Conflict-intelligent leadership in a messy, anxious world: Development and validation of a multidimensional assessment of leaders' conflict capacities

1 Introduction

Modern workplaces operate within a broader social landscape marked by political extremity, social fragmentation, and increasingly complex patterns of conflict (Coleman, 2024). Leaders today confront polycrises—simultaneous and interconnected disruptions spanning identity conflict, moral divides, systemic injustice, and institutional volatility (Coleman, 2024; Tooze, 2022; Whiting and Park, 2023; World Economic Forum., 2023). These conditions amplify tensions within teams and organizations, intensify emotional reactivity, and increase the likelihood that leaders' decisions will be scrutinized under conditions of heightened public sensitivity.

Research consistently shows that unresolved conflict carries substantial costs, including reduced work engagement, diminished psychological safety, lowered productivity, and organizational paralysis (Edmondson and Lei, 2014; Society for Human Resource Management (SHRM), 2024, 2025). Yet conflict, when properly harnessed, can also be a source of creativity, learning, and systemic renewal (Coleman, 2011; Coleman and Ferguson, 2014; Coleman and Lim, 2001; Deutsch, 1973; Gelfand et al., 2012). This dual nature of conflict—potentially destructive or generative—has led scholars and managers to seek frameworks that distinguish between maladaptive and adaptive forms of conflict engagement.

2 Theoretical foundation

2.1 The emergence of conflict intelligence and conflict-intelligent leadership

Derived from decades of research on constructive conflict engagement, conflict intelligence (CIQ) was initially defined as the capacity to “facilitate the management of qualitatively different types of conflicts in diverse or changing situations, effectively and constructively” (Coleman, 2018). CIQ expands traditional approaches to conflict resolution by integrating complexity-science insights into a multilevel competency model that explains why some employees remain effective under pressure while others collapse (Coleman, 2024, 2025b).

Building on this foundation, conflict-intelligent leadership (CIQ-L) was subsequently defined as a leaders' ability to engage with and transform conflicts at multiple levels—self, social, situational, structural, and systemic. Fundamentally, CIQ-L is about leveraging conflict energy for innovation and cohesion rather than merely resolving it (Coleman, 2024, 2025b,a). CIQ-L includes regulating one's own emotional reactivity, utilizing transparent communication, reading contextual shifts, balancing tension levels, and framing conflict as a productive force for organizational improvement.

For the purposes on the current paper, we define Conflict Intelligence as the capacity to understand, navigate, and transform conflict in ways that produce stronger relationships, better decisions, and more resilient systems. Conflict-Intelligent Leadership applies CIQ to leaders—those responsible for guiding others through conflict, uncertainty, and change—by engaging with and transforming conflict dynamics at multiple levels—self, social, situational, structural, and systemic. It recognizes that leading through conflict requires not only self-awareness and adaptability, but also the ability to manage public perception, maintain team stability, and engage systemic challenges effectively.

Although CIQ-L has strong theoretical grounding across psychology, conflict studies, and complexity science (Coleman, 2018, 2024, 2025b; Vallacher et al., 2010), the field has lacked an empirically validated measurement instrument to directly evaluate its effects. Identifying and measuring the behavioral markers of CIQ-L is essential for assessing leadership effectiveness, designing training, and advancing theory on conflict engagement.

2.2 Conceptual framework and model

CIQ-L is grounded in a multilevel theoretical model (Coleman, 2018, 2024) in which leaders must simultaneously manage:

• Self-regulation: emotional regulation, conflict anxiety management
• Social dynamics: interpersonal conflict engagement, transparent communication
• Situational adaptivity: matching strategies to context and power dynamics
• Structural stewardship: shaping norms, incentives, and conflict processes
• Systemic wisdom: understanding sociopolitical forces shaping conflict

Drawing from mathematical modeling, CIQ-L can be represented as an evolving dynamic system (Coleman, 2025a), which formalizes the following interacting components:

dCILdt=αA+βE+γS+δO+θP-λR

According to the AESOPR model, growth in conflict-intelligent leadership is shaped by:

• Adaptive leadership (A)—adjusting approaches to fit shifting conflict demands
• Emotional regulation (E)—maintaining steady presence under pressure
• Systemic awareness (S)—sensing broader forces shaping conflict
• Optimal tension management (O)—balancing too much vs. too little heat
• Public perception management (P)—communicating transparently and responsibly
• Resistance to change (R) inhibits these dynamics, positioning CIQ-L as an evolving system rather than a static trait.

As such, Conflict-Intelligent Leadership is not viewed as a trait, but rather as a dynamic capacity that evolves over time depending on how leaders regulate themselves, adapt to complexity, manage tension, understand systems, and navigate public narratives. CIL grows or declines depending on five interacting forces: Adaptive Leadership, Emotional Regulation, Systemic Awareness, Optimal Tension Management, Public Perception—minus Resistance to Change. So, in plain terms, CIQ Leadership effectiveness in conflict rises when leaders adapt wisely, stay regulated, understand systems, manage tension skillfully, and handle public narratives—and it falls when resistance, inertia, or backlash overwhelm them.

While Conflict-Intelligent Leadership can be operationalized and developed as a multi-level set of interactive competencies, which includes knowledge, attitudes and skills at the self, social, situational, structural, and systemic levels, it culminates in what we describe as a “meta-competency,” the effects of which can be measured through the perceptions and experiences of the employees supervised by the leader.

However, despite extensive theoretical grounding, CIQ-L currently lacks an empirically validated measure—limiting research, assessment, and training. The present work attempts to fill this gap by developing and validating the Conflict-Intelligent Leadership Scale (CIQ-L). To achieve this goal, we conducted two studies, discussed in the following section:

Study 1: scale development, exploratory and confirmatory factor analyses, and predictive validity with key organizational outcomes.

Study 2: cross-sectional test of convergent and discriminant validity against related leadership constructs.

