting for those with little political knowledge and interest in decreasing interest over time, in what they also called a downward spiral process, and conversely for those with greater interest an upward spiral might be expected, and proposed this mechanism might help explain knowledge gaps with respect to political information. Positive feedback loops. In the language of cybernetics and system theory, the pattern in which some type of media exposure would lead to some attitude or behavior leading to greater exposure to that type of media, and so on over time, would be called a positive feedback loop. A loop exists only as a graphic convenience when drawing a system diagram, however. The arrow of time in physical and social experience, of course, goes only one way. A process of mutual reinforcement over time can be conceptualized as a spiral, in which influences tend to strengthen each other. In principle, according to theories of system dynamics (Shea & Howell, 2000; von Bertalanffy, 1968; Wiener, 1961), a positive feedback loop, or a spiral of influence as illustrated in Figure 1, should produce the following: Proposition 4. In a perfectly closed system (free from the effects of competing outside social, psychological, or environmental influences) the spiral of influence of media selectivity and effect should work to maximize use of a given type of media to the maximum permitted by available time and access, and maximize levels of the cognitions or behaviors impacted to those permitted by available capacity. Paired spirals of influence page 10 However, there are always competing social, psychological, and environmental influences present, though to varying degrees: Proposition 4A. Social subsystems (or subcultures) in which media effect processes take place vary in the extent to which they are open and closed along various dimensions; none are fully closed, and spirals of influence are limited by environmental or other constraints. Dampening positive feedback loops and achieving, homeostasis in spirals of influence. Among the most important research questions that arise when studying spirals of influence in media effects regards factors that limit these feedback loops, as suggested by Proposition 4A: Research question set 1: What are the a) environmental and b) psychological forces that dampen these positive feedback loops and that eventually bring them to some kind of steady state or homeostasis, or even extinguish their effects' In many cases, of course, these forces are relatively easy to suggest. With regard to violent media content use and violence, environmental factors presumably include family, peer, and institutional or legal consequences for aggressive behavior. Internal factors may include maturation and assumption of social roles that lead to reduced consumption of such media and of expression of aggression, habituation to violent media stimuli, and beliefs and values internalized from other sources, mediated and unmediated, inconsistent with enacting aggression. To some extent, such processes were evidenced by a study of moderators of the violent media content and aggression relationship. Using the same data set described above, researchers found that this relationship was strengthened among adolescents who were particularly vulnerable to mediated influence, such as high sensation-seekers who Paired spirals of influence page 11 have less inhibition about enacting risky behavior and youth who were being bullied by others, who likely were at such times feeling more isolated and less integrated into prosocial cultures (Slater, Henry, Swaim & Cardador, 2004). The effect of the influence spiral, then, may be to achieve homeostasis at least temporarily, so that, in this example, aggressive tendencies may be more likely to survive despite increasing social and developmental factors tending to extinguish such aggression. In other words, the cumulative effects of a spiral of influence may appear as non-cumulative or even weakening, depending on whether countervailing forces are becoming increasingly strong and the extent to which the immediate social environment (including patterns of media use) for an individual is relatively open or closed. Influences also may be self-limiting, as in habituation to a media stimulus or when ceiling effects intervene with respect to practical opportunity to access or spend time with the stimulus, or simply the point at which overexposure to the media stimulus type simply gets repetitive and boring. Spiral of silence: Positive feedback leading to paralysis. The most prominent spiral effects theory in communication is, of course, spiral of silence (Noelle-Neumann, 1974). The spiral of silence is a theory of the silent majority. It asserts that members of the public with conservative viewpoints, in a nation with a predominantly liberal press, refrain from political expression and participation because they believe their positions are marginal and isolated. This lack of expression leads to further invisibility in the media and continued reluctance to express conservative viewpoints. I have deferred discussing this theory in detail for two reasons. One is because it is highly controversial given its ideological implications and possible