conflict escalation to violence and escalation of violent conflicts

14
Conflict escalation to violence and escalation of violent conflicts Zeev Winstok ,1 The Center for the Study of Society, University of Haifa, Mount Carmel, Haifa 31905, Israel Available online 13 October 2007 Abstract The objective of this article is to propose definitional, theoretical, and operational frameworks as a first step in the study of escalation of conflicts in intimate relationships. The second part of the article discusses the escalation of violent conflicts (across conflicts); the third and fourth parts address the individual and interpersonal aspects of conflict escalation to violence (within a single conflict). In the fifth and final part a new approach is proposed for the study of escalation in intimate partner violence. © 2007 Elsevier Ltd. All rights reserved. Keywords: Conflict; Escalation; Aggression; Intimate partner violence 1. Introduction Conflicts between intimate partners arise in a setting of discord, but not every disagreement necessarily brings about conflict. Conflicts may be resolved by normative means such as discussion, agreement, compromise, or concession, but this is not always the case. A conflict escalates to violence when one or both parties repeatedly attempt to force their position on the other. Over time, conflicts may escalate to even more severe violence. In both cases (i.e., single conflict escalation to violence and escalation of successive violent conflicts), escalation is a concept that describes a behavioral pattern of increasing severity of aggression. Although there is broad agreement about the importance of studying these patterns, they are seldom addressed theoretically or examined empirically (Winstok, Eisikovits, & Gelles, 2002). In most cases escalation is perceived and presented as a self-explanatory concept (Winstok, 2007). The aim of the present article is to propose definitional, theoretical, and operational frameworks as a step towards understanding escalation in intimate relationships. Escalation can be explored using various frames of reference. Two frames of reference are addressed below: escalation patterns of a single conflict and escalation patterns across successive conflicts. The former is narrower in its time span than the latter. Escalation patterns across conflicts received moderate theoretical and empirical attention, but the escalation patterns of a single conflict has remained almost unexplored. Despite the significant difference between these frames of reference, they may share resembling conceptual and theoretical principles. Theoretically it would be Available online at www.sciencedirect.com Children and Youth Services Review 30 (2008) 297 310 www.elsevier.com/locate/childyouth Tel.: +972 4 8240195; fax: +972 4 8249268. E-mail address: [email protected]. 1 Zeev Winstok, Ph.D., is a Senior Lecturer in the Faculty of Social Welfare & Health Science, University of Haifa and a research fellow at the Center for the Study of Society, University of Haifa, Israel. 0190-7409/$ - see front matter © 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.childyouth.2007.10.007

Upload: zeev-winstok

Post on 05-Sep-2016

219 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: Conflict escalation to violence and escalation of violent conflicts

Available online at www.sciencedirect.com

Children and Youth Services Review 30 (2008) 297–310www.elsevier.com/locate/childyouth

Conflict escalation to violence and escalation of violent conflicts

Zeev Winstok ⁎,1

The Center for the Study of Society, University of Haifa, Mount Carmel, Haifa 31905, Israel

Available online 13 October 2007

Abstract

The objective of this article is to propose definitional, theoretical, and operational frameworks as a first step in the study ofescalation of conflicts in intimate relationships. The second part of the article discusses the escalation of violent conflicts (acrossconflicts); the third and fourth parts address the individual and interpersonal aspects of conflict escalation to violence (within asingle conflict). In the fifth and final part a new approach is proposed for the study of escalation in intimate partner violence.© 2007 Elsevier Ltd. All rights reserved.

Keywords: Conflict; Escalation; Aggression; Intimate partner violence

1. Introduction

Conflicts between intimate partners arise in a setting of discord, but not every disagreement necessarily brings aboutconflict. Conflicts may be resolved by normative means such as discussion, agreement, compromise, or concession, butthis is not always the case. A conflict escalates to violence when one or both parties repeatedly attempt to force theirposition on the other. Over time, conflicts may escalate to even more severe violence. In both cases (i.e., single conflictescalation to violence and escalation of successive violent conflicts), escalation is a concept that describes a behavioralpattern of increasing severity of aggression. Although there is broad agreement about the importance of studying thesepatterns, they are seldom addressed theoretically or examined empirically (Winstok, Eisikovits, & Gelles, 2002). Inmost cases escalation is perceived and presented as a self-explanatory concept (Winstok, 2007). The aim of the presentarticle is to propose definitional, theoretical, and operational frameworks as a step towards understanding escalation inintimate relationships.

Escalation can be explored using various frames of reference. Two frames of reference are addressed below:escalation patterns of a single conflict and escalation patterns across successive conflicts. The former is narrower in itstime span than the latter. Escalation patterns across conflicts received moderate theoretical and empirical attention, butthe escalation patterns of a single conflict has remained almost unexplored. Despite the significant difference betweenthese frames of reference, they may share resembling conceptual and theoretical principles. Theoretically it would be

⁎ Tel.: +972 4 8240195; fax: +972 4 8249268.E-mail address: [email protected].

1 Zeev Winstok, Ph.D., is a Senior Lecturer in the Faculty of Social Welfare & Health Science, University of Haifa and a research fellow at theCenter for the Study of Society, University of Haifa, Israel.

0190-7409/$ - see front matter © 2007 Elsevier Ltd. All rights reserved.doi:10.1016/j.childyouth.2007.10.007

Page 2: Conflict escalation to violence and escalation of violent conflicts

298 Z. Winstok / Children and Youth Services Review 30 (2008) 297–310

more convenient to address escalation patterns of a single conflict and then move to escalation patterns across conflicts,using an inductive approach. But because the latter received broader theoretical and empirical attention, we follow adeducting approach and begin the discussion with escalation patterns across successive conflicts. The discussion thenproceeds to address the individual and interpersonal aspects of escalation patterns within a single conflict. Finally, wepropose a new approach for studying escalation in the field of intimate partner violence.

2. Escalation patterns across successive conflicts (violent conflict escalation)

Escalation patterns across conflicts represent an increase in the most severe violent behavior directed by individualsat their partners from one conflict to the next. For example, the man's most severe act against his wife in the firstconflict consisted of cursing. In the second conflict, several days later, his most severe act against her included pushing.A third conflict occurred in the following days during which the man attacked his wife with his fists. In this example theman's aggression escalated over time.

An early seminal work by Feld and Straus (1989), “Escalation and desistance of wife assault in marriage,” providesan in-depth conceptual and theoretical foundation for the study of escalation of violent conflicts in intimaterelationships. The approach of Feld and Straus to wife assault relied on a criminal career perspective that emphasizedthe pattern of the individuals' criminal activity over time (see Blumstein, Cohen, & Farrington, 1988; Gottfredson &Hirschi, 1990). They suggested that wife assault is similar to other forms of deviance and crime in that desistance iscommon and that engaging in minor forms of deviance is a risk factor for engaging in major forms of deviance andcrime. One extreme conclusion that can follow from this approach is that each case of wife assault is independent, sothat nothing can be learned from examining a criminal career because knowing about one event reveals nothing aboutthe likelihood of another (there is no pattern). Another extreme conclusion is that wife assault strictly follows one of afew typical patterns of succession. Thus, to determine whether and to what extent a man will assault his wife is todetermine his current position in a particular pattern (succession) of events. It was (and still is) widely accepted thatwife assault that occurs more than once persists until the termination of the relationship (by separation or death), andthat battering escalates in severity (frequency, intensity, and form) over time (Giles-Sims, 1983; Pagelow, 1981;Walker, 1979). Eisikovits and Buchbinder (2000) argued that over time situational violence is transformed into aviolent way of life. Despite the widespread beliefs about the persistence of wife assault, the quantitative evidencesuggests that most wife assaults subside (Dutton, 1988; Fagan, 1989; Sherman & Berk, 1984).