3 Empirical studies

This section first provides an overview of the scale development and item generation process used across studies. It then summarizes three pilot studies that informed the development of the current scale. The next section discusses in detail two validation studies of the Conflict-Intelligent Leadership Scale (CIQ-L). In Study 1 (N = 496), a longitudinal design was utilized to evaluate the internal structure of the CIQ-L scale along with its predictive validity (i.e., a measure's ability to predict an outcome taking place at a later date; Lin and Yao, 2024). Finally, Study 2 (N = 314) employed a cross-sectional design to replicate the internal structure through confirmatory factor analysis and to assess convergent and discriminant validity.

3.1 Scale development and item generation

The first step in developing the CIQ-L scale involved formulating a clear definition of the measured construct (DeVellis and Thorpe, 2021; Kyriazos and Stalikas, 2018). Thus, the definition for CIQ-L, the multilevel framework (i.e., Self-regulation, Social dynamics, Systemic awareness, Situational adaptivity, Structural stewardship, and Systemic wisdom), and the six theoretical dimensions from the AESOPR mathematical model (i.e., Adaptive leadership, Emotional regulation, Systemic awareness, Optimal tension management, Public perception management, and Resistance to change) guided item generation and scale design. In this series of studies, leadership conflict intelligence referred to leaders' capacity to navigate conflicts adaptively and effectively across levels (Coleman, 2024, 2025a). To ensure specificity and enhance clarity (DeVellis and Thorpe, 2021), we outlined the context for the CIQ-L measure to capture employees' perceptions of their managers' conduct during workplace conflicts. As this instrument was reported from employees' perspectives, items for the CIQ-L scales focused exclusively on observable, behavioral aspects of leadership conflict intelligence. The target sample comprised adults over the age of 18 and working full-time under a direct supervisor.

The second step for scale development process was generating scale items, where the research team utilized a deductive approach. Adhering to methodologists' recommendations of a minimum of three to five items for each factor (MacCallum et al., 1999; Velicer and Fava, 1998), five to eight items were generated for each dimension. Increasingly incorporated in psychological and educational assessment research (e.g., Fyffe et al., 2024; Speer et al., 2023; Thompson, 2007), generative artificial intelligence (AI) models can be beneficial in the brainstorming process (Nah et al., 2023), with AI-generated psychological measurements performing as well as those developed using traditional methods (Hernandez and Nie, 2023; Hommel et al., 2022; Terry et al., 2025). Specifically, GPT-based models have demonstrated item generation capabilities in the literature (Hernandez and Nie, 2023; Hommel et al., 2022; Jia and Lee, 2025; Salah et al., 2025). In addition, these AI large language models can reduce the resources required (e.g., cost, time) needed for the item-generation process in creating assessments (Hao et al., 2024; Lee and Jia, 2026). However, to ensure efficiency without compromising quality, it is highly recommended to combine AI-generated items with expert review (Hao et al., 2024; Salah et al., 2025; Shneiderman, 2022). Thus, the items were derived from two primary sources: (1) items written by humans with references from existing leadership and conflict management measures that were subsequently included in Study 2 and (2) items generated by AI (i.e., ChatGPT o4-mini; Salah et al., 2025) corresponding to each CIQ-L dimension. Items such as “My manager handles conflicts involving significant cultural differences effectively” and “My manager seems to align conflict decisions with our long-term values and purpose” were created by the research team, while “My manager maintains steadiness even when our team is under intense pressure” and “My manager communicates the benefits of addressing conflict rather than postponing it” were AI-generated.

Furthermore, both positively and negatively worded items were included to avoid agreement bias, defined as respondents' tendencies to agree with the statements irrespective of the actual content (DeVellis and Thorpe, 2021). Moreover, using mixed wording allowed detection of straight-lining (i.e., rating all items as “4 = Somewhat agree”; Podsakoff et al., 2003). To further control for acquiescence bias, we utilized Qualtrics' built-in detection for straight lining. However, although some scale development literature recommended having reverse-coded items, others have argued that they reduced factor reliability and confounded the internal structure due to a method effect rather than content (Sonderen et al., 2013; Van Dam et al., 2012; Weijters et al., 2013; Widaman et al., 1999; Roszkowski and Soven, 2010). Following recommendations from (Dodeen 2023), we included the reverse-coded items in the survey as fillers but excluded them from our analyses.

Finally, two attention checks, “For quality control, please select always for this statement” and “For quality control, please select never for this statement,” were included in the survey to ensure that participants were reading the prompts carefully throughout the survey.

3.2 Pilot studies

A series of three pilot studies was first conducted to test items capturing leaders' conflict intelligence across the multilevel theoretical model (i.e., Self-regulation, Social dynamics, Systemic awareness, Situational adaptivity, Structural stewardship, and Systemic wisdom; Coleman, 2025a). For each iteration, the item pool was refined based on the research team's expertise and empirical results from the previous studies: Pilot 1 (N = 202) had 12 items, Pilot 2 (N = 202) had 33, and Pilot 3 (N = 386) had 26 items. Items that did not have normal distribution, little variance, or did not load clearly on any factors were removed (full process discussed in Study 1 below). Thus, the pilot studies results contributed to enhancing the item clarity and factor interpretability.