ideological roots. I did not want to obscure the key ideas of this essay by linking them to a theory that is Paired spirals of influence page 12 distasteful to many scholars given its ideological cast, and with somewhat problematic empirical support (Scheufele & Moy, 2000). Second, the phenomenon it proposes represents a more unusual version of the positive feedback or spiral process, in the context of media effects. Instead of a feedback process in which media use reinforces a belief or behavior which leads to increased media use, it is the opposite type of positive feedback loop: one in which the spiral process leads to paralysis, a regression to zero instead of towards infinity. (Note, however, that Eveland, et al. [2003], in their discussion of the evolution of the political “chronic know-nothing”, also describes a similar regression towards zero.) Several points regarding this theory are relevant here. First is acknowledgement of its technical achievement. The theory addresses the reciprocal relationship between media use and the outcome of media use as a dynamic, over-time process, key elements of media effects theorizing argued for here. In so doing, explanation moves seamlessly from media’s effects on the behavior of individuals to influence on segments or subcultures within society—which I will argue is likely often to be a feature of spiral of influence theorizing and research. There is much to be learned here. Second is the closed system assumption that underlies claims of such a strong spiral effect. Such an assumption would require both a uniformly hostile media environment and a lack of competing alternative media or other collective communication channels through which competing beliefs and norms can be shared. In an era of Internet communication and multiple radio and cable television channels, such an assumption is unlikely to often hold in contemporary democratic societies, which spiral of silence theory is intended to describe. Paired spirals of influence page 13 Third is the recognition that the previous critique itself suggests a distinctive dynamic, in which alternative mediated communication channels, as well as associated interpersonal and group channels, can be actively selected by people in society who do not fully identify with values expressed in the mainstream media, and which can provide a more conventional reinforcement spiral leading to increased identification with some subculture’s set of values. Paired Spirals of Influence and Maintenance of Political and Social Identity and Subcultures Another contribution of a spiral approach is the focus on the role of communication and media on the formation and maintenance of subculture within the larger society. In some cases, paired spirals of influence may take place as incidental sequelae of a preference for a media type—violent, erotic, consumer-oriented—that arises from idiosyncratic, individual interests and preferences, though perhaps with nontrivial consequences given the number of people affected over a national or international population. In other cases, as suggested by Price and Allen (1990), the spirals may help define and sustain communities or subcultures of opinion; I would add the same is oftentimes true for communities of religion, lifestyle, place, and class. It also should be noted that while my discussion, consistent with my own interests and training, has emphasized mediated communication influences, in the context of subculture, the spiral of influence process must be understood in terms of a full range of communicative choice, including group affiliation and individual communication networks, as well as use of mediated communication (Price & Allen, 1990). In general, we are likely to find that members of a given subculture will share preferred media outlets and channels, and selectively attend to programming and content Paired spirals of influence page 14 that reflect and share the values of that subculture as well as providing instrumental information of use to members of that subculture (Kelly & Donohew, 1999). Such choices are consistent with and would potentially sustain the process of social and personal identity formation theorized to be integral to maintenance of social groups (Blanton & Christie, 2003; Tajfel & Turner, 1986; Turner, Oakes, Haslam & McGarty, 1994). For example, distinctive patterns of media and other communicative choice can easily be discerned in the U.S. for the evangelical, religious right, for gay activist community, for marginalized youth, for middle-class, high-consumption teen girls (Brown, 2000; Morton & Duck, 2000; Wilson, 2002). Identification with such subcultures itself may be viewed as a spiral process of media and communicative influence, in which identification leads to media selection reinforcing social identity, ad infinitum: Proposition 5. Political and social subcultures, and individual participation in such subcultures, are in part defined and are largely maintained by spiral patterns of media selection and influence (as well as by associated preferences in group and individual association and communication). Open and closed communication subcultures: some implications of spiral models of influence. As discussed earlier, such