Research Report: Conflict + Anxiety = Turmoil! Introducing a Measure of Conflict Response Derailers Coleman, Peter T 1 ; Chan, Anthea 2 1 Morton Deutsch International Center for Cooperation and Conflict Resolution at Teachers College, Columbia University 2 Senior Research Associate at Coqual ProQuest document link ABSTRACT (ENGLISH) Experiences of social conflict often trigger anxiety, which is associated with more reactive, extreme, and problematic responses to conflict. However, individuals respond to conflict anxiety in different ways. This article presents the findings from a series of scale development studies that sought to create and test a measurefor assessing common behavioral response tendencies in interpersonal conflict. The Conflict Anxiety Response Scale (CARS) is based on a theoretical framework proposed by Morton Deutsch-a pioneering conflict research scholar and clinical practitionerthat outlines a set of tendencies identified as common manifestations of anxiety management in conflict. The present line of research aims to develop a valid and reliable scale to assess respondents on these inclinations and explore their consequences. Ultimately, we seek to offer a practical assessment tool of conflict anxiety responses to enhance self-awareness, professional development, and well-being. FULL TEXT Headnote Experiences of social conflict often trigger anxiety, which is associated with more reactive, extreme, and problematic responses to conflict. However, individuals respond to conflict anxiety in different ways. This article presents the findings from a series of scale development studies that sought to create and test a measurefor assessing common behavioral response tendencies in interpersonal conflict. The Conflict Anxiety Response Scale (CARS) is based on a theoretical framework proposed by Morton Deutsch-a pioneering conflict research scholar and clinical practitionerthat outlines a set of tendencies identified as common manifestations of anxiety management in conflict. The present line of research aims to develop a valid and reliable scale to assess respondents on these inclinations and explore their consequences. Ultimately, we seek to offer a practical assessment tool of conflict anxiety responses to enhance self-awareness, professional development, and well-being. Keywords: conflict, coping, conflict derailers, anxiety, scale development, Morton Deutsch Introduction Conflict is notorious for making most people anxious. When asked to free associate to the term, most of us respond with words such as problem, aggression, tension, violence, confrontation, or war. When we find ourselves face-toface with a tricky conflict, often our stomach sinks, our blood pressure rises, and we start to feel a creeping sense of dread that the worst will happen (Heffner et al. 2004). And when our conflicts mingle with our anxiety… watch out! Thus, it seems odd that most of the more popular conflict style assessments in use today seem to neglect the role of anxiety. For that matter, most of these assessments avoid altogether the effects of emotions in conflict (Shapiro 2004). For example, the Thomas-Kilmann Conflict Mode Instrument (Thomas 1976), based on the dual concern framework originally developed by Blake and Mouton (1964) and arguably the most utilized conflict style assessment of our time, focuses on different behavioral responses to disputes but neglects the foundational role of emotions completely. The same can be said for other commonly employed tools, such as the Intercultural Conflict Style Inventory (Hammer 2005) and the Rahim Organizational Conflict Inventory-II (ROCI-II) (Rahim 1983; WeiderPDF GENERATED BY PROQUEST.COM Page 1 of 14 Hatfield 1988). This article seeks to address this missing link in our understanding of our emotive selves in conflict. It builds on the clinical observations and insights of one of the founders of the field of conflict resolution, and one of the most eminent psychologists of all time, Morton Deutsch (APA Monitor 2002; Coleman 2018). It offers a summary of Deutsch’s model of conflict anxiety and then presents an overview of a series of studies conducted to develop and test a new assessment tool, the Conflict Anxiety Response Scale (CARS). The article summarizes the history of this research and discusses in detail two recent studies that offer reliability estimates of the scales and initial validation of the instrument. It also presents preliminary findings on the relationship of scores on the scale to differences in gender, positive and negative affect, generalized anxiety, and well-being. Conflict Anxiety Response Scale: The Origin Story Social conflict is defined as an encounter between people over differences that matter (Follett 1924; Coleman and Ferguson 2014). It can be more destructive, involving harms and losses for some or all disputants, or more constructive, where the disputants walk away feeling mostly satisfied or enhanced by the dispute, depending on a variety of factors (Deutsch 1973; Coleman, Deutsch, and Marcus 2014). However, most conflictual encounters trigger tension and anxiety initially (Larkin, Frazer, and Semenchuk 1996), which, in turn, makes us more inclined to perceive or