Message/Author 


How do I specify a binary covariate in a Monte Carlo simulation that does not have a 50/50 split, but instead, for example, a 75/25 split? 


You use the CUTPOINT option of the MONTECARLO command. See Example 11.1 in the Mplus User's Guide. The value given is a zscore. So you use a zscore table to select the value that corresponds to a 75/25 split. 

June Zhou posted on Thursday, February 07, 2013  3:17 pm



I have a similar question about generating a binary independent variable in a Monte Carlo simulation. I'd like to generate "Gender" variable that with population mean of 0.5 and variance of 0.25. What code should I use? Thank you in advance. 


The continuous variable you generate should have mean zero and variance 1. You should use a cutpoint of zero which cuts the sample 50/50 with a variance of .25. 

Jamie Stagl posted on Wednesday, February 12, 2014  10:49 am



Hello, I am doing a Monte Carlo simulation of a growth model with 3 time points and a nominal predictor (3 intervention groups). I specified 2 binary dummy variables (x2 and x3) to represent my 3 groups. What CUTPOINT value should I use for these 2 dummy variables, considering they are not a 50/50 split (1/3 of sample gets intervention A, 1/3 gets intervention B, and 1/3 gets intervention C)? Thank you. 


If you specify mean and variance = 0, 1 for the variable that you apply Cutpoints to, you can use a table for a standard normal distribution function to get the cutpoints. For an example, see 12.1. 

Jamie Stagl posted on Sunday, February 16, 2014  9:06 pm



Thank you, that was very helpful. In general, we expect to see that 2 of the 3 groups do not change over the 3 time points, while the third group improves on the outcome. Would you say that the use of 2 dummy variables is an accurate way to estimate the necessary sample size (does the simulation know there are 3 linked groups with these 2 dummy variables or is it only doing a 2group comparison)? On a related note, would you suggest setting the slope growth factors at different values to reflect our hypothesis, and reference the power associated with the smaller parameter estimate to determine sample size? 


Actually, you are better off doing a multiplegroup analysis with 3 groups, where you control the number of observations in the groups and have freedom to vary any parameter across the groups. So don't use dummy variables. Your hypothesis sounds like you would have the slope mean at zero in two of the groups. 

Tao Yang posted on Monday, January 23, 2017  10:58 pm



Hello, I'd like to run Monte Carlo power analysis of the interactions of x (binary predictor) with z and w (continuous variables). I generated data sets for x, z, w, and y to be used for external Monte Carlo in Mplus (so that I can create interaction terms using DEFINE). Syntax below. Without "MODEL: y ON x", I got an error message that "Only xvariables may have cutpoints in this type of analysis..." So I included MODEL command line only to tell the program to treat x as an exogenous variable. Does the MODEL command influence data generation? In other words, are data sets generated based on specifications in MODEL POPULATION or MODEL? MONTECARLO: NAMES = x z w y; NOBSERVATIONS = 500; CUTPOINTS = x(0); NREPS = 100;REPSAVE = ALL; SAVE = rep*.dat; ANALYSIS: TYPE = RANDOM; MODEL POPULATION: zb BY z@1; zb@.50; z@.01; wb BY w@1; wb@.30; w@.01; xzb  zb XWITH x; xwb  wb XWITH x; y ON x*.20 zb*.10 wb*.30 xzb*.21 xwb*.35; [x@.50 zb@0 wb@0 z*.35 w*.12]; x@.25; [y*.12]; y*.02; MODEL: y ON x; 


Answered elsewhere. 

Back to top 