Specifically, for item development, all 12 items from the Pilot 1 study were drafted and refined qualitatively three times by two of the authors (a tenured professor and a doctoral candidate in organizational psychology and conflict resolution) for scale consistency and readability. For example, “How wide is the gap between how well your manager think they respond to conflict, and how you experience them in conflict (1 = vast gap to 5 = no gap)?” became “There is a large gap between how well my manager think they respond to conflict and how I experience them in conflict (1 = strongly disagree to 5 = strongly agree).” After initial analysis of Pilot 1 that resulted in a one-dimension CIQ-L scale, we decided to increase the number of items to 33 in Pilot 2 to capture additional aspects of conflict intelligence. In addition to human-written items, we also provided the theoretical dimensions and our definition of conflict intelligence to ChatGPT o4-mini for machine-generated items. The first iteration for Pilot 2 consisted of 48 items, which were then evaluated qualitatively by a team of three organizational psychology doctoral students and one tenured professor. After removing items that did not perform well in Pilot 1, overlapped theoretically, and those that did not fully align with our definition of CIQ, we tested a total of 33 items in the Pilot 2's survey. Finally, for Pilot 3, we retained the items that loaded onto the Pilot 2's factor analysis scale. Any items that failed the preliminary checks (see Study 1 for more information) and has low loadings (< 0.40) onto the factors were entered into large language models for rephrasing. The machine-generated revisions went through final edits by the research team before tested in the Pilot 3 survey.

All studies were reviewed and approved by the Institutional Review Board at the authors' home institution. The data was collected via questionnaires on Qualtrics¹, which asked participants to report the extent to which each statement described their immediate managers' behavior in workplace conflict. All studies used a longitudinal design, where outcome variables (e.g., Psychological Safety, Leader-Member Exchange) were collected at a later point in time to establish predictive validity. Across all three pilot studies, employees' perceptions of their managers' CIQ-L scores were positively correlated with their Job Satisfaction, Work Engagement, and Psychological Safety, providing initial evidence of the practical implications of improving leaders' conflict intelligence at work. However, despite strong evidence of predictive validity, the factor structures and subscale reliabilities varied across iterations of the studies. Thus, we reviewed the factors emerging from the pilot studies to identify the underlying theoretical dimensions for the AESOPR mathematical model. This process resulted in an initial pool of 36 items across the six hypothesized dimensions of leadership conflict intelligence (i.e., Adaptive leadership, Emotional regulation, Systemic awareness, Optimal tension management, Public perception management, and Resistance to change). The psychometric property of these items was subsequently evaluated in Study 1.

3.3 Study 1

Study 1 had two main objectives. First, it sought to determine the internal structure of the 36-item CIQ-L through exploratory and confirmatory factor analysis. Second, to establish the real-world relevance and value of CIQ-L, predictive validity was examined using prior predictors of organizational success in the literature.

Specifically, we propose that Conflict-Intelligent Leadership Scale will be positively associated with Psychological Safety, Work Engagement, Job Satisfaction, and Leader-Member Exchange (LMX). Psychological safety has been identified as foundational for constructive conflict resolution (Edmondson, 1999; Edmondson and Lei, 2014); without it, even well-designed interventions may fail. Furthermore, given that high work engagement and job satisfaction are linked to greater productivity and reduced turnover intentions (Stajkovic and Luthans, 1998), understanding their relationship with CIQ-L may help organizations mitigate the hidden costs of conflict and enhance overall organizational resilience. Finally, LMX, a construct examining managers' relationships with their employees, has been shown to be an important outcome for conflict leadership (Zhao et al., 2019). Taken together, these constructs can provide a robust assessment of the predictive validity of the CIQ-L scale.

3.3.1 Participants

Participants were found through Prolific, an online recruitment platform for research that tended to provide higher quality data than other platforms such as Amazon's MTurk (Douglas et al., 2023). To establish predictive validity, a longitudinal design was employed by collecting data at two time points a week apart (i.e., Time 1 and Time 2). At Time 1, 496 participants (N = 496) completed the survey after excluding those with incomplete responses (n = 16), failed attention checks (n = 10), or without a direct manager to rate (n = 19). The final sample consisted of 49.8% women, 49.8% men, and 0.4% identifying as non-binary or third gender. Racial demographics were as follows: 40.1% White/Caucasian, 22.4% Black/African American, 19.2% East/South/Southeast Asian, 8.0% Mixed/Multiracial, 6.0% non-White Hispanic/Latino, and 2.5% identifying as other racial groups (e.g., Indigenous, Middle Eastern/North African). When asked how long they had worked under their current manager, 11.7% said less than a year, 38.9% indicated 1–3 years, 24.4% for 3–5 years, and 25% had worked with their manager for over 5 years. Finally, more than two-thirds of the respondents worked in the private sector (68.8%), 27.0% in the public sector, and 3.8% were in the non-profit sector. In Time 2, 200 participants who took part in Time 1 were randomly sampled to take the survey. Tableone's chi-square statistics (Yoshida, 2022) indicated no significant difference between the 200 participants in Time 2 and the remaining 296 participants across race (p = 0.47), age (p = 0.22), gender (p = 0.49), levels of education (p = 0.31), and mean scores across the CIQ-L items (p values ranged from 0.27 to 0.35).

3.3.2 Procedures

In Time 1′s survey, participants were first asked to rate the frequency with which they experienced their managers exhibiting each behavior in conflict. All 36 items for the CIQ-L scale were presented in randomized order along with two attention checks. The survey then ended with demographic questions, and upon completion, participants were compensated $2.50 for their time (median completion time was 10 min).

A week later in Time 2, a random subsample of 200 participants from Time 1 completed additional measures to assess predictive validity. The Time 2 survey included measures of Psychological Safety, Work Engagement, Job Satisfaction, and LMX. Completion time averaged 8 min, and participants were compensated $2.00.

3.3.3 Measures

3.3.3.1 Conflict-intelligent leadership

The 36-item CIQ-L scale was tested in Time 1′s survey. Participants were asked to indicate the frequency with which their managers exhibited each behavior in the past 6 months on a 5-point Likert scale from 1 = Never to 5 = Always. Sample items were “My manager proactively addresses communication barriers that can cause team conflict” and “My manager stays composed and respectful when facing disagreements at work.”

The following measures were included in Time 2′s survey to establish predictive validity.

3.3.3.2 Psychological safety

The 60 item measure from (Andersson et al. 2020); Cronbach's alpha = 0.81 was used to measure psychological safety. Response options ranged from 1 = Strongly disagree to 5 = Strongly agree to statements such as “In my team, employees value and respect each other's contributions” and “In my team, if an employee makes a mistake, it is often held against them” (reverse coded).