In their study, Feld and Straus sought to determine whether intimate partner violence was similar to other forms ofdeviance in two ways: high desistance and minor forms serving as a risk factor for severe ones. Their study was basedon a panel study that was part of a series of national surveys of violence in American families. It consisted of 380married respondents who reported one or more assaults in their marriage in 1985 and were re-interviewed one yearlater. Findings supported the hypothesis that predicted a high rate of desistance, even for husbands who used violencefrequently. The findings also supported the hypothesis that minor assaults and assaults by wives are associated withsubsequent severe assaults by husbands.

Feld and Straus compared the frequency and highest severity of violent behaviors (ordinal variables) of twoconsecutive time periods (dichotomous ordinal variable). Theoretically, the frequency and severity of violent behaviorare complementary aspects of dangerousness (harm potential). Operationally, they can be represented in ways otherthan ordinal. But the time variable is of higher relevance to the present discussion as it is an independent variable thatmay represent change. Feld and Straus measured and scored violence across two consecutive time periods of one yeareach. In this case time had no specific meaning except for its calendar significance. A different division of time couldproduce different results. Feld and Straus used time as a sampling criterion for comparing behaviors over equivalentperiods. The assumption that equal time periods mean an approximately equal opportunity to act aggressively may notnecessarily withstand the test of reality. Conflicts are an actual opportunity for aggressive behavior.

There is evidence that time has a moderating effect on conflict occurrence (Winstok, 2006a). A possible explanationis that over time couples come to better understand their own and each other's needs, wishes, and boundaries, whichcan reduce friction and conflicts between them. Hence, the probability for conflicts, especially escalatory ones,subsides over time. Additional reasons for a decrease in the number of conflicts can be avoidance on the part of thewoman to engage in conflicts in light of past experience with violent response (Walker, 1979). A battered wife maylearn which conditions predict her partner's violence against her and take action to prevent them. In sum, equalconsecutive time periods is one sampling option; the other option is equal number of opportunities, or even better,

Page 3: Conflict escalation to violence and escalation of violent conflicts

299Z. Winstok / Children and Youth Services Review 30 (2008) 297–310

actual conflicts. The work of Feld and Straus is to be regarded as an indirect attempt to study the escalation of violentconflicts over time.

Consistent with the criminal career perspective, Feld and Straus also suggested four reasons for which minor assaultsindicate a likelihood of subsequent more severe assaults (escalation). First, neutralization of countervailing norms (e.g.,“I slapped her… slapping is not a big deal, it does not inflict significant damage”); second, some stability in the externalfactors causing deviance (e.g., ”I don't know how to talk with her and sometimes hitting her is much easier”); third, theeffectiveness of violence for meeting the goals of the individual (e.g., “After I slap her she becomes just like a disciplinedchild”); fourth, minor deviance may provoke negative responses from others (authorities, friends, relatives) that provokefurther deliberate acts of rebellion or spite (e.g., “The police came… it only fueled my anger”). Feld and Straus alsosuggested four parallel reasons why violent assaults may desist. First, continued normative pressures (e.g., “Now afterthe police visit everybody knows about our problem and I need to be holier than the pope”); second, motivation foracting violently may change (e.g., “Since our child was born things have changed… I need to think more about thefuture…”); third, violence does not accomplish the goals of the individual and may result in undesired consequences(e.g., “After I hit her things between us became worse”); fourth, responses of others (whether punitive or permissive)may deter or discourage further deviance (e.g., “The policeman told me that if I hit her I would find myself in jail”).

In the Feld and Straus theory both the phenomenon and its factors are described as change. There is no effort topredict or explain behavior at a certain point in time based on previous behaviors. There is, however, an attempt toexplain changes in behavior over time. Thus, the predictors of behavioral change are not given factors but ratherchanges that take place within them. Taking this theory one step back would result in the suggestion that four factorspredict violent behavior; taking it one step forward would result in a theory of change. In other words, a change in thesefactors predicts or explains changes in violent behavior. For example, pro-violence attitudes may promote violentbehavior whereas anti-violence attitudes may inhibit it; hence an individual's attitudes toward violence at a certain timemay affect his violent behavior. A change in his attitudes may bring about a change in behavior. When pro-violentattitudes increase, violent behavior may increase, and a decline in these attitudes may reduce violent behavior.

Furthermore, a change in one factor (e.g., attitudes) may result in changes in other factors (e.g., negotiation skills).Therefore changes in violent behavior may attest in general to changes in one or more of the four factors suggested byFeld and Straus. The escalation and desistance factors proposed by them are associated, and maybe even partiallyredundant. These limitations may be overcome by a broader theoretical framework such as cost and benefitconsiderations. Desistance and escalation factors can indicate higher costs and lower benefits. In this conceptualizationescalation over time results from a decrease in the cost of violent behavior and an increase in the benefits it provides toperpetrators. A reverse dynamic promotes desistance.

It is widely accepted that human behavior is regulated to a large extent by anticipated consequences of prospectiveactions, aimed to minimize undesirable outcomes (costs) and maximize desirable ones (benefits) (Bandura, 1973). Costand benefit considerations are evaluated with reference to goals that reflect social values (Ohbuchi & Tedeschi, 1997;Tedeschi & Felson, 1994). Goals pursued during interpersonal conflict have motivational effects on behavior (Murphy& Eisenberg, 2002). They provide a framework for processing a social situation by directing attention and affectinginterpretations that subsequently influence strategy generation, evaluation, and selection (Chung & Asher, 1996; Crick,& Dodge, 1994; Graham, Argyle, & Furnham, 1980). Campbell (2005) referred to aggression as a contingent strategywhere the value of the benefits multiplied by the probability of obtaining them exceeds the value of the costs multipliedby the likelihood of incurring them. The costs include the possibility of injury or death; the benefits include theacquisition or retention of valued resources. Applying this approach to violence in intimate relationships isstraightforward and has the potential to explain violent behaviors as well as changes in them. Naturally, in this casebehavioral changes are also to be attributed to changes in the attribution of costs and benefits.

Fig. 1 is a heuristic flow chart attempting to describe how past experiences of a conflict influence future use (or non-use) of aggression. In the first phase (i.e., examining the cost/reward ratio of aggression), the aggressor evaluates thecost and benefit of using violence against his partner. If the benefit outweighs the cost, it is probable that in the secondphase (enactment) he will proceed to aggression. In the third phase (reexamining the cost/reward ratio of aggression),the consequences of the use of aggression are evaluated. This evaluation has implications for the use of aggression infuture conflicts: if the consequences are consistent with the expectations preceding the aggressive act the resultingexperience reinforces the original perception of the costs and benefits of aggression; if they do not meet theexpectations, the underlying assumptions are undermined and in the next incident this may affect the decision whetheror not to use violence, and if yes to what degree.