instigate conflict (Etkin et al. 2010). When lingering, this state can stress our relationships, lower our selfesteem (Ozdemir 2014), and increase our blood pressure (Honeycutt, Hample, and Hatcher 2016), thereby making us feel even more anxious and disagreeable. Anxiety is a psychological, physiological, and behavioral state induced in animals and humans by a real or perceived threat to well-being or survival (Steimer 2002). It is associated with increased arousal, autonomic and neuroendocrine activation, and chronic behavioral-response patterns, which are believed to facilitate coping with adverse circumstances. Anxiety can be a normal and healthy reaction to worrisome events, and moderate amounts have been shown to improve performance on some tasks by helping to focus attention (Teigen 1994). But prolonged states of anxiety, particularly when coupled with an underlying anxiety disorder, can have serious negative consequences for our physical and psychological health and social relationships. In the 1970s, Morton Deutsch, an experimental social psychologist and psychoanalytically trained psychotherapist, observed an intriguing pattern of responses to conflict in his clinical practice with romantic couples (Deutsch 1993). Initially, Deutsch noted that many of the individuals in his practice typically evidenced a palpable level of anxiety when disputes would emerge between them and their partners. Trained in psychoanalysis, Deutsch understood this angst as the result of the partners’ early experiences with their respective families of origin, and their related fears of causing or suffering from severe harm in response to conflict. He wrote, “In clinical work, I have found that the anxiety is often based on unconscious fantasies of being overwhelmed and helpless in the face of the other’s aggression or of being so angry and aggressive that you will destroy the other” (Deutsch 1993: 514). Over time, Deutsch also began to observe that the individuals comprising these couples tended to respond to the anxiety instigated by their conflicts in one or more of the following ways: * chronically avoiding conflict or obsessively seeking out conflict; * becoming hard and unyielding or becoming overly soft and unassertive; * overintellectualizing and rationalizing or becoming overwhelmingly saturated in emotion; * standing rigid and controlling or falling into a state of loose disorganization; * skyrocketing into escalation or shrinking back into minimization of problems; and * compulsively revealing everything or stonewalling and concealing everything. Deutsch observed that moderate tendencies along these dimensions were quite normal-and often adaptiveresponses to the feelings and circumstances his clients faced. However, the more anxious the disputants became in response to conflict and the longer their distress persisted, the more chronic and extreme their responses, and the more problematic it was for their relationships and psychosocial well-being. Deutsch, the social-psychological theorist, found this pattern of dynamics fascinating and instructive, and described it later in an article in American Psychologist (Deutsch 1993). Deutsch, the psychotherapist, found that helping PDF GENERATED BY PROQUEST.COM Page 2 of 14 individuals and couples begin to recognize their own response tendencies along these dimensions often helped them to keep such tendencies in check when they found themselves becoming derailed by their own “go to” responses (Table One). Hypotheses Although Deutsch’s clinical observations of his clients’ responses to conflict-induced anxiety proved useful in his clinical practice, he never empirically tested the basic assumptions of the model that he articulated (Deutsch 1993). They included: 1. Individuals experiencing higher levels of conflict anxiety will evidence more of at least some of the six-dimensional reactions to conflict. 2. The twelve types of reactions to conflict anxiety fall into six bipolar dimensions (e.g., individuals will tend to either avoid conflict or become overly involved). 3. The more extreme the responses to conflict anxiety along each dimension, the more problematic it will be for the affect, well-being, and relationships of the individual. 4. The more extreme the responses to conflict anxiety along each dimension, the more likely individuals will evidence less constructive and more destructive conflict dynamics. 5. Increasing awareness of one’s more chronic conflict-anxiety response tendencies may allow one to modify them when they are inappropriate in a given conflict. 