3.3.3.3 Work engagement

The UWES-9 (α = 0.92; Seppälä et al., 2009) asked respondents to rate how often they felt each of the nine statements about their job (e.g., “I am enthusiastic about my job”) on a 6-point Likert scale from 1 = A few times a year or less to 6 = Every day.

3.3.3.4 Job satisfaction

Five items from the Job Satisfaction scale (e.g., “I find real enjoyment in my work;” α =0.85; Brayfield and Rothe, 1951; Judge et al., 1998) were included, with options ranging from 1 = Strongly disagree to 5 = Strongly agree.

3.3.3.5 Leader-member exchange

The LMX-7 (α = 0.89; Graen and Uhl-Bien, 1995; Liden et al., 2015) consisted of 7 items on a 5-point Likert scale from 1 = Strongly disagree to 5 = Strongly agree. Some of the items were “I can count on my manager to ‘bail me out' even at their own expense when I really need it,” “My manager recognizes my potential,” and “I usually know where I stand with my manager.”

3.3.4 Results

3.3.4.1 Preliminary analysis

After excluding the 5 reverse-coded items (one from each dimension) as fillers from our analyses (Dodeen, 2023), we started with initial evaluations of the data. Preliminary analyses involved examining the mean, standard deviation, variance, and normality for the remaining 31 CIQ-L items. The items' kurtosis ranged from −1.06 to 0.93, passing the normality check (George and Mallery, 2024). The inter-item correlations ranged from 0.19 to 0.69—items that were correlated at 0.80 and higher should be examined to avoid overlapping content. Thus, no item was removed to control for multicollinearity (Beavers et al., 2013; DeVellis and Thorpe, 2021). Finally, all item-total correlations exceeded 0.40 (DeVellis and Thorpe, 2021; Ladhari, 2010; Loiacono, 2002; Wolfinbarger and Gilly, 2003). All of the 31 positively worded CIQ-L items moved onto the next steps.

3.3.4.2 Data preparation for factor analysis

To investigate construct validity and avoid model overfitting (Fokkema and Greiff, 2017), the data from Time 1 (N = 496) were randomly divided into two subsamples (n₁ = 246 and n₂ = 250) for factor analyses. Tableone's chi-square statistics for categorical variables (Yoshida, 2022) indicated no significant difference between the two subsets across race, age, gender, and levels of education.

3.3.4.3 Exploratory factor analysis

Exploratory factor analysis (EFA) was conducted on the first sample (n₁ = 246) with principal axis factoring on Promax rotation, allowing for inter-factor correlations (Finch, 2020). Assumptions check confirmed that the n₁sample was suitable for factor analysis: Kaiser-Meyer-Olkin (KMO) = 0.97 and a significant Bartlett's test of sphericity, χ²(630) = 5,855.75, p < 0.001 (Bartlett, 1954; Kaiser, 1970). Examination of eigenvalues, scree plot (Figure 1), and parallel analysis suggested either a two- or three-factor solution, diverging from the theoretically proposed six-dimensional AESOPR model. The three-factor solution was prioritized to capture more variance and retain as many meaningful dimensions as possible (Fabrigar et al., 1999), explaining 59% of the variance.

The three resulting factors were: Stabilizing Presence, Context-Strategic Alignment, and Conflict-Positive Orientation. Each factor demonstrated good internal reliability, with Cronbach's alphas of 0.92, 0.90, and 0.78, and McDonald's omegas of 0.94, 0.92, and 0.79, respectively. Inter-factor correlations ranged from 0.73 to 0.81, suggesting that while the three factors were interrelated, they still measured distinct components of Conflict-Intelligent Leadership (Cheung et al., 2024).

Item retention adhered to the 0.40-0.30-0.20 rule (Howard, 2016), keeping only items with primary loadings >0.40, secondary factor loadings < 0.30, and the loading differences between the primary and secondary factors must be >0.20. One item, “Q7. My manager invites people to surface hard truths and opposing views,” loaded onto their main factors at.38 and removed due to being lower than the threshold. Three items (i.e., “Q13. My manager actively tests their assumptions in disagreements rather than just defends them”, “Q14. My manager welcomes feedback even when it challenges their preferred solution”, and “Q32. My manager anticipates how their messages and actions will land with employees and other important stakeholders”) did not pass the secondary factor loadings < 0.30 criterion. Finally, five items (i.e., “Q20. My manager assesses new or changing conflict situations effectively and adjusts their responses accordingly”, “Q24. My manager seeks optimal solutions to disputes that balance their own interests and those of others involved”, “Q25. My manager structures our team's work in ways that encourage cooperation between us rather than competition”, “Q28. My manager seeks input from people with multiple perspectives when making sense of complex conflicts”, and “Q34. My manager chooses the right channels or spokespeople to share information about important conflicts”) had loadings between their primary and secondary factors < 0.20. Another round of EFA was conducted on the remaining 22 items. The final CIQ-L scale consisting of 18 items, with results presented in Table 1. Out of those 18 items, 5 were generated by AI (i.e., Q3, Q6, Q10, Q33, and Q36), 6 were written by humans (i.e., Q5, Q16, Q19, Q22, Q23, and Q36), and the remaining 7 items (i.e., Q1, Q2, Q12, Q 21, Q29, Q30, and Q31) were first generated by machines and then revised by the research team.

Table 1. Exploratory factor analysis results for CIQ-L.

The bolded values indicate each item loading onto their main factor.