Page 4: Conflict escalation to violence and escalation of violent conflicts

Fig. 1. From present experience to future use of aggression.

300 Z. Winstok / Children and Youth Services Review 30 (2008) 297–310

The process described above can have five possible conclusions: a non-aggressive one, a conclusion in whichaggression pays, and three in which aggression does not pay. The following example is of a case in which theconclusion is that aggression pays. A man in conflict with his wife considers using aggression. At this stage he needs todetermine whether using a certain form of aggression would gain the desired results. Possibly he is considering severalforms of aggression of varying severity, and finally chooses what he believes to produce maximum gain at minimumcost. He dismisses cursing as not being sufficiently effective (its cost is not significant, but it is unlikely to do any“good”). He also dismisses slapping because its cost is too high (his wife may comply for some time but then mayconsider leaving him). He finally decides on a shove because he thinks that it is not severe enough to incur a high costbut it is sufficiently so to produce the expected benefit. After shoving his wife, the man must determine whether theexpectations that led him to choose shoving were realized. If they were, the conflict ends with the understanding that inthe given situation shoving is an effective behavior. This experience enhances the man's readiness to use violence ingeneral and shoving specifically, as a means of manipulating his wife in future conflicts.

This process appears to be a perfectly rational decision making procedure rather than an emotional one. But inescalatory conflicts, especially those escalating to violence, the decision making process is clearly steeped in emotion.

Page 5: Conflict escalation to violence and escalation of violent conflicts

301Z. Winstok / Children and Youth Services Review 30 (2008) 297–310

Emotions have a direct effect on the process (Loewenstein, 1996) and at times play a central part in its costs andbenefits. Damasio (1994) suggested that emotions must be engaged to some extent for individuals to make effectivedecisions, especially in the personal and social domains. He viewed emotion as vital for rationality because it allowseffective filtering and processing of information. From this perspective, emotional reactions to environmental andinternal stimuli operate to quickly assign values or attach preferences to those stimuli and to the possible responses tothem. Damasio suggested that a rational consideration of all the options and consequences of a decision would makeeffective (even if non-normative) decision-making in daily life awkward and almost impossible.

A central conclusion that can be drawn from the ongoing discussion about escalation patterns across successiveconflicts is that in order to predict or explain them an understanding of the escalation patterns of a single conflict isrequired. Escalation across conflicts is based exclusively on an individual pattern, whereas escalation of a singleconflict is based on both individual and interpersonal patterns. The individual patterns for both levels of reference (i.e.,across successive conflicts and within a single conflict) are essentially similar. In both cases individual escalatorypatterns show how the aggressive actions of an individual become increasingly severe.

3. Individual escalatory patterns in a single conflict (conflict escalation to violence)

Most of the research on violence in intimate relationships focuses on individuals' single and isolated aggressiveactions (context free). For example, a man slapping his wife is treated with no reference to crucial information such aswhat preceded the strike, what followed, and what were the man's motivation and objective. The individual escalatorypattern is a broader frame of reference then an individual's single and isolated aggressive action. The individualescalatory pattern describes how the aggressive actions of an individual become increasingly severe. Designating aseverity level for each violent action is a prerequisite for the evaluation of a behavioral pattern (e.g., a punch is a moresevere form of aggression then a shove). Individual escalatory patterns across successive conflicts may be expressed bythe intervals between the most severe aggressive behaviors manifested in consecutive conflicts. When examiningindividual escalatory patterns in a single conflict this pattern may be expressed by the intervals between the severitylevels of two consecutive actions. For example, the man shoved his partner (t1) and following her response (which isirrelevant in this case because it is an individual pattern) he punched her in the face (t2). In this example, the first action(at t1) is more moderate than the second one (at t2).

Individual escalatory patterns were addressed indirectly and identified in studies of the co-occurrence of variousforms of violence. Gordon (2000) suggested that studies have usually found that many individuals engage in verbalaggression or psychological abuse but do not engage in physical aggression toward their spouses. Individuals whoengage in serious physical aggression against partners tend to engage in frequent abusive behaviors of all types andlevels of severity. Stets (1990) also noted this tendency and referred to it as escalation. But it is difficult to establish inthese studies whether and to what extent escalation takes place within or across conflicts because most of these studiesare based on cross-sectional data.

Within conflicts, individuals attempting to resolve a disagreement may systematically employ aggressive means,starting with milder ones and gradually moving toward increasingly severe ones. When mild actions such as verbalaggression fail to resolve the conflict, they resort to mild and then to severe physical aggression. Unresolved situationscan thus result in death, but more often than not even extremely violent individuals have certain boundaries they refrainfrom crossing. This suggestion is an interpretation of observations of isolated single behaviors and not a directobservation of escalation.

The correlation matrix presented as part of the validation of the revised Conflict Tactics Scales (Straus, Hamby,Boney-McCoy, & Sugerman, 1996) is another example of an indirect indication of an individual pattern. Table 1presents this correlation matrix and shows clearly that the negotiation scale is linked mostly to the verbal aggressionscale; the verbal aggression scale is associated mostly with the physical aggression scale; and the physical aggressionscale correlates mostly with the injury scale. These relationships are consistent with the idea of an individual escalatorypattern: aggression of a certain level of severity increases the probability that aggression of a higher level of severitywill be employed. Straus and colleagues did not address the positive association between the negotiation scale and theverbal aggression scale. Their view at the time was probably that whereas negotiation was a normative approach toconflict resolution, aggression was not. Thus, the relationship between the two scales should have been negative. Butfindings indicated a positive association between the negotiation and the verbal aggression scales. From the perspectiveof an individual pattern of conflict escalation to violence, a positive association is to be expected. Negotiations exist

Page 6: Conflict escalation to violence and escalation of violent conflicts

Table 1Correlation among approximations to CTS2 scales for self-report of perpetration

CTS2 scale Gender Psychological aggression Physical assault Sexual coercion Injury

Negotiation Men 0.22 −0.05 0.03 0.01Women 0.40 0.21 0.10 0.16

Psychological aggression Men 0.71 0.66 0.63Women 0.67 0.25 0.41

Physical assault Men 0.90 0.91Women 0.26 0.42

Sexual coercion Men 0.87Women 0.29

Source: Straus, M., Hamby, S., Boney-McCoy, S., & Sugerman, D. (1996). The revised conflict tactics scales (CTS2): Development and preliminarypsychometric data. Journal of Family Issues, 17(3), 283–316.

302 Z. Winstok / Children and Youth Services Review 30 (2008) 297–310

when there is disagreement; therefore it is a proxy representation of a conflict between partners. A conflict is aprerequisite of verbal aggression, so a positive association is to be expected between the two as evident from thecorrelation matrix. This last point supports the notion that escalation begins when normative attempts at conflictresolution fail, even before the evident manifestations of aggression.

