Rating Scale Analysis of the Leadership Practices Inventory

Purpose/Hypothesis : Development of leadership skills is an increasing focus in physical therapy education and practice. The Leadership Practices Inventory-Self (LPI-Self) was developed to assess five dimensions of leadership characteristics. No previous studies have assessed the psychometric properties of the LPI-Self in a sample of physical therapist students or clinicians. The purpose of this study was to investigate properties of two published versions of the LPI-Self. Specifically, the original version of the LPI-Self (5 point Likert scale, or 5-LPI-Self) and current version (10 point Likert scale, or 10-LPI-Self) were compared to assess the respective category structure (category fit, step calibration and average measure values), person and item reliability values, floor and ceiling effects, and contribution of items to the unidimensionality of the scale construct.Number of Subjects : A total of 220 LPI-Self questionnaires were included for analysis from 194 physical therapist students and 26 physical therapists.Materials/Methods : The partial credit Rasch measurement model was used to analyze raw scores and examine the psychometric properties of the LPI-Self.Results : The findings demonstrated disordered step calibrations, average measures and/or response category misfit for 100% of the items of the 10-LPI-Self compared to 6 items (20%) for the 5-LPI-Self. Both versions of the LPI-Self demonstrated high person reliability (.92 and .93 respectively) and item separation reliability (.98 and .98 respectively). The 5-LPI-Self demonstrated a slightly greater logit range (-1.57 to 1.13) compared to the 10-LPI-Self (-1.14 to .56). Both versions demonstrated a large ceiling effect (person range beyond range of item or scale). Finally, both versions of the LPI-Self demonstrated acceptable item fit, suggesting items contribute to assessment of the unidimensional scale construct regardless of the scoring structure.Conclusions : The results demonstrate more robust psychometric properties of the 5-LPI-Self compared to the 10-LPI-Self for assessment of leadership characteristics in physical therapist students and clinicians. The higher frequency of disordered step calibrations, average measures and item fit in the 10 versus 5 point scale suggests raters may have difficulty discriminating leadership characteristics across 10 versus 5 category options. The ceiling effect that was observed suggests the scale may be more appropriate for assessment of leadership early in the careers of students and therapists, and should be complimented with other assessments of leadership that target higher leadership skill level.Clinical Relevance : This study has demonstrated that giving a subject fewer choices when using a Likert scale on a survey allows for better discrimination of results of the survey tool. This will assist in the development of new Likert scale survey tools, as well as a better understanding of the results of survey tools that use higher than 5 Likert scale choices.

BACK to Abstract Results

  • Control #: 2012714
  • Type: Poster
  • Event/Year: CSM2015
  • Authors: Laura LoVasco, Don D. Straube, Kristine Thompson
  • Keywords: Leadership|Rasch Method|Leadership Practices Inventory

BACK to Abstract Results