Psychometric (Rasch) testing of a professional identity scale for pharmacy students

Title
Psychometric (Rasch) testing of a professional identity scale for pharmacy students
Publication Date
2023-07
Author(s)
Spark, Joy
( author )
OrcID: https://orcid.org/0000-0001-5240-8217
Email: jspark@une.edu.au
UNE Id une-id:jspark
Barwick, Anna
( author )
OrcID: https://orcid.org/0000-0003-4504-3453
Email: aunger2@une.edu.au
UNE Id une-id:aunger2
Type of document
Conference Publication
Language
en
Entity Type
Publication
Publisher
Elsevier BV
Place of publication
United States of America
Edition
21
DOI
10.1016/j.sapharm.2023.04.023
UNE publication id
une:1959.11/74929
Abstract

Introduction. The Macleod Clark Professional Identity Scale (MCPIS-9) is a tool to assess Professional identity (PI) in pharmacy students.1 Despite validation, a recent review found it had undergone limited psychometric testing.2

Aims. To examine the construct validity of the MCPIS-9 using Rasch analysis.

Methods. Undergraduate pharmacy students across all year levels completed an adjusted MCPIS-9 with a seven point strongly disagree/ strongly agree response scale and specific referral to the pharmacy profession. The fit of the MCPIS-9 to the polytomous Rasch model was evaluated using RUMM2030.

Results. The MCPIS-9 was completed by 109 pharmacy students. PI scores were positively skewed, ranging from 27 to the maximum of 54 (mean: 43±6.8). Initial assessment of the MCPIS-9 fit to the Rasch model identified severe miss-targeting and disorder thresholds; these were not improved by collapsing response categories. Consequently the mid-point was removed and response categories collapsed into 4. There was evidence of response dependence for the negatively worded questions. The two questions relating specifically to students were most redundant and removed. Unidimensionality of the 7-item scale was explored using principle components analysis of the residuals. Investigation of the 2 highly positive items and 3 highly negative items using a t-test showed the number of responses that differed by more than 5% was 6.5% (95CI 2.5% to 11.4%) of responses indicating unidimensionality. The resulting scale showed good fit to the Rasch model with a non-significant item-trait interaction total χ2=10.3 (df=14, p=0.74) and PSI=0.74. Targeting is reasonable except for one threshold on an item that had no disagree responses.

Discussion. The high Cronbach’s α reported for the MCPIS-92 is artificially inflated by the response dependence between 3 of the items. There is a large positive skew to responses that can be controlled by removing the mid-point and collapsing categories. In future, consideration should be given to using a 7-item tool with an even number of response categories that may also be tested in health professionals as well as students.

Link
Citation
Research in Social and Administrative Pharmacy, 19(7), p. e1080-e1080
Start page
e1080
End page
e1080

Files:

NameSizeformatDescriptionLink