In the present article we suggest that the individual escalatory pattern evolves through a series of five potentialdistinct steps, each one promoting the next. In the first step individuals may feel the need to control their partner; in thesecond their self-control may be challenged or even lost; in the third they use verbal aggression against their partner; inthe fourth they threaten their partner with physical aggression; and in the fifth step they put these threats into effect. Ingeneral, this sequence is always maintained, although some steps may be skipped; for example, when the need tocontrol arises, the individual may proceed directly to verbal aggression, without losing self-control.

The need to control the partner and the ability for self-control were identified as important factors affecting violencein general and in intimate relationships in particular. Despite the importance of these two control-related factors, theassociation between them has rarely been examined (Winstok, 2007). There has been support for the positiveassociation between control needs and aggression and the negative association between the ability for self-control andaggression (Edelson, Eisikovits, Guttmann, & Sela-Amit, 1991; Mason and Blankenship, 1987; Prince and Arias,1994). Subject to the relationship between these control factors and aggression, it may be shown that there is a negativeassociation between the two: the higher the need of individuals to control their partners, the lower their ability to controltheir own behavior. Individuals' need to control their partners may arise and increase when their partners are unwillingto comply with their requests. In their normative attempts to convince their partners to comply, these individuals mayrun out of self-control resources. The longer their need to control their partner is unsatisfied, the more their ability tocontrol their own behavior can decrease, and they may resort to non-normative means. Therefore, control needs andabilities (covert aspects) are also subject to escalatory dynamics, as are the various forms of aggressive behavior (overtaspects). Losing control does not mark the end of the conflict but rather fuels it. Here aggression becomes a pseudo-means to re-acquire a sense of power and control (Denzin, 1984). In principle, any conflict can continue to escalate tothe bitter end: severe injury or even murder. But in most cases it escalates until one party appears to be defeated or untilit becomes clear that the cost of further escalation would be too high (Winstok & Eisikovits, 2006).

Discussing individual escalation patterns across conflicts, Feld and Straus (1989) provided evidence for bothescalation and desistance of violence over time. They did not address directly de-escalation (the reversed pattern ofescalation), in which aggressive actions across conflicts decrease in severity. Desistance is not a pattern and does notnecessarily represent an extreme case of de-escalation. The question is whether within a single conflict such reversepattern of escalation exists. There is no empirical evidence, but it seems that within a given conflict applying suchpattern (as opposed to an escalatory one) is ineffective. Moreover, within a given conflict this pattern would require ahigh level of self-control. Thus it may be concluded that conflict escalation ends when violence desists. In most cases, ifnot all, de-escalation within a given conflict is not an applicable pattern.

Despite these suggestions, there are situations in which within a single conflict a man threatens to hit a woman, hitsher, threatens to hit her again, and hits her again. These four actions consist of two pairs of escalatory actions: the firsttwo and the last two. The two middle actions (hitting followed by a threat) appear to be de-escalatory. The term“appear” implies that de-escalation within conflicts is an artificial concept. In the above example the first action broughtforth the second (escalation) and the third brought about the fourth (again, escalation), but the second action did not

Page 7: Conflict escalation to violence and escalation of violent conflicts

303Z. Winstok / Children and Youth Services Review 30 (2008) 297–310

bring forth the third one (allegedly, de-escalation). More precisely, the first pair of actions describes escalation and thesecond pair re-escalation; one should not regard any phase of the process as de-escalatory. In sum, in an escalatoryconflict aggressive behavior becomes increasingly severe. For reasons already mentioned, the process ends whenaggression is terminated and not as a result of a reverse process in which the severity of aggression decreases.

The individual pattern of conflict that escalates to violence described in this section is based on qualitative studiesand on the interpretation of quantitative ones measuring single, isolated behaviors. Such indirect approach is limitedand does not provide solid support for the theoretical framework presented in this paper. Panel design is a prerequisitefor properly testing and supporting the suggestions made here for the individual escalatory pattern. But a design of thisnature creates multiple methodological and ethical difficulties. Until an applicable methodology appropriate for directexamination of the theoretical framework of an individual escalatory pattern is developed, we can formulate hypothesesthat can be explored based on existing, mostly cross-sectional databases measuring single isolated behaviors. Thesehypotheses are limited, however, in the extent to which they address the process because we cannot establish causalityusing cross-sectional data.

We can extrapolate hypotheses from the theoretical framework of individual patterns of conflict escalation toviolence, as described in this section, and we can test them using cross-sectional data about men's control needs andcapabilities, and the frequency of aggressive behavior with varying degrees of severity. (1) “Loss of self-control”mediates the association between “Need to control women” and “Acts of verbal aggression.” (2) “Acts of verbalaggression” mediates the association between “Loss of self-control” and “Threats of physical aggression.” (3) “Threatsof physical aggression” mediates the association between “Acts of verbal aggression” and “Acts of physicalaggression.” Even if these hypotheses are confirmed, however, there are still limitations to the conclusions that can bedrawn and to the extent to which they support the theoretical framework. For instance, when using cross-sectional datait is impossible to differentiate between personal escalatory patterns within and across conflicts. Moreover, falsealternative hypotheses, such as reversed ones, are mathematically equivalent and may also be supported. Under theselimitations, testing the mediation hypotheses using cross-sectional data should be regarded as an initial attempt atsubstantiating the feasibility of the individual escalatory pattern and as a small and temporal step toward developing anescalation theory and methodology.

4. Interpersonal escalatory patterns in a single conflict (conflict escalation to violence)

The individual escalatory pattern is but one aspect within a given conflict. Another one is the interpersonal pattern ofescalation, where the frame of reference is dyadic rather than individual. Its time span is narrower than that of theindividual escalatory pattern. The interpersonal pattern begins with one party's act and ends with the subsequent act ofthe other party. Each act by each party has an immediate reason (the perception of the previous act of the other party),and an immediate result (the next act of the other party); therefore, each act is simultaneously both an action and areaction. Similar to the individual pattern of escalation, the interactional one is also based on the interval between theseverity levels of two consecutive actions. For example, the woman cursed her partner (t1), and he punched her (t2). Inthis example, the first action taken by one party (t1) is more moderate than the second one taken by the other (t2).

The idea of exploring aggression from an interactive perspective is not new. Carins, Santoyo and Holly (1994)argued that the “hypothesis that social interchanges escalate to aggression and violence is a simple and powerful idea.Cast into an interactional framework, the proposition is that hostile acts of one person towards another provide thestimuli for more hostile counter-reactions. All things equal, aggression begets aggression” (p. 227). In a study assessingaggressive dyads entitled “It takes two to fight,” Coie et al. (1999) maintained that previous studies of aggression hadbeen dominated by the individual aggressor, with aggression being the unit of analysis. The researchers recognized theimportance of individual factors but maintained that interactional and dyadic factors may play an important role in thedevelopment of aggression given that aggression is invariably the result of the participants' views and interpretations ofeach other's actions and of the context in which these occur (Mummendey, Bornewasser, & Loschper, 1982).

