Perbandingan Tingkat Konsistensi Normalitas Distribusi Metode Kolmogorov-Smirnov, Lilliefors, Shapiro-Wilk, dan Skewness-Kurtosis

ABSTRACT There are many kinds of normality test method in determining the data whether has the normal distribution or not. Some of these methods result can make different decision so it can be misleading and confusing practitioners in performing statistical tests. Normality test needed a method that can produce consistent decisions. The aim of this research was to compare the results of distribution normality analysis with the Kolmogorov-Smirnov, Lilliefors, and Shapiro-Wilk toward Skewness-Kurtosis test and analyze which one has the best consistency level. This research used secondary data from the report of activity result of one of posyandu in Surabaya. The data is about Weight (W), Height (H), and Index W/H. Total population of the posyandu was 80 subject, then doing simulation of the data so that obtained samples with multiples of 10, there are 10, 20, 30, 40, 50, 60, and 70. Samples were selected by simple random sampling method. This research can be performed the results of normality test decision on each samples and each repetition. This research resulted in the percentage consistency level of the results in three method of normality test toward Skewness-Kurtosis test result, there are Kolmogorov-Smirnov test was 68.26%, the Lilliefors test was 82.54%, and the Shapiro-Wilk test was 90.48%. The conclusion is Shapiro-Wilk test has the best consistency level and then followed by Lilliefors test and Kolmogorov-Smirnov test. The next researchers can analyzed the normal distribution suggested using other methods such as the Anderson-Darling test, Cramer-von Mises test, or Fisher’s test cumulat. Keywords normality test, Kolmogorov-Smirnov, Lilliefors, Shapiro-Wilk, Skewness-Kurtosis