Monte Carlo PC has the following capabilities:
- Generates data sets with up to 99 variables (this limit can be increased on a custom basis).
- Generates an unlimited* number of observations per variable (* limited only by the storage capacity of the output drive).
- Generates data having either uniform or normal distributional form.
- User-specified accuracy of exact (i.e., population) parameters (i.e., means, SDs, and correlations) up to 8 decimal places.
- Uses the most current version of the Mersenne Twister pseudo-random number generation algorithm, with a period of 219937 − 1; automatic seed generation with high degree of randomness.
- Data set specifications may be saved in reloadable files for repeated use.
- Data may be generated either as a sample of the population with the specified parameters (with sampling error), or as an exact representation of the population, with parameters reproduced with up to 8 decimal places of accuracy, as specified by the user.
- If a singular (i.e., non-invertible) correlational structure is entered, the program will attempt to iteratively reduce correlations between the fully-determined variable and the k-1 others to achieve a condition of nonsingularity; the user is required to inspect the changed correlations and decide whether or not to proceed.
- Monte_Carlo Pc can be run iteratively from the MC Mode Dashboard, with no practical limit on the number of iterations that can be specified (other than human life expectancy!). Calls to any external programs, along with all necessary command line arguments, can be composed in an easy-to-use editing tool integrated into the user interface.
- Data file output can be cut-and-pasted directly into Excel and into many standard statistical software systems (e.g., SPSS, SAS, Minitab, STATA).
- A random sampling generator that can draw a virtually unlimited number of samples of any specified size, with or without replacement, from a group or population of a specified size.
- Within the random sampling subsytem, there is also the capability to generate estimates of standard errors, using bootstrapping (i.e., repeated sampling of a sample with replacement) or the more accurate Monte Carlo method (repeated sampling of an astronomically large population without replacement). For these methods the user need only specify the number of samples, and the size, mean, and standard deviation of the observed sample.