3.3.4.4 Confirmatory factor analysis

To further evaluate the factor structure, a series of five confirmatory factor analysis (CFA) models were conducted using the second subsample to identify the best fit (n₂ = 250; Credé and Harms, 2015). The first model was the correlated three-factor model suggested by the EFA results. This model yielded an acceptable fit to the data according to established cutoff criteria (Hu and Bentler, 1999; Marsh et al., 2004): CFI = 0.93, TLI = 0.92, RMSEA = 0.07, and SRMR = 0.05. Standardized factor loadings ranged from 0.62 to 0.88, with Average Variance Extracted (AVE) values for each factor meeting the recommended 0.50 threshold, ranging from 0.50 to 0.54 (Fornell and Larcker, 1981; Shiu et al., 2011). These results indicated strong relationships between observed items and their latent constructs, supporting the stability and replicability of the three-factor structure identified in the EFA.

To identify the most optimal model, four more CFA analyses were conducted. The second model assumed that the three factors for the CIQ-L scale were uncorrelated (i.e., orthogonal). The third was the unidimensional model, where all items loaded onto one single CIQ-L general factor with no distinct sub-dimensions. The fourth, the second-order model, assumed that the items first loaded onto their specific factors, which then loaded onto a higher-order CIQ-L construct. Finally, for the bifactor model, all of the items simultaneously loaded onto a general CIQ-L factor and their respective sub-dimensions.

3.3.4.5 Model selection

χ², CFI, TLI, RMSEA, SRMR, Akaike Information Criterion (AIC), and Bayesian Information Criterion (BIC) were utilized to compare fit across the different models (see Table 2 for all model fit indices). Given that lower AIC and BIC values were indicative of better fit (Preacher and Yaremych, 2023; West et al., 2023), the bifactor model was the recommended model for the CIQ-L scale.

Table 2. Confirmatory factor analysis model fit indices.

To determine whether the CIQ-L should be calculated using a single aggregated score or by individual subscales, model-based internal consistency statistics were evaluated Dueber, 2017). From the BifactorIndicesCalculator output in R (Dueber, xbib2020), the general CIQ-L factor explained 83.71% of the total variance (i.e., ECV). In comparison, Factors 1, 2, and 3 explained only 6.10, 6.73, and 3.46% of the total variance, respectively. Supplementing ECV, Omega Hierarchical (OmegaH) was used to determine the percentage of systematic variance in the overall score attributed exclusively to the general factor (Rodriguez et al., 2016). OmegaH for the general factor was 0.92, exceeding the 0.75 recommended by (Reise et al. 2013). This means that even though there are multidimensionality across items for the CIQ-L, the total score of all the 18 items should be used to represent the leadership conflict intelligence construct. Finally, the OmegaHS for subscales ranged from 0.02 to 0.15 (below 0.50), further supporting the interpretation of the general factor for the CIQ-L rather than subscale scores.

3.3.4.6 Predictive validity

Predictive validity refers to a measure's ability to predict outcomes measured at a later point in time (Lin and Yao, 2024). The data from the 200 participants who completed both Time 1 and Time 2 surveys were analyzed. A series of simple linear regressions was conducted to determine whether the 18-item refined CIQ-L scale predicted theoretically relevant outcomes: Psychological Safety, Work Engagement, Job Satisfaction, and Leader–Member Exchange (LMX).

Supporting our propositions, the results indicated that higher levels of CIQ-L as assessed by the three-factor scale significantly predicted better work outcomes: Psychological Safety, b = 0.58, R² = 0.41, F(1, 199) = 137.3, p < 0.001, Work Engagement, b = 0.38, R² = 0.08, F(1, 199) = 17.89, p < 0.001, Job Satisfaction, b = 0.39, R² = 0.13, F(1, 199) = 30.78, p < 0.001, and Leader-Member Exchange, b = 0.64, R² = 0.50, F(1, 199) = 198.2, p < 0.001.

Stepwise regression was additionally conducted to control for the length of supervisor-supervisee relationship. In Step 1, Length of Relationship was regressed onto the DVs. In Step 2, CIQ-L scale was added to the base models. Model comparison results showed that higher levels of CIQ-L significantly predicted more positive job-related outcomes even when controlling for the length of the relationship: Psychological Safety, F(1, 196) = 138.50, p < 0.001, Work Engagement, 18.05, p < 0.001, Job Satisfaction, F(1, 196) = 30.01, p < 0.001, and Leader-Member Exchange, F(1, 196) = 202.95, p < 0.001.

3.3.5 Brief discussion

Study 1 aimed to test, develop, and empirically validate the Conflict Intelligent Leadership Scale (CIQ-L) grounded in the multilevel conceptual model and the subsequent mathematical model (Coleman, 2025a). Overall, findings provided support for the reliability, construct, and predictive validity of the CIQ-L as a multidimensional measure of leaders' conflict intelligence.

Diverging from the six-dimensional AESOPR dynamical-systems theoretical model, the exploratory factor analysis revealed a three-factor structure (see Figure 2). Stabilizing Presence includes the capacities to maintain steadiness under pressure, stay composed and respectful when facing disagreements, and signal calm (steady tone, pace) to others in heated moments. Contextual-Strategic Alignment entails the ability to explain how broader societal forces may be contributing to tensions at work, consider the effects of broader forces like policies, incentives, and culture when diagnosing more demanding conflicts, and align conflict decisions with longer-term values and purposes. Conflict-Positive Framing involves communicating the benefits of addressing conflict rather than postponing it and framing conflict for others as an opportunity for improvement (Coleman and Phan, 2024).

