Monte Carlo Analysis - Advanced Tab

Select the Advanced tab of the Monte Carlo Analysis dialog box to access options to enter the parameters and conditions for a Monte Carlo experiment. Begin by entering the seed values and Number of Replications; then set the analysis parameters by selecting from a variety of other options.

Seed1

Enter an integer between 1 and 2,147,483,647 in the Seed1 box. This seed is used to generate random numbers.

Seed2

Enter another integer between 1 and 2,147,483,647 in the Seed2 box. This seed is used in Contaminated Normal generation only.

Number of replications

Specify the number of Monte Carlo replications in the Number of replications box. The number can be no greater than 1000.

Group characteristics

Enter the characteristics for your Monte Carlo groups under Group characteristics.

Sample Sizes

Click the Sample Sizes button to display the Set Monte Carlo Sample Sizes dialog box, which is used to enter the sample sizes for each group in the analysis. Sample sizes should be greater than the number of variables in the analysis, but they can be different for each group.

Contamination Factors

Click the Contamination Factors button to display the Set Contamination Parameters dialog box where you enter, for each group, a proportion of outliers, and a multiplier factor. Outlier contamination is simulated in the Monte Carlo module by a standard mixture distribution technique. Suppose your data come from a population with mean μ and covariance matrix σ. We simulate contamination by substituting, a certain proportion p of the time, observations with a mean μ and a covariance matrix kσ, where k is a moderately large multiplier (say, 10). In effect, the outliers are obtained by taking "what they would have been before the mean was added on" and multiplying by the square root of k.

For each sample, enter contamination factors k and p. For each sample, a proportion p of the time, the observations will then have a mean m and a covariance matrix kS.

Store extra information

Select the options under Store extra information to store additional information about the results of each analysis and then display it in the Overall Results spreadsheet.  

Parameter Estimates

Select the Parameter Estimates check box to store the values for free parameters. They are given names corresponding to the numbers actually assigned in the PATH1 program. The names are PAR_# , where # is the parameter number. So, if a program had free parameters 1, 2, 3, 6, the variables PAR_1, PAR_2, PAR_3, and PAR_6 would be added to the Overall Results spreadsheet.

Standard Errors

Select the Standard Errors check box to store the estimated standard errors for each free parameter, using the following naming scheme. Each standard error is named SE_#, where # is the free parameter number.

Fit Indices

Check the Fit Indices check box to store the available fit indices for each analysis.

Get population from

Select either the Model, Data, or Bootstrap option button under Get population from to specify a particular population structure. STATISTICA generates data that simulates samples from the selected population structure. These structures are described in Population Structures.

Special data types

Select one or both of the non-normal data types listed under Special data types.

Categorical Data

Select the Categorical Data check box to change the data from continuous to categorical, using categorization rules that you have specified. You can select between 2 and 10 categories, and you select the boundaries for the categorization. (See the Categories option under the Adjust Distribution for details.) Note that STATISTICA performs this change as a last step.

Outlier Contamination

Select the Outlier Contamination check box to create a distribution contaminated by outliers, using a standard mixture distribution technique. For details, see Contamination Factors (above).

Adjust distribution

Use the options under Adjust distribution to adjust special distributional characteristics if you are simulating data that are not multivariate normal. Characteristics can be adjusted separately by group.

Group

In the Group box, enter the number of the group whose characteristics you want to alter.

Skewnesses

Click the Skewnesses button to display the Skewnesses for Group dialog box in which you enter the desired skewnesses for variables in each group.

Kurtoses

Click the Kurtoses button to display the Kurtoses for Group dialog box in which you enter the desired kurtoses for variables in each group.

Categories

Click the Categories button to display an "interactive spreadsheet" (Categorical Data Setup) in which you enter the parameters for converting each continuous variable to categorical form. For each variable, enter the number of categories. If you have k categories, you will have k-1 cutoff points. All continuous values below the first cutoff will be given the categorical score 1. All continuous values exceeding the first cutoff but below the second cutoff will be given the categorical score 2, etc.

Restore Defaults

Click the Restore Defaults button to restore the various options to their original default values.