psych (version 2.6.1)

fparse: Parse and extend formula input from a model and return the DV(s), IV(s), and associated terms.

Description

Formula input from e.g., lm, may be extended to include mediators, quadratic and partial terms using a standard syntax. This is used by (e.g.,) lmCor, mediate, and error.bars.by. lavParse will take a limited number of lavaan syntax commands and return as a model matrix for CFA, CFA.bifactor, sim, and sim.structure

Usage

fparse(expr)
lavParse(model, phi=FALSE)

Value

v.mat

A matrix of variable weights for each factor

y

A list of elements from the left side of the formula

x

A list of elements from the right side of the formula

m

A list of those elements of the formula included in ()

prod

A list of elements separated by a * sign

ex

A list of elements marked by I()

model

A matrix with cols for the factors, rows for the variable, values specified

Arguments

expr

A legitimate expression in the form y ~ x1 , etc. (see details)

model

A lavaan style factor specification. (see examples)

phi

If TRUE, the model can include the factor correlations.

Author

William Revelle

Details

The basic formula input given as DV1 + DV2 ~ IV1 + IV2 + (IV3) + I(IV4^2) - IV5 will be parsed to return 2 DVs (1 and 2), two normal IVs (1 and 2), a mediator (IV3) a quadratic (IV4) and a variable to be partialed (IV5). See the various examples in lmCor and mediate.

To allow rapid comparisons with sem and cfa output from lavaan, it is convenient to use lavaan style syntax.

CFA andCFA can specify the model using lavaan syntax as well as the item keys format.

Examples

fparse(DV  ~ IV1 + IV2 * IV2*IV3 + (IV4) + I(IV5^2) )
#somewhat more complicated
fparse(DV1 + DV2 ~ IV1 + IV2 + IV3*IV4 + I(IV5^2) + I(Iv6^2) + (IV7) + (IV8) - IV9)


model <- '	F1 =~ .9*V1 + .8*V2 + .7*V3 
         	F2 =~ .8 * V4 + .7*V5 + .6*V6
        	F3 =~ .9*V7 + .7*V8 + .5*V9
       		F1 ~ .3 * F2 + .6* F3
        	F2 ~ .5*F3'
lavParse(model,phi=TRUE)   #will return a 9 x 3 matrix loadings and 3x3 phi matrix for sim 
 
#by specifying the coefficients, we can run sim more readily
sim.data <-sim(model,n=500)
cx <- CFA(model,sim.data$observed) 

#The CFA function will not use the coefficients nor the phi matrix, just the path model
#So we can use the same model for simulations and CFA
#But, lavaan needs to drop the coefficients
modelg <- 'g =~ .4*V1 +.4*V2+.4*V3 +.4*V4+.4*V5+.4*V6 +.4*V7+.4*V8+.4*V9
F1 =~ .9*V1 + .8*V2 + .7*V3 
         	F2 =~ .8 * V4 + .7*V5 + .6*V6
        	F3 =~ .9*V7 + .7*V8 + .5*V9   '
  
modelg <- 'g=~V1 + V2 + V3 + V4 + V5 + V6 + V7 + V8 +V9
    	F1 =~  V1 + V2 + V3 
     	F2 =~  V4 + V5 + V6
        F3 =~ V7 + V8 + V9'
lavParse(modelg)
#model without coefficients to run in lavaan