Simulate a Column data table

To simulate a family of column data sets with random error, start from any data table or graph, click Analyze, open the Simulate data  category, and then select Simulate Column Data.

On the Experimental design tab, choose the number of data sets, and the mean of each data set. For each data set, enter the number of values you wish to simulate for that data set (number of rows of data).

On the Random error tab, choose among several methods for generating random scatter and also adding outliers. You must choose one setting for the random values for all the data sets. For example, if you choose Gaussian error (the most common), you can only choose one standard deviation, which applies to all the data sets.

Simulate a 2x2 contingency table

To simulate a contingency table, start from any data table or graph, click Analyze, open the Simulate data  category, and then select Simulate Contingency Table.

On the Experimental design tab, choose the sample size (total number of subjects for both rows and both columns). Also specify which of four experimental designs you wish to simulate.

On the Rows and columns tab, name the two rows and two columns, and specify (on average) how many subjects go in each.

Prism will use the binomial random values to decide how many subjects go into each cell, maintaining the total you entered.

Note that this analysis only can simulate a 2x2 contingency table.

How Prism generates random numbers

Prism generates pseudo random numbers from the binomial or Poisson distribution, using ideas adapted from pages 372-377 of Numerical Recipes, third edition, by  WH Press and colleagues.