spirals of influence are typically limited to the extent to which participants in a subculture exist within an open system. Mainstream values and perspectives, in a contemporary democratic society, are difficult to shut out entirely. However, the effects of a spiral of influence are likely to be particularly strong in subcultures that seek with some success to motivate closure, often through a culture of suspicion of outside influences especially as conveyed via the media, use of subculturespecific media such as websites, books and magazines that consistently reiterate a Paired spirals of influence page 15 consistent and distinctive worldview, and of course interpersonal and group networks that largely or completely exclude non-participants in the subculture. In the U.S., the more dramatic examples might include militias and white supremacy groups, and to a lesser extent very conservative evangelical communities which are often distinguished by their desire to school their children separately, limit access to mass media influence, and to encourage near-exclusive use of radio, web, and cable outlets reflecting their outlooks (Freilich, 2003; Keddie, 1998). Similar patterns, of course, are likely to be found in subcultures of the left. One feature of such relatively-closed systems—perhaps particularly among those participants who voluntarily opt for a closed communication experience—is that the resulting isolation or separation may find expression in increased suspicion and hostility towards members of out-groups (Ellemers, Spears & Doosje, 2002; Tajfel & Turner, 1986). Some of those in America who have enacted violence against perceived outgroups, whether the Ku Klux Klan members through mid-century, some Students for a Democratic Society members in the late sixties, or various school shooters, Tim McVeigh, and Eric Rudolph in recent years, appear to have been participants in subcultures that maintained largely closed communicative systems. Contemporary Islamic fundamentalism, of course, is a particularly dramatic example of a relatively closed communicative system. It has proven particularly effective with respect to developing interpersonal and media channels that reinforce a spiral of influence and maintain a closed, self-sufficient communicative system, by birthing the radical madrasa that socializes these perspectives, and providing a base of support for a highly professional news network, Al Jazeera, that provides broadcast news more congenial to their worldview than otherwise available (Ayish, 2002; Zaman, 1999). Paired spirals of influence page 16 Proposition 6. The more closed the communication system within a subculture, the stronger the effect of spirals of influence (i.e., the less dampening there will be of positive feedback loop mechanisms) and the more likely participants in the subculture will view out-group members with hostility and perhaps even as legitimate targets of violence. The effects of a closed communicative environment, as in the case of extreme Islamic fundamentalism, are likely to be magnified in the absence of a diverse and open communication environment that can provide competing influences that would serve to dampen the influence spiral. Proposition 7. The effect of closure in subculture communication patterns on isolation and inter-group hostility will be exacerbated to the extent the national system also tends to be a closed one. It is also noteworthy that, given the increased movement towards audiencespecific media and channels in the era of Internet and cable television, and away from a few major broadcast television sources that characterized the 50s through the 80s (MacDonald, 1994), it may be that such subculture identification becomes increasingly the norm rather than a statement of perceived marginalization. In other words, two potentially countervailing forces exist with respect to spirals of influence within subcultures: the increased diversity and accessibility of communication content and exchange, and the ability to identify and access sources closely reflecting one’s own views and values that can become an exclusive or near-exclusive source of information and conversation. There is little new, of course, in the larger points being made here about openness and closure in political, religious, and lifestyle subcultures. The value of the paired Paired spirals of influence page 17 spirals of influence framework should instead be in the questions we as researchers and scholars are encouraged to explore from the paired spirals perspective: Research question set 2. What are the structural forces which militate against, punish, or reward openness and communicative exchange within a subculture, and how does this influence the action of a spiral of influence' What are the individual differences (psychological, demographic, etc.) that lead to greater or lesser willingness to seek or process perspectives from outside of a given subculture with which an individual may identify, and therefore dampen or fail to dampen that spiral' What is the impact of place—being a participant in a subculture in a physical environment in which one is a majority versus being a minority or a true isolate—on use of media and other communication channels to