Swann, Pelham and Roberts (1987) proposed that people simplify their perceptions of their interactions byorganizing them into discrete causal chunks. Once formed, these chunks presumably affect the extent to which peopleare aware of their effect on others and of their impressions of others. Findings of a study they conducted indicate thatpeople formed self-causal chunks when they possessed an offensive set (e.g., my action causes my partner's action),and other-causal chunks when they possessed a defensive set (e.g., my partner's action causes my own). In the contextof intimate partner violence, in most cases both parties perceive themselves as reactors rather than initiators (Eisikovits,

Page 8: Conflict escalation to violence and escalation of violent conflicts

304 Z. Winstok / Children and Youth Services Review 30 (2008) 297–310

Winstok, & Gelles, 2002; Winstok et al., 2002). Thus, response patterns should refer mainly to a defensive interaction.Here too, as in the previously mentioned behavioral patterns, designating a severity level for each violent action is aprerequisite for evaluating escalation. The action completing the interaction (the second action) can be more severe(increasing), equally severe (conserving), or less severe (decreasing) than the action that started it (the first one). Theseoptions are located on an escalation continuum, expressing the degree of shift in the severity of violent actions (i.e.,increasing, decreasing, or conserving the severity of violence). The terms escalation, escalator, and de-escalator canreplace the terms aggression (violence), perpetrator (aggressor), and victim. The question of who is the perpetrator andwho is the victim becomes a question of who increases, decreases, or conserves the severity of violence between theparties.

Research on gender differences in intimate partner violence indirectly addresses issues relevant to the study ofescalation, especially to its interpersonal aspect, as it refers to the unilateral/bilateral patterns of violence. Thus, genderasymmetry/symmetry can be considered as a proxy or partial representation of the “interpersonal pattern.” A majordebate among researchers in intimate partner violence focuses on the question whether it is appropriate to distinguishmen-to-women violence from women-to-men violence. Those who support such distinction reject symmetry andclaim that violence in intimate relationships is the unilateral violence of men against their female partners (see forexample, Dobash & Dobash, 1979). This approach is supported by studies showing that men use violencesignificantly more than women to “show the partner who is boss.”Women use violence significantly more than men“to protect themselves” (Barnett, Lee, and Thelen, 1997; Cascardi and Vivian, 1995; Hamberger, Lohr, & Bonge,1994; Vivian, 1998). Those who do not support this distinction between men and women's violence maintain thatviolence in intimate relationships is bilateral and symmetrical, and accumulating empirical evidence increasinglysupports this notion (Kim, 2003; Medeiros & Straus, 2006; So-Kum Tang, 1999; Stets & Pirog-Good, 1987; Straus,Gelles, & Steinmetz, 1980; Straus & Gelles, 1990; Sugihara & Warner, 2002).

Johnson (1995, 2001) tried to bridge these approaches by arguing that unilateral aggression is valid mostly forservice populations and bilateral aggression for the general population. He labeled unilateral aggression of men againsttheir female partners “intimate terrorism,” and claimed that it is a characteristic of men who view their gender asauthorizing them to dominate their female partners. These men perceive the gender power balance as inherent in natureand not open to negotiation. Johnson named bilateral violence “situational couple violence.”He suggested that this typeof violence is limited to situational power struggles and is not embedded in an overall worldview.

The “asymmetry” as well as “symmetry” concepts imply a framework in which parties in a violent event areperceived as perpetrator and victim. These are important basic terms in studying violence, but they are compounds ofmore than single, isolated acts of violence such as a curse, a shove, or a punch. The definition of perpetrator (oraggressor) as one who acts aggressively, and of victim as one who is subject to aggressive action is acceptable but toosimplistic, sometimes inaccurate, and even wrong. Using the term aggression to define perpetrator and victim is oneoption; the other is to define aggression by using the terms victim and perpetrator. For example, aggression is what isused by perpetrators against their victims. These are relative terms that constitute each other. Researchers that supportthe symmetry assumption define relations based on behaviors, whereas those who reject symmetry define behaviorsbased on relations. They both take for granted the meaning of the terms perpetrator and victim, but for one it is the cause(an independent variable) and for the other the consequence (a dependent variable). The terms victim and perpetratorare not obvious. They are abstract, subjective, interactional, and dynamic: abstract because they cannot be measureddirectly; subjective because they are the outcome of social discourse and construction; interactional because they aremutually definable and interdependent; and dynamic because they are not specific behaviors but rather patterns ofconduct. All this makes them problematic.

To directly study behavioral patterns (individual and interpersonal), a step-by-step description of conflictdevelopment and escalation is required. Creating situations that produce such data raises theoretical, methodological,and ethical problems. During the discussion in the previous section on the individual pattern of escalation within agiven conflict it was suggested that until such time when a methodology enabling direct examination of thephenomenon is developed and tested, it is possible to use an indirect and limited approach based on cross-sectionaldatabases of measurements of individuals' single and isolated aggressive actions. This suggestion cannot be applied tointerpersonal escalatory patterns (interactional measurement). But developing a direct interactional measurementwithin a cross-sectional design is quite straightforward and it requires only to measure interactional behavior ratherthan single and isolated behaviors. Various methods can be used for this purpose. One method was recently proposedby Winstok (submitted for publication), suggesting that instead of measuring the frequency of aggressive behaviors of

Page 9: Conflict escalation to violence and escalation of violent conflicts

305Z. Winstok / Children and Youth Services Review 30 (2008) 297–310

varying severity levels, as was done in the CTS-2 (Straus et al., 1996), one should measure the frequency of a subject'saggressive reactions of varying severity levels to aggressive actions of varying severity levels employed by the partner,if at all.

5. Shifting the theoretical focus from violence to conflict escalation

Despite the strong link between violence and escalation, they are as different in essence as still and motion pictures.Violence refers to a single isolated act in the behavior of an individual. Escalation refers to the dynamic aspects of theconflict (Eisikovits et al., 2002; Winstok et al., 2002). The present article addresses two frames of references: escalationpatterns across successive conflicts and escalation patterns within a single conflict. The across-conflicts frame ofreference is represented with an individual escalation pattern and the within-conflict frame of reference with bothindividual and interpersonal escalatory patterns. Note, however, that this is merely a convenient construction. Atpresent it is difficult to determine the contribution of the individual and interpersonal patterns and of the interactionbetween the two to the development of conflicts and their escalation to violence. Fig. 2 describes an example of anindividual and interpersonal escalation process within a single conflict. The example begins with a man shoving hiswife. She then curses him. The example ends with the man punching his wife. Along the individual course the manmoved from a less severe aggressive action to a more severe one. It can be stated that the man's aggression escalated.Along the interpersonal course the situation was similar: the man responded with greater aggression than that employedby his wife. But what predicts the end of the episode? To what extent does the interaction that took place between theman and his wife before the man's last action predicts that this action would be a punch? The example shows also thatthe wife's interpersonal pattern is de-escalatory because her reaction, albeit aggressive, was more moderate than theman's action. This example does not provide sufficient information to determine the wife's individual course. The twoprevious actions, that of the man and that of the wife, raise the probability that the man's next action would beescalatory.

The cost–benefit assumption, as an individual, subjective, and temporal product regulating behaviors in general andaggressive ones in particular, may be found applicable to understanding intimate partners' conflicts that escalate toviolence. The above example can be examined with this assumption in mind. Because the man shoved his wife, it may

Fig. 2. Personal and interpersonal escalatory pattern.