This outcome suggests that, while the full six-variable mathematical model likely captures nuanced aspects of conflict intelligence from a leader's perspective, these dimensions may differ between employees' perceptions of their leaders, which were captured in the current research, and their leaders' self-reports. Furthermore, from the employees' perspectives, some of the variables were seen as combined into one (such as supervisors' Emotional Regulation and Transparent Communication into a leader's “Stabilizing Presence”, Systemic Awareness and Adaptivity into their “Context-Strategic Alignment”, and Conflict-Positive Orientation and Intellectual Humility into their capacity for “Conflict-Positive Framing”). In fact, the three factors include items across all the “levels” of the model (Self, Social, Situational, Structural, and Systemic) and “variables” (AESOPR) from the CIQ-L theoretical model, suggesting that, although they are not seen as coherent entities, they are important aspects of CIQ Leadership. Given that the Dependent Variables in this study were also collected as seen from the employee's perspective, it is critical that we focus on what they (employees) view as central components of their supervisor's CIQ. The three retained factors—Stabilizing Presence, Context-Strategic Alignment, and Conflict-Positive Framing—demonstrated good internal reliability. Also, given that the bifactor model outperformed other models such as the unidimensional model, these findings suggest that Conflict-Intelligent Leadership is a construct that can be represented by a strong general factor while still maintaining meaningful underlying dimensions.

Regression analyses further showed the predictive validity of the CIQ-L scale on Psychological Safety, Work Engagement, Job Satisfaction, and Leader–Member Exchange (LMX). These results aligned with prior empirical findings that effective conflict management could lead to a trusting, collaborative work climate and better supervisor-supervisee relationships (Edmondson and Lei, 2014; Zhao et al., 2019). Furthermore, the positive link between CIQ-L and Job Satisfaction and Work Engagement indicated that conflict intelligence may facilitate a more positive workplace experience for employees. Thus, improving leaders' conflict intelligence can have crucial practical implications.

3.4 Study 2

Building upon the findings of Study 1, Study 2 aimed to further validate the CIQ-L scale by examining its convergent and discriminant validity (Grimm and Widaman, 2023), employing a cross-sectional design. Specifically, the study assessed how the CIQ-L related to theoretically similar (i.e., convergent) and dissimilar (i.e., discriminant) leadership constructs to determine its conceptual distinctiveness and theoretical positioning within the broader leadership literature.

We conceptualize CIQ-L as a meta-competency that integrates and extends leadership capacity to navigate conflict and transform tensions. CIQ-L builds and extends upon a few related but distinct constructs. First, cultural competency and intelligence (Gelfand et al., 2008; Lederach, 1996) emphasizes sensitivity to cultural context, whereas CIQ-L advances this by incorporating adaptability to conflict across cultural boundaries (Coleman, 2024; Phan and Coleman, 2025). Second, adaptive leadership (Heifetz et al., 2009), an area of significant overlap with CIQ-L, addresses management through times of uncertainty and organizational change (adaptive challenges). CIQ-L complements this by explicitly naming specific conflict-engagement skills, such as self- social, situational, structural, and systemic competencies. Finally, transformational leadership (Bass and Riggio, 2006) emphasizes profound, ethical, and inspiring leadership behaviors. CIQ-L distinguishes itself by focusing on the mechanisms through which leaders navigate conflict. On the other hand, abusive leadership (Tepper, 2000), which describes toxic behaviors, was included for discriminant validity. Thus, demonstrating positive correlations with cultural intelligence, adaptive leadership, and transformational leadership, and a negative relationship with abusive leadership would suggest the convergent and discriminant validity for the CIQ-L scale.

3.4.1 Participants

Study 2 recruited completely new participants via Prolific to avoid repetition bias. After excluding 11 incomplete responses, 5 failed attention checks, and 7 participants without a direct manager, the final sample consisted of N = 314 respondents. In terms of gender, 47.3% identified as women, 51.6% as men, and 1.1% as non-binary or third gender. Many participants identified as White/ Caucasian (36.7%), followed by East Asian/ South Asian/ Southeast Asian at 20.7%, Black/ African American at 20.6%, Mixed/ Multiracial and non-White Hispanic/ Latino each took up 9.4%, and 13.2% identified with other racial groups. For the length of time worked with their managers, 17.8% of participants worked less than a year, 32.8% for 1–3 years, 24.2% for 3–5 years, and 31.21% had their managers for over 5 years.

3.4.2 Procedures

In the first section of the survey, participants were first asked to rate the frequency with which they experienced their managers exhibiting each behavior in conflict. The following scales were presented in randomized order: the Conflict-Intelligent Leadership scale (i.e., CIQ-L), adapted 5-item Cultural Intelligence scale, 15-item Adaptive Leadership scale, 7-item Transformational Leadership scale, and 15-item Abusive Leadership scale. The survey then asked participants to complete the Positive Affect subscale to control for common method bias in a cross-sectional design. Finally, it ended with a demographic questionnaire. The survey was estimated to take participants around 15 min, and they were then directed to receive their payment of $2.75 upon completion.

3.4.3 Measures

3.4.3.1 Conflict-intelligent leadership

The 18-item CIQ-L from Study 1 was included in this study as a measure of managers' conflict intelligence. Participants were asked to rate their managers from 1 = Never to 5 = Always.

The following measures were included for convergent, discriminant validity, and as a control variable for common method bias. Unless specified, they are all rated on a 5-point Likert scale from 1 = Strongly disagree to 5 = Strongly agree.

3.4.3.2 Cultural intelligence

Manager's cultural intelligence was measured by adapting the scale from Lima and colleagues (Lima et al., 2016). Participants were asked to rate five statements, such as “My manager checks the accuracy of cultural knowledge when interacting with those from different backgrounds.”

3.4.3.3 Adaptive leadership scale

The Adaptive Leadership measure (Nöthel et al., 2023) consisted of 15 items capturing leaders' ability to adapt to different situations. Sample items were “My manager quickly grasps what kind of leadership behavior is optimal for a specific situation” and “My manager is capable of adjusting their leadership style based on the needs of their subordinates.”

3.4.3.4 Global transformational leadership

To measure transformational leadership, all 7 items from the GTL scale (e.g., “My manager communicates a clear and positive vision of the future;” (Carless et al., 2000) were included.

3.4.3.5 Abusive leadership

As discussed, we expected leaders with CIQ-L not to possess abusive leadership behaviors. Tepper's Abusive Leadership scale (2000) utilized 15 items to portray toxic management behaviors, including but not limited to “My manager ridicules me,” “My manager tells me my thoughts or feelings are stupid,” and “My manager blames me to save themselves embarrassment.” All items were rated from 1 = Never to 5 = Very often.