sustain social identity with the subculture' How does this interact with the structural characteristics of the subculture or the individual characteristics of the group participant' What is the role of the Internet and other media innovations in encouraging openness or closure of various subcultures as well as of national systems' Substantive implications—the role of subculture diversity and openness and closure to communicative influences within subcultures. The substantive implications of the comments above regarding the implications of subculture openness versus closure are of course readily apparent. Closure in subcultures increases the potential of conflict with out-groups, sometimes to the point that people get hurt or die. Diversity of subculture within a relatively open social or national system is sometimes celebrated and sometimes bemoaned. There may be an objective sense, however, in which such diversity is advantageous. Social, technological, historical and Paired spirals of influence page 18 environmental changes take societies in unexpected directions. How does evolution in social norm and social practice happen' As Price and Allen (1990) note, traditional models of social influence emphasize conformity, and do not provide good explanations of how change in a society takes place. They cite Moscovici’s (1985) discussion of group innovation, in which change can happen in a group to accommodate a visible and inflexible minority. This suggests a possible advantage to diversity: a society with diverse subcultures may offer the opportunity for some to thrive and grow when historical circumstances favor their development over others, making a society more able to adapt and change over time. However, subcultures themselves vary in the extent to which they are open or closed. There is no guarantee that relatively open subcultures will prove more adaptive in a changing environment than a closed one that happens to be well-matched to the newly-emerging environmental context. In fact, Moscovici’s work even suggests that a relatively closed or inflexible subculture may have particular influence on the larger system—at least if their small group experimental results can be generalized to larger social systems (Moscovici & Lage, 1976; Moscovici, 1985). As a closed system grows in power, the larger system presumably becomes more closed and, given the lesser acceptance of out-groups and willingness to engage in or permit communicative exchange, potential for conflict or oppressiveness increases. This perspective, too, may lead to some useful research directions: Research question set 3. Does the spiral of media and other communicative choice, and corresponding impacts on beliefs and behavior, become more influential and less dampened as a subculture becomes itself more influential in a larger society' Under what circumstances does the openness or closure of a Paired spirals of influence page 19 subculture to external communicative influences make it a more likely candidate for successful adaptation to changing historical conditions' Identifying Lag Times and Possible Asymmetry of Spirals. Earlier, a number of other issues relevant to describing the distinctive patterns of various spirals of influence were briefly noted. Principal among these are the nature of the lags themselves. Time-ordered relationships can only be detected if measurement points are no further apart than the plausible duration of influence from a causal factor to a potential effect. Typically, panel research designs must choose time lags based on the logistics of staffing and data collection, and on the very sensible desire to minimize sensitization that would likely result from closely repeated data collection points. Unfortunately, such lags often are months apart at best; the effect of mediated stimuli such as advertisements or the influence of fluctuations of life situation on media use choices may be far shorter-lived than this, and therefore be detectable only or primarily as cross-sectional associations. Price and Allen (1990) suggest that processes of social norm change (and perhaps with them changes in patterns of mediated and group communication) can be expected to take years to unfold; in contrast, a study of the relationships between perceived efficacy and performance assessed four trials in a period of two hours (Shea & Howell, 2000). Social network researchers have argued that the diffusion of an innovation depends on the number and timing of feedback cycles leading to the innovation becoming normative; a fully diffused innovation, they suggest, is the result of the establishment of a positive feedback loop (Abrahamson & Rosenkopf, 1997). It should also be noted, as a methodological concern, that claims of lagged effects may often be artifacts of the fact that the time 1 predictor is correlated with the time 2 predictor; the effect of the predictor at time 1 may be because the time 1 predictor in a Paired spirals of influence page 20 sense serves as a proxy for time 2 effects. Therefore, it is important to match lags in data collection where possible (perhaps through observational data collection to reduce sensitization) to plausible duration of causal effects, or one must accept having