6. More optimal responses to conflict anxiety will likely fall in between the more extreme poles of each dimension, thus incorporating the beneficial aspects of some of the responses. The studies discussed below seek to develop the empirical means to begin to test these and other related hypotheses. Ultimately, we seek to develop an assessment tool for individuals, couples, and families that could help to increase awareness of each person’s chronic conflict “derailer” tendencies-thus providing them with useful information feedback for their personal and professional development. This research program seeks to parallel similar initiatives, such as the decades of research conducted on the Hogan Development Survey (Hogan et al. 2021), a well-validated leadership assessment tool that offers managers feedback on their “Dark Side”-the more problematics aspects of their leadership tendencies that have been shown to have negative consequences for job performance. The Development of the Conflict Anxiety Response Scale Over the past few years, we at the Morton Deutsch International Center for Cooperation and Conflict Resolution at Columbia University have been studying these hypothesized conflict tendencies to better understand their psychometric properties and test the basic assumptions of Deutsch’s model and its consequences for mental health, conflict management, and well-being. Here, we briefly summarize a series of studies leading up to the development of the current scale. We then report in depth on two studies that examine the factor structure, reliabilities, and predictive validity of the current version of the Conflict Anxiety Response Scale. Study 1 was conducted to examine overlaps in the 12 CARS dimensions and enhance the distinctiveness of the subscales. Study 2 tested the factor structure, interitem reliability, and predictive validity of the scale in relation to standard measures of affect, anxiety, and well-being, thereby allowing for the examination of Hypotheses 1-3 from above (Figure One). Overview of Previous Scale Development Studies We previously conducted seven iterations of studies on the CARS survey development designed to assess respondents’ conflict-anxiety tendencies along the proposed six bipolar dimensions identified by Deutsch (1993). For each study, the participants were asked questions about how they typically respond to important conflicts in their personal lives. Based on Deutsch’s framework (1993), we drafted item sets for each of the twelve hypothesized tendencies included in the six bipolar dimensions. Over each iteration, we honed the scales and their psychometric properties to increase construct clarity, reliability, and validity. Each iteration resulted in a separate empirical study, leading to further refinement of the scale’s items and format. Across these seven studies, we revised our scale from 53 items in Study 1 (N =100), to 83 items in Study 2 (N =180), 75 items in Study 3 (N = 215), 53 items in Study 4 (N PDF GENERATED BY PROQUEST.COM Page 3 of 14 = 204), 61 items in Study 5 (N = 212), 60 items in Study 6 (N = 298), and 62 items in Study 7 (N = 620). The format of each study was an online questionnaire, and data were collected from participants through Amazon’s Mechanical Turk (mTurk) online platform, a Qualtrics Panel, or the Prolific recruitment system. For each study, confirmatory factor analyses were run to determine whether each item set fit with the twelve dimensions as described by Deutsch’s framework. Over the course of these seven studies, we improved model fit with each iteration. We want RMSEA to be close to zero, preferably at 0.05. MacCallum, Browne, and Sugawara (1996) have used 0.01, 0.05, and 0.08 to indicate excellent, good, and medium fit, respectively. Thus, the latest versions of our models can be considered tween and medium fit. Fit statistics, descriptive statistics, and Cronbach’s alphas for each iteration are presented in Tables Two and Three. Study 1: Exploratory Analysis of Internal Structure of Scale The goal of this new study (Study 8 in our sequence but Study 1 for this article) was to build on prior iterations of studies of the questionnaire to assess the twelve response tendencies to interpersonal conflict. In this study, we describe the items and the process by which we identify items negatively impacting interitem reliability of the subscales and the factor structure. Participants Subjects were 760 participants recruited via Prolific, an online survey recruitment platform. Subjects ranged in age from 18 to 80 years (M = 33.12, SD = 11.89). 50.3% of the participants were women (N=383), while 48.1% of the participants were men (N = 366), and 1.2% identified as a gender not listed (N= 9). Our sample consisted of people of various ethnic identities: American Indian/Alaskan Native (N = 10, 1.3%); Asian/Pacific Islander (N= 90, 11.8%); Black/African American (N= 59, 7.8%); Hispanic/Latin American (N= 76, 10%), White (N= 567, 74.5%); and other (N= 12, 1.6%). Two hundred and fourteen participants have a high school diploma or GED (28.1%), 115 have an associate degree (15.1%), 295 have a bachelor’s degree (38.8%), 89 have a master’s degree (11.7%), 13 have a doctorate (1.7%), 20 have a professional degree (2.6%), and 15 have some form of education not listed (2.0%). Subjects were paid $2.50 for their time in completing a 15-minute survey. Measures Conflict Anxiety Response Scale Version 8.0. Seventy-six items, identified and revised through the prior seven studies, were tested in this iteration of the scale. The introduction to the scale read, “People face many different types of conflict in their lives. Some conflicts are minor, others are more serious; some are with strangers, others with colleagues or within intimate relationships; some are in public places, others at home or at work. Below is a list of statements