jaotustest

Jaotustest is a statistical term used in Estonian to refer to tests that assess whether a data sample conforms to a specified theoretical distribution, such as the normal distribution. These tests are used in model diagnostics, quality control, and research to verify the assumed distributional form of data before applying further analyses.

The basic idea of a jaotustest is to compare the observed data to what would be expected under a chosen distribution. The null hypothesis typically states that the sample comes from the specified distribution, while the alternative suggests a deviation. A test statistic is computed from the data, and a p-value is obtained to decide whether to reject the null hypothesis at a chosen significance level. Interpretation should consider the test’s power, sample size, and whether data meet assumptions such as independence and continuity.

Common examples include the Kolmogorov-Smirnov test, the Anderson-Darling test, and the Cramér-von Mises test, which are used to assess goodness-of-fit to a reference distribution. For normality specifically, the Shapiro-Wilk test is widely used, while the Lilliefors correction adjusts the KS test when distribution parameters are estimated from the data. Pearson’s chi-squared test is applicable when data are binned. Different tests have varying sensitivity to deviations in centers, tails, or both, and the choice depends on data type and sample size.

In practice, jaotustest are implemented in statistical software and help guide decisions about model assumptions and subsequent analyses.