One-sample Nonparametric Tests
See Also
This menu allows you to perform one-sample nonparametric tests. You can choose the type of analysis to perform by selecting the test from the list given.

Wilcoxon test

The Wilcoxon Matched-Pairs (or Signed-Ranks) test is a nonparametric test of location for two related samples (for example a before-and-after study). The null hypothesis is that the samples arise from exactly the same distribution, and this is tested against the alternative that the underlying distributions differ in their locations.

The data for the one-sample Wilcoxon test is a variate holding differences between the two samples. The name of the variate should be entered into Variate box, and can be selected from those listed in the Available Data window.

Sign test

This menu allows you to perform a nonparametric two-sided test of location of a single sample of observations. The data for the test are in a GenStat variate structure whose name must be entered into the Variate box. This can be selected from those listed in the Available Data window.

The default null hypothesis is that the sample has a median value of zero, but you can test against other values by changing the Median Value box. Missing values and observations that equal the specified Median Value are ignored and the effective sample-size reduced accordingly.

You can choose to perform a one-sided or two-sided test from the Type of Test options.

Runs test

This menu allows you to perform a test of the randomness of a sequence of observations. The data for the test is assumed to be an ordered sequence of observations of two types, and a run is defined to be a succession of observations of the same type. A clue to lack of randomness is provided by the total number of runs in the sequence. A low number of runs might indicate positive serial correlation while a high number might arise from negative serial correlation.

The data for the test are in a GenStat variate structure whose name must be entered into the Variate box. This can be selected from those listed in the Available Data window.

Observations larger than the value specified by the Boundary box are considered to be of the first type, while observation smaller than this are taken to be of the second type. Missing values and observations that equal the boundary value are ignored.

See Also

These statistical tests are performed by the following procedures which may also be used in command mode with additional parameters in some cases:

The descriptions of these commands contain full details of the methods used to calculate the test statistics, along with suitable references.