Page 10: Conflict escalation to violence and escalation of violent conflicts

306 Z. Winstok / Children and Youth Services Review 30 (2008) 297–310

be assumed that this act existed as an option for him even before its actual perpetration and that he had considered itscost and benefit (within situational limitations such as his emotional state) and found that the latter outweighed theformer. The model does not address the achievement of goals but rather the evaluation of cost vs. benefit when usingaggression, and its consequences for the severity of the aggression being perpetrated. Probably the man did not achievehis goals with the first action, because the woman responded by cursing. Because the man next took more severe action,it may be assumed that he evaluated his previous action as being too moderate (not severe enough). This analysis showsthat the model is limited in its treatment of cost–benefit considerations to the actual use of aggression, rather thanstarting at the point where aggression is being considered. The model does not address the goals of the conflict, itsdevelopment to violence, or its end. It also has some weakness in addressing the interpersonal aspect of the conflict. Amodel is required that can treat conflicts more comprehensively.

Fig. 3 is a heuristic flow-chart describing a conflict from initiation to desistance. It is argued that conflicts have amotive and a goal. These may change during the conflict but they exist at every point in time. The starting point of themodel is expectation, a derivative of conflict goals. If expectations are realized, the conflict ends; if not, an action isneeded to advance the realization of the expectation. The action used to realize the expectation may be normative (e.g.,discussion) or non-normative (i.e., violence). The model describes first normative attempts aimed at realizingexpectations. If these are realized, the conflict ends. If not, additional normative actions may be employed. At a certainpoint in the conflict, when all normative options seem to have been exhausted, two courses of action are available: oneis to adjust expectations, which may result in a non-violent termination of conflict; the other is to start employing non-normative actions such as violence to try to realize the expectations. This course of action is marked with a thick line inthe model. It is not necessarily one action, and could also represent cycles of non-normative actions that continue untilthe expectation is realized or adjusted. The model described in Fig. 1 refers to attempts to illustrate how pastexperiences of a conflict influence future use (or non-use) of aggression. It was presented as part of escalation patternsacross successive conflicts. But it may also be applied to successive interactions. As such, it can be incorporated in theloop of the presently discussed model describing a conflict from its initiation to its desistance and providing a betterunderstanding of the mechanism that regulates violent actions. The example described above starts with a loop in whichthe man makes non-normative efforts to realize expectations (shoves his wife). The wife's reaction to her husband'saction (a curse) probably does not promote the realization of his goal (unspecified in this example). At this point theprevious model is incorporated, that is, the man concludes that he should enhance (i.e., escalate) his aggression(punches his wife). This process continues until such point that the man's expectations are realized (when his wife

Fig. 3. Between initiation of conflicts and desistance.

Page 11: Conflict escalation to violence and escalation of violent conflicts

307Z. Winstok / Children and Youth Services Review 30 (2008) 297–310

submits to his dictates). It may be expected that the manner in which this conflict is “resolved” affects, at least partially,future conflicts.

The approach proposed here is based in principle on decision-making processes driven by cost–benefitconsiderations. Two principles that guide the escalation of conflicts to violence may support this approach. First,severity of aggression in a conflict escalates gradually (from moderate to relatively more severe actions); second, aviolent conflict ends at the peak of severity of aggressive acts. In both cases, the dynamics are derived from estimationsof cost and benefit. Even if the benefit of using aggression is perceived to be high, violence has its price. This cost is notfixed and is related to the severity of violence used: the more severe the violence, the higher the cost. Because the costof violence is usually more certain than its benefits, and the price of moderate violence is lowest, it makes moreeconomic sense to employ moderate violence first. If this provides the perpetrator with the expected benefit, he “got agood deal.” If not, he learned (and perhaps taught the other party) that violence is ineffective. He may also have learnedthat for violence to be effective it should be more severe. In either case, this tactic ensures that benefits are obtainedusing violence with the minimum cost.

The tactic described above reinforces the process of escalation in another way. Using moderate and ineffectiveviolence incurs a cost that did not produce the expected benefit. At first the motivation for violence was its benefits, butlater it transforms into an attempt to cut losses. The culmination of the severity of violence marks one of three options:the perpetrator achieved his goals, the perpetrator understands that even the most severe violence cannot bring himcloser to these goals (emphasis on benefits), or the perpetrator realizes that even if more severe violence were to bringhim closer to his goals, its cost would be too high. Therefore even a violent conflict ends at this point: if a goal was notreached through escalating violence, it cannot be reached with de-escalation.

Could rationality, even if it is defined as an individual, subjective, and temporal concept, serve as the starting pointfor the study of conflicts escalating to violence? Where is the process of escalation evident in cases of extremely severeviolence perpetrated as if “out of nowhere” rather than evolving gradually? What is rational about such behavior? Insuch cases it can be argued that if there is no perceivable cost to any form or severity of violence, the cost–benefit ratiois extremely high. The perpetrator feels that his situation could not become any worse (he has nothing to lose) and thatthere is no way out of this situation. In such cases violence can be lethal. But this is not the only possible answer to thesequestions. The phrase “conflict escalation to violence” describes a process in which there is first a conflict between theparties reflecting discord, followed by an eruption of violence. This is an escalatory process. The question is, what is itthat escalates? Is it the discord itself? In other words, do the differences between the parties grow larger? Or is it themanifestation of discord that escalates, that is, the parties' actions become increasingly aggressive? The discussion ofescalation focused up to this point on its overt behavioral aspects. Because these are affected by covert cognitive andemotional processes, such as cost–benefit considerations, they also serve as objects of escalation (cost decreases andbenefit increases). If escalation has overt and covert aspects, it is possible that part if not most of the process ofescalation is intra-psychic and covert, and it becomes overtly expressed in behavior only at its peak. Therefore it can beargued that severe violence does not spring out of nowhere, but rather evolves through an inner, covert escalationprocess.

In recent years the study of aggression has been shifting toward an approach based on event perspective. The presentarticle presents such an approach. From this perspective it is argued that conflicts are shaped to a large extent by theconfluence of the parties' perceptions of opportunities and risks (Felson, 1993; Oliver, 1994; Tedeschi & Felson, 1994;Wilkinson, 2003; Wilkinson & Hamerschlag, 2005; Winstok, 2006b). Future studies aimed at expanding anddeepening the understanding of the escalation process in the field of intimate partner violence need to be conductedalong two lines, with focus on the overt expressions of escalation (e.g., various forms of legitimate as well as non-legitimate behaviors) and on its covert aspects (e.g., motivations and goals). In both cases theoretical andmethodological frameworks should be further developed and tested empirically.

A fundamental suggestion of the present study is that most research on violence in intimate relationships focuses onindividuals' single and isolated aggressive actions (context free). This unit of reference impairs our ability to gaininsight into the dynamics of interpersonal conflicts in intimate relationships, a limitation that can be reduced byemploying more dynamic units of reference such as interactional ones (see Gottman & Notarius, 2002, who reviewsstudies that adopt this approach). But the reference unit is only a means and not the objective of the examination.

