Revisiting Leader-Member Exchange: A Review Through the Lens of the "Too Much of a Good Thing" Effect

A groundbreaking review of existing research is challenging a long-held assumption in organizational psychology: that high-quality leader-member exchange (LMX) relationships are universally beneficial. The study, published in Frontiers in Psychology, systematically re-examines empirical evidence suggesting that an excessive level of leader-member closeness can lead to diminishing returns and even negative outcomes, a phenomenon termed the "too much of a good thing" (TMGT) effect. This research moves beyond the traditional "more is better" paradigm, proposing a new theoretical framework that integrates resource management, emotional costs, and motivational drivers to explain these complex dynamics.

For decades, Leader-Member Exchange (LMX) theory has been a cornerstone of understanding workplace relationships. It posits that leaders develop unique, qualitatively different relationships with each of their direct reports. High-quality LMX relationships, characterized by trust, respect, and mutual obligation, have consistently been linked to a wide array of positive outcomes. These include enhanced job performance, increased job satisfaction, stronger organizational commitment, and reduced intentions to leave an organization. Consequently, both academic research and management practices have largely operated under the assumption that cultivating the highest possible LMX quality is the optimal strategy for both individual employees and the organization as a whole.

However, a growing body of recent empirical work is casting doubt on this linear progression. The relationship between LMX quality and employee outcomes appears to be more nuanced, potentially exhibiting a curvilinear or inverted U-shaped pattern. This suggests that while moderate to high levels of LMX are beneficial, excessively strong relationships may trigger unintended negative consequences, such as increased psychological strain and emotional exhaustion. This observation aligns with the "too much of a good thing" (TMGT) framework, which proposes that even beneficial concepts can become detrimental beyond an optimal threshold.

The review highlights that while LMX is undeniably crucial for individual and workplace experiences, the subtle psychological trade-offs associated with extremely high LMX have often been overlooked. The study’s authors argue that the traditional focus on the positive aspects of LMX has obscured potential downsides. To address this, the researchers conducted a systematic scoping review of existing literature that empirically tested the TMGT effect within the LMX framework. As a formal meta-analysis was not yet warranted due to insufficient studies, this review aims to map the key concepts, evidence, and research gaps.

The core research question driving this review is whether LMX can indeed produce non-linear effects. Theoretically, the authors propose an integrated framework, dubbed the Resource-Emotion-Motivation (REM) model, to explain this phenomenon. Their argument is that when the relational benefits of LMX become exceptionally high, the associated emotional and social costs can exceed an individual’s available resources, disrupting the optimal relational equilibrium and leading to the TMGT effect.

Methodological Rigor in Exploring LMX Paradoxes

To ensure a comprehensive and reproducible analysis, the researchers adopted a structured literature search methodology. In November 2025, they scoured academic databases including PsycINFO, PsycARTICLES, ProQuest, and Web of Science, supplemented by Google Scholar for broader coverage. The search terms focused on "leader-member exchange" (or LMX) in conjunction with concepts like "curvilinear," "non-linear," and "too much of a good thing" (TMGT). The review was limited to peer-reviewed academic journal articles published in English up to 2024. Notably, the concept of "guanxi," a culturally specific form of social exchange prevalent in China, was intentionally excluded to maintain focus on the general LMX construct, as it has been shown to possess distinct patterns beyond standard LMX.

The screening process involved two independent researchers who blinded themselves to each other’s judgments, ensuring objectivity. After initial screening of titles and abstracts, 31 articles were deemed relevant for full-text review. The final selection criteria were stringent: studies had to examine LMX as an independent variable influencing psychological, behavioral, or performance outcomes; explicitly analyze non-linear or curvilinear effects (using methods like quadratic regression or polynomial terms); be available in full text and English; and offer theoretical justification or relevance to paradoxical LMX theories. Theoretical commentaries, editorials, meta-analyses without new empirical data, and studies focusing solely on linear relationships were excluded. Furthermore, studies on LMX differentiation, which examine group-level phenomena, were also excluded to maintain the focus on dyadic LMX effects. After a thorough review by the lead author, 16 studies were included, collectively representing 19 distinct datasets.