to try to interpret causal direction in concurrent relationships, perhaps through use of instrumental variable techniques. Another implication of the problem of lag to effect is the likelihood that the two types of effects driving a spiral of influence do not have the same duration to effect. Some outcomes of communication use, such as those due to information gain or social modeling, may have more lasting effects. Effects such as the impact of an attitude or behavior (especially ones that aren’t highly stable over time) on media selection are likely to be more immediate. Such a pattern would show up as an asymmetric spiral, in which the selectivity paths were primarily detectable with respect to concurrent relationships, but effects of media use might be detectable in lagged analyses, as was found as described earlier in Slater, et al. (2003). There are, then, a series of questions that arise with respect to the nature and dynamics of any instance of a spiral influence process. These might include: Research question set 4. What is the duration of a media or other communication effect, or the effect of this outcome on communication content or context selection' What is the comparative duration and strength of each path' Does the relative strength vary as a function of the duration measured' How stable, too, are these paths over time' Does the strength of an association decrease or increase as a cumulative effect, or as a function of maturation or of external influences such as accessibility of media content, changes in social context and group membership, etc.' Paired spirals of influence page 21 Analyzing Spirals of Influence: Old and New Methodological Tools A full discussion of the alternative statistical and research design methods for probing spirals of influence is beyond the scope of this paper, and, if pursued in technical detail, beyond the competence of the present author. However, some useful approaches and alternatives should be acknowledged at least in brief, especially as emergence of newer statistical techniques has made possible more sophisticated and complete analyses of these relationships. Figure 1 illustrates a three-wave, cross-lagged model. Autoregressive crosslagged regression analyses, the traditional method for analyzing such models, have been strongly criticized (Rogosa & Willett, 1985). However, other authorities have pointed out that such analyses conducted in a regression or structural equation modeling context, while involving more restrictive assumptions than growth modeling methods, can be appropriate and informative (Curran & Bollen, 2001). Certainly, structural equation models using chi-square difference tests to examine the incremental contribution of each lag (Bentler, 1990) of a cross-lagged model such as that illustrated in Figure 1, along with examination of various asymmetric alternatives still consistent with an influence spiral, is a straightforward and easily interpreted approach to such a test. Ideally, it could be triangulated with some of the other methods noted below. Another approach is use of hierarchical, multi-level models to analyze panel data (Raudenbush & Bryk, 2002). This approach has been preferred to growth modeling in a structural equation modeling format when time intervals between points of data collection among members of a panel vary, a commonly-occurring circumstance that structural equation growth modeling was ill-equipped to manage. (It should be noted that more recently software packages such as M-plus Version 3 [ (Muthen & Muthen, 1998- Paired spirals of influence page 22 2004)] have become available that can combine capabilities of multi-level and structural equation modeling.) A latent-difference score approach to such models has also been proposed (Ferrer & McArdle, 2003). This approach appears promising but has not been widely used or replicated as yet. A multi-level model approach was used in Slater, et al. (2003), described above. Perhaps the most noteworthy aspect of this analysis from a methodological viewpoint was the analysis of intra-individual or within-person variability over time. Such analyses permit more rigorous tests of relations in spiral influence models. For example, one way of understanding the Slater et al. findings from a methodological viewpoint is that when simple between-subject effects were studied, prospective paths of both types were relatively easy to find in part because one is analyzing the stable element of exposure and behavior in an individual. Apparent prospective effects may be confounded with all sorts of individual differences that give rise to these between person differences. When one looks at within-person fluctuation, as in any repeated-measures analysis, the betweenperson confounds (as well as age and maturational issues, because the analysis looks at deviation from a slope over time) are controlled. It should be noted that analyses of the predictive effects of variability from an overall norm or average could also be employed to assess the effects on individuals of being a minority or majority in a community of place or other social group, examining how that moderates spiral influence processes, as suggested in question set 2. Another approach that has considerable potential for examination of spiral processes and dynamics involve use of rolling cross-sectional data collection and lagged time-series analyses (Fan, 2002), though to my knowledge they have not been widely used explicitly to examine such spiral effects as such. The rolling cross-sections used in Paired spirals of influence page 23 such analyses make possible examination of much tighter time lags without risk of sensitization, and can examine societal influences of, for example, media content on various health related belief and policy outcomes (Yanovitzky, 2002). However, the rolling cross-sections limit the ability to explore individual psychological mechanisms and moderators. Taken together, such methods could do much to elucidate specific spiral processes within a given literature. Conclusion: Spirals of Influence and the Role of Mediated, Group, and Interpersonal Communication The paired spirals of influence approach may have appeal as a general model of many social phenomena for a simple reason: Any reciprocal relationship involving attitudes or behaviors as influenced by external factors, carried forward in time, will tend to have spiral tendencies or elements. The general set of considerations proposed here, regarding dampening effects, internal and external moderators, effects of openness or closure of a subculture or surrounding society, are likely to be in large measure applicable. However, the role of communication appears to be of particular importance for those social and individual processes that can be viewed as spirals. A spiral, after all, can be conceptualized as a feedback system for creating and maintaining personal and social identity. As external influences may help shape such identities, increasing identification with a religious, political, lifestyle, or other kind of subculture should lead to selectively seeking out social relationships, communication exchanges, and mediated presentations and interpretations of the larger social world that express and reaffirm that identity. In particular, a paired spirals of influence model is a conceptual framework for exploring how specific subsets of external influences, typically communicative ones, may Paired spirals of influence page 24 shape individual and group identity and behavior. It also poses specific questions about how such identities are negotiated in the context of a larger society with competing communicative options and identities, and the circumstances under which that negotiation may break down and result in uncivil or destructive outcomes. It highlights research design and data collection issues and opportunities that may facilitate such investigations. In a more immediate and practical way, a paired spirals of influence model may highlight conceptual and methodological tools to better understand the specifics of the relationships between mediated and other communicative experiences and political, health and lifestyle, aggressive, and other social behaviors. Paired spirals of influence page 25 References Abrahamson, E., & Rosenkopf, L. (1997). Social network effects on the extent of innovation diffusion: A computer simulation. Organization Science, 8, 289-309. Anderson, C. A., & Bushman, B. J. (2002). The effects of media violence on society. Science, 295, 2377-2379. Ayish, M. I. (2002). Political communication on Arab world television: Evolving patterns. Political Communication, 19, 137-154. Baron, R. M., & Kenny, D. A. (1986, December). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personal Social Psychology, 51(6), 1173-1182. Bentler, P. (1990). Comparative Fit Indexes in Structural Models. Psychological Bulletin, 107, 238-246. Berelson, B., Lazarsfeld, P. F., & McPhee, W. N. (1954). A Study of Opinion Formation in a Presidential Campaign. Chicago: University of Chicago Press. Blanton, H., & Christie, C. (2003). Deviance regulation: A theory of action and identity. Review of General Psychology, 7, 115-149. Brown, J. D. (2000). Adolescents' sexual media diets. Journal of Adolescent Health, 27, 35-40. Bryant, J., & Zillman, D. (2002). Media Effects: Advances in Theory and Research. Mahwah, NJ: Lawrence Erlbaum. Comstock, G., Chaffee, S., Katzman, N., McCombs, M., & Roberts, D. (1978). Television and human behavior. New York: Columbia University Press. Paired spirals of influence page 26 Curran, P. J., & Bollen, K. A. (2001). The best of both worlds: Combining autoregressive and latent growth models. In L. M. Collins & A. G. Sayer (Eds.), New methods for the analysis of change (pp. 107-135). Washington, D.C.: APA. Ellemers, N., Spears, R., & Doosje, B. (2002). Self and social identity. Annual Review of Psychology, 53, 161-186. Eveland, W. P., Shah, D., & Kwak, N. (2003). Assessing causality in the cognitive mediation model: A panel study of motivations, information processing, and learning during campaign 2000. Communication Research, 30, 359-386. Fan, D. P. (2002). Impact of persuasive information on secular trends in health-related behaviors. In R. C. Hornik (Ed.), Public health communication: Evidence for behavior change (pp. 251-264). Mahwah, NJ: Lawrence Erlbaum Associates. Ferrer, E., & McArdle, J. J. (2003). Alternative structural models for multivariate longitudinal