The primary purpose of the present study is to encourage scholars to focus their research of violence in intimaterelationships on the couple's dynamics, which is what causes and sustains conflicts that escalate to violence. The studyproposes several initial insights into these patterns and their interpretation, for example the escalation of conflict at the

Page 12: Conflict escalation to violence and escalation of violent conflicts

308 Z. Winstok / Children and Youth Services Review 30 (2008) 297–310

behavioral level. But these are merely preliminary suggestions. Further research is needed to verify these suggestionsand examine the pattern distribution in various populations, as well as their reasons and results at the personal,situational, and social levels. Fundamental questions in this field include descriptive matters such as the commonpersonal and interpersonal escalation patterns addressing the form of escalation, its extreme level of severity, its speed,and the consistency of these patterns. Further research should compare the recognizable escalation patterns of variousgroups, for example men and women of different age groups from the general and service populations. More advancedresearch should study the reasons behind these patterns and their outcomes in a variety of context levels, from thepersonal to the situational and the social levels. Progress in these studies requires a broader theoretical framework thatwill map the components relevant to such research and their possible associations. One optional framework that isconsistent with the present study and has proved to be efficient in the study of violence (but had not received properattention in the study of violence in intimate relationships) is social information processing (Crick & Dodge, 1994).

The social information processing model (Crick & Dodge, 1994; Dodge, 1980) describes a cyclical process startingwith acquired information about a given social situation, proceeding with the processing of this information, andending with an action taken to manipulate the situation. It consists of six interrelated steps: (1) encoding situational andinternal cues; (2) interpretation of cues; (3) selecting or clarifying a goal; (4) generating or accessing possibleresponses; (5) choosing a response; (6) and behavioral enactment. This cyclical process is guided by mental structures.Crick and Dodge suggested that: (1) social experiences lead to the generation of latent mental structures that are storedand carried forward over time in memory in the form of social knowledge; (2) the mental structures constitute the“database” in processing models and influence an individual's on-line processing of social cues; (3) on-line processingdirectly influences social behavior; and (4) social behavior and its outcomes are mentally represented and stored inmemory, and they become part of the individual's general social knowledge that will influence future actions. Thesocial information processing model can be placed behind each block in Fig. 2 that describes an example of anindividual and interpersonal escalation process within a single conflict (each block represents a behavior). Each actionby each party is part of the social information processing input and output.

Instead of discussing the model that describes the process leading to action (the last phase) in conventional terms, itcan be treated as a behavior that is relative to another behavior. Then one can perhaps address previous phases as well inthe same way (for example, changes in interpretation, targets, and tactic selection and evaluation). In the case of anindividual pattern, it may be possible to refer to behavior that is relative to the previous behavior of the subject; in thecase of an interpersonal pattern, it may be possible to refer to behavior that is relative to the partner's behavior. Thisformulation enhances the ability of the model to examine escalation. The conceptualization lays the foundation of thestudy of escalation in intimate relationships by mapping the critical process components as well as their possiblerelationships. Many questions and hypotheses can be derived from this conceptualization, as for example, how dodifferent types interpret and evaluate social situations, what are their targets and how do they construct them, what aretheir action skills, and how do they assess and choose their actions.

Finally, the study of courses of escalation can have practical implications as well. Current practice dealing withfamily violence is based mainly on inefficient definitions of victim and aggressor and of myths derived in part fromthese definitions, such as the myth of violent men (aggressors) and of the women (victims) who defend themselves.There is accumulating evidence showing that in many cases the benefit of dealing with violence in intimaterelationships is doubtful. The escalatory perspective, instead of addressing behaviors, deals with the aggressivedynamics and eliminates the problematic need to define the victim and the aggressor without assigning guilt andwithout absolving of responsibility. This perspective shifts attention from the behaviors of the parties in the conflict totheir conflict management. It views the parties primarily as a couple rather than as two cohabiting individuals. Itaddresses the escalatory conflict as a joint, rather than a personal product. It acknowledges the ability of both parties inreplacing devastating patterns with constructive ones in resolving the conflict normatively. This perspective has manyimplications for treatment, but for now it is possible to only acknowledge their potential. In the future it may becomepossible to translate it to intervention procedures subject to established knowledge.

References

Bandura, A. (1973). Aggression: A social learning analysis. Englewood Cliffs, N J: Prentice– Hall.Barnett, O. W., Lee, C. Y., & Thelen, R. E. (1997). Gender differences in attributions of self-defense and control in interpartner aggression. Violence

Against Women, 3(5), 462−481.

Page 13: Conflict escalation to violence and escalation of violent conflicts

309Z. Winstok / Children and Youth Services Review 30 (2008) 297–310

Blumstein, A., Cohen, J., & Farrington, D. P. (1988). Criminal career research: It's value for criminology. Criminology, 26, 1−35.Campbell, A. (2005). Sex differences in direct aggression: What are the psychological mediators? Aggression and Violent Behavior, 11, 237−264.Carins, R. B., Santoyo, V., & Holly, K. A. (1994). Aggressive escalation: Toward a developmental analysis. In M. Potegal & J. F. Knutson (Eds.), The

dynamics of aggression. Biological and social processes in dyads and groups (pp. 227−253). Hillsdale, New Jersey: Lawrence ErlbaumAssociates, Publishers.

Cascardi, M., & Vivian, D. (1995). Context for specific episodes of marital violence: Gender and severity of violence differences. Journal of FamilyViolence, 10, 265−293.

Chung, T. Y., & Asher, S. R. (1996). Children's goals and strategies in peer conflict situations. Merrill-Palmer Quarterly, 42, 125−147.Coie, J. D., Cillessen, A. H. N., Dodge, K. A., Hobbard, J. A., Schwartz, D., Lemerise, E. A., & Bateman, H. (1999). It takes two to fight: A test of

relational factors and a method for assessing aggressive dyads. Developmental Psychology, 35(5), 1179−1188.Crick, N. R., & Dodge, K. A. (1994). A review and reformulation of social information processing mechanisms in children's adjustment. Psy-

chological Bulletin, 115, 74−101.Damasio, A. R. (1994). Descartes' error: Emotion, reason, and the human brain. New York, NY: Putnam.Denzin, N. K. (1984). On understanding emotions. San Francisco: Jossey-Bass.Dobash, R., & Dobash, R. P. (1979). Violence against wives: A case against patriarchy. New-York: Free Press.Dodge, K. A. (1980). Social cognition and children’s aggressive behavior. Child Development, 53, 620−635.Dutton, D. G. (1988). The domestic assault of woman: Psychological and criminal justice perspectives. Boston: Allyn and Bacon.Edelson, J. L., Eisikovits, Z. C., Guttmann, E., & Sela-Amit, M. (1991). Cognitive and interpersonal factors in woman abuse. Journal of Family

