Sample Size in Reliability Studies: A Practical Guide Based on Cronbach's Alpha


Karakaya S. P., ALPARSLAN Z. N.

PSYCHIATRY AND BEHAVIORAL SCIENCES, vol.12, no.3, pp.150-157, 2022 (ESCI) identifier

  • Publication Type: Article / Article
  • Volume: 12 Issue: 3
  • Publication Date: 2022
  • Doi Number: 10.5455/pbs.20220127074618
  • Journal Name: PSYCHIATRY AND BEHAVIORAL SCIENCES
  • Journal Indexes: Emerging Sources Citation Index (ESCI)
  • Page Numbers: pp.150-157
  • Keywords: consistency Internal Consistency, Cronbach?s Alpha, Sample Size, Precision, Power, COEFFICIENT-ALPHA, REQUIREMENTS
  • Çukurova University Affiliated: Yes

Abstract

Objective: Internal consistency of scales is crucial for the quality of clinical practices, and scales that measure various aspects of clinical disorders must possess acceptable internal consistency. In the assessment of the internal consistency of the scales, the present study considered the Cronbach's alpha's coefficient. This study aims (1) to provide a practical sample size guide based on Cronbach's alpha for medical researchers; (2) to investigate the factors related to the determination of the sample size needed for a correct estimation of Cronbach's alpha. Methods: The calculation steps of the two sample size approaches (desired precision and desired power) proposed by Bonett are detailed and the sample size tables of these approaches are tabulated. Results: Findings indicate that as the number of items in the scale increases, the sample size decreases to a certain level but reaches a plateau when the number of items is about 30. Furthermore, it is claimed that the sample size should be at least 45 (and more) in order to calculate a meaningful internal consistency coefficient. Conclusion: In the present study, we extracted detailed information about the key components of sample size calculation for Cronbach's alpha and provided an easy and practical tables, which will enable clinicians to determine the appropriate sample size for their internal consistency studies.