3184 Development of a Leadership Assessment Scale in Translational Science

Abstract

OBJECTIVES/SPECIFIC AIMS: To create the instrument, we employed a modified Delphi approach by conducting a thorough literature review on Leadership to help concretize the relevant constructs, and then usied these extracted constructs as a springboard for the Rockefeller Team Science Educators (TSE’s) to discuss and refine the leadership domain areas, collectively creating domain-specific survey items, and then further discussed and refining the number, grouping, and wording of the items. METHODS/STUDY POPULATION: We piloted the Leadership Survey by having all of the Rockefeller TSEs rate Clinical Scholars. Each item was answered using a six-point Likert scale where a low score indicated poor expression of the specific leadership attribute and a high score represented excellent expression of the specific leadership attribute. RESULTS/ANTICIPATED RESULTS: Means, medians, standard deviations, and ranges of each item were calculated and tabulated. A complete (Pearson) correlation matrix was computed so that the raw inter-item relationships can be observed. For each a priori Domain an equal weighted summary scale was created and tabulated for review. The internal consistency of each a priori scale was assessed by calculating Cronbach’s Alpha (α). Items with low Item to Construct coefficients were candidates for elimination or modification, and overall scales with low’s will undergo further discussion. To challenge our assumptions of the construction and integrity of each domain, we employed exploratory Principal Components Analysis (PCA), followed by orthogonally rotated Factor Analysis (FA). We also forced the PCA / FA analysis to extract the a priori dimensions that allowed us to compare if the empirical and a priori structures match. DISCUSSION/SIGNIFICANCE OF IMPACT: We are partnering with the CTSA programs at Penn and Yale to assess issues of generalizability and scalability. We are working with Vanderbilt to install survey onto REDCap for ease of dissemination. Will continue to assess psychometric properties and refine as we receive more input.