that reflect different ways people respond to conflict. Please indicate how often you tend to respond to conflicts in your personal life in the following ways.” The response options were 5-point Likert scales with choice options from “Never to Always.” Analysis The dataset of 760 respondents was randomly split in half, leaving two datasets with 380 participants each. On the first dataset, we conducted exploratory factor analyses with principal axis factoring and promax rotation on every possible combination of the 12 factors, resulting in 66 EFAs. For example, we ran EFAs on all the Avoiding items in combination with the set of items for each of the Soft, Hard, Involved, Escalating, etc. subscales, one at a time. These paired analyses were conducted because the 76 items of the 12 hypothesized dimensions failed to resolve after countless iterations. This alternative paired EFA is recommended by Watkins (2018). Factors were identified for each EFA based on having eigenvalues >1. In each of these EFAs, we examined the loadings of each item to factors in the pattern matrix, using a minimum loading of 0.40. These loadings represent the relationship between a factor and the item, when controlling for the other factors. We conducted these 66 EFAs to examine whether any of our items cross-loaded with other factors, causing fuzzy demarcations between subscales. If an item loaded on a factor other than the one it was hypothesized to (e.g., if an Avoiding item loaded on a factor containing mostly Soft items), we dropped it from the scale. Finally, we estimated Cronbach’s alpha for each subscale and looked for items that were negatively influencing interitem reliability. Based on these analyses, we arrived at a smaller set of items. PDF GENERATED BY PROQUEST.COM Page 4 of 14 These analyses were conducted in SPSS Version 26. On the testing dataset (N= 380), confirmatory factor analysis (CFA) was performed using R’s lavaan package. Six CFAs were conducted between each hypothesized opposing pair (e.g., Avoid versus Involve), using the trimmeddown set of items resulting from our analyses of the first dataset of 380 respondents (Exploratory Factor Analysis). We attempted to run CFAs on all twelve dimensions together, but the model did not converge. Thus, we ran the CFAs in pairs and a CFA for all twelve dimensions using the full dataset. Confirmatory factor analysis models were estimated using the lavaan package in R studio version 3.6.3. Fit was determined based on several indices: the comparative fit index (CFI), Tucker-Lewis Index (TLI), root mean square error of approximation (RMSEA), and AIC (Akaike’s Information Criteria) and BIC (Bayesian Information Criteria) to compare models. Hu and Bentler (1999) recommend a cutoff of 0.95 or higher for CFI and TLI and close to or below 0.05 on the RMSEA to indicate excellent fit between the model and the data. We also examined significance tests for factor loadings and R2 values. When the models were finalized, composite scores were calculated for each of the twelve dimensions. Pearson’s correlations were calculated between each of the hypothesized opposing dimensions and between each dimension and Generalized Anxiety (Table Four). Results Here, we report findings for each of the six bipolar dimensions proposed by Deutsch from the EFA, CFA, and various tests of fit. See Table Five for items that were kept after analyses and Table Six for items that were dropped. Avoiding and Involved Six items were tested for the Avoiding subscale and six for the Involved subscale. Items that cross-loaded with another CARS dimension are marked with a †, while items that decreased interitem reliability as measured by Cronbach’s alpha are marked with a ®. Three models were tested: one with all the items, one without items that reduced Cronbach’s alpha, and the final one without cross-loading items. Our final model had an RMSEA of 0.086, which would be considered below medium fit by Browne and Cudeck (1993) but was an improvement on the first model, which had an RMSEA of 0.109. This model was close to meeting standards for CFI (0.926) and TLI (0.897) of >0.95, as set by Hu and Bentler (1999). This indicated that the Avoiding and Involved dimensions needed additional honing. Avoiding had good interitem reliability after removal of items (a before removal = 0.835, final a = 0.808). Involved had improved interitem reliability after removal of items (a before removal = 0.755, final a = 0.789). The final subscales for Avoiding and Involved had an insignificant negative correlation (r = -0.063, p = 0.226), suggesting they are orthogonal dimensions. Soft and Hard Four items were tested in Study 1 for Soft and six items were tested for Hard. The final model had an RMSEA of 0.067, which would be considered medium fit and was an improvement on the first model, which had an RMSEA of 0.091. This model met standards for CFI (0.96) and TLI (0.945). Soft had sufficient interitem reliability after removal of items (a before removal = 0.707, final a = 0.752). No items needed to be removed for Hard (a = 0.861). The final subscales for Soft and Hard had a significant but moderate negative correlation (r = -0.302, p
Purchase answer to see full attachment