Fix the Latent Heywood Model
In the last exercise, you found that the adoption survey had a correlation > 1 on the latent variable. You should fix the Heywood case by collapsing the two latent variables into one latent variable. Then you should analyze and summarize the model to explore if merging these two factors into one has solved the problematic correlation.
Este exercício faz parte do curso
Structural Equation Modeling with lavaan in R
Instruções do exercício
- Change the model to create only one
goodstory
factor that is measured by all six manifest variables in theadoptsurvey
dataset. - Analyze the model with the
cfa()
function to ensure there are no error messages. - Run the
summary()
for the model to view the final model fit.
Exercício interativo prático
Experimente este exercício completando este código de exemplo.
# Look at the data
head(adoptsurvey)
# Edit the original model
adopt.model <- 'goodstory =~ pictures + background + loveskids
inperson =~ energy + wagstail + playful'
# Analyze the model
adopt.fit <- cfa(model = ___, data = adoptsurvey)
# Look for Heywood cases
summary(___, standardized = TRUE, fit.measures = TRUE)