ABSTRACT
Marginal homogeneity has usually been tested using both McNamar’s test and Liddell’s exact test, but in this research, we aim to determine the preferred test and the conditions under each test that can be used. The Monte-Carlo approach was used for all the test’s development and analysis. To compute both tests’ power and type I error, considerations were given to the various sample sizes, effect sizes, α- level. A random sample of size 20-1000 was used under 500 iterations. Three different hypothetical schemes were employed to examine the behaviour of the power and the type 1 error of both McNamara’s test and Liddell’s exact test under different values of P12 & P21 and using 5% and 1% significance level levels. At the end of the analysis, it was observed that Liddell’s exact test is preferable when the effect sizes are high and moderate for almost all the sample sizes considered. McNamar’s test is not preferable when the sample sizes are ≤ 20. Liddell’s test is preferred when the proportions are close and small, and the sample sizes are high.
References
- [1] Aberson, C.L.: Applied Power Analysis for Behavioural Science, 2010.
- [2] Agresti, A. (2002). Categorical Data Analysis, Wiley, New York.
- [3] Barlow W. (1998). Modeling of Categorical Agreement. The Encyclopedia of Biostatistics, Wiley, New York.
- [4] Bayo, A. Lawal (2012). Categorical Data Analysis with SAS and SPSS Applications.
- [5] Cohen, J. Statistical Power Analysis for the Behavioural Sciences (2nd ed.) Lawrence Erlbaum Associates, New Jersey, (1988).
- [6] Ellis, Paul (2010). The Essential Guide to Effect Sizes: Statistical Power, Meta-Analysis, and the Interpretation of Research Results. Cambridge University Press. p. 52. ISBN 978-0521142465.
- [7] Fleiss, J.L. (1981). Statistical Methods for Rates and Proportions (2nd edition), Wiley, New York.
- [8] Liddell, D. (1976). “Practical Tests of 2 x 2 Contingency Tables”. Journal of Royal Statistics Society 25(4) : 295 – 304
- [9] McNemar, Quinn (1947). “Noteon the Sampling Error of the Difference between Correlated Proportions on Percentages”. Psychometrika 12(2): 153 – 157
- Sun, X.Z. (2008): “ Generalized McNemar’s Test for Homogeneity of Marginal Distributions”. SAS Global
- Yates, F.(1934): Contingency table involving small numbers and the X2 Supplement to the Journal of the Royal Statistics Society 1(2): 217 – 235
Download all article in PDF
