Dependent Probability Test

The dependent probability test, depend measures the distribution of the random number generator used in datagen in a matrix. This test is similar to the independence test, indep. However, in this test, each probability in the contingency table is different. In the independence test, each probability is the same.

First, the row probabilities are generated at random. The sum of the row probabilities is equal to 1.0.

For each row, the contingent column probabilities are generated at random. The sum of each row of column probabilities is equal to 1.0.

Given these contingent probabilities, the test is ready to start. A random number from 0.0 to 1.0 is generated to select a row. For that row, a second random number from 0.0 to 1.0 is generated. The column in that row is selected to match the second random fraction, based on the contingent probabilities for that row. The observed contingency table is updated for that column and row. After the test is over, the tallies in the observed contingency table are summed to create a chi-square value.

If the chi-square test passes, the locations produced by datagen are distributed properly across the matrix. The count in each cell of the matrix should match the expected frequency for that cell.

The test will fail once in twenty trials, on the average. If the test fails, re-run the test. It should pass in the next trial.

Syntax

The syntax for depend is:

      depend size columns rows

Example

      depend 10000 6 6


Table of Contents     Job and Record Level Definitions
Field Level Definitions     Datagen Syntax
Supporting Routines     FAQ     Copying