Power Analysis
Most statistical methods test if the null hypothesis is true or false. The analyst usually sets a small value known as alpha as the significance criterion. The alpha prevents us from rejecting the null hypothesis when it is indeed true. This usually leads to a type I error. Power analysis only deals with type II errors, accepting the null hypothesis yet it is not true. Statistix supports power analysis for several designs of balanced ANOVA. You can use the power of the F test or the power of multiple comparisons to perform a power analysis. Some of the power analysis procedures in Statistix include one-sample and paired T-test, factorial designs, two-sample T-test, strip-plot design, etc.
Association Tests
Association tests are used to determine if there is a relationship or similarity between more than one variable. Statistix offers the following association tests:
Multinomial test
It is a goodness of fit test that checks if frequencies of categories that are mutually exclusive fit in the hypothesized distribution.
Chi-square test
Statistix’s chi-square test is a goodness of fit test for two-way tables.
Kolmogorov-Smirnov Test
This test compares the distributions of two samples from two different population
Other association tests available in Statistix are Spearman rank correlation, log-linear models, two-by-two tables.
Nonlinear Regression
Nonlinear regression is a statistical technique that analyzes data by fitting it into a model and then expressing it as a mathematical function. In Statistix nonlinear regression models are fitted using the Levenberg-Marquardt-Nash algorithm. The independent variables and the parameters in the models can each be up to 20. You can create your model by choosing from a list of 30 models that have already been predefined in Statistix.
Time series
Time series constitutes a group of sequentially collected observations. These observations are often gathered over a specific period and are sequentially dependent. Time series is a vital tool employed in several fields to forecast ideas that have not yet been implemented. Statistix has the Time series plot procedure for plotting one or more variables. If you want to create an autocorrelation plot for a specific variable, you can use the autocorrelation plot procedure. Other time series procedures available in Statistix include exponential smoothing, moving averages, partial autocorrelation plots, and cross-correlation.