Violence, 6, 167−182.Eisikovits, Z., & Buchbinder, E. (2000). Locked in a violent embrace. Thousand Oaks, CA: Sage.Eisikovits, Z., Winstok, Z., & Gelles, R. (2002). Structure and dynamics of escalation from the victim's perspective. Families in Society, 83(2), 142−152.Fagan, J. (1989). Cessation of family violence: Deterrence and dissuasion. In Llyod Ohlin & Michael Tonry (Eds.), Crime and justice: An annual

review of research, Vol. 11. Chicago: University of Chicago Press.Feld, S. L., & Straus, M. A. (1989). Escalation and desistance of wife assault in marriage. Criminology, 27, 141−161.Felson, R. B. (1993). Predatory and dispute-related violence: A social interactionist approach. In R. V. Clarke & M. Felson (Eds.), Routine activity

and rational choice, advances in criminological theory, Vol. 5 (pp. 103−126). New Brunswick, NJ: Transaction Press.Giles-Sims, J. (1983). Wife battering: A systems theory approach. New York: Guilford.Gordon, G. (2000). Definitional issues in violence against women: Surveillance and research from a violence research perspective. Violence against

Women, 6(7), 747−783.Gottfredson, M. R., & Hirschi, T. (1990). A general theory of crime. Stanford, CA: Stanford University Press.Gottman, J. M., &Notarius, C. I. (2002). Marital research in the 20th century and a research agenda for the 21st century. Family Process, 41(2), 159−197.Graham, J., Argyle, M., & Furnham, A. (1980). The goal structure of situations. European Journal of Social Psychology, 10, 345−366.Hamberger, L. K., Lohr, J. M., & Bonge, D. (1994). The intended function of domestic violence is different for arrested male and female perpetrators.

Family Violence and Sexual Assault Bulletin, 10, 40−44.Johnson, M. P. (1995). Patriarchal terrorism and common couple violence: Two forms of violence against women. Journal of Marriage and the

Family, 57, 283−294.Johnson, M. P. (2001). Conflict and control: Symmetry and asymmetry in domestic violence. In A. Booth, A. C. Crouter, & M. Clements (Eds.),

Couples in conflict (pp. 95−104). Mahwah, NJ: Lawrence Erlbaum.Kim, J. -Y. a., & Clifton, E. (2003). Marital power, conflict, norm consensus, and marital violence in a nationally representative sample of Korean

couples. Journal of Interpersonal Violence, 18, 197−219.Loewenstein, G. (1996). Out of control: Visceral influences on behavior. Organizational Behavior and Human Decision Processes, 65, 272−292.Mason, A., & Blankenship, V. (1987). Power and affiliation motivation, stress, and abuse in intimate relationships. Journal of Personality and Social

Psychology, 52, 203−210.Medeiros, R. A., & Straus, M. A. (2006). Risk factors for physical violence between dating partners: Implications for gender-inclusive prevention and

treatment of family violence. In J. C. Hamel & T. Nicholls (Eds.), Family approaches to domestic violence: A practitioners guide to gender-inclusive research and treatment. Springer (also available at http://pubpages.unh.edu/~mas2).

Mummendey, A., Bornewasser, M., & Loschper, G. (1982). Defining interactions aggressive in specific social contexts. Aggressive Behavior, 8(2),224−228.

Murphy, B. C., & Eisenberg, N. (2002). An integrative examination of peer conflict: Children's reported goals, emotions, and behaviors. SocialDevelopment, 11, 534−557.

Ohbuchi, K., & Tedeschi, J. T. (1997). Multiple goals and tactical behavior in social conflicts. Journal of Applied Social Psychology, 27, 2177−2199.Oliver, W. (1994). The violent social world of black men. New York: Lexington Books.Pagelow, M. D. (1981). Woman-battering: Victims and their experiences. Beverly Hills, Calif.: Sage.Prince, J. E., & Arias, I. (1994). The role of perceived control and the desirability of control among abusive and nonabusive husbands. American

Journal of Family Therapy, 22, 126−134.Sherman, L., & Berk, R. A. (1984). The specific deterrent effects of arrest for domestic assault. American Sociological Review, 49, 261−272.So-Kum Tang, C. (1999). Marital power and aggression in a community sample of Hong Kong Chinese families. Journal of Interpersonal Violence,

14, 586−602.Stets, J. E. (1990). Verbal and physical aggression in marriage. Journal of Marriage and the family, 52, 501−514.Stets, J. E., & Pirog-Good, M. A. (1987). Violence in dating relationships. Social Psychology Quarterly, 50, 237−246.Straus, M. A., Gelles, R. J., & Steinmetz, S. K. (1980). Behind closed doors: Violence in the American family. New York: Doubleday/Anchor Books

(Re-issued with a new forward and a post-script chapter by Transaction Publications, 2006).

Page 14: Conflict escalation to violence and escalation of violent conflicts

310 Z. Winstok / Children and Youth Services Review 30 (2008) 297–310

Straus, M., Hamby, S., Boney-McCoy, S., & Sugerman, D. (1996). The revised conflict tactics scales (CTS2): Development and preliminarypsychometric data. Journal of Family Issues, 17(3), 283−316.

Straus, M. A., & Gelles, R. J. (1990). How violent are American families ? Estimates from the National Family Violence resurvey and other studies. InM. A. Straus & R. J. Gelles (Eds.), Physical violence in American families: Risk factors and adaptation to violence in 8,145 families (pp. 95−112).New Brunswick, NJ: Transaction Press.

Sugihara, Y., & Warner, J. A. (2002). Dominance and domestic abuse among Mexican Americans: Gender differences in the etiology of violence inintimate relationships. Journal of Family Violence, 17, 315−339.

Swann, W. B., Jr., Pelham, B. W., & Roberts, D. (1987). Causal chunking: Memory and inference in ongoing interaction. Journal of Personality andSocial Psychology, 53, 858−865.

Tedeschi, J. T., & Felson, R. B. (1994). Violence, aggression, and coercive actions. Washington, DC: American Psychological Association.Vivian, D. (1998, November). The context and characteristics of partner violence in couples seeking marital therapy. Paper presented at the meeting

of the Association for the Advancement of Behavior Therapy.Walker, L. E. (1979). The Battered Woman. New York: Harper and Row.Wilkinson, D. L. (2003). Guns, violence, and identity among African American and Latino youth. New York: LFB Scholarly Publications, LLC.Wilkinson, D. L., & Hamerschlag, S. J. (2005). Situational determinants in intimate partner violence. Aggression and Violent Behavior, 10, 333−361.Winstok, Z. (2006). The why and what of intimate conflicts: The effect of partners' divergent perceptions on verbal aggression. Journal of Family

Violence, 21, 461−468.Winstok, Z. (2006). Sex differences in the structure and content of escalatory tendencies at the home and workplace. Aggressive Behavior, 32,

433−441.Winstok, Z. (submitted for publication). Control and violence in intimate relationships.Winstok, Z. (2007). Toward an interactional perspective on intimate partner violence. Aggression and Violent Behavior. doi:10.1016/j.avb.2006.12.001Winstok, Z., & Eisikovits, Z. (2006, May). Motives and control in escalatory conflicts in intimate. relationships. Paper presented at a study group on

Recent Trends in Intimate Violence Intervention. New-York: NYC.Winstok, Z., Eisikovits, Z., & Gelles, R. (2002). Structures and dynamics of escalation from the batterer's perspective. Families in Society. Journal of

Contemporary Human Service, 129−141.