Kean University

PurposeThere has been a growing emphasis in both academic and industry spheres on the importance of efficient supply chain management. However, supply chains are facing unprecedented levels of upheaval and unforeseen events. From natural disasters to artificial crises, political and economic turmoil and the ever-changing landscape, uncertainties abound. Therefore, this study investigates the effects of structural risk sharing, structural collaboration, supply chain ambidexterity, dynamic reconfiguration, dynamic sensing and supply chain flexibility on supply chain resilience.Design/methodology/approachData were collected from 2,301 manufacturers across various industries in the USA. The hypotheses were evaluated using Structural Equation Modeling through Smart PLS version 4.0.FindingsAll direct and sequential mediation hypotheses were supported. Structural collaboration exerted a stronger effect on supply chain ambidexterity than structural risk sharing. Supply chain ambidexterity strongly enhanced dynamic sensing and reconfiguration, thereby improving supply chain flexibility. Notably, supply chain flexibility emerged as the strongest direct predictor of resilience. The most influential indirect pathway was structural collaboration -> ambidexterity -> dynamic sensing -> flexibility -> resilience, underscoring flexibility as the primary mechanism through which structural practices translate into resilience outcomes.Originality/valueSequential mediation introduces innovation by highlighting the impact of supply chain ambidexterity, which benefits both the firm and the broader supply chain network. Prior investigations concentrated on the internal implications of supply chain ambidexterity within a firm. Nonetheless, informed by the dynamic capability view, the variables studied indicate that supply chain ambidexterity empowers firms to navigate intricate supply chain challenges and capitalize on opportunities. This implies that firms can adapt to evolving market conditions, harmonize their operations with their objectives and enhance the resilience of their supply chains against disruptions.

Purpose This study examines how AI literacy shapes workplace equality and competitive advantage, with particular attention to the mediating role of psychological capital. Design/methodology/approach Using survey data from 467 owner-managers, managers and senior staff in small- and medium-sized enterprises in Surabaya, Indonesia, the study applies partial least squares structural equation modeling (PLS-SEM) to test the proposed relationships. Findings AI literacy exhibits a statistically significant yet modest direct effect (ss = 0.133, p < 0.05) on workplace equality and competitive advantage. However, its primary influence operates indirectly through psychological capital (strong path: ss = 0.600, p < 0.001), which fosters self-efficacy, resilience, optimism and hope, thereby driving equality (ss = 0.531, p < 0.001) and firm performance. Psychological capital, in turn, significantly promotes workplace equality, emerging as a pivotal antecedent to competitive advantage (ss = 0.494, p < 0.001). These patterns underscore AI literacy's role as a human-centered amplifier in resource-constrained SMEs. Practical implications Beyond technical training, integrate PsyCap-building interventions (e.g. resilience workshops paired with AI ethics modules) into SME support programs, drawing on evidence from Indonesian contexts to foster inclusive AI adoption and align with UNESCO equity goals. Originality/value By conceptualizing AI literacy as a socially embedded capability, this study demonstrates that AI functions as a social amplifier: its performance and equality outcomes depend less on technology itself than on how it reshapes human agency and psychological capacity in organizational contexts.

Exclusionary discipline has been a well-documented and growing problem in early childhood programs in the United States, with potentially long-term negative consequences for young children. Although expulsion from early learning settings is widely examined in the literature, far less attention has been given to pre-enrollment exclusion, or the practice of denying admission to a child based on their actual or perceived behaviors. This study reports findings from an online, anonymous survey of 112 community childcare program administrators from a single state located in the northeastern region of the United States regarding their use of pre-enrollment exclusion practices. Most administrators (65.2

Over the past fifty years, scholars have developed a number of sexual identity development models in an effort to improve clinical knowledge regarding these processes among lesbian, gay, bisexual, trans*, and queer (LGBTQ+) people. These models, while important in helping bring attention to the diverse experiences of LGBTQ+ people, have largely been developed through deductive approaches that rely on contested assumptions about the nature of sex, gender, and sexual orientation. In doing so, they both replicate several problematic assumptions regarding the underlying nature of sex, gender, and sexual orientation and fail to account for the identities and experiences of many LGBTQ+ people. This pilot constructivist grounded theoretical study adds to a growing body of scholarship that is grounded in the language and experiences of the research participants and seeks to understand how sexual identities are developed and expressed in relation to social context in order to address the failure of popular sexual identity models to capture the lived experiences of diversely identifying LGBTQ+ people. Borrowing from Deleuze and Guattari's rhizomatic (1980/1987), I propose that an alternative framing of gendered rhizomatic assemblages might better convey the identities of LGBTQ+ people who use multiple and changing identity labels across time and context.

Reinforcement Learning with Verifiable Rewards (RLVR) has become an effective post-training paradigm for improving the reasoning abilities of large language models. However, existing RLVR methods typically rely on final-answer correctness to assign trajectory-level rewards, providing sparse supervision and treating all tokens uniformly regardless of their actual contribution to reasoning. Although recent studies introduce intermediate signals such as process rewards, high-entropy tokens, and semantic uncertainty, these signals are often not inherently verifiable and may fail to distinguish beneficial strategic patterns from harmful ones. To address this limitation, we propose STRIDE (Strategic Trajectory Reasoning with Discriminative Estimation), a fine-grained RLVR framework that derives strategic reasoning supervision from verifiable outcomes. STRIDE contrasts successful and failed trajectories within each response group to estimate the outcome-discriminative preference of each $n$-gram strategic pattern, and further combines this signal with reasoning saliency entropy to identify decision-relevant strategic patterns. These patterns are assigned differentiated advantage values during RL optimization, enabling more precise credit assignment while preserving the verifiability of RLVR. Extensive experiments demonstrate that STRIDE consistently improves reasoning performance across diverse models, tasks, and extended settings, including VLMs and agent-based systems.