3.4.3.6 Positive affect scale

Common method bias can take place when data is collected using the same method at the same time. In this case, as the CIQ-L, convergent, and discriminant measures were all collected in one cross-sectional survey, the Positive Affect subscale was included to control for common method bias (see Podsakoff et al., 2003). The 5-item Positive Affect subscale (Thompson, 2007) asked participants to think about themselves and rate the extent to which they generally felt alert, inspired, determined, attentive, and active.

3.4.4 Results

3.4.4.1 Confirmatory factor analysis

CFA was conducted to confirm the internal structure of the CIQ-L scale. Adhering to cutoffs of 0.90 for CFI and TLI, 0.06 for RMSEA, and 0.08 for SRMR (Hu and Bentler, 1999; Marsh et al., 2004), the three-factor model was an excellent fit to the data, with CFI = 0.97, TLI = 0.96, RMSEA = 0.05, and SRMR = 0.04. These results strongly supported the internal structure and reliability of the current CIQ-L scale.

3.4.4.2 Convergent validity

Convergent validity investigated whether the CIQ-L scale was correlated with other existing leadership measures (Carlson and Herdman, 2012). Aligning with predictions, the CIQ-L was strongly correlated with Cultural Intelligence (r = 0.65; p < 0.001), Global Transformational Leadership (r = 0.82; p < 0.001), and Adaptive Leadership (r = 0.82; p < 0.001) when controlling for Positive Affect.

3.4.4.3 Discriminant validity

Similar to convergent validity, correlation controlling for common method bias was applied to establish discriminant validity. CIQ-L was negatively correlated with Abusive Leadership at r = −0.65, p < 0.001 when controlling for Positive Affect.

3.4.5 Brief discussion

The findings from Study 2 reinforced the construct validity of the CIQ-L by demonstrating both convergent and discriminant validity with related leadership measures. The strong positive correlations with Cultural Intelligence, Global Transformational, Leadership and Adaptive Leadership suggested that CIQ-L reflected a strong sense of self-awareness, adaptivity, and sensitivity for leaders. Additionally, although significant, these correlations suggested that CIQ-L added to these frameworks by focusing specifically on the capacity to engage in conflict transparently and constructively and to ultimately transform conflict contexts.

For instance, Adaptive Leadership focuses on mainly on mobilizing people to confront and adapt to difficult realities by regulating productive disequilibrium and shifting values, roles, and behaviors. In contrast, Conflict-Intelligent Leadership centers both on managing oneself in conflict, and on diagnosing, redesigning, and managing the underlying conflict system, optimizing multi-level dynamics (emotional, relational, structural, and narrative) to prevent destructive escalation and cultivate durable, constructive tension. Similarly, the central behaviors relevant to Global Transformational Leadership (communicates a vision, develops staff, provides support, empowers staff, is innovative, leads by example, and is charismatic) all capture aspects of supportive and effective leaders, but do not focus on specific conditions where leaders are facing pressure or tension from conflict, which is core to CIQ Leadership.

Finally, the negative relationship with Abusive Leadership further reflected that CIQ-L was conceptually opposed to more toxic, negative management behaviors.

4 General discussion

Informed by the three pilots and two studies, we developed and validated the Conflict-Intelligent Leadership Scale (CIQ-L), an empirically grounded measure of leaders' ability to navigate conflict constructively in organizational settings. The development of valid and reliable metrics for CIQ and CIQL involve multi-phase processes, which begin with the testing and development of measures of the core variables that constitute each of the main parameters in the mathematical models, and their effects on relevant dependent variables. This was the focus of the current study. Once these variable measures are shown to be valid and reliable, they can be incorporated into dynamical measures to track incremental growth or decay patterns of individual and group changes in both CIQ and CIQ-L over time.

The findings from the current employee perception surveys of their supervisor's CIQ support CIQ-L as a multidimensional construct capturing three primary dimensions: Stabilizing Presence (affective regulation), Context-Strategic Alignment (adaptive systems cognition), and Conflict-Positive Framing (normative meaning construction). This three-factor model diverges from the initial proposition of six factors (higher levels of Adaptive leadership, Emotional regulation, Systemic awareness, Optimal tension management, Public perception management, and lower Resistance to change), suggesting that, while the full six-variable model likely captures nuanced aspects of conflict intelligence from a leader's perspective, fewer dimensions resonate with employees. However, given that employees viewed some of the full set of variables as combined (such as supervisors' Emotional Regulation and Transparent Communication into a leader's “Stabilizing Presence”, Systemic Awareness and Adaptivity into their “Context-Strategic Alignment”, and Conflict-Positive Orientation and Intellectual Humility into their capacity for “Conflict-Positive Framing”), subsequent research with leaders should continue to test the full model in order to capture a comprehensive understanding of CIQ-L from both perspectives. The three-factor CIQ-L model was found to be predictive of outcomes central to organizational health, including Psychological Safety, Work Engagement, Job Satisfaction, and the quality of Leader–Member Exchange. It was also found to be strongly related to—but not redundant with—other constructive leadership constructs (i.e., Cultural Intelligence, Adaptive, and Transformational Leadership), and negatively related to Abusive Leadership. Specifically, we found:

• CIQ-L positively predicted employees' psychological safety. This follows research indicating that emotional steadiness and transparent communication, which is a dimension of CIQ-L, are foundational for psychologically safe climates in teams (Edmondson, 1999; Study 1 results).
• CIQ-L positively predicted employees' work engagement. Adaptive leaders who diagnose conflict systemically and respond flexibly enable meaning-making, direction, and agency, thus enhancing engagement (Coleman, 2024).
• CIQ-L positively predicted employees' job satisfaction. Framing conflict as a constructive opportunity fosters empowerment, learning, and relational trust, which contribute to satisfaction (Coleman and Lim, 2001).
• CIQ-L positively predicted leader–member exchange (LMX). CIQ-L behaviors—including transparency, fairness, responsiveness, and openness—are core antecedents of high-quality leader–follower relationships (Zhao et al., 2019; Study 1).
• CIQ-L exhibited convergent validity with cultural intelligence, adaptive, and transformational leadership, and negative associations with abusive leadership. Study 2 provided initial support for these relationships; future studies should extend these findings by exploring CIQ-L's relationships with other constructs.

