Objective: Covariance analysis (ANCOVA) is a method used in biomedical and health research, but when the assumption of normality is not satisfied or the dependent variable is bivariate or on ordinal scale, various procedures are presented in the literature for covariance analysis. If assumptions are not satisfied and parametric methods are used, Type I error increases and the power of the test decreases. To overcome these issues, the researcher needs to look for an alternative approach to the analysis of covariance. For non-parametric ANCOVA, many methods are presented in the literature that can be applied to different types of data. Material and Methods: In our study, by analyzing various non-parametric covariance methods; the analysis of repeated categorical data obtained from the clinic is to be analyzed with the non-parametric randomization-based analysis of covariance (NPANCOVA) method. Results: The application of the method was performed on repeated clinical data obtained from 5 consecutive visits from pediatric patients diagnosed with Crimean-Congo hemorrhagic fever disease in the pediatric health and diseases service. Conclusion: As a result, we can state that the NPANCOVA method can be used in clinical analysis due to its many superior advantages in the analysis of categorically repeated non-parametric data.
Keywords: Non-parametric; non-parametric randomization-based analysis of covariance; non-parametric methods
Amaç: Kovaryans analizi (ANCOVA), biyomedikal ve sağlık araştırmalarında kullanılan bir yöntemdir, fakat normallik varsayımı sağlanmadığında veya bağımlı değişkenin 2 değerli veya sıralı değişken olması durumunda literatürde kovaryans analizi için çeşitli prosedürler sunulmuştur. Varsayımlar sağlanmadığı hâlde, eğer parametrik yöntemler kullanılırsa Tip I hatanın artmasına ve testin gücünün azalmasına neden olmaktadır. Parametrik olmayan ANCOVA için literatürde farklı türden verilere uygulanabilecek pek çok yöntem sunulmuştur. Gereç ve Yöntemler: Çalışmamızda, parametrik olmayan çeşitli kovaryans analizi yöntemleri incelenerek, klinikten elde edilen tekrarlanan kategorik verilerinin analizinin non-parametrik randomizasyon tabanlı kovaryans analizi (NPANCOVA) metoduyla yapılması amaçlanmıştır. Bulgular: Metodun uygulaması ise çocuk sağlığı ve hastalıkları servisinde yatan Kırım-Kongo kanamalı ateşi hastalığı tanısı olan çocuk hastalardan art arda 5 vizitden elde edilen, tekrarlanan klinik veriler üzerinde yapılmıştır. Sonuç: Sonuç olarak, non-parametrik randomizasyon tabanlı kovaryans analizi metodunu kategorik tekrarlanan parametrik olmayan verilerin analizinde, pek çok üstün avantajları sebebiyle klinik analizlerde kullanılabileceğini ifade edebiliriz.
Anahtar Kelimeler: Non-parametrik; non-parametrik randomizasyon tabanlı kovaryans analizi; parametrik olmayan yöntemler
- Huitema BE. The Analysis of Covariance and Alternatives. 2nd ed. New York: John Wiley; 2011. [Crossref]
- Keppel G. Design and Analysis: A Researcher's Handbook. 3rd ed. Englewood Cliffs, NJ: Prentice Hall; 1991.
- Dorsey SG, Soeken KL. Use of the Johnson-Neyman technique as an alternative to analysis of covariance. Nurs Res. 1996;45(6):363-6. [Crossref] [PubMed]
- Quinn GP, Keough MJ. Experimental Design and Data Analysis for Biologists. 1st ed. New York: Cambridge University Press; 2002. [Crossref]
- Barret TJ. Computations using analysis of covariance.WIREs Computational Statistics. 2011;3(3):260-8. [Crossref]
- Quade D. Nonparametric analysis of covariance by matching. Biometrics. 1982;38(3):597-611. [Crossref] [PubMed]
- Koch GG, Tangen CM. Non-parametric analysis of covariance and its role in non-inferiority clinical trials. Drug Information J. 1999;33(4):1145-59. [Crossref]
- Koch GG, Tangen CM, Jung JW, Amara IA. Issues for covariance analysis of dichotomous and ordered categorical data from randomized clinical trials and non-parametric strategies for addressing them. Stat Med. 1998;15-30;17(15-16):1863-92. [Crossref] [PubMed]
- Koch GG, Carr GJ, Amara IA, Stokes ME, Uryniak TJ. Categorical data analysis. In: DA Berry, ed. Statistical Methodology in the Pharmaceutical Sciences. 1st ed. New York: Marcel Dekker; 1990. p.389-73.
- McKean JW, Vidmar TJ. A comparison of two rank-based methods for the analysis of linear models. American Statistician.1994;48(3):220-9. [Crossref]
- Kloke JD, McKean JW. Rfit: Rank-based estimation for linear models. The R Journal 2012;4(2):57-64. [Crossref]
- Zink RC, Koch GG. NParCov3: A SAS/IML macro for non-parametric randomization-based analysis of covariance. J Statistical Software. 2012;50(3):1-17. [Crossref]
- Sievers GL, McKean JW. On the robust rank analysis of linear models with non-symmetric error distributions. J Statistical Inference Planning. 1986;13:215-30. [Link]
- Fan C, Zhang D, Wei L, Koch G. Methods for missing data handling in randomized clinical trials with nonnormal endpoints with application to a phase III clinical trial. Statistics in Biopharmaceutical Research. 2016;8(2):179-93. [Crossref]
- Zhao P, Tang N, Qu A, Jiang D. Semiparametric estimating equations inference with nonignorable missing data. Statistica Sinica. 2017;27(1):89-113. [Link]
- Hussey MA, Koch GG, Preisser JS, Saville BR. Analysis of matched studies with dichotomous outcomes using nonparametric randomization-based analysis of covariance. Statistics in Biopharmaceutical Research. 2013;5(3):194-203. [Crossref]
- YU Jing. Statistical methods for topics involving repeated measures for categorical data. 2019. [Link]
.: Process List