Predict factor score using Bayes estimation or the maximum posterior likelihood method

72 views

Skip to first unread message

Dai Duong

unread,

Mar 25, 2018, 7:20:01 PM3/25/18

to lavaan

I need to predict score for the latent variable in my model by using either Bayes estimation or the maximum posterior likelihood method. How can I do that with lavaan and blavaan?

My model is:

ctrl.labor <- "

control_labor =~ lincome+lwh++mjob+pos

control_labor~ urbanrate+lhincome+spland_hec+owing+poor+manu_mine+uti+construction+lowser+hiser+lm1ac5+ltedu+ lf_prov14

lwh~lm1ac5+m1ac2+major+migrated+ttnt+married

lincome~m1ac2+major+migrated++ttnt+married

mjob+pos~lm1ac5+m1ac2+major+ttnt+married

lincome~~lwh

pos~~lwh

mjob~~lwh

pos~~lincome

The result is below. I know fit indices are not good but I want to know how to predict scores for latent variable. I am going to improve the model later.

lavaan (0.5-23.1097) converged normally after 78 iterations

Used Total

Number of observations 19600 21365

Estimator DWLS Robust

Minimum Function Test Statistic 4659.730 4582.862

Degrees of freedom 36 36

P-value (Chi-square) 0.000 0.000

Scaling correction factor 1.017

Shift parameter 2.682

for simple second-order correction (Mplus variant)

Model test baseline model:

Minimum Function Test Statistic 25726.117 22863.878

Degrees of freedom 78 78

P-value 0.000 0.000

User model versus baseline model:

Comparative Fit Index (CFI) 0.820 0.800

Tucker-Lewis Index (TLI) 0.609 0.568

Robust Comparative Fit Index (CFI) NA

Robust Tucker-Lewis Index (TLI) NA

Root Mean Square Error of Approximation:

RMSEA 0.081 0.080

90 Percent Confidence Interval 0.079 0.083 0.078 0.082

P-value RMSEA <= 0.05 0.000 0.000

Robust RMSEA NA

90 Percent Confidence Interval NA NA

Standardized Root Mean Square Residual:

SRMR 0.012 0.012

Weighted Root Mean Square Residual:

WRMR 7.318 7.318

Parameter Estimates:

Information Expected

Standard Errors Robust.sem

Latent Variables:

Estimate Std.Err z-value P(>|z|)

control_labor =~

lincome 1.000

lwh 0.478 0.009 51.026 0.000

mjob -0.469 0.022 -21.766 0.000

pos 0.881 0.024 36.387 0.000

Regressions:

Estimate Std.Err z-value P(>|z|)

control_labor ~

urbanrate 0.355 0.026 13.605 0.000

lhincome 0.502 0.007 74.291 0.000

spland_hec 0.009 0.002 4.613 0.000

owing -0.033 0.009 -3.789 0.000

poor -0.008 0.002 -3.930 0.000

manu_mine 0.512 0.014 35.416 0.000

uti 0.632 0.064 9.901 0.000

construction 0.399 0.027 15.023 0.000

lowser 0.583 0.013 46.266 0.000

hiser 0.849 0.017 51.421 0.000

lm1ac5 0.015 0.015 1.005 0.315

ltedu 0.087 0.008 10.569 0.000

lf_prov14 -0.010 0.005 -2.040 0.041

lwh ~

lm1ac5 -0.146 0.013 -11.283 0.000

m1ac2 0.118 0.009 13.477 0.000

major -0.136 0.013 -10.645 0.000

migrated 0.075 0.014 5.495 0.000

ttnt 0.012 0.012 1.045 0.296

married 0.296 0.013 22.101 0.000

lincome ~

m1ac2 0.216 0.009 23.088 0.000

major 0.164 0.014 11.892 0.000

migrated 0.112 0.014 8.117 0.000

ttnt -0.001 0.012 -0.104 0.917

married 0.148 0.016 9.404 0.000

mjob ~

lm1ac5 0.024 0.034 0.713 0.476

m1ac2 0.059 0.020 2.902 0.004

major -0.236 0.029 -8.197 0.000

ttnt -0.579 0.026 -22.481 0.000

married 0.545 0.034 16.252 0.000

pos ~

lm1ac5 0.258 0.038 6.804 0.000

m1ac2 -0.149 0.020 -7.366 0.000

major 0.228 0.033 6.856 0.000

ttnt 0.078 0.023 3.312 0.001

married 0.050 0.032 1.574 0.116

Covariances:

Estimate Std.Err z-value P(>|z|)

.lincome ~~

.lwh 0.118 0.007 17.398 0.000

.lwh ~~

.pos 0.051 0.008 6.196 0.000

.mjob 0.157 0.006 27.567 0.000

.lincome ~~

.pos 0.131 0.013 10.436 0.000

Intercepts:

Estimate Std.Err z-value P(>|z|)

.lincome -0.274 0.064 -4.303 0.000

.lwh 6.819 0.056 121.468 0.000

.mjob 0.000

.pos 0.000

.control_labor 0.000

Thresholds:

Estimate Std.Err z-value P(>|z|)

mjob|t1 -0.837 0.142 -5.898 0.000

pos|t1 0.222 0.167 1.327 0.185

pos|t2 3.800 0.158 24.105 0.000

pos|t3 5.177 0.162 32.021 0.000

pos|t4 6.585 0.168 39.223 0.000

Variances:

Estimate Std.Err z-value P(>|z|)

.lincome 0.476 0.013 36.445 0.000

.lwh 0.367 0.004 94.552 0.000

.mjob 1.018

.pos 1.064

.control_labor -0.083 0.013 -6.616 0.000

Scales y*:

Estimate Std.Err z-value P(>|z|)

mjob 1.000

pos 1.000

Mauricio Garnier-Villarreal

unread,

Mar 26, 2018, 12:30:22 AM3/26/18

to lavaan

You cannot do this model with blavaan so far, blavaan doesnt support models with categorical indicators yet