Keith Warren

Substance abuse as a dynamical disease: Evidence and clinical implications of nonlinearity in a time series of daily alcohol consumption

Several authors have suggested that chaos theory, the study of nonlinear dynamics and the application of the knowledge gained to natural and social phenomena, might yield insight into substance-related disorders. In this article, we examine the dynamics of substance abuse by fitting a nonlinear model to a time series of the amount of alcohol, which an adult male with a diagnosis of substance abuse consumed on a daily basis. The nonlinear model shows a statistically superior fit when compared to a linear model. We then use the model to explore a question that is pertinent to the treatment of substance abuse, whether controlled drinking or abstinence is a preferred strategy for maintaining sobriety.

The Emergence of Racial Segregation in an Agent-Based Model of Residential Location: The Role of Competing Preferences

Models of segregation dynamics have examined how individual preferences over neighborhood racial composition determine macroscopic patterns of segregation. Many fewer models have considered the role of household preferences over other location attributes, which may compete with preferences over racial composition. We hypothesize that household preferences over location characteristics other than racial composition affect segregation dynamics in nonlinear ways and that, for a critical range of parameter values, these competing preferences can qualitatively affect segregation outcomes. To test this hypothesis, we develop a dynamic agent-based model that examines macro-level patterns of segregation as the result of interdependent household location choices. The model incorporates household preferences over multiple neighborhood features, some of which are endogenous to residential location patterns, and allows for income heterogeneity across races and among households of the same race. Preliminary findings indicate that patterns of segregation can emerge even when individuals are wholly indifferent to neighborhood racial composition, due to competing preferences over neighborhood density. Further, the model shows a strong tendency to concentrate affluent families in a small number of suburbs, potentially mimicking recent empirical findings on favored quarters in metropolitan areas.

Offense cycles, thresholds and bifurcations : Applying dynamical systems theory to the behaviors of adolescent sex offenders

Abstract In recent years there has been much discussion of the possible application of dynamical systems theory, sometimes called nonlinear dynamics, to social science. One plausible application would be to the antisocial behaviors of sex offenders, since common clinical offense cycle models appear to imply nonlinearity. In this article, we fit linear and nonlinear models to three time series of antisocial behaviors of three adolescent sex offenders. Nonlinear models best characterize two of the three time series. The implications of nonlinearity in offender behaviors include difficulty in demonstrating the effectiveness of interventions, the possibility of endogenously generated swings in behavior, and the possibility that attempts to control offending behaviors might actually lead to larger numbers of those behaviors.

Verbal feedback in therapeutic communities: pull-ups and reciprocated pull-ups as predictors of graduation.

The most important proximal outcomes for residents of therapeutic communities (TCs) are retention and successful completion of the program. At this point there has been no quantitative analysis of the relationship between the exchange of corrective reminders, or pull-ups, between peers in TCs and graduation. This study draws on a database of pull-ups exchanged between 5464 residents of three midwestern TCs. Residents who send more pull-ups to peers and who reciprocate pull-ups with a larger percentage of peers are more likely to graduate. Residents who receive more pull-ups from peers and staff and a larger percentage of whose peers reciprocate pull-ups that they send are less likely to graduate. Implications of these findings for program theory and program improvement are discussed.

Short-Run Prosocial Behavior in Response to Receiving Corrections and Affirmations in Three Therapeutic Communities

Therapeutic communities (TCs) have a strong record of maintaining high quality social climates in prison units. One possible reason for this is the system of mutual monitoring among TC residents, based on the assumption that peer affirmation of behavior in accord with TC norms, and peer correction of behavior contrary to TC norms, will lead to increased resident prosocial behavior. Laboratory experiments have demonstrated that such peer monitoring can lead to cooperation, but there has been no quantitative test of this hypothesis in an actual TC. In this article we test this assumption by using the affirmations that residents of three different TCs send as a measure of prosocial behavior following the reception of peer affirmations and corrections. At all three facilities residents send more affirmations following the reception of both affirmations and corrections, with this relationship being stronger and longer lasting after receiving affirmations. No other variable consistently predicts the number of affirmations that residents send to peers. These findings imply that mutual monitoring among TC residents can lead to increased levels of prosocial behavior within the facility, and that prosocial behavior in response to peer affirmations plays a key role.

The Interaction of Segregation and Suburbanization in an Agent-Based Model of Residential Location

We present a model of the interaction of segregation and suburbanization in determining residential location. The model incorporates differential income between two classes of agents, a simplified market mechanism for the purchase of housing, and a simple geographic structure of one central city and four symmetrically arranged suburbs. Agents derive utility from neighborhood racial composition, the size of their lot, private amenities that are specific to neighborhoods, and public amenities that stretch across municipalities. We find that the public-amenities term leads to a positive-feedback loop in which migration to suburbs increases the public amenities in those municipalities while lowering amenities in the central city, thus sparking further migration. When the minority agents are uniformly less affluent than the majority agents, this dynamic produces discontinuity in segregation as measured by centralization. Such discontinuities are typical of first-order phase transitions. When minority and majority incomes overlap, significant regions appear over which there are multistable equilibria at high and low levels of segregation, along with considerable sensitivity to the initial distribution of minority agents. We discuss the implications of these findings.

The Spirit Is Willing: Nonlinearity, Bifurcations, and Mental Control

In recent years there has been considerable interest in the construction of nonlinear models of the dynamics of human behavior. In this exploratory article we argue that attempts at controlling problematic thoughts, emotions, or behaviors can lead to nonlinearity in mental/behavioral dynamics. We illustrate our model by fitting threshold autoregression models to self-recorded time series of the daily highs in intensity of anxiety and obsessive ruminations, kept by an individual in therapy for this problem. In our discussion, we raise the possibility that bifurcations that occur in this nonlinear model may offer insight into mental control paradoxes.

He Did It First! A Path Model of Interactions in the Aggressive Behaviors of Elementary School Boys during Three Successive Recess Periods.

Abstract This exploratory study uses observations of the aggressive behavior of elementary school boys during three successive recess periods to test the hypothesis that aggressive behavior can spread between recess periods through peer contagion. If this hypothesis is correct, aggressive behaviors during the second recess period should mediate any correlation between aggressive behaviors during the first and third periods. Structural equation modeling supports the mediating role of the second period. This finding suggests that peer contagion in aggressive behaviors can occur quite rapidly and adds theoretical support to the use of proactive interventions within schools that curtail aggressive behaviors before they have the opportunity to spread. Implications for schools related to supervision in unstructured settings throughout the school day are discussed.

Thresholds and the Abstinence Violation Effect A Nonlinear Dynamical Model of the Behaviors of Intellectually Disabled Sex Offenders

The idea that sexual offenses occur as the culmination of a chain of actions on the part of offenders is central to contemporary cognitive behavioral models of relapse prevention. However, quantitative tests of this idea have remained rare. This article presents a mathematical model of the offense chain that allows a quantitative hypothesis test and tests the model using archival records of the daily behaviors of 7 developmentally disabled sex offenders. The results lend statistical support to models of the offense chain and allow consideration of the implications of the dynamics of offender behaviors for treatment

A network of helping: generalized reciprocity and cooperative behavior in response to peer and staff affirmations and corrections among therapeutic community residents

Abstract Background: Clinical theory in therapeutic communities (TCs) for substance abuse treatment emphasizes the importance of peer interactions in bringing about change. This implies that residents will respond in a more prosocial manner to peer versus staff intervention and that residents will interact in such a way as to maintain cooperation. Method: The data consist of electronic records of peer and staff affirmations and corrections at four corrections-based TC units. We treat the data as a directed social network of affirmations. We sampled 100 resident days from each unit (n = 400) and used a generalized linear mixed effects network time series model to analyze the predictors of sending and receiving affirmations and corrections. The model allowed us to control for characteristics of individuals as well as network-related dependencies. Results: Residents show generalized reciprocity following peer affirmations, but not following staff affirmations. Residents did not respond to peer corrections by increasing affirmations, but responded to staff corrections by decreasing affirmations. Residents directly reciprocated peer affirmations. Residents were more likely to affirm a peer whom they had recently corrected. Residents were homophilous with respect to race, age and program entry time. Conclusion: This analysis demonstrates that TC residents react more prosocially to behavioral intervention by peers than by staff. Further, the community exhibits generalized and direct reciprocity, mechanisms known to foster cooperation in groups. Multiple forms of homophily influence resident interactions. These findings validate TC clinical theory while suggesting paths to improved outcomes.