NORMALITY TESTS AND DISTRIBUTIONAL FITTING

Several statistical tests exist for deciding if a sample set of data comes from a specific distribution. The most commonly used are the Kolmogorov–Smirnov test and the chi-square test. Each test has its advantages and disadvantages. The following sections detail the specifics of these tests as applied in distributional fitting in Monte Carlo simulation analysis. Other less powerful tests such as the Jacque–Bera and Wilkes–Shapiro are not used in Risk Simulator as these are parametric tests and their accuracy depends on the dataset being normal or near-normal. Therefore, the results of these tests are oftentimes suspect or yield inconsistent results.

 

error: Content is protected !!