pwSEM (version 1.0.0)

vanishing.tetrads: The vanishing.tetrads function

Description

This function implements the vanishing tetrads theorem of Spirtes, Glymour & Scheines (1993). If a set of four variables in dat has a saturated unoriented dependency graph in CI.algorithm, and a tetrad equation is zero, then this is evidence for a latent variable.

Usage

vanishing.tetrads(dat, sig = 0.05, bootstrap = FALSE, B = 1000)

Value

Just output to the screen listing each tetrad equation, its value and its significance level.

Arguments

dat

A data frame containing the observed variables. No other variables can be in this file, such as ones describing the nesting structure.

sig

A numerical value between 0 and 1 giving the significance level to use when judging (conditional) independence. The default value is 0.05.

bootstrap

A logical value specifying if you want bootstrap probabilities or not. Defaults to FALSE

B

The number of bootstrap samples required. Defaults to 1000.

Examples

#Determines which of the three tetrad equations are zero in this data set
#having 500 observations and 4 variables.  Since this set of 4 variables
#has a saturated partially oriented dependency graph,
#as shown using the CI.algorithm function, the tetrad equations
#that are zero (i.e. vanish) identify where latent variables occur that
#are common causes of these variables
#Since this is a saturated partially oriented dependency graph:
vanishing.tetrads(dat=sim_tetrads,sig=0.05)