Randomization Test Options

Use this dialog to select the output generated by a randomization test of the difference between two groups, and the options used in the test.

Display

Specifies which items are displayed in the Output window or the Graphics window.

Summary and estimates	A summary, estimates and the randomization test probability. Additional information is displayed when the t-test method is used which includes the lower and upper confidence limits.
Histogram of samples	A histogram of the randomization distribution of the statistic with the observed statistic marked on the graph.

Type of test

Lets you choose either a one- or two-sided t-test. A two-sided test looks for differences either side of 0, whereas a one-sided test checks whether the mean is greater than or less than the given mean.

Estimate of variance for t-test

This option is only available when the two-sample t-test method is selected. For a two-sample t-test you can choose which method is used for estimate of variances for the test statistic.

Automatic	Uses a pooled estimate unless there is evidence of unequal variances.
Pooled	Always uses a pooled estimate of the standard deviation for the test, i.e. a common standard deviation that is formed from the average variance over the two groups.
Separate	Estimate the standard deviation for the test using separate variances.

Confidence limits (%)

This option is only available when the two-sample t-test method is selected. This lets you control the limits for the calculation of the confidence interval for the t-test. The limit is expressed as a percentage in the range 0-100.

Seed for randomization

This gives a seed to initialize the random number generation used for resampling the data. Using zero initializes this from the computer’s clock: specifying a non-zero value gives a repeatable analysis.