What are the outputs of the inspect in lavaan?

233 views
Skip to first unread message

Abc Abc

unread,
Jun 22, 2019, 12:49:33 PM6/22/19
to lavaan
Dear all,

According to the tutorial, inspect "returns a list of the model matrices that are used internally to represent the model". But what exactly are those matrices?


For CFA, with these code:
model <- 'O  =~ o1 + o2 + o3 + o4 + o5
T  =~  t1 + t2 + t3 + t4
G =~ g1 + g2 + g3'

fit <- cfa(model, data=data)  #data is just a table of data

inspect(fit, "standardized") 

Output: 
$lambda                                         
      O    T     G
o1 0.870 0.000 0.000
o2 0.909 0.000 0.000
o3 0.834 0.000 0.000
o4 0.830 0.000 0.000
o5 0.881 0.000 0.000
t1 0.000 0.834 0.000
t2 0.000 0.861 0.000
t3 0.000 0.804 0.000
t4 0.000 0.799 0.000
g1 0.000 0.000 0.811
g2 0.000 0.000 0.876
g3 0.000 0.000 0.893

$theta
   o1    o2    o3    o4    o5    t1    t2    t3    t4    g1    g2    g3   
o1 0.243                                                                  
o2 0.000 0.173                                                            
o3 0.000 0.000 0.304                                                      
o4 0.000 0.000 0.000 0.311                                                
o5 0.000 0.000 0.000 0.000 0.223                                          
t1 0.000 0.000 0.000 0.000 0.000 0.305                                    
t2 0.000 0.000 0.000 0.000 0.000 0.000 0.259                              
t3 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.353                        
t4 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.362                  
g1 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.343            
g2 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.233      
g3 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.202

$psi
    OC    TC    TGC  
OC  1.000            
TC  0.148 1.000      
TGC 0.346 0.447 1.000


I believe lambda is factor loading; psi is the correlation between constructs. But what is theta?


For SEM, with these code:

model1 <- 'O  =~ o1 + o2 + o3 + o4 + o5
T  =~  t1 + t2 + t3 + t4
G =~ g1 + g2 + g3
P =~ p1 + p2 + p3
P ~ OC + TC + TGC'

fit1 <- sem(model1, data=data)
inspect(fit1)

Output:
$lambda
    O  T   G  P
o1  0  0   0  0
o2  1  0   0  0
o3  2  0   0  0
o4  3  0   0  0
o5  4  0   0  0
t1  0  0   0  0
t2  0  5   0  0
t3  0  6   0  0
t4  0  7   0  0
g1  0  0   0  0
g2  0  0   8  0
g3  0  0   9  0
p1  0  0   0  0
p2  0  0   0 10
p3  0  0   0 11

$theta
   o1 o2 o3 o4 o5 t1 t2 t3 t4 g1 g2 g3 p1 p2 p3
o1 15                                          
o2  0 16                                       
o3  0  0 17                                    
o4  0  0  0 18                                 
o5  0  0  0  0 19                              
t1  0  0  0  0  0 20                           
t2  0  0  0  0  0  0 21                        
t3  0  0  0  0  0  0  0 22                     
t4  0  0  0  0  0  0  0  0 23                  
g1  0  0  0  0  0  0  0  0  0 24               
g2  0  0  0  0  0  0  0  0  0  0 25            
g3  0  0  0  0  0  0  0  0  0  0  0 26         
p1  0  0  0  0  0  0  0  0  0  0  0  0 27      
p2  0  0  0  0  0  0  0  0  0  0  0  0  0 28   
p3  0  0  0  0  0  0  0  0  0  0  0  0  0  0 29

$psi
    OC TC TGC P 
OC  30          
TC  34 31       
TGC 35 36 32    
P    0  0  0  33

$beta
    OC TC TGC P
OC   0  0   0 0
TC   0  0   0 0
TGC  0  0   0 0
P   12 13  14 0

What do lambda, theta, psi and beta respectively represent?

Thank you in advance for your help!

Mauricio Garnier-Villarreal

unread,
Jun 22, 2019, 2:26:43 PM6/22/19
to lavaan
lambda is the factor loadings
psi is the factor covariance/correlation
beta is the regressions
theta is the residual covariance/correlation

Abc Abc

unread,
Jun 22, 2019, 3:33:21 PM6/22/19
to lavaan
Thank you! 
Do you know where I can find the information regarding the output of inspect?

Terrence Jorgensen

unread,
Jun 23, 2019, 2:38:54 PM6/23/19
to lavaan
Do you know where I can find the information regarding the output of inspect?

The ?lavInspect help page, and reading introductory SEM textbooks.

Terrence D. Jorgensen
Assistant Professor, Methods and Statistics
Research Institute for Child Development and Education, the University of Amsterdam

Reply all
Reply to author
Forward
0 new messages