Development and psychometric validation of caring leadership scale in nursing

Background: Caring leadership plays a pivotal role in supporting nurses’ well-being, fostering patient recovery, and advancing organizational performance. Despite its recognized importance, a validated instrument for measuring caring leadership has been lacking in the literature. Objective: This study aimed to develop a Caring Leadership Scale and evaluate its validity, reliability, and practical applicability. Methods: Guided by a conceptual framework of caring leadership, scale items were initially generated through qualitative interviews, open-ended questionnaires, and a comprehensive literature review. The Delphi technique was used to refine these items. A pilot survey informed further modifications before a large-scale survey was conducted with 2,125 frontline nurses from six tertiary hospitals across China. Item screening was carried out using classical test theory, including assessments of item discrimination, item-total correlations, and internal consistency (Cronbach’s alpha). Content validity, factor structure, internal reliability, and acceptability were investigated using content validity indices, exploratory and confirmatory factor analyses, correlation coefficients, and descriptive statistics. Results: An initial pool of 40 items across five dimensions was developed. Following two Delphi rounds and preliminary testing, 39 items were retained. Item analysis led to the elimination of 12 items that did not meet predetermined criteria. The final scale consisted of 27 items across five dimensions, rated on a 5-point Likert scale. The content validity indices were strong, with item-level CVIs ranging from 0.90 to 1.00, a scale-level content validity index universal agreement (S-CVI/UA) of 0.89, and a scale-level content validity index average (S-CVI/Ave) of 0.99. Exploratory factor analysis (EFA) supported a well-defined five-factor structure, with common factor variances ranging from 0.849 to 0.949. Confirmatory factor analysis demonstrated good model fit (RMSEA = 0.078, CFI = 0.964, NFI = 0.958, TLI = 0.959, IFI = 0.964). The average variance extracted (AVE) ranged from 0.834 to 0.925, and composite reliability (CR) ranged from 0.965 to 0.989. The scale showed moderate correlations with job satisfaction (r = 0.463–0.484, p