Skip to content

U Test

Jump to bottom
Simone Maurizio La Cava edited this page Apr 22, 2020 · 2 revisions

Also known as Wilcoxon–Mann–Whitney test, the U Test is a nonparametric statistical test which verify if there are differences between the two groups' set of data.

In particular, this test can be used to investigate whether the two groups of data shows the same distribution or not, and the result is given it terms of probability: if the p-value (probability that the data of the two sets belong to the same distribution) that the two groups comes from different distributions is higher than a threshold, called α-value, the two sets' distributions can be considered as significantly different.

After the measure's directory, you have to select what kind of spatial analysis you want to analyze between:

  • Areas, to study one of the macroareas (frontal area, temporal area, central area, parietal area or occipital area)
  • Total, to study the single locations
  • Asymmetry, to study the differences on the measure between the right and the left neural hemisphere
  • Global, to study the overall measure value

Before to start the analysis, you also have to select the conservativeness level:

  • the minimum let you to use an alpha value equal to 0.05 (so, a comparison can be significant if the relative p-value is less than this value)
  • the maximum reduce this value, by considering it as equal to 0.05 divided by the number of different comparisons, for example if will be compared n frequency bands and m locations, you will have a p-value equals to:

Now you can RUN your test to discover if the two groups can be considered as significantly different.

When the computation is finished, the toolbox will show a figure containing a p-value table for each comparison.

If there is any significant comparison, it will be also shown a table which contains some informations about it, such as which group shows an higher value on average.

The data relative to the only significant comparisons will be saved in a file with their information, allowing you to export them and to use them for a classification analysis.

Furthermore, il will be shown a table which consists of all the data used in such comparisons (they may be copied for external analysis).

Now you can execute other tests or return to the previous interface in order to execute other statistical analysis.

Clone this wiki locally