Premium
Development and validation of a feline abdominal palpation model and scoring rubric
Author(s) -
Williamson J. A.,
Hecker K.,
Yvorchuk K.,
Artemiou E.,
French H.,
Fuentealba C.
Publication year - 2015
Publication title -
veterinary record
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.261
H-Index - 99
eISSN - 2042-7670
pISSN - 0042-4900
DOI - 10.1136/vr.103212
Subject(s) - rubric , palpation , likert scale , medicine , content validity , reliability (semiconductor) , internal consistency , variance (accounting) , medical education , veterinary medicine , medical physics , radiology , psychology , clinical psychology , statistics , psychometrics , mathematics education , mathematics , power (physics) , physics , quantum mechanics , accounting , business
Simulation in veterinary education enables clinical skills practice without animal use. A feline abdominal palpation model was created that allows practice in this fractious species. This study assessed the model and rubric using a validation framework of content evidence, internal structure and relationship with level of training. Content Evidence: Veterinarians accepted this model as a helpful training tool for students (median=4 on five‐point Likert scale). Internal Structure Evidence: G‐coefficients were low for first‐ and second‐year students (0.28 and 0.23), but were acceptable for veterinarians (0.61). Internal consistency values (0.24, 0.42 and 0.67) followed a similar pattern. Thus, scores were more reliable for veterinarians than for the students. Evidence of Relationship with Level of Training: Although level of training impacted reliability, its effect on performance scores was inconsistent. Analysis of variance (ANOVA) identified no differences among the groups of students and veterinarians. However, effect size between first‐ and third‐year students was medium to large (0.62). Effect sizes between the veterinarians and student groups were small. Although the model and rubric appeared valid for experts, modifications would be necessary to generate reliable scores for students. These results allow greater understanding of the needs of students utilising a low‐fidelity model.