4.1 Limitations and future research

This series of research is not without limitations. First, although our studies have been able to show initial evidence of convergent, discriminant, and predictive validity, further research should examine other types of validity (e.g., incremental to test whether CIQ-L predicts beyond other constructs). Second, although we utilized longitudinal design and included a control variable for response bias, we could not eliminate all of common method variance as our data comes solely from employees' self-reported perception of their leaders' conflict intelligence. Although Prolific provides better quality data than other platforms (Douglas et al., 2023), it cannot fully capture an organizational context where employee-leader interactions take place. The next set of studies should be conducted in a field setting where both employees' and leaders' self-ratings are measured and tested for discrepancies, allowing for a more accurate and holistic assessment of managers' conflict intelligence level to inform development trainings. In addition, although there was a time lag in between Times 1 and 2, Study 1 can benefit from a longer interval more than 1 week. Despite analysis showing no significant differences across the groups who either participated or not in the Time 2 survey, respondents' self-selection process might introduce some bias to the results. Finally, although CIQ-L's relationship to other constructs (i.e., Adaptive Leadership, Cultural Intelligence) provided support for its convergent validity, the correlations were relatively significant. Evaluations of when and how CIQ-L might be more suitable can be an avenue for future studies. Theoretically, the present research extends multilevel models of conflict and leadership by demonstrating how conflict-related competencies cohere into a dynamic pattern of behaviors that shape the social climate of teams and organizations.

4.2 Practical implications

Beyond its value as a research instrument, the Conflict-Intelligent Leadership Scale (CIQ-L) provides organizations with a practical diagnostic tool for identifying and developing leaders' capacity to engage conflict constructively. The scale can be incorporated into leadership assessment systems, including 360-degree feedback processes, executive coaching engagements, and leadership development programs. Although the CIQ-L scale should be interpreted with just the general dimension, the CIQ-L items capture observable behavioral patterns across the dimensions of Stabilizing Presence, Contextual–Strategic Alignment, and Conflict-Positive Framing, it enables organizations to identify specific areas in which leaders may require support. For example, leaders scoring lower on Stabilizing Presence may benefit from training focused on emotional regulation, conflict communication, and managing reactivity under pressure, while leaders scoring lower on Contextual–Strategic Alignment may require development in systems thinking, stakeholder analysis, and adaptive conflict diagnosis. In this way, the CIQ-L can function not merely as an evaluative measure but as a developmental roadmap for cultivating more conflict-intelligent leadership.

Organizations can also use the CIQ-L to guide the design of leadership development curricula and conflict management interventions. Training programs may be structured around the empirically identified items of CIQ-L, helping leaders strengthen their capacity to remain composed and transparent during tense interactions, diagnose conflict within broader organizational and societal systems, and frame disagreement as a source of learning and improvement rather than a threat to encourage dialogue (e.g., O'Connell, 2022; O'Connell and Adams, 2024; Phan, 2025). Integrating CIQ-L assessment results into leadership development programs allows organizations to tailor coaching and training to leaders' specific developmental profiles. For instance, leaders who struggle with conflict-positive framing may be coached in practices that normalize disagreement, surface diverse perspectives, and encourage constructive debate within teams. Over time, repeated CIQ-L assessments can allow organizations to track growth in leaders' conflict capabilities and evaluate the effectiveness of training initiatives.

At the organizational level, the CIQ-L can also support efforts to cultivate conflict-positive cultures. Research suggests that leaders play a central role in shaping norms for how conflict is expressed and managed within teams and institutions (Gelfand et al., 2012). By systematically assessing CIQ-L across managerial ranks, organizations can identify patterns in how conflict is currently experienced and addressed within their systems. These insights can inform broader organizational interventions, such as revising conflict management policies, redesigning communication norms, or embedding constructive conflict practices into leadership expectations. In this sense, the CIQ-L offers organizations not only a means of evaluating individual leaders but also a lens for understanding and improving the broader conflict culture of the organization, thereby helping transform conflict from a source of dysfunction into a driver of innovation, resilience, and collective learning.

Training and coaching programs can leverage the three CIQ-L dimensions as focal areas, helping leaders remain calm under pressure, communicate with clarity and candor, read the systemic context of conflict, adapt their strategies, and frame conflict as an opportunity for shared learning rather than a threat.

5 Conclusion

Conflict-Intelligent Leadership represents a critical meta-competency for a world in which conflict is both ubiquitous and consequential. By providing a validated measure of CIQ-L, this research enables scholars and practitioners to better understand, assess, and cultivate leaders who can transform conflict from a liability into a resource for resilience, innovation, and collective well being. Future research should examine the development of CIQ-L over time, its role in diverse cultural and sectoral contexts, and interventions that can strengthen Conflict-Intelligent Leadership at scale.

Generative AI statement

The author(s) declared that generative AI was used in the creation of this manuscript. Generative AI was used to generate some items for the study, as discussed in detail in the manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: WDNSM Tennakoon, Wayamba University of Sri Lanka, Sri Lanka

Reviewed by: Ashenafi Woldemichael Woime, Hossana College of Health Sciences, Ethiopia

Shajara Ul-Durar, University of Sunderland, United Kingdom

Footnotes

The de-identified data can be accessed in the following link: https://osf.io/ft3q4?view_only=574d961fd08b44949dec02167f3d9af3.