Statistics
Statistics objects generally contain information about statistical tests or pattern classification, but they can store information about any analysis. They simply have a name and a reference to a MAT-file that contains the details about the analysis.
This section also describes an important part of the toolbox, regressors. Regressors are used for many tasks, including calculation of event-related averages, creating statistical maps, and running pattern classification analysis.
Univariate statistical tests produce statistical maps, or statmaps. A separate statistical test is run for each channel, time, and frequency in a pattern to produce a map of significance values. Statmaps may be created by pattern_statmap, and significance is plotted by many of the plotting functions in the toolbox. Some tests, such as multi-way ANOVAs, return multiple significance values. In this case, different effects may be named by a variable called names saved with the stat object.
stat_R is a general wrapper for running statistical tests in R, a free program that includes support for many advanced statistical tests. There are a few R scripts included in the toolbox, such as rmaov1.R, which can be used as templates to write new tests. Currently the interface between MATLAB and R is rather primitive, however, so making statmaps using R can be slow.
Pattern classification results are saved in standard res structures. These results structures follow the format of MVPA results structures, except the iterations substruct is 4-D, corresponding to the dimensions of the pattern that was classified.