Reliability reconsidered: Cronbach's alpha and paediatric assessment in occupational therapy

• Published in: Australian occupational therapy journal Vol. 56; no. 3; pp. 150 - 155
• Main Author: Spiliotopoulou, Georgia
• Format: Journal Article
• Language: English
• Published: Melbourne, Australia Blackwell Publishing Asia 01.06.2009
• Subjects: Assessment; Child; Child, Preschool; coefficient alpha; Coefficients; Data Interpretation, Statistical; Disability Evaluation; Humans; internal consistency; Occupational Therapy; outcome measures; Paediatrics; Psychometrics; Reliability; Reproducibility of Results
• ISSN: 0045-0766; 1440-1630
• DOI: 10.1111/j.1440-1630.2009.00785.x

Background/aim:  Using reliable outcome measures is a necessity for the occupational therapy profession in enabling valid assessments of clients. Although Cronbach's alpha is the most widely applied index of internal consistency reliability, there are misconceptions about its use and interpretation. This paper aims to guide assessment developers in paediatric occupational therapy, as well as practitioners who are evaluating outcome measures in using and interpreting the Cronbach's alpha estimates appropriately. This will enable them to decide on the tools’ clinical value and incorporate them into their practice with children. Method:  Previously published papers reporting on internal consistency issues of outcome measures in paediatric occupational therapy were searched through the Allied and Complementary Medicine database. These papers were used as a basis to discuss possible reasons for reporting of low internal consistency. Results:  The analysis demonstrates that Cronbach's alpha reports are not always interpreted in a sound way. The paper emphasises that one should be cautious about judging estimates of internal consistency. Low size of the coefficient alpha might not always indicate problems with the construction of the tool; whereas large sizes do not always suggest adequate reliability. Instead, these reports might be related to the data characteristics of the construct. Conclusion:  In judging an outcome measure's internal consistency, researchers and practitioners in occupational therapy should report and consider the nature of data, the scale's length and width, the linearity and the normality of response distribution, the central response tendency, the sample response variability and the sample size.