Key Findings: When More Becomes Less

The review unearthed several critical insights into the TMGT effect of LMX. Of the 16 studies analyzed, the earliest to demonstrate this effect was Kramer’s (1995) work, which qualitatively categorized leader-member relationships. Subsequent quantitative studies explored outcomes such as member stress and turnover intentions. The publication timeline revealed a growing interest in this phenomenon, with a notable increase in research from 2020 onwards. Geographically, the United States dominated the research landscape (56.3%), with a smaller but present contribution from Europe (18.8%) and Asia (25.0%). This suggests a potential bias towards Western cultural contexts in current research.

The studies covered a diverse range of organizational settings, from high-technology firms and public organizations to police departments, hotels, and military units. Research specifically targeting stress or turnover intentions often drew data from industries with historically high turnover rates or where workload and complexity were significant concerns, such as the hotel and healthcare sectors.

A significant methodological observation was the predominance of cross-sectional survey designs (63.2% of the 19 datasets), relying on self-reported questionnaires. While seven studies (36.8%) employed longitudinal designs, none utilized psychophysiological measures, indicating a potential gap in capturing the dynamic and physiological aspects of LMX. The LMX scale most frequently used was the LMX-7 scale, a common practice in previous LMX research.

Unpacking the Paradoxical Outcomes

The review categorized the outcome variables studied into cognitive/perceptual, affective, and behavioral/attitudinal. Behavioral tendencies, particularly concerning turnover intention and its composite outcomes (like stress), emerged as the primary areas where the TMGT effect of LMX was observed.

Notably, only one study, examining turnover intentions among nurses, failed to fully support the TMGT effect, showing only a significant negative association with LMX. Several other studies indicated that the TMGT effect was contingent or limited. For instance, one study found an inverted U-shaped effect of LMX on task performance, but only when considering the affective dimension of LMX and under conditions of high supervisory interpersonal justice. Another study revealed that while role ambiguity showed a negative linear relationship with LMX, role conflict and overload exhibited inverted U-shaped effects.

Furthermore, eight studies explored cubic curves, with only one finding a significant increase in variance for the cubed LMX term, suggesting a complex, S-shaped relationship between LMX and ostracism. This study posited that both very low and excessively high LMX could lead to ostracism, albeit for different reasons, highlighting the intricate social dynamics at play.

Moderating Factors and Underlying Mechanisms

The review strongly suggests that the TMGT effect of LMX is not a universal constant but rather a phenomenon that appears under specific conditions, influenced by individual differences and contextual factors. For example, the relationship between LMX and task performance exhibited an inverted U-shape only when individuals displayed high negative affectivity. Similarly, the TMGT effect of LMX on task performance was more pronounced when perceptions of leader interpersonal justice were high.

However, findings regarding moderation by job level were inconsistent. One study found a U-shaped effect of LMX on job pressure only among senior nurses, while another found the TMGT effect on turnover intentions among non-managerial but not managerial employees. These discrepancies point to the complexity introduced by variations in job type, autonomy, role ambiguity, and potentially cultural context.

Regarding the underlying mechanisms, the research offered a nuanced perspective. One hypothesis that high LMX creates high demands leading to stress was not consistently supported. Instead, an interpretation from a socioemotional exchange perspective suggested that as LMX quality increases, unclear obligations and role expectations might initially increase stress. Beyond a certain point, however, the symbolic meaning of LMX might become clearer, potentially reducing role stress.

Another significant finding highlighted the role of emotional events. Leaders’ behaviors, such as safety enforcement, might be mediated by the emotional favorability component of LMX. This suggests that socio-emotional factors play a crucial role in the psychological mechanisms driving the TMGT effect, moving beyond purely transactional explanations.

