Results of misusage of nonparametric methods: a simulation study
Abstract
Purpose: Visual Analogue Scale has a narrow range which is used to evaluate the pain severity. In general, such type of data don’t provide the parametric test hypothesis. But at the group comparisons the number of outlier could be excessive at one of the groups. Consequently, it is analyzed by using non-parametric methods. Therefore, in this study as comparing two groups which are narrow ranged rates and also have outliers (skew) it has been aimed that evaluating the performance of Student t test, Logistic transformed Student t test, Mann Whitney-U test and Mood’s Median tests.
Material and Methods: The study was planned as a simulation study. Accordingly, it has been formed distributions which are at interval of 0-10, based on the mostly used Visual Analogue Scale rates which were used in evaluation of pain. Simulations were repeated 1000 times for 20 combinations which consisted of 4 different distributions and 5 different sample size.
Results: Type I error rates which obtained have varied as the sample size and the distribution type is changed and the test results haven't been changed though normalization method used in the type of data that is not concordance with normal distribution.
Conclusion: Misusing and overusing of non-parametric tests gives biased results in scientific studies. Although the general approach was using non-parametric tests in which datas were recorded at the interval of 0-10 cm, the simulation study showed that such datas could also be evaluated by using parametric tests
Keywords
Kaynakça
- 1. Day S. Dictionary for Clinical Trials, 2nd ed. USA: John Wiley & Sons, LTD. 2007:23. 2. Hawker GA, Mian S, Kendzerska T, French M. Measures of adult pain: Visual Analog Scale for Pain (VAS Pain), Numeric Rating Scale for Pain (NRS Pain), McGill Pain Questionnaire (MPQ),Short-Form McGill Pain Questionnaire (SF-MPQ), Chronic Pain Grade Scale (CPGS), Short Form-36 Bodily Pain Scale (SF-36 BPS) and Measure of Intermittent and Constant Osteoarthritis Pain (ICOAP).Arthritis Care Res 2011;63:240-52. doi: 10.1002/acr.20543. 3. Sullivan L. Nonparametric tests. http://sphweb.bumc.bu.edu/otlt/MPH-Modules/BS/BS704_Nonparametric/index.html. 4. Vickers AJ. Parametric versus non-parametric statistics in the analysis of randomized trials with non-normally distributed data. BMC Medical Research Methodology 2005;5:1-12. doi:10.1186/1471-2288-5-35. 5. Cardone R. Nonparametric: distribution-free, not assumption-free. Erişim adresi: http://www.isixsigma.com/tools-templates/hypothesis-testing/nonparametric-distribution-free-not-assumption-free/ 6. Alpar R. Uygulamalı İstatistik ve Geçerlik-Güvenirlik, Ankara: Detay Yayıncılık. 2012:215-25. 7. Senn S. Cross Over Trials in Clinical Research. 2nd ed. London: John Wiley & Sons LTD. 2002:92-6. 8. Dexter F, Chestnut DH. Analysis of statistical tests to compare visual analog scale measurements among groups. Anesthesiology. 1995;82:896-902. 9. Kan İ. Biyoistatistik. Ankara: Nobel Yayın Dağıtım. 2006:193. 10. Mood’s median test. http://www2.hawaii.edu/~taylor/z631/moods.pdf. 11. Dexter F, Chestnut DH. Analysis of statistical tests to compare visual analog scale measurements among groups. Anesthesiology 1995;82:896‑902. 12. Price DD, Staud R, Robinson ME. How should we use the visual analogue scale (VAS) in rehabilitation outcomes? II: Visual analogue scales as ratio scales: An alternative to the viıew of Kersten et al. J Rehabil Med. 2012;44:800-4. 13. Bishop PA, Herron RL. Use and Misuse of the Likert Item Responses and Other Ordinal Measures. Int J Exerc Sci Science. 2015;8:297-302