data analysis. Structural Equation Modeling, 10, 493-524. Freilich, J. D. (2003). American Militias: State-Level Variations in Militia Activities. New York: LFB Scholarly Publishing. Gerbner, G., Gross, L., Morgan, M., Signorielli, N., & Shanahan, J. (2002). Growing up with television: Cultivation processes. In J. Bryant & D. Zillman (Eds.), Media Effects: Advances in Theory and Research (pp. 43-68). Mahwah, NJ: Lawrence Erlbaum. Graber, D. A. (2002). Mass media and American Politics. Washington, D.C.: CQ Press. Keddie, N. R. (1998). The new religious politics: Where, when, and why do "fundamentalisms" appear' Comparative Studies in Society and History, 40, 696723. Paired spirals of influence page 27 Kelly, K., & Donohew, L. (1999). Media and primary socialization theory. Substance Use & Misuse, 34, 1033-1045. Kim, J., & Rubin, A. M. (1997). The variable influence of audience activity on media effects. Communication Research, 24, 107-135. Klapper, J. T. (1960). The effects of mass communication. New York: The Free Press. MacDonald, J. F. (1994). One Nation Under Television: The Rise and Decline of Network TV. New York: Pantheon Books. Morton, T. A., & Duck, J. M. (2000). Social identity and media dependency in the gay community: The prediction of safe sex attitudes. Communication Research, 27, 438-460. Moscovici, S. (1985). Innovation and minority influence. In Serge Moscovici, Gabriel Mugny & Eddy Van Avermaet (Eds.), Perspectives on Minority Influence (pp. 951). Cambridge: Cambridge University Press. Moscovici, S., & Lage, E. (1976). Studies in social influence: III. Majority and minority influence in a group. European Journal of Social Psychology, 6, 149-174. Muthen, L. K., & Muthen, B. O. (1998-2004). Mplus User's Guide. Third Edition. Los Angeles: Muthen & Muthen. Noelle-Neumann, E. (1974). The spiral of silence: A theory of public opinion. Journal of Communication, 24, 43-51. Price, V. & Allen, T. (1990). Opinion spirals, silent and otherwise: Applying smallgroup research to public opinion phenomena. Communication Research, 17, pp. 369-392. Raudenbush, S., & Bryk, A. (2002). Hierarchical linear models: Applications and data analysis methods (2nd ed.). Thousand Oaks, CA: Sage Publications, Inc. Paired spirals of influence page 28 Rogosa, D., & Willett, J. B. (1985). Satisfying simplex structure is simpler than it should be. Journal of Educational Statistics, 10, 99-107. Rubin, A. M. (2002). The uses-and-gratifications perspective of media effects. In J. Bryant & D. Zillman (Eds.), Media Effects: Advances in Theory and Research (pp. 525-548). Mahwah, NJ: Lawrence Erlbaum. Scheufele, D. A., & Moy, P. (2000). Twenty-five years of the spiral of silence: A conceptual review and empirical outlook. International Journal of Public Opinion Research, 12, 3-28. Shea, C. M., & Howell, J. M. (2000). Efficacy-performance spirals: An empirical test. Journal of Management, 24, 791-812. Slater, M. D., Henry, K. L., Swaim, R. C., & Cardador, J. M. (2004). Vulnerable teens, vulnerable times: How sensation seeking, alienation, and victimization moderate the violent media content-aggressiveness relation. Communication Research, 21, 642-668. Slater, M. D., Henry, K. L., Swaim, R., & Anderson, L. (2003). Violent media content and aggression in adolescents: A downward-spiral model. Communication Research, 30, 713-736. Tajfel, H., & Turner, J. C. (1986). The social identity theory of intergroup behavior. In S. Worchel & W. G. Austin (Eds.), Psychology of Intergroup Relations (pp. 7-24). Chicago: Nelson-Hall. Turner, J. C., Oakes, P. J., Haslam, S. A., & McGarty, C. (1994). Self and collective: Cognition and social context. Personality and Social Psychology Bulletin, 20, 454-463. Paired spirals of influence page 29 von Bertalanffy, L. (1968). General Systems Theory: Foundations, Development, Applications. New York: George Braziller. Wiener, N. (1961). Cybernetics or Control and Communication in the Animal and the Machine. Cambridge, MA: The M.I.T. Press. WIlson, B. (2002). The "anti-jock" movement: Reconsidering youth resistance, masculinity, and sport culture in the age of the Internet. Sociology of Sport Journal, 19, 206-233. Windahl, S. (1981). Uses and gratifications at the crossroads. Mass Communication Review, 2, 174-185. Yanovitzky, I. (2002). Effects of news coverage on policy attention and actions: A closer look at the media-policy connection. Communication Research, 29, 422-451. Zaman, M. Q. (1999). Religious education and the rhetoric of reform: The madrasa in British India and Pakistan. Comparative Studies in Society and History, 41, 294323. Zillman, D., & Bryant, J. (1985). Selective exposure to communication. Mahwah, NJ: Erlbaum. Paired spirals of influence page 30 Figure 1. A minimal path representation of paired spirals of influence. Note that while prospective prediction is normally of primary interest, a wide variety of alternative and indirect paths exist. Concurrent paths can be conceptualized as correlations (or correlated disturbance terms) or as directional paths, subject to problems of identification. Paired spirals of influence page 31 Media use Media use Media use Belief, attitude, behavior Belief, attitude, behavior Belief, attitude, behavior Time 1 Time 2 Time 3