An Integrated Model: The REM Framework

Building on the existing evidence, the review proposes an integrated theoretical framework, the Resource-Emotion-Motivation (REM) model, to explain the TMGT effect in LMX. This model connects LMX theory with prospect theory, organizational justice theory, and self-determination theory.

From the perspective of prospect theory, individuals with high LMX may perceive their abundant resources as a baseline. Any perceived threat to this privileged relationship can diminish satisfaction from further gains and foster vigilance, anxiety, and negative emotions like fear of losing face or status. This can lead to stress and a suppression of dissent or proactive behaviors like constructive voice.

Organizational justice theory offers insights into why high LMX individuals might face exclusion from peers. Perceived workplace unfairness due to preferential treatment can breed envy and anger among lower-LMX colleagues, leading to negative interactions for high-LMX individuals.

Self-determination theory emphasizes the importance of autonomous motivation. While high LMX should theoretically foster autonomy, an over-reliance on socio-emotional resources from a leader can lead to dependence, hindering competence fulfillment and creating negative interpersonal spirals. Hostility and envy from colleagues further undermine autonomous motivation, leading individuals to refrain from proactive behaviors.

In essence, the REM model posits that when LMX quality exceeds a threshold, resource scarcity and associated psychological strain, particularly socio-emotional costs, can outweigh the benefits. This strain, driven by resource depletion fears, negative emotions, and inhibited autonomous motivation, can lead to stress induction and a suppression of proactive behaviors.

Future Directions and Practical Implications

While this review offers valuable new perspectives, the authors acknowledge limitations. The mechanisms underlying the TMGT effect are still underexplored, with few studies directly demonstrating non-linear effects and mediating or moderating variables. Future research should focus on behavioral outcomes and the role of emotional exhaustion and social comparison. Moderators such as industry type, competitive environments, individual attachment styles, and cultural factors are also crucial areas for investigation. The evolving nature of work, particularly the rise of virtual environments, raises questions about whether the TMGT effect of LMX can manifest in digital interactions.

Methodologically, the reliance on cross-sectional surveys needs to be addressed. The review advocates for longitudinal designs, experiential sampling, psychophysiological monitoring, and behavioral trace data to capture the dynamic nature of emotions and causality. Multimodal measurements combining subjective reports with objective physiological responses can elucidate the socio-emotional costs and stress accumulation in leader-member interactions.

From a practical standpoint, the findings challenge the simplistic notion that "high-quality LMX always leads to favorable outcomes." Organizations must move beyond this assumption. Leadership development programs should equip leaders with the skills to regulate emotions, manage relational asymmetry, and recognize signs of over-involvement that can inadvertently increase psychological burdens. Human resource professionals and managers need to foster awareness of the potential risks associated with extreme LMX and implement boundary management strategies for job roles.

Emerging technologies, such as AI-powered communication analysis and digital coaching, offer potential avenues for monitoring LMX in remote settings, though ethical and privacy concerns must be carefully managed. The core message for practice is that sustained functionality in relationships requires regular recovery, flexibility, and moderation. Timely identification and response to imbalances between LMX strength and job roles or organizational goals are critical for optimizing team relationships and fostering healthier, more sustainable leader-member dynamics.

Conclusion: Towards a Balanced Understanding of LMX

The "too much of a good thing" effect provides a critical lens for understanding the paradoxes within leader-member exchange. This review underscores that while high-quality LMX relationships are generally beneficial, excessively strong bonds can incur unintended psychological costs. By highlighting methodological limitations and proposing new research directions, this study contributes to a more balanced and nuanced understanding of LMX dynamics. Future research that integrates longitudinal designs, psychophysiological assessments, and practical interventions, considering both member and leader perspectives, will be essential. Furthermore, interdisciplinary research spanning psychology, organizational behavior, information systems, and ethics is needed to foster healthier leader-member relationships and develop effective educational programs and intervention methods.