StatPowers
File
? ✖ Enter the population name and select the number of variables.
If you want to remove a variable, you just reduce the number of variables - variables are always removed from the end of the list. Each variable must be named and assigned a type:
  • Categorical: descriptive variables with a finite number of levels
  • Discrete: Integer variables
  • Continuous: Decimal variables
Population Name:
Number of Variables
? ✖ Order the variables. Independent variables should be at the top of the list.
A variable can only depend of the values of variables above it in the list.
? ✖ Select a variable. Depending on the variable type, you will have different options:
Independent Variables will be the simplest.
  • Categorical: Specify the number of levels, name the levels and define the probability for each level. If the probabilities don't add up to 1, click "Scale Probabilities" to fix them.
  • Discrete: For distribution shapes (uniform, skew, symmetric) you just specify the minimum and maximum value. For a custom distribution, specify the number of values possible, define the values and the associated probabilities. If the probabilities don't add up to 1, click "Scale Probabilities" to fix them.
  • Continuous: Select a distribution shape, as with a discrete variable.
For Dependent variables things get more complicated.
  • Categorical: You may specify discrete dependency or a log-odds model. For discrete dependencies, dependent on categorical and discrete variables, you create a dependency, specify the precise values of the dependency variables. For a log-odds model, you specify a log-odds model for each level of the categorical variable. For 2 levels, the second level will be the complement. For 3+ levels, the probabilities will be scaled at the time of sampling so they all add up to 1. Syntax details below.
  • Discrete: You may only define discrete depenencies. All dependencies must have the same distribution type (e.g. uniform, symmetric, custom)
  • Continuous: You may select either discrete dependencies (As long as no variable dependencies are continuous) or a generalized linear model.
Linear model syntax
Terms should be separated by + or -
Error term is in the form σz (e.g. 5.2z)
Variable transformations allowed: log (e.g. lnv3) or power (e.g. v4^2, v2^.5)
Interaction terms are in the form vivj (e.g. 5v1v3)
Entire model may be in the exponent by surrounding with exp(...) (e.g. exp(5+2v1+2.3z))
Examples: 
4+2v1+3.5z
4+2v1+3v2-.23v1v2+.4z
exp(5-v1+2z)
3+2lnv1+3lnv2+55z
Variables To Report
Sampling Method
Random Number Seed:
Sample Size